Methods in

Molecular Biology 1892

Nese Sreenivasulu Editor

Rice Grain
Quality
Methods and Protocols
Methods in M o l e c u l a r B i o lo g y

Series Editor:
John M. Walker
School of Life and Medical Sciences
University of Hertfordshire
Hatfield, Hertfordshire, United Kingdom

For further volumes:

http://www.springer.com/series/7651
 Rice Grain Quality
Methods and Protocols

Editors

Nese Sreenivasulu
Grain Quality and Nutrition Center, International Rice Research Institute, Manila, Philippines
Editor
Nese Sreenivasulu
Grain Quality and Nutrition Center
International Rice Research Institute
Manila, Philippines

ISSN 1064-3745     ISSN 1940-6029 (electronic)

Methods in Molecular Biology
ISBN 978-1-4939-8912-6    ISBN 978-1-4939-8914-0 (eBook)
https://doi.org/10.1007/978-1-4939-8914-0

Library of Congress Control Number: 2018960445

© Springer Science+Business Media, LLC, part of Springer Nature 2019

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction
on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation,
computer software, or by similar or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not
imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and
regulations and therefore free for general use.
The publisher, the authors, and the editors are safe to assume that the advice and information in this book are believed
to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty,
express or implied, with respect to the material contained herein or for any errors or omissions that may have been
made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

This Humana Press imprint is published by the registered company Springer Science+Business Media, LLC part of
Springer Nature.
The registered company address is: 233 Spring Street, New York, NY 10013, U.S.A.
Dedication

I dedicate this book to Dr. Bienvenido O. Juliano (August 15, 1936–February 21, 2018),
who immensely contributed significant knowledge in the area of cereal chemistry research.
Dr. Juliano pioneered the research of grain quality at the International Rice Research Institute
(IRRI) in 1961; he went on to spend more than 32 years at IRRI, who made significant
research contributions in the area of cooking, eating, and nutritional qualities of rice grain.
He also helped to build the grain-quality research capability of PhilRice, where he continued to
pursue his rice research as a senior consultant/expert.

v
Preface

Acceptance of new rice genotypes requires their ability to satisfy consumer preferences for
premium grain quality, besides being high yielding. As rice consumers become increasingly
particular about the quality of the rice, we need to ensure that modern varieties were less
susceptible to breaking during milling and assure premium cooking quality with optimum
texture, flavor, and aroma. Translating human health benefits by enhancing nutritional
properties includes enriching micronutrients, ensuring food safety, and introducing slower
digestibility factors. Recent advances made in introducing various high-throughput pheno-
typing methods to screen milling quality, cooking quality, and nutritional quality in the
pool of breeding material are discussed in depth. This volume presents the relevant meth-
ods and detailed protocols with appropriate instructions outlined by experts to facilitate
grain quality and nutritional screening in the germplasm. Detailed protocols to define seed
development stages, panicle architectural traits to understand yield components, and mea-
sure physical traits such as grain dimensions using imaging techniques and chalk and head
rice yield have been discussed in Chapters 1, 3, 4, 5, and 6. We need to link initial indicators
of cooking quality (amylose, gel consistency, and gelatinization temperature) with texture
and viscoelastic properties to capture distinct cooking quality classes. Also covered are bio-
chemical methods that measure properties related to cooking such as starch structure and
protein properties in Chapters 2, 7, 8, 9, and 10. Holistic understanding of grain quality
traits by exploring metabolomics platforms to screen primary metabolites and volatiles has
been described in Chapter 11. In addition, health and nutritional aspects need to be fac-
tored into rice breeding programs to contribute to the overall wellness of rice consumers.
Therefore, we need to take into account breeding healthier target traits by capturing the
diversity for low glycemic index and high-resistant starch and enriching nonstructural poly-
saccharides and micronutrients with acceptable cooking quality, texture, and palatability.
Methods related to micronutrient profiling, screening heavy metals, and identifying rice
cultivars with lower glycemic index have been addressed in Chapters 13–15.
We need to establish the genetic basis of the variation of cooking and eating quality
traits through genome-wide association studies by studying both diverse set of lines and
pre-breeding core collection. Genome-wide -omics analyses have provided efficient
approaches to identify key genomic regions that control grain quality traits. Knowledge of
these genes and the influence of specific alleles present in both domesticated and wild rice
gene pools provide a robust platform for marker-assisted selection in breeding to introgress
premium grain quality traits in high-yielding backgrounds. Emphasis has been given to
utilizing resequencing resources for the gene discovery programs via structural genomics,
exploring transcriptome, epigenetics, and genome editing technologies to unravel grain
quality, and nutritional traits have been discussed in Chapters 12 and 16–18. In summary,
overall emphasis has been given to holistic understanding of grain quality traits covering
various phenomics technologies and links it through to gene discovery via QTL cloning
and structural-functional genomics strategies.

Manila, Philippines Nese Sreenivasulu

vii
Contents

Preface. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
1 Improving Head Rice Yield and Milling Quality: State-of-­the-Art
and Future Prospects. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .    1
Vito M. Butardo Jr. and Nese Sreenivasulu
2 Improving Rice Grain Quality: State-of-the-Art and Future Prospects. . . . . . .   19
Vito M. Butardo Jr., Nese Sreenivasulu, and Bienvenido O. Juliano
3 Morphology of Rice Seed Development and Its Influence
on Grain Quality. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .   57
Paul A. Counce and Karen A. K. Moldenhauer
4 Novel Imaging Techniques to Analyze Panicle Architecture. . . . . . . . . . . . . . .   75
Erstelle Pasion, Roinand Aguila, Nese Sreenivasulu, and Roslen Anacleto
5 Measuring Head Rice Recovery in Rice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .   89
Jennine Rose Lapis, Rosa Paula O. Cuevas, Nese Sreenivasulu,
and Lilia Molina
6 Measurement of Rice Grain Dimensions and Chalkiness,
and Rice Grain Elongation Using Image Analysis. . . . . . . . . . . . . . . . . . . . . . .   99
Marnol V. Santos, Rosa Paula O. Cuevas, Nese Sreenivasulu,
and Lilia Molina
7 Method Development of Near-Infrared Spectroscopy Approaches
for Nondestructive and Rapid Estimation of Total Protein in Brown
Rice Flour. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  109
Rosario Jimenez, Lilia Molina, Iman Zarei, Jennine Rose Lapis,
Ruben Chavez, Rosa Paula O. Cuevas, and Nese Sreenivasulu
8 Multi-Dimensional Cooking Quality Classification Using Routine
Quality Evaluation Methods. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  137
Lilia Molina, Rosario Jimenez, Nese Sreenivasulu, and Rosa Paula O. Cuevas
9 Characterization of Mechanical Texture Attributes of Cooked Milled
Rice by Texture Profile Analyses and Unraveling Viscoelasticity
Properties Through Rheometry. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  151
Rosa Paula O. Cuevas, Pawan S. Takhar, and Nese Sreenivasulu
10 Characterizing Starch Molecular Structure of Rice. . . . . . . . . . . . . . . . . . . . . .  169
Cheng Li, Hongyan Li, and Robert G. Gilbert
11 Rice Grain Quality Benchmarking Through Profiling of Volatiles
and Metabolites in Grains Using Gas Chromatography Mass Spectrometry. . . .  187
Cindy Llorente, Rosario Jimenez, Jackie, Yariv Brotman, Alisdair R. Fernie,
and Nese Sreenivasulu
12 Re-sequencing Resources to Improve Starch and Grain Quality in Rice . . . . . .  201
Gopala Krishnan Subbaiyan, Ardashir K. Masouleh, Agnelo Furtado,
Daniel L. E. Waters, and Robert J. Henry

ix
x Contents

13 Quantifying Grain Digestibility of Starch Fractions in Milled Rice . . . . . . . . . .  241

Crisline Mae Alhambra, Sushil Dhital, Nese Sreenivasulu,
and Vito M. Butardo Jr.
14 Determination of Macronutrient and Micronutrient Content
in Rice Grains Using Inductively Coupled Plasma-Optical Emission
Spectrometry (ICP-OES). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  253
Lilia Molina, Jennine Rose Lapis, Nese Sreenivasulu,
and Rosa Paula O. Cuevas
15 Determination of Cadmium Concentration in Milled and Brown
Rice Grains Using Graphite Furnace Atomic Absorption Spectrometry. . . . . . .  265
Lilia Molina, Jennine Rose Lapis, Nese Sreenivasulu,
and Rosa Paula O. Cuevas
16 Analysis of Developing Rice Grain Transcriptome Using the Agilent
Microarray Platform. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  277
Mandy Püffeld, Christiane Seiler, Markus Kuhlmann, Nese Sreenivasulu,
and Vito M. Butardo Jr.
17 Quantification of DNA Methylation as Biomarker for Grain Quality. . . . . . . . .  301
Christiane Seiler and Markus Kuhlmann
18 CRISPR-Cas9-Mediated Genome Editing of Rice Towards Better
Grain Quality. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  311
Anindya Bandyopadhyay, Xiaojia Yin, Akshaya Biswal, Robert Coe,
and William Paul Quick

Index. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  337
Contributors

Roinand Aguila • International Rice Research Institute, Los Baños, Laguna, Philippines
Crisline Mae Alhambra • International Rice Research Institute, Los Baños, Laguna,
Philippines
Roslen Anacleto • International Rice Research Institute, Los Baños, Laguna,
Philippines
Anindya Bandyopadhyay • International Rice Research Institute, Los Baños, Laguna,
Philippines; Syngenta Beijing Innovation Center, Changping District, Beijing, China
Akshaya Biswal • International Rice Research Institute, Los Baños, Laguna, Philippines
Yariv Brotman • Max-Planck-Institute of Molecular Plant Physiology, Potsdam-Golm,
Germany; Department of Life Sciences, Ben-Gurion University of the Negev, Beersheba,
Israel
Vito M. Butardo Jr. • International Rice Research Institute, Los Baños, Laguna,
Philippines; Department of Chemistry and Biotechnology, Faculty of Science, Engineering
and Technology, Swinburne University of Technology, Hawthorn, VIC, Australia; Faculty
of Science, Engineering and Technology, Swinburne University of Technology, Hawthorn,
VIC, Australia
Ruben Chavez • International Rice Research Institute, Los Baños, Laguna, Philippines
Robert Coe • International Rice Research Institute, Los Baños, Laguna, Philippines
Paul A. Counce • University of Arkansas, Rice Research and Extension Center,
Stuttgart, AR, USA
Rosa Paula O. Cuevas • International Rice Research Institute, Los Baños, Laguna,
Philippines
Sushil Dhital • ARC Centre of Excellence in Plant Cell Walls, Centre for Nutrition and
Food Sciences, Queensland Alliance for Agriculture and Food Innovation, The University
of Queensland, St Lucia, QLD, Australia
Alisdair R. Fernie • Max-Planck-Institute of Molecular Plant Physiology, Potsdam-Golm,
Germany; Center of Plant System Biology and Biotechnology, Plovdiv, Bulgaria
Agnelo Furtado • Queensland Alliance for Agriculture and Food Innovation, University
of Queensland, Brisbane, QLD, Australia
Robert G. Gilbert • Joint International Research Laboratory of Agriculture and Agri-­
Product Safety of Ministry of Education of China, Yangzhou University, Yangzhou,
China; Centre for Nutrition & Food Sciences, QAAFI, University of Queensland,
Brisbane, Australia
Gopala Krishnan Subbaiyan • Division of Genetics, ICAR-Indian Agricultural Research
Institute, New Delhi, India
Robert J. Henry • Queensland Alliance for Agriculture and Food Innovation, University
of Queensland, Brisbane, QLD, Australia
Jackie • Shimadzu (Asia Pacific) Pte. Ltd., Singapore, Singapore
Rosario Jimenez • International Rice Research Institute, Los Baños, Laguna, Philippines
Bienvenido O. Juliano • Philippine Rice Research Institute, Los Baños, Laguna,
Philippines

xi
xii Contributors

Markus Kuhlmann • Leibniz Institute of Plant Genetics and Crop Plant Research,
Gatersleben, Germany
Jennine Rose Lapis • International Rice Research Institute, Los Baños, Laguna,
Philippines
Cheng Li • Joint International Research Laboratory of Agriculture and Agri-Product
Safety of Ministry of Education of China, Yangzhou University, Yangzhou, China
Hongyan Li • School of Food and Chemical Engineering, Beijing Technology and Business
University, Beijing, China
Cindy Llorente • International Rice Research Institute, Los Baños, Laguna, Philippines
Ardashir K. Masouleh • Queensland Alliance for Agriculture and Food Innovation,
University of Queensland, Brisbane, QLD, Australia
Karen A. K. Moldenhauer • University of Arkansas, Rice Research and Extension
Center, Stuttgart, AR, USA
Lilia Molina • International Rice Research Institute, Los Baños, Laguna, Philippines
Erstelle Pasion • International Rice Research Institute, Los Baños, Laguna, Philippines
Mandy Püffeld • Leibniz Institute of Plant Genetics and Crop Plant Research,
Gatersleben, Germany
William Paul Quick • International Rice Research Institute, Los Baños, Laguna,
Philippines
Marnol V. Santos • International Rice Research Institute, Los Baños, Laguna,
Philippines
Christiane Seiler • Leibniz Institute of Plant Genetics and Crop Plant Research,
Gatersleben, Germany
Nese Sreenivasulu • International Rice Research Institute, Los Baños, Laguna,
Philippines
Pawan S. Takhar • Department of Food Science and Human Nutrition, University of
Illinois at Urbana-Champaign, Urbana, IL, USA
Daniel L. E. Waters • Southern Cross Plant Science, Southern Cross University, Lismore,
NSW, Australia; ARC ITTC for Functional Grains, Charles Sturt University, Wagga,
NSW, Australia
Xiaojia Yin • International Rice Research Institute, Los Baños, Laguna, Philippines
Iman Zarei • International Rice Research Institute, Los Baños, Laguna, Philippines
 Chapter 1

Improving Head Rice Yield and Milling Quality:

State-of-­the-Art and Future Prospects
Vito M. Butardo Jr. and Nese Sreenivasulu

Abstract
Increasing paddy yield in rice does not directly translate to enhancing food security because significant
decrease in grain yield can happen during postharvest processing of the rice paddy. In parallel with enhanc-
ing paddy yield, improving the milling quality of rice is essential in ensuring food security by mitigating the
impact of significant losses during the postharvest processing of rice grains. From an industrial standpoint,
maximizing the milling recovery of whole grain polished rice is crucial in fetching higher revenues to rice
farmers. Significant advances in rice postharvest processing technology have been achieved which are
geared toward reducing the incidence of fissures and chalkiness to increase head rice yield (HRY) in rice.
The genetic bases of kernel development and grain dimension are also characterized. In addition to these
advancements, an integrated phenotyping suite to simultaneously characterize phenotypes related to mill-
ing quality will help in screening for breeding lines with high HRY. Toward this goal, modern imaging
tools and computer algorithms are currently being developed for high-throughput characterization of rice
milling quality. With the availability of more sophisticated, affordable, automated, and nondestructive
phenotyping methods of milling quality, it is envisioned that significant improvement in HRY will be made
possible to ensure rice food security in the future.

Key words Breeding, Chalk, Fissure, Genomics, Grain quality, Head rice yield, Milling quality

1 Introduction

Breeders traditionally focus on increasing paddy yield coupled with

enhanced stress tolerance to ensure food sustainability and security
for rice. However, focusing on paddy yield alone is not enough
because economic losses occur as a result of quantity reduction due
to poor milling quality and cooking quality deterioration can hap-
pen at any stage along the postharvest value chain [1]. It is esti-
mated that up to 30% loss in rice grain can happen due to rice
breakage as they are subjected to mechanical stress from harvesting
to grain processing. Rice grain breaks when the extrinsic mechanical
stresses applied to the rice grain during processing are greater than
the intrinsic grain strength. This can be estimated by bending or

Nese Sreenivasulu (ed.), Rice Grain Quality: Methods and Protocols, Methods in Molecular Biology, vol. 1892,
https://doi.org/10.1007/978-1-4939-8914-0_1, © Springer Science+Business Media, LLC, part of Springer Nature 2019

1
2 Vito M. Butardo Jr. and Nese Sreenivasulu

fracture strength, which is the maximum force a brittle material like

rice grain can tolerate before it cracks due to flexural load [2]. Loss
in kernel biomass is usually assessed by quantifying head rice yield
(HRY), which is the proportion of rough rice that remains as head
rice (intact grain) after complete milling. Head rice, on the other
hand, is conventionally defined as intact grains that have ¾ of the
original kernel length after complete milling. HRY is the gold stan-
dard of rice millers to quantify milling quality. Elevated proportion
of broken rice grains are usually avoided because it reduces the
commercial value of broken rice by half.
Usually, the milling yield in rice breeding lines varies from 70 to
79% while HRY potential is estimated at 24–74%. The significant
loss in seed biomass due to kernel breakage which can otherwise
feed the hungry is primarily mitigated by observing best posthar-
vest practices as part of the rice quality management system [3].
There are currently few research and breeding efforts made to
develop new cultivars with grains that have superior HRY proper-
ties. Previous studies revealed that a major way of improving HRY
quality of rice is by reducing kernel weakness and susceptibility to
breakage, which is usually addressed at the postharvest level by
optimizing the grain drying process. The impact of dehulling and
milling on breakage susceptibility of rice is well characterized [2].
This trait is associated with grain fissuring (cracking), chalkiness,
maturity of rice grains, and kernel dimensions. All these factors will
be discussed below after a short discussion on the best postharvest
method to reduce grain loss and increase HRY.

2 Observing Best Postharvest Practices

The quality management practices in the rice postproduction sec-

tor are already comprehensively defined [3]. Based on voluminous
comprehensive studies summarized in IRRI Knowledge Bank, sim-
ple harvest and postharvest practices are usually promoted to farm-
ers and local rice processors to ensure ease of compliance. Harvest
should be properly scheduled to coincide with 22–24% moisture
content (MC). If at all possible, combine harvesting is encouraged
to avoid delays and the harvested rice panicles should not be
stacked in the field. After threshing, the harvest should be immedi-
ately dried within 24 h after harvesting. During drying, the tem-
perature should not exceed 43 °C in a flatbed dryer or any other
fixed bed batch dryer. For milling in a mixing type dryer with tem-
pering between drying passes, temperature should not exceed
50 °C. For sun drying, the grains should have a depth between 2
and 4 cm and they should be mixed every 30 min. The aim is to
slowly dry to 14% MC which ensures minimal grain breakage.
After drying, the seeds should be stored safely to avoid pests, water,
and re-wetting of the grains.
 Improving Head Rice Yield and Milling Quality: State-of-the-Art and Future Prospects 3

3 Reducing Grain Fissuring

Fissuring of rice is a major problem in the rice industry because it

results in the reduction of HRY and commercial value of the grain.
Rice fissures are large internal fractures that usually develop per-
pendicular to the length of the grain. This trait introduces weak-
ness in the grain due to hairline cracks which usually lead to grain
breakage upon dehulling and milling. Higher incidence of grain
breakage leads to significant reduction in HRY. The formation of
fissures is influenced by environmental conditions in the field
before and during harvest, as well as during grain processing and
storage. At the end of grain-filling stage, the supply of water from
the vegetative tissues in rice is cut-off to initiate grain maturation.
Farmers also withhold irrigated water from the farm plots to expe-
dite grain desiccation. During this stage, MC is no longer con-
trolled by moisture transfer within the panicle but is susceptible
and varies in response to environmental fluctuations. Rain, air tem-
perature, and humidity fluctuations are the three major environ-
mental factors in the field that can influence the formation of
fissures during and prior to harvest. At this stage, the thermal and
material properties of rice grain tend to change based on relative
humidity (RH), temperature (T), and MC. Grain fissures are usu-
ally formed in the field when the moisture MC suddenly drops
below 15% [4]. The MC of individual kernels within the same pan-
icle can vary up to 10% [2] because rice grains belonging to a single
inflorescence do not mature synchronously. This variation in MC is
expected to be higher among different rice plants growing on the
same field. Because grains of different MC are hygroscopic, kernels
with variations in MC desorb or adsorb moisture from the air until
an equilibrium MC is reached. This equilibrium MC depends on
both the RH and air temperature. Moisture gradient induces swell-
ing in the grain and induces force within the endosperm, which
can cause fissures. The response of rice kernels to tensile and com-
pressive stress in a MC gradient will determine whether the grain
will form fissure during grain ripening and desiccation stage [5].
Not all grains containing fissures break during milling but the role
of fissure in increasing the incidence of grain breakage and reduc-
tion in HRY is well known [6, 7].
Rice grains from different varieties are usually mixed by farm-
ers or grain processors during harvest and storage. This factor
additionally contributes to the large variations in MC of grains,
ranging from 16 to 26%. Suboptimal drying and storage at the
postharvest stage can also lead to significant fissuring. Exposing
rice kernels to extreme drying conditions usually increases the pro-
portion of kernels with fissures. However, it is commonly observed
that most rice grains do not form fissures immediately after drying
[8]. Instead, the fissures emerge when the rice grains have already
4 Vito M. Butardo Jr. and Nese Sreenivasulu

cooled down after drying. As a result, post-drying treatment like

optimum high temperature tempering can be implemented to
reduce the incidence of fissuring [9, 10]. Tempering in commercial
mills involves holding the grains for a certain time in between dry-
ing cycles to decrease the MC gradient and hence reduce the inci-
dence of fissuring and subsequent rice breakage. A modern method
of tracking the grains MC gradient and the presence of fissures
during the tempering processes can be facilitated by magnetic reso-
nance imaging (MRI) analyses [11]. This technology is one of the
more promising techniques that can be employed to track milling
quality at every stage of the value chain.
Glass transition temperature (Tg) can be used to explain rice
kernel fissuring during the course of seed desiccation where drying
and tempering are being accounted [12]. Tg is the temperature
range at which polymers shift from a soft “rubbery” state to a hard
“glassy” material and vice versa. This can be measured by differen-
tial scanning calorimetry (DSC) and dynamic mechanical analysis
(DMA), but thermomechanical analysis (TMA) is the most widely-­
accepted method [13–15]. Taking into account the Tg range dur-
ing rice drying and tempering is crucial because it affects milling
quality [12].
The relationship of kernel MC gradients and Tg to HRY is well
established [16, 17]. The major polymer in rice kernel is starch,
which can reach up to 90% in milled grains. It is a partially crystal-
line and partially amorphous polymer of glucose composed of
essentially linear amylose and highly branched amylopectin. The
amylose and amylopectin branch points form the amorphous layer
while the amylopectin outer chains form the crystalline region.
The semi-crystalline property of starch dynamically changes
according to T and MC [18, 19]. During the drying process, the
kernel temperature increases causing each grain to dry unevenly
from the surface to the inside [20, 21]. As a result of this moisture
gradient, the rice kernel undergoes glass transition at different
parts of the grain. During this phase transition, a temperature and
moisture gradient can develop along the rice kernel. The tempera-
ture gradient has a negligible effect on the integrity of the rice
kernel but the MC gradient has a significant impact during and
after drying, highlighting the importance of water as plasticizer of
rice kernel [15]. Moisture absorption and desorption in the rice
grain can significantly affect the resulting HRY [22, 23]. The
reversible shifts between the glassy and rubbery states of starch
influence the formation of fissures and grain breakage due to dif-
ferential tensile and compressive stresses resulting from MC gradi-
ent within the rice kernel [5, 24]. This glass transition of starch
typically occurs during the desiccation process and commercial
drying of rice grain [13]. It is possible that the outer layer of a
growing grain is already in the glass state while the inner kernel is
still in the rubbery state, which can induce fissure and subsequent
 Improving Head Rice Yield and Milling Quality: State-of-the-Art and Future Prospects 5

breakage as a result of differential stresses [9, 13]. Drying the rice

grain in the glassy region is preferred because it does not cause
significant reduction in HRY [16]. In contrast, drying in the rub-
bery state and cooling down immediately without tempering can
cause increase in MC gradients leading to significant reduction in
HRY [16].
The differential tensile and compressive stresses in rice grain
during the drying process have viscous and elastic components due
to the viscoelastic properties of rice grains. The viscous stress com-
ponent can be minimized by reducing moisture gradient inside the
rice kernel while the elastic component can be reduced by temper-
ing above Tg [25]. To enhance milling recovery, infrared (IR) dry-
ing with tempering has been demonstrated to be more effective
than conventional drying [26]. However, the associated cost and
the feasibility of this technology need to be further explored in
large-scale rice milling operations. To circumvent this problem,
low-cost optimal drying methods for freshly-harvested rice grains
were developed. One recent innovation is the Solar Bubble Dryer
(SBD), a low-cost drying method developed by IRRI, Hohenheim
University and GrainPro. SBD can process up to one ton of rice
compared to mechanical dryer, and has a very low operating cost
because it relies on solar energy. Drying occurs in a buffered plastic
tunnel which regulates the temperature thereby protecting the
grains from overheating. This technology improves upon the tradi-
tional sun drying method because the drying process happens in a
completely sealed environment. Consequently, postharvest losses
are minimized and the grains are protected from contamination
and pests.

4 Reducing Grain Chalkiness

Chalk is the opaque white streak in a translucent rice grain. Chalk is

influenced by spikelet position [27], it is usually induced by ele-
vated temperature during grain filling [28, 29], and it negatively
affects milling quality and HRY [2]. The formation of chalk is
related to the disturbed translocation of assimilates [27] leading to
the abortion of starch and protein biosynthesis in the developing
grain [30]. This produces incompletely filled starch granules with
many air spaces in between, which is visible as opaque spots along
the translucent grain [31]. These air spaces, which can be visualized
by X-ray microtomography [32], are responsible for making chalky
grains more brittle. Chalky grains crack more easily compared to
translucent grains [33], significantly elevating the proportion of
broken grains which are usually discarded [34]. The susceptibility
of chalky grains to breakage is due to softer kernel [31, 35] and
increased tendency to form fissures [36]. These make chalky grains
more prone to mechanical stresses during grain dehulling and
6 Vito M. Butardo Jr. and Nese Sreenivasulu

polishing. Differences in endosperm compactness and micro-poros-

ity are also thought to explain differences in breakage susceptibility
between chalky and non-chalky grains [2]. Even if the chalky grains
resist breakage and survive the milling process, their presence sig-
nificantly lowers the market value and consumer acceptance of
milled rice [37] due to alteration in cooking quality [38] and reduc-
tion in sensory properties [39]. A more in-depth study on chalk is
crucial because it has negative impacts on the value of the rice along
the supply chain, from farmer to consumer. This is exhaustively dis-
cussed in a recent review article [30].
A more accurate phenotyping method for chalk is needed. The
conventional method using Cervitec (Foss) is limited because bro-
ken grains are discarded prior to analyses. This introduces bias
because only the whole grains are characterized for grain dimen-
sion and chalk. Consequently, the percent chalkiness detected by
this method is only applicable for the breakage-resistant kernels, as
the breakage susceptible proportion of grain is not subjected to
imaging. Therefore Cervitec can only detect the chalkiness which
affects appearance quality, not the chalkiness that can impact
HRY. A newer method based on SeedCount (Next Instrument) is
available that uses flatbed scanner for transmission and reflectance
image analysis. It offers a better method for chalkiness trait quan-
tification coupled with accurate dimension, color, degree of white-
ness, immature grain impurity, and broken grain analyses. Novel
quantitative image processing technology that accurately detects
location and type of chalk along the grain is available using com-
puter vision which relies on support vector machine (SVM) cou-
pled with principal component analysis (PCA) or convex point
matching [40, 41]. However, these methods are not yet commer-
cially available. In addition, they do not simultaneously measure
other traits like kernel dimension, percent broken grains, and per-
cent immature grains that are necessary to obtain a holistic under-
standing of milling quality.

5 Tailoring Grain Maturity and Kernel Dimension

Asynchronous kernel maturation leads to elevated proportion of

immature rice grains. This is a problem not just in deterioration of
appearance quality but also in milling quality because immature
grains tend to be more susceptible to breakage during grain process-
ing. Breakage susceptibility due to immature grains is comparatively
less significant for photosensitive rice varieties. This is because pho-
tosensitive rice cultivars tend to have more uniform and synchro-
nous anthesis resulting in uniform grain maturation compared to
non-photosensitive rice accessions. For non-­photosensitive rice cul-
tivars, early-maturing varieties (90–100 days) have more immature
grains than medium-maturity cultivars (130–140 days). Hence, tai-
Improving Head Rice Yield and Milling Quality: State-of-the-Art and Future Prospects 7

loring days to flowering (DTF) and synchronizing anthesis are two

important targets to minimize HRY and reduce percentage of chalk.
There appears to be a link between flowering time, heading date,
panicle development, and grain yield in rice [42, 43]. This is an
active area of research which is proving to be very challenging
because it involves genetic, epigenetic mechanisms, and complex
regulatory network [44, 45]. This complexity is better addressed
using grain quality genomics and systems genetics techniques as
recently demonstrated in rice [46].
With respect to kernel dimension, milled rice can be clustered
into four main types based on size: short grain, medium-grain,
and long grain (Table 1). Similarly, it can also be grouped into
four distinct categories based on shape: slender, medium, bold,
and round (Table 2). Kernel size and shape is one of the most
stable properties of a rice variety. Kernel dimensions impact HRY,
particularly in long slender background. It is hard to isolate the
impact of grain size trait from other kernel defects like fissures,
chalk, and immature grains [47, 48]. It is therefore crucial to
develop a robust method that can simultaneously detect kernel
dimension, detects grain breakage, fissure, and chalk percentage
of grain that cumulatively affects HRY. Attempts to combine
image processing, discriminant analysis, and artificial neural net-
work were successful in morphometric and varietal identification
of corn [49] and wheat [50]. In line with this, nondestructive rice
quality control, grading, and milling quality scoring is currently
being tested in laboratory-­scale conditions. For example, machine
vision was used to identify grain defects in rice [51]. Breakage and
cracks were identified in parboiled rice using a Java program based
on an ImageJ improvement which introduces gap-filling segmen-
tation technique [52]. In another study, velocity representation
method was used to identity broken kernels using pattern recog-
nition of contour characteristics of rice after image acquisition
[53]. Similar studies used flatbed scanning and image analysis to
determine size distribution and proportion of broken rice kernels
[54], as well as multiple rice quality parameters including grain
dimension and aspect ratio, head rice yield, percent chalkiness,
and transparency grading [55]. Monitoring of milling quality is
also possible by characteristic dimension ratio (CDR) which com-
pares the dimensional features of all head rice kernels to that of
broken rice grains in the sample [56]. It is therefore possible to
develop a more comprehensive algorithm for integrated detection
of rice physical- and milling quality parameters. Grain quality eval-
uation by computer vision as well as hyperspectral and multispec-
tral imaging techniques as applied in other cereals [57, 58] can
now be employed in rice. Toward this goal, hyperspectral imaging
was recently used to discriminate rice quality [59]. In this study,
the dimension reduction technique was performed on spectral
and image information using PCA and back propagation neural
 8
Table 1
Summary of general routine rice grain quality tests [87–89]

Quality
parameter Measure Principle Standard method Alternative method
A. Physical quality
Moisture Moisture content Evaporation of Around 2–3 g of rice flour dried in oven at 130 ± 1 °C for Moisture meter; Brabender
content moisture an hour. For paddy rice, 10–15 g of samples are dried moisture tester; Infrared moisture
for 24 h. Samples are cooled in a desiccator for analyzer
45–60 min. Moisture content computed by difference
between original weight and dry weight.
Grain Length Measurement of Manual measurement of 50 grains using millimeter ruler Cervitec (Foss); SeedCount (Next
dimension Width grain dimension and vernier calliper; photographic enlargement or Instrument); ImageJ
Thickness and appearance projection of rice grains followed by manual
Size measurement; image scanning followed by image
Shape analysis.
Vito M. Butardo Jr. and Nese Sreenivasulu

Grain Broken grains Visual observation, counting and scoring; image scanning SeedCount (Next Instrument); Rice
appearance Chalky grains followed by image analysis. translucency meter; ImageJ
Grain color Grain color Grain color Determination of grain color and impurity using Kett SeedCount (Next Instrument);
assessment whiteness meter. color meter or chromameter;
milling meter
B. Milling quality
Milling Head rice yield Milling yield The paddy is dehulled to obtain brown rice. The hull is None
quality and Broken rice determination by separated from the brown rice using a blower. Brown
yield Milled rice yield successive grain rice is then polished to produce milled rice. Broken rice
1000 grain weight processing grains are separated from whole grains by sieving using
grain separator. The processed grains are then weighed
successively. The percentage weight of brown rice
removed as bran is usually used to quantify degree of
milling. Grain counter can be used for 1000 grain
weight prior to weight for ease of analysis.
Grain Grain hardness Measurement of The hardness of at least 25 individual grains is measured Hand-operated hardness tester;
hardness hardness of using Instron or Kiya Hardness Tester. This can also be Vicker microhardness distribution
individual grains assessed by ease of grinding using Brabender
farinograph with resistograph attachment.
Grain cracks Grain cracks and Visualization of Visualization of cracks and fissures in rice grains using X-ray visualization
and fissures fissures cracks and transmitted light, hand-held torch, or microbial colony
fissures counter. Crack resistance is measured by soaking grains
for 1–3 h in 30 °C water bath and subjecting the
samples in Kett Pearlest mill.
Degree of Degree of milling Estimation of Five grams of grain is washed three times and stained with NIRS, Satake milling meter
milling degree of milling 5 mL May-Gruenwald reagent (eosin and methylene
blue in methanol solution) for 1–2 min. The grains are
destained three times with water. Staining pattern is
then measured.
Improving Head Rice Yield and Milling Quality: State-of-the-Art and Future Prospects
9
10 Vito M. Butardo Jr. and Nese Sreenivasulu

Table 2
Classification of rice grain based on length

Size IRRI (1966) FAO (1972) USDA ISO (1995)

Extra long >7.5 ≥7.00 None None
Long 6.61–7.50 6.00–6.99 Approximately 3× longer than it is wide ≥6.6
Medium 5.51–6.60 5.00–5.99 2.1–2.9 times longer than it is wide 6.2–6.5
Short ≤5.5 <5.00 Less than 2× longer than it is wide <6.2

network (BPNN). In addition to all of these image analyses tech-

niques, in situ nondestructive methods to characterize rice kernel
during development and at the mature paddy level using X-ray,
NMR, and MRI technologies can be used to properly screen and
identify breakage resistant rice lines prior to dehulling and polish-
ing. Nondestructive detection of cracks in paddy rice is now pos-
sible through X-ray imaging without the need to dehull the
samples [60]. However, all these techniques are not generally
accessible for routine rice grain quality laboratories because they
are not yet commercialized and validated by international collab-
orative testing panels.
Another emerging field in the area of physical and milling
quality research related to kernel dimension and maturity involves
designing equipment for processing, sorting, and sizing of grains
to separate defective from superior quality grains. A new machine
called QSorter (Qualisense) is capable of sorting rice grains at 50
kernels per seconds. Sorting is automatically done by the machine
to provide faster and more accurate quantification of HRY and
milling quality. It is also capable of doing single kernel analyses of
grain dimension, chalk and fissures/cracks. In addition, more inte-
grated machine vision grain quality inspection system that sorts
out grain according to milling and physical appearance is also now
available that can process an average of 1200 kernels per minute
[61, 62]. This prototype system that consists of an automatic
inspection machine and an image-processing unit is used to classify
and sort head rice, chalky and cracked grains. These technological
revolutions help overcome the traditional human inspection which
is more laborious, inconsistent, and highly subjective. This system
can find possible application in export-quality rice to satisfy rice
grading system requirements of importing countries.

6 Physiological Mechanisms and Genetic Basis of Grain Breakage Resistance,

Chalk, and Grain Dimensions

When rice is subjected to best preharvest and postharvest practices

with the aim of minimizing grain loss during grain processing, each
variety will still exhibit a cultivar-specific susceptibility to fissuring.
Improving Head Rice Yield and Milling Quality: State-of-the-Art and Future Prospects 11

Some cultivars are inherently resistant to fissuring because of endo-

sperm chemical composition, hull and bran diffusivity, and grain
size and shape. This indicates that aside from environmental, farm-
ing, postharvest, and grain processing factors, inherent genotypic
traits can also contribute in avoiding postharvest loss due to fis-
sures. In general, rice grains genetically predisposed to form fis-
sures have critical moisture content (CMC) of 15–16%. Below
CMC, the grains will crack. This critical range is lower for fissure-­
resistant cultivars, which usually crack below 12–14% MC.
The physical and biochemical properties of the hull, bran, and
endosperm can affect the movement of moisture through the ker-
nel and are therefore important traits to consider in developing
breakage resistant rice grains. The hull is the outer covering of rice
composed of interlocking lemma and palea which serves as water
barrier protecting the endosperm from rapid changes in moisture
content. This depends on hull composition, i.e., amount of silica
present, structural integrity of the interlocking lemma and palea.
In general, moisture movement is reduced when the hull is thicker
and tighter. On the other hand, the bran is a layer composed of
aleurone that covers the starchy endosperm. Aside from being
nutritious, it acts as a protective moisture barrier between the hull,
pericarp, and the endosperm. Thicker bran can reduce the mois-
ture movement and therefore moisture gradient along the endo-
sperm. Lastly, the endosperm is the main component of the rice
grain composed of storage starch and seed storage protein. The
central core of the grain is mechanically weaker than the periphery
and this depends on the shape, orientation, and packing of starch
granules.
Fissure resistance is exhibited in different rice cultivars by dif-
ferent structural mechanisms. For example, the fissure resistance of
Cypress is due to protective hull barrier while that of Saber is due
to high endosperm diffusivity [63]. It is also believed that rice
grains with compact starch granules can resist breakage and there-
fore increase HRY. Hence, the development of integrated methods
to measure hull thickness and tightness, bran thickness and diffu-
sivity, and endosperm starch granule packing is crucial in screening
for fissure resistance in rice germplasm collection. An induce-­
fissuring method was developed to identify fissure resistance from
early-generation breeding lines [64]. Using this method, studies
identified QTLs related to breakage resistance of Cypress, where
one is linked to the semidwarf sd1 locus [63–65]. However, addi-
tional follow-up studies are needed to determine the exact genetic
basis of fissure resistance based on hull, bran, and endosperm prop-
erties. This can be expedited by the creation of novel phenotyping
methods that can be used to make detailed analyses for these traits.
Similar to fissure resistance, chalk formation is also influenced
by the genotype, aside from environmental, postharvest, and pro-
cessing factors. Because chalk is easier to phenotype, more studies
have been conducted on this subject. The genetic basis of chalk has
12 Vito M. Butardo Jr. and Nese Sreenivasulu

been investigated both in japonica and indica germplasm [66–72].

As reviewed by Sreenivasulu et al. [30], more than 100 QTLs and
7 genes are linked with the chalkiness trait [73–79]. However,
because of the issue on standardization of method to phenotype
chalk discussed above, there is a need to distinguish between chalk-
iness that only affects appearance quality without resulting in grain
breakage from chalkiness that actually reduce milling yield. This
can pave the way for determining the actual contribution of chalk
in reducing HRY.
Rice grain dimension and shape is also easy to phenotype. As a
result, the genetic basis of rice grain size is well characterized from
QTL mapping and mutant analyses [80]. Recent studies are reveal-
ing a more detailed understanding on the molecular events related
to developmental regulation of grain size and shape. Almost all of
the identified genes so far (GW2, GS3, qGL3, GW5, GS5, GW8,
and TGW6) are involved in regulating grain size by modulating the
number of cells through enhanced cell division [81]. More inter-
estingly, a newly-identified gene, GL7/GW7, is involved in regu-
lating grain size by modulating grain length through cell elongation,
and also influences chalk [82, 83]. It appears that genes for cell
division have been preferentially selected by rice breeders com-
pared to genes related to cell elongation [84]. It is therefore pos-
sible to identify more novel genes for cell elongation in rice diversity
panel than in breeding lines. Integrated phenotyping methods that
allow for concurrent detection of fissures, chalk, grain dimension,
and panicle architecture are crucial. Breeding against chalk and fis-
sures with synchronous flowering, medium-maturity, and opti-
mum grain dimension might be effective in reducing the incidence
of breakage susceptibility of rice grain. Moreover, focus should also
be on screening for perfect grains, not imperfect grains. In this
connection, a method to screen for translucency needs to also be
developed and included in the proposed integrated phenotyping
method.

7 Additional Notes

The first-ever meeting of the International Network of Quality

Rice (INQR) was held at the International Rice Research Institute
on December 1–2, 2015. This meeting was attended by an inter-
national panel of experts from rice grain quality, breeding, and
industry sectors. Based on this 2-day meeting, the panel recom-
mended thrust areas to be addressed in research for enhancing
HRY/HRR and to minimize postharvest grain loss. These include:
1. Accessing the 3000 rice genome accession [85] and develop
molecular marker database for key traits related to milling
quality.
Improving Head Rice Yield and Milling Quality: State-of-the-Art and Future Prospects 13

2. Exploring X-ray, NIR, MRI, NMR, polarized light microscopy

and other modern technology that can be used to characterize
chalk, translucency, cracks, breakage susceptibility, and other
grain imperfections that affects head rice and milling yield.
3. Determining the exact location of chalk. Chalk in the dorsal
and the ventral side can affect HRY more than chalk in the
core. For example, Yamadashi and Arborio rice have core
chalkiness but it does not break.
4. Comparing data from commercial, laboratory, and hand mills
to determine adjustment factors. Several factors can influence
the difference in milling including equipment and brand vari-
ability. Various practices that affect storage conditions like
moisture content should be factored in.
5. Breeding for higher HRY and translucency instead of breed-
ing alone for kernel dimension and chalk. Focus should be
more on the enhancing HRY which can command better price
in the market.
6. Using systems biology to link genetics with environmental
responses and to identify the underlying biological regulatory
networks to improve complex traits such as uniform flowering,
synchronous seed filling, finding the ideal panicle architecture,
reduce fissures grain breakage and lower chalk. This strategy is
already being applied to fruits for regulation of processes and
responses, quality prediction, genetic improvement, and
development of virtual fruit model [86].
Because HRY and milling quality is becoming a central
issue in the hybrid rice industry, the following strategies can be
applied to this rapidly expanding rice sector:
7. Germplasm screening for high HRY and translucency in
parental lines.
8. Incorporation of high HRY in female and male parents in
hybrid breeding programs with fissure-resistant and non-­
chalky (translucent) grain appearance.
9. Determination of genetic, biochemical, physiological, and
physical factors influencing high HRY.
10. Defining the genetics of high HRY across grain types.
11. Understanding the genetic basis of kernel hardness/breaking
strength.
12. Development of high-throughput screening methods for
quick determination of head rice recovery in small samples
(e.g., determination of techniques to identify fissures in paddy
grains rather than milled rice, development of nondestructive
methods to quantify fissures and chalk).
13. Validated robust markers to hasten the breeding program to
reduce grain breakage.
14 Vito M. Butardo Jr. and Nese Sreenivasulu

8 Conclusion

Improving the milling recovery of rice has been an active area of

research for several decades but focus is mostly on postharvest pro-
cessing and technology. The genetic improvement of head rice
yield is severely unexplored probably because of the lack of cost-­
effective, reliable, and high-throughput screening method. With
the availability of state-of-the-art imaging, visualization and ana-
lytical technologies for milling quality, it is envisioned that rice
breeding programs will be geared toward improving the head rice
recovery of rice. Improving the milling quality of rice is crucial in
ensuring food security for half of the world’s population in an era
of dwindling natural resources and climate unpredictability.
Head rice yield and grain translucency are very important pre-
ferred factors universally selected by breeders, traders, millers, and
consumers. Therefore, HRY plays an important role in determin-
ing the acceptability and price of rice in the market. The aim there-
fore is to achieve higher HRY. Strategies to improve HRY include
breeding against breakage susceptibility, targeting synchronous
flowering, reduced chalk, postharvest improvement at the farm
level, and grain processing technology improvement. Breeding
strategies for improving HRY include optimization of genetic gains
for reduced grain breakage with reduced chalk and optimized
postharvest handling and storage. For stakeholder strategies, mill-
ers should have customized machinery and milling procedures.
Parboiling can be considered whenever applicable to enhance HRY
in long slender grains, but this strategy is strictly limited to con-
sumer segments that prefer parboiled rice. Stakeholders and
­breeders should also be working hand-in-hand to find solutions at
the genetic level and to tackle better management practices during
the post-harvest value chain. Additional strategies include develop-
ment of methods to correlate laboratory results for HRY with that
of commercial mills, and development of procedures to instantly
determine HRY of farmers produce and congruence with results
from millers.

Acknowledgments

This work has been supported under the CGIAR thematic area
Global Rice Agri-Food System CRP, RICE, Stress-Tolerant Rice
for Africa and South Asia (STRASA) Phase III, and Australian
Centre for International Agricultural Research (Project ID
CIM/2016/046) funding.
 Improving Head Rice Yield and Milling Quality: State-of-the-Art and Future Prospects 15

References

1. Hodges RJ, Buzby JC, Bennett B (2011) transition and rice drying: development of
Postharvest losses and waste in developed a brown rice state diagram. Cereal Chem
and less developed countries: opportunities to 77(6):708–713
improve resource use. J Agric Sci 149:37–45 14. Siebenmorgen TJ, Yang W, Sun Z (2004) Glass
2. Buggenhout J, Brijs K, Celus I, Delcour JA transition temperature of rice kernels deter-
(2013) The breakage susceptibility of raw mined by dynamic mechanical thermal analysis.
and parboiled rice: a review. J Food Eng Trans ASAE 47(3):835–839
117(3):304–315 15. Sun ZH, Yang WD, Siebenmorgen T, Stelwagen
3. Bell M, Bakker R, De Padua D, Rickman A, Cnossen A (2002) Thermomechanical
J (1998) Rice quality management-principles transitions of rice kernels. Cereal Chem
and some lessons. In: ACIAR proceedings, 79(3):349–353. https://doi.org/10.1094/
2000. ACIAR. pp 255–263 Cchem.2002.79.3.349
4. Otto RK, David LC (2004) Rough-rice dry- 16. Yang W, Jia CC, Siebenmorgen TJ, Pan Z,
ing - moisture adsorption and desorption. Cnossen AG (2003) Relationship of kernel
In: Champagne ET (ed) Rice: chemistry moisture content gradients and glass transition
and technology. Grain science references. temperatures to head rice yield. Biosyst Eng
American Association of Cereal Chemists, 85(4):467–476
Inc., St. Paul, MN, pp 223–268. https://doi. 17. Yang W, Jia CC, Howell TA (2003) Relationship
org/10.1094/1891127349.009 of moisture content gradients and glass transi-
5. Kunze O, Choudhury M (1972) Moisture tion temperatures to head rice yield during
adsorption related to the tensile strength of cross-flow drying. Biosyst Eng 86(2):199–206
rice. Cereal Chem 49:684–696 18. Slade L, Levine H (1991) A food polymer
6. Zhang Q, Yang W, Sun Z (2005) Mechanical science approach to structure-property rela-
properties of sound and fissured rice kernels tionships in aqueous food systems: non-­
and their implications for rice breakage. J Food equilibrium behavior of carbohydrate-water
Eng 68(1):65–72 systems. In: Levine H, Slade L (eds) Water
7. Iguaz A, Rodríguez M, Vírseda P (2006) relationships in foods, Advances in experi-
Influence of handling and processing of rough mental medicine and biology, vol 302.
rice on fissures and head rice yields. J Food Eng Springer, New York, pp 29–101. https://doi.
77(4):803–809. https://doi.org/10.1016/j. org/10.1007/978-1-4899-0664-9_3
jfoodeng.2005.08.006 19. Slade L, Levine H (1995) Glass transitions and
8. Sharma AD, Kunze OR (1982) Post-drying fis- water-food structure interactions. Adv Food
sure developments in rough rice. Trans ASAE Nutr Res 38(2):103–179
25(2):465–468 20. Sarker NN, Kunze OR, Strouboulis T (1996)
9. Cnossen AG, Jimenez MJ, Siebenmorgen TJ Transient moisture gradients in rough rice
(2003) Rice fissuring response to high dry- mapped with finite element model and related
ing and tempering temperatures. J Food Eng to fissures after heated air drying. Trans ASAE
59(1):61–69 39(2):625–631
10. Cnossen AG, Siebenmorgan TJ, Yang W, 21. Yang W, Jia CC, Siebenmorgen TJ, Howell
Bautista RC (2001) An application of glass TA, Cnossen AG (2002) Intra-kernel mois-
transition temperature to explain rice kernel ture responses of rice to drying and temper-
fissure occurrence during the drying process. ing treatments by finite element simulation.
Dry Technol 19(8):1661–1682. https://doi. J Long Form Workform 45(4):1037–1044
org/10.1081/Drt-100107265 22. Banaszek MM, Siebenmorgen TJ (1993)
11. Hwang SS, Cheng YC, Chang C, Lur HS, Lin Individual Rice Kernel Drying Curves.
TT (2009) Magnetic resonance imaging and Trans ASAE 36(2):521–528. https://doi.
analyses of tempering processes in rice ker- org/10.13031/2013.28368
nels. J Cereal Sci 50(1):36–42. https://doi. 23. Chen H, Siebenmorgen TJ, Marks BP
org/10.1016/j.jcs.2008.10.012 (1997) Relating drying rate constant to
12. Cnossen AG, Siebenmorgen TJ (2000) The head rice yield reduction of long-grain rice.
glass transition temperature concept in rice Trans ASAE 40(4):1133–1139. https://doi.
drying and tempering: effect on milling qual- org/10.13031/2013.21331
ity. Trans ASAE 43(6):1661–1667 24. Kunze O (1979) Fissuring of the rice
13. Perdon AA, Siebenmorgen TJ, grain after heated air drying. Trans
Mauromoustakos A (2000) Glassy state ASEA 22(5):1197–1201. https://doi.
org/10.13031/2013.35183
16 Vito M. Butardo Jr. and Nese Sreenivasulu

25. Zhang Q, Yang W, Jia C (2003) Preservation 36. Swamy YMI, Bhattacharya KR (1982)
of head rice yield under high-temperature tem- Breakage of rice during milling. 4. Effect of
pering as explained by the glass transition of rice kernel chalkiness. J Food Sci Technol Mysore
kernels. Cereal Chem 80(6):684–688. https:// 19(3):125–126
doi.org/10.1094/CCHEM.2003.80.6.684 37. Fitzgerald MA, McCouch SR, Hall RD (2009)
26. Ding C, Khir R, Zhongli P, Zhang J, Not just a grain of rice: the quest for quality.
El-Mashad H, Tu K (2015) Effect of infrared Trends Plant Sci 14(3):133–139. https://doi.
and conventional drying methods on physico- org/10.1016/j.tplants.2008.12.004
chemical characteristics of stored white rice. 38. Cheng FM, Zhong LJ, Wang F, Zhang
Cereal Chem J. https://doi.org/10.1094/ GP (2005) Differences in cooking and eat-
CCHEM-11-14-0232-R ing properties between chalky and trans-
27. Fitzgerald MA, Resurreccion AP (2009) lucent parts in rice grains. Food Chem
Maintaining the yield of edible rice in a warm- 90(1–2):39–46. https://doi.org/10.1016/j.
ing world. Funct Plant Biol 36(12):1037– foodchem.2004.03.018
1045. https://doi.org/10.1071/fp09055 39. Chun A, Song J, Kim K-J, Lee H-J (2009)
28. Zakaria S, Matsuda T, Tajima S, Niita Y (2002) Quality of head and chalky rice and deterio-
Effect of high temperature at ripening stage on ration of eating quality by chalky rice. J Crop
the reserve accumulation in seed in some rice Sci Biotechnol 12(4):239–244. https://doi.
cultivars. Plant Prod Sci 5(2):160–168 org/10.1007/s12892-009-0142-4
29. Tabata M, Hirabayashi H, Takeuchi Y, Ando 40. Yoshioka Y, Iwata H, Tabata M, Ninomiya S,
I, Iida Y, Ohsawa R (2007) Mapping of quan- Ohsawa R (2007) Chalkiness in rice: potential
titative trait loci for the occurrence of white-­ for evaluation with image analysis. Crop Sci
back kernels associated with high temperatures 47(5):2113–2120
during the ripening period of rice(Oryza sativa 41. Sun CM, Liu T, Ji CX, Jiang M, Tian T, Guo
L.). Breed Sci 57(1):47–52. https://doi. DD, Wang LJ, Chen YY, Liang XM (2014)
org/10.1270/Jsbbs.57.47 Evaluation and analysis the chalkiness of con-
30. Sreenivasulu N, Butardo VM, Misra G, Cuevas nected rice kernels based on image process-
RP, Anacleto R, Kavi Kishor PB (2015) ing technology and support vector machine.
Designing climate-resilient rice with ideal grain J Cereal Sci 60(2):426–432
quality suited for high-temperature stress. 42. Endo-Higashi N, Izawa T (2011) Flowering
J Exp Bot 66(7):1737–1748. https://doi. time genes heading date 1 and early heading
org/10.1093/jxb/eru544 date 1 together control panicle development in
31. Ashida K, Iida S, Yasui T (2009) Morphological, rice. Plant Cell Physiol 52(6):1083–1094
physical, and chemical properties of grain and 43. Huang XH, Zhao Y, Wei XH, Li CY, Wang A,
flour from chalky rice mutants. Cereal Chem Zhao Q, Li WJ, Guo YL, Deng LW, Zhu CR,
86(2):225–231. https://doi.org/10.1094/ Fan DL, Lu YQ, Weng QJ, Liu KY, Zhou TY,
cchem-86-2-0225 Jing YF, Si LZ, Dong GJ, Huang T, Lu TT,
32. Zhu LJ, Dogan H, Gajula H, Gu MH, Liu Feng Q, Qian Q, Li JY, Han B (2012) Genome-­
QQ, Shi YC (2012) Study of kernel structure wide association study of flowering time and
of high-amylose and wild-type rice by X-ray grain yield traits in a worldwide collection of
microtomography and SEM. J Cereal Sci rice germplasm. Nat Genet 44(1):32–U53
55(1):1–5 44. Sun CH, Chen D, Fang J, Wang PR, Deng XJ,
33. Lyman NB, Jagadish KSV, Nalley LL, Dixon Chu CC (2014) Understanding the genetic
BL, Siebenmorgen T (2013) Neglecting rice and epigenetic architecture in complex net-
milling yield and quality underestimates eco- work of rice flowering pathways. Protein Cell
nomic losses from high-temperature stress. 5(12):889–898
PLoS One 8 (8): ARTN e72157. https://doi. 45. Tsuji H, Taoka K, Shimamoto K (2011)
org/10.1371/journal.pone.0072157 Regulation of flowering in rice: two florigen
34. Lisle AJ, Martin M, Fitzgerald MA (2000) genes, a complex gene network, and natural
Chalky and translucent rice grains differ in variation. Curr Opin Plant Biol 14(1):45–52
starch composition and structure and cooking 46. Butardo VM Jr, Anacleto R, Parween S,
properties. Cereal Chem 77(5):627–632 Samson I, de Guzman K, Alhambra CM, Misra
35. Singh N, Sodhi NS, Kaur M, Saxena SK G, Sreenivasulu N (2017) Systems genetics
(2003) Physico-chemical, morphological, ther- identifies a novel regulatory domain of amy-
mal, cooking and textural properties of chalky lose synthesis. Plant Physiol 173(1):887–906.
and translucent rice kernels. Food Chem https://doi.org/10.1104/pp.16.01248
82(3):433–439
 Improving Head Rice Yield and Milling Quality: State-of-the-Art and Future Prospects 17

47. Sun H, Siebenmorgen TJ (1993) Milling char- 60. Kumar PA, Bal S (2007) Automatic unhulled
acteristics of various rough rice kernel thickness rice grain crack detection by X-ray imaging.
fractions. Cereal Chem 70(6):727–733 Trans ASABE 50(5):1907–1911
48. Indudhara Swamy YM, Indudhara Swamy YM, 61. Wan YN, Lin CM, Chiou JF (2002) Rice
Bhattacharya KR (1980) Breakage of rice dur- quality classification using an automatic
ing milling - effect of kernel defects and grain grain quality inspection system. Trans ASAE
dimension. J Food Process Eng 3(1):29–42 45(2):379–387
49. Chen X, Xun Y, Li W, Zhang JX (2010) 62. Wan YN (2002) Kernel handling performance
Combining discriminant analysis and neu- of an automatic grain quality inspection sys-
ral networks for corn variety identification. tem. Trans ASAE 45(2):369–377
Comput Electron Agric 71:S48–S53 63. Pinson SRM, Jia YL, Gibbons JW (2013)
50. Dubey BP, Bhagwat SG, Shouche SP, Sainis JK Three quantitative trait loci conferring resis-
(2006) Potential of artificial neural networks tance to kernel fissuring in rice identified by
in varietal identification using morphometry of selective genotyping in two tropical Japonica
wheat grains. Biosyst Eng 95(1):61–67 populations. Crop Sci 53(6):2434–2443
51. Cheng F, Ying YB, Li YB (2006) Detection 64. Pinson SRM, Jia YL, Gibbons J (2012)
of defects in rice seeds using machine vision. Response to early generation selection for
Trans ASABE 49(6):1929–1934 resistance to rice kernel fissuring. Crop Sci
52. Courtois F, Faessel M, Bonazzi C (2010) 52(4):1482–1492
Assessing breakage and cracks of parboiled rice 65. Nelson JC, McClung AM, Fjellstrom RG,
kernels by image analysis techniques. Food Moldenhauer KA, Boza E, Jodari F, Oard JH,
Control 21(4):567–572 Linscombe S, Scheffler BE, Yeater KM (2011)
53. Lin P, Chen YM, He Y (2012) Identification Mapping QTL main and interaction influences
of broken rice kernels using image analysis on milling quality in elite US rice germplasm.
techniques combined with velocity repre- Theor Appl Genet 122(2):291–309. https://
sentation method. Food Bioprocess Technol doi.org/10.1007/s00122-010-1445-z
5(2):796–802 66. Liu X, Hua ZT, Wang Y (2011) Quantitative
54. van Dalen G (2004) Determination of the size trait locus (QTL) analysis of percentage grains
distribution and percentage of broken kernels chalkiness using AFLP in rice (Oryza sativa L.).
of rice using flatbed scanning and image analy- Afr J Biotechnol 10(13):2399–2405
sis. Food Res Int 37(1):51–58 67. Liu X, Wang Y, Wang SW (2012) QTL
55. Fang CY, Hu XQ, Sun CC, Duan BW, Xie LH, analysis of percentage of grains with chalki-
Zhou P (2015) Simultaneous determination of ness in Japonica rice (Oryza sativa). Genet
multi rice quality parameters using image anal- Mol Res 11(1):717–724. https://doi.
ysis method. Food Anal Method 8(1):70–78 org/10.4238/2012.March.22.1
56. Yadav BK, Jindal VK (2001) Monitoring mill- 68. Guo T, Liu XL, Wan XY, Weng JF, Liu SJ,
ing quality of rice by image analysis. Comput Liu X, Chen MJ, Li JJ, Su N, Wu FQ, Cheng
Electron Agric 33(1):19–33 ZJ, Guo XP, Lei CL, Wang JL, Jiang L, Wan
57. Jayas DS, Singh CB (2012) 15 - Grain quality JM (2011) Identification of a stable quanti-
evaluation by computer vision. In: Sun D-W tative trait locus for percentage grains with
(ed) Computer vision technology in the food white chalkiness in rice (Oryza sativa). J Integr
and beverage industries. Woodhead Publishing, Plant Biol 53(8):598–607. https://doi.
Cambridge, UK, pp 400–421. https://doi. org/10.1111/j.1744-7909.2011.01041.x
org/10.1533/9780857095770.3.400 69. Zhou LJ, Chen LM, Jiang L, Zhang WW,
58. Shahin MA, Hatcher DW, Symons SJ (2012) Liu LL, Liu X, Zhao ZG, Liu SJ, Zhang LJ,
17 - Development of multispectral imag- Wang JK, Wan JM (2009) Fine mapping
ing systems for quality evaluation of cereal of the grain chalkiness QTL qPGWC-7 in
grains and grain products. In: Sun D-W (ed). rice (Oryza sativa L.). Theor Appl Genet
Computer vision technology in the food and 118(3):581–590. https://doi.org/10.1007/
beverage industries. Woodhead Publishing, s00122-008-0922-0
Cambridge, UK, pp 451–482. https://doi. 70. Liu XL, Wan XY, Ma XD, Wan JM (2011)
org/10.1533/9780857095770.3.451 Dissecting the genetic basis for the effect of rice
59. Wang L, Liu D, Pu HB, Sun DW, Gao WH, chalkiness, amylose content, protein content,
Xiong ZJ (2015) Use of hyperspectral imaging and rapid viscosity analyzer profile characteris-
to discriminate the variety and quality of rice. tics on the eating quality of cooked rice using
Food Anal Method 8(2):515–523 the chromosome segment substitution line
population across eight e­ nvironments. Genome
18 Vito M. Butardo Jr. and Nese Sreenivasulu

54(1):64–80. https://doi.org/10.1139/ Cell 22(10):3280–3294. https://doi.

G10-070 org/10.1105/tpc.109.070821
71. Wan XY, Wan JM, Weng JF, Jiang L, Bi JC, 79. Sun WQ, Zhou QL, Yao Y, Qiu XJ, Xie K, Yu
Wang CM, Zhai HQ (2005) Stability of SB (2015) Identification of genomic regions
QTLs for rice grain dimension and endo- and the isoamylase gene for reduced grain
sperm chalkiness characteristics across chalkiness in rice. PLoS One 10(3):e0122013
eight environments. Theor Appl Genet 80. Zuo JR, Li JY (2014) Molecular genetic dis-
110(7):1334–1346 section of quantitative trait loci regulating rice
72. Tan YF, Xing YZ, Li JX, Yu SB, Xu CG, Zhang grain size. Annu Rev Genet 48:99–118
QF (2000) Genetic bases of appearance qual- 81. Zheng J, Zhang YD, Wang CL (2015)
ity of rice grains in Shanyou 63, an elite rice Molecular functions of genes related to grain
hybrid. Theor Appl Genet 101(5–6):823–829. shape in rice. Breed Sci 65(2):120–126
https://doi.org/10.1007/s001220051549 82. Wang Y, Xiong G, Hu J, Jiang L, Yu H, Xu J,
73. Kang H-G, Park S, Matsuoka M, An G (2005) Fang Y, Zeng L, Xu E, Xu J, Ye W, Meng X,
White-core endosperm floury endosperm-4 in Liu R, Chen H, Jing Y, Wang Y, Zhu X, Li J,
rice is generated by knockout mutations in the Qian Q (2015) Copy number variation at the
C4-type pyruvate orthophosphate dikinase gene GL7 locus contributes to grain size diversity in
(OsPPDKB). Plant J 42(6):901–911. https:// rice. Nat Genet. https://doi.org/10.1038/
doi.org/10.1111/j.1365-313X.2005.02423.x ng.3346 http://www.nature.com/ng/
74. Ryoo N, Yu C, Park C-S, Baik M-Y, Park I-M, journal/vaop/ncurrent/abs/ng.3346.
Cho M-H, Bhoo SH, An G, Hahn T-R, Jeon html - supplementary-information
J-S (2007) Knockout of a starch synthase gene 83. Wang S, Li S, Liu Q, Wu K, Zhang J, Wang S,
OsSSIIIa/Flo5 causes white-core floury endo- Wang Y, Chen X, Zhang Y, Gao C, Wang F,
sperm in rice (Oryza sativa L.). Plant Cell Rep Huang H, Fu X (2015) The OsSPL16-GW7
26:1083–1095. https://doi.org/10.1007/ regulatory module determines grain shape
s00299-007-0309-8 and simultaneously improves rice yield
75. Woo M-O, Ham T-H, Ji H-S, Choi M-S, and grain quality. Nat Genet. https://doi.
Jiang W, Chu S-H, Piao R, Chin J-H, Kim org/10.1038/ng.3352 http://www.nature.
J-A, Park BS, Seo HS, Jwa N-S, McCouch com/ng/jour nal/vaop/ncur r ent/abs/
S, Koh H-J (2008) Inactivation of the ng.3352.html - supplementary-information
UGPase1 gene causes genic male sterility and 84. Miura K, Matsuoka M (2015) Rice genet-
endosperm chalkiness in rice (Oryza sativa ics: control of grain length and quality. Nat
L.). Plant J 54(2):190–204. https://doi. Plants 1:15112. https://doi.org/10.1038/
org/10.1111/j.1365-313X.2008.03405.x nplants.2015.112
76. Wang E, Wang J, Zhu X, Hao W, Wang L, Li Q, 85. The 3000 Rice Genomes Project (2014) The
Zhang L, He W, Lu B, Lin H, Ma H, Zhang G, 3,000 rice genomes project. GigaScience 3:7.
He Z (2008) Control of rice grain-filling and https://doi.org/10.1186/2047-217X-3-7
yield by a gene with a potential signature of 86. Hertog MLATM, Rudell DR, Pedreschi
domestication. Nat Genet 40(11):1370–1374. R, Schaffer RJ, Geeraerd AH, Nicolai BM,
https://doi.org/10.1038/ng.220 Ferguson I (2011) Where systems biology
77. Li Y, Fan C, Xing Y, Yun P, Luo L, Yan B, meets postharvest. Postharvest Biol Technol
Peng B, Xie W, Wang G, Li X, Xiao J, Xu C, 62(3):223–237
He Y (2014) Chalk5 encodes a vacuolar H(+)- 87. Bhattacharya KR (2011) 13 - Analysis
translocating pyrophosphatase influencing of rice quality. In: Bhattacharya KR (ed)
grain chalkiness in rice. Nat Genet 46(4):398– Rice quality. Woodhead Publishing,
404. https://doi.org/10.1038/ng.2923 Cambridge, UK, pp 431–530. https://doi.
78. She K-C, Kusano H, Koizumi K, Yamakawa org/10.1533/9780857092793.431
H, Hakata M, Imamura T, Fukuda M, 88. Juliano BO (2007) Rice chemistry and qual-
Naito N, Tsurumaki Y, Yaeshima M, ity. Philippine Rice Research Institute, Munoz,
Tsuge T, Matsumoto K, Kudoh M, Itoh Nueva Ecija
E, Kikuchi S, Kishimoto N, Yazaki J, Ando
T, Yano M, Aoyama T, Sasaki T, Satoh H, 89. Bergman CJ, Bhattacharya KR, Ohtsubo
Shimada H (2010) A novel factor FLOURY K (2004) Rice end-use quality analysis. In:
ENDOSPERM2 is involved in regulation Champagne ET (ed) Rice chemistry and tech-
of rice grain size and starch quality. Plant nology, 3rd edn. The American Association of
Cereal Chemists, St, Paul. MN
 Chapter 2

Improving Rice Grain Quality: State-of-the-Art and Future

Prospects
Vito M. Butardo Jr., Nese Sreenivasulu, and Bienvenido O. Juliano

Abstract
Rice grain quality encompasses complex interrelated traits that cover biochemical composition, cooking,
eating, nutritional, and sensory properties. Because rice endosperm is composed mainly of starch, rice
grain quality is traditionally defined by characterizing starch structure and composition, which is then
subsequently correlated with functional properties of the grain. The current proxy tests routinely used to
describe rice grain quality preferences are rather limited to the estimation of apparent amylose content,
gelatinization temperature, and gel consistency. Additional tests that characterize starch property, visco-
elasticity, grain texture, and aroma are also employed in more advanced laboratories. However, these tests
are not routinely applied in breeding programs to distinguish cooking quality classes to reflect evolving
consumer preference and market demand. As consumer preferences in Asia and all over the world are
diverse due to varied demographics and culture, defining uniform attributes to capture regional grain qual-
ity preferences becomes more challenging. Hence, novel and innovative proxy tests are needed to charac-
terize rice grain quality to meet the demand for consumer preferences of commercially-released cultivars.
In this chapter, the current methods employed in rice grain quality monitoring are succinctly reviewed.
Future prospects for improvement are identified, introducing cutting edge technologies that can facilitate
high-throughput screening of rice diversity panels and breeding lines. Aside from addressing the require-
ments for quality improvement in the traditional inbred rice breeding programs, we also tackled the need
to enhance grain quality in the hybrid rice sector.

Key words Aroma, Appearance, Cooking, Eating, Grain quality, Milling, Nutrition, Physical, Rice,
Sensory

1 Introduction

The current challenge for rice research is not only to increase rice
productivity through higher yield (see Chapter 1) but also to add
value through improved grain quality. Grain quality is a critical
component in breeding for high yielding varieties to ensure accep-
tance by an ever-growing population of rice consumers. In the
past, rice is considered a subsistence crop, primarily consumed
worldwide as a major caloric source. Hence, the primary focus of
rice breeding programs in IRRI during the 1960s up to the 1970s

Nese Sreenivasulu (ed.), Rice Grain Quality: Methods and Protocols, Methods in Molecular Biology, vol. 1892,
https://doi.org/10.1007/978-1-4939-8914-0_2, © Springer Science+Business Media, LLC, part of Springer Nature 2019

19
20 Vito M. Butardo Jr. et al.

is on increasing productivity and yield [1]. As developing countries

achieve rice self-sufficiency and the ability to export surplus rice,
consumers became selective in preferring high quality in the suc-
ceeding decades [2]. Consequently, improving rice grain quality
has become crucial in ensuring consumer acceptance. It is also
important in alleviating the lives of marginalized farmers by pro-
viding options for rice cultivars with value-added premium traits.
The two factors that can influence sustained farmer adoption of
new rice cultivars are high yield and superior grain quality. An
emerging challenge therefore is to maintain rice grain quality and
yield in an era of dwindling natural resources and uncertain envi-
ronmental fluctuations.
Rice grain quality is hard to define and quantify, primarily
because it depends on regional consumer preference, market
demand, and intended functional use. Its definition means differ-
ent things to different demographics such that what is acceptable
in one region may not generally be preferred in another. For
instance, South Asians generally prefer long-grain rice with a high
amylose content and hard gel consistency, while Southeast Asians
favor long grains with intermediate amylose content and soft gel
consistency [3]. In some cases, top-quality rice in one region may
be considered a low-quality variety in another region. Furthermore,
because rice is composed of diverse germplasm, its grains have
huge variations in physical, biochemical, cooking, sensory, and
processing properties [4], making grain quality classification
challenging.
A generally accepted consensus among grain quality experts
and cereal chemists is that rice grain quality covers the physico-
chemical properties of rice kernel which influences their milling,
cooking, eating, sensory, and nutritional properties (Fig. 1). All
these properties are linked by the biochemical properties of the rice
kernel. The general challenge for rice grain quality research is to
come up with specific quantitative proxy measures which are sim-
ple, accurate, robust and can be universally applicable for all rice
varieties for each specific trait. This is critical to distinguish low,
medium, and premium classes of rice. Ideally, the method should
be portable, high-throughput, and measurable from the farm to
the market to facilitate ease of tracking of quality at all stages of the
value chain.
The physical and milling quality of rice is reviewed in the first
chapter. In this chapter, we review the traditional, advanced, and
emerging techniques used to characterize the cooking, eating, sen-
sory, and nutritional traits of rice. We wrap up the chapter by dis-
cussing future prospects for improvement in rice grain quality
analyses for inbred and hybrid rice.
 Improving Rice Grain Quality: State-of-the-Art and Future Prospects 21

Milling

Nutritional Cooking
Rice
Grain
Quality

Sensory Eating

Fig. 1 The scope of rice grain quality encompasses interrelated physicochemical

properties, which influence milling, cooking, eating, sensory, and nutritional traits

2 Conventional Rice Grain Quality Tests

The cooking, eating, textural, and nutritional quality of rice is rou-

tinely measured using three main physicochemical tests based pri-
marily on endosperm starch properties. These traditional tests
include estimation of apparent amylose content (AAC), gelatiniza-
ton temperature (GT), and gel consistency (GC). Whereas AAC is
directly related to the proportion and structure of amylose, GT
and GC are amylopectin properties. Low GT starch has shorter
chain amylopectin branches, whereas intermediate/high GT have
more long-chain branches. They will be discussed one by one in
the following subsections.

2.1 Apparent The first and most important physicochemical test is apparent amy-
Amylose Content lose content (AAC) estimation which relies on iodine binding to
gelatinized rice flour to produce a blue iodine-amylose complex
which is routinely quantified using a visible-light spectrophotom-
eter [5]. AAC of milled rice grains can be ranked as waxy (0–2%),
very low (3–9%), low (10–19%), intermediate (20–25%), and high
(>25%) using the AAC classification [6]. Milled rice can also be
classified as non-glutinous (non-waxy) or glutinous (waxy),
depending on the presence or absence of amylose chains, respec-
tively. Improvements were proposed in the 1970s to reduce sources
of variations identified in the original assay [7, 8]. The reproduc-
ibility of the simplified method was thoroughly validated by several
international testing laboratories and additional enhancements
were proposed, including the introduction of an automated colo-
rimetric method using AutoAnalyzer (Technicon and Bran Luebbe)
22 Vito M. Butardo Jr. et al.

[9]. This simplified method gained widespread adoption in several

routine rice breeding programs to screen for AAC of thousands of
samples per year due to its generally strong correlation with cook-
ing, textural, and digestibility attributes of cooked milled rice [10].
For instance, the method played prominent role in characterizing
the grain quality of world rice accessions in the germplasm collec-
tion of IRRI in the 1990s [11].
The IRRI Grain Quality and Nutrition Center (GQNC) cur-
rently uses a Continuous Flow Analyzer (CFA) to estimate amylose
content based on the simplified method using San++ Automated
Wet Chemistry Analyzer Model 3000 (Skalar Analytical, The
Netherlands). This new high-throughput system consists of an
auto-sampler, an amylose chemistry unit (manifold, proportioning
pump, and colorimeter with 620 nm filter), an interface, and a
computer for data processing. It offers flexibility for increasing
daily sample throughput by optionally installing additional amy-
lose chemistry units. The current system in GQNC can estimate
the amylose content of more than 300 samples per run using two
chemistry units (see Chapter 8).
There are two approved methods for the estimation of amylose
content in milled rice. One was established by the Association for
American Cereal Chemists (AACC) International (AACCI Method
61-03.01) and the other by the International Organization for
Standardization (ISO Method 6647-1:2015) [12, 13]. Despite
these standard methods, one major drawback of iodine binding for
amylose determination is the increased iodine affinity to longer
amylopectin chains that overestimates actual amylose values [14–
16]. This is the reason why the result obtained from this method is
technically known as “apparent” amylose content, to signify that it
just provides a blue value (BV) estimate of the true amylose con-
tent of the flour sample. An international cooperative testing in
2009 identified additional sources of inter-laboratory variations in
the conduct of the amylose assay [17]. Additional improvements
were proposed including the use of calibrated rice standards, the
use of longer wavelengths (e.g., 720 nm) to minimize the interfer-
ence of iodine-amylopectin complex [17], and the replacement of
acetate with ammonium buffer to reduce amylopectin-iodine
absorbance [18]. Using ammonium buffer (pH 9) for amylose
estimation provides more accurate result compared to the tradi-
tional acetate buffer because it reduces amylopectin-iodine absorp-
tion and gives AAC values close to the DSC method (see below).
Proposal to improve the ease, portability, and affordability of the
assay are also published, including the use of standard color charts
[19], portable colorimeter [20], and direct iodine staining of
milled rice grains [21]. It must be pointed out, however, that no
international cooperative testing has been published on these new
proposed modifications to date. Additional modifications might be
harder to implement because these radical changes in the p ­ rocedure
 Improving Rice Grain Quality: State-of-the-Art and Future Prospects 23

can alter the amylose classification of previously-released commer-

cial rice varieties. For example, the use of ammonium buffer
pH 9 in place of acetate buffer pH 4.5 lowered apparent amylose
content of milled rice by about 3%, but closer to DSC values [18,
22]. Implementation of this new method can potentially have his-
torical ramifications on rice breeding programs worldwide.

2.2 Gelatinization The second physicochemical test is gelatinization temperature

Temperature (GT) to estimate the time and temperature of cooking of milled
rice grains. GT is the temperature at which the semicrystalline
structure of starch begins to melt upon cooking of rice flour in a
specific ratio of water. It is traditionally measured by determining
the loss of birefringence of starch granules as viewed under a polar-
ized light microscope. The GT of milled rice grains are ranked as
low (55–69.5 °C), intermediate (70–74 °C), and high (74.5–
80 °C) [23]. There are several techniques to measure GT in rice
based on the changes in optical, hydration, swelling, crystallinity,
and viscosity parameters of flour slurry during cooking [24]. The
two methods that gained popularity are based on grain digestion
by alkali and differential scanning calorimetry (DSC). The method
based on alkali digestion is an indirect proxy measure which relies
on the tendency of starch granules of milled rice to disintegrate
when soaked in dilute alkali such as potassium hydroxide [25].
This is numerically represented as alkali spreading value (ASV)
[25] or alkali score (AS) [26]. In comparison, the DSC method is
more accurate and direct because it relies on measuring the first
peak of the endotherm in real time as the starch granules gelatinize
[27, 28]. This peak endotherm is associated with variations in amy-
lopectin fine structure [29, 30] which was validated using more
than 4000 rice samples [31]. However, the DSC method cannot
be routinely used for screening in rice breeding programs because
the aluminum pans and lids used in the assay are expensive.
Consequently, the ASV is the routine method of choice to screen
thousands of lines in the breeding pipeline. Validation of GT using
DSC is usually conducted only on advanced breeding lines before
their release as new rice varieties. Waxy and low-AC rices with ASV
score of 4–5 typically have high instead of intermediate GT [32].
The gelatinization temperature of rice can also be estimated by
measuring the viscosity of starch pastes. A method based on
Brabender amylograph viscosity can estimate final GT of rice flour
paste as validated by an international cooperative test [33].
However, this method is low-throughput and requires large quan-
tities of samples (50 g) which is not suitable for routine breeding
programs. This method is replaced by Rapid Visco Analyzer (RVA)
which offers speed, convenience, and lower sample size require-
ment (4 g). More recently, the AACC International approved an
RVA method (AACCI Method 61-04.01) to precisely estimate the
GT of milled rice flour based on a recent collaborative
24 Vito M. Butardo Jr. et al.

i­nter-­laboratory study. This method provides more precise and

accurate results compared to those obtained from parallel testing
using the amylograph and the DSC methods [34].

2.3 Gel Consistency The third physicochemical method is the measurement of gel con-
sistency (GC) to reflect the range of texture observed in high amy-
lose rice grains. The eating quality of rice can vary between amylose
classes. The original test was developed to exclusively distinguish
between textural attributes of high amylose rice grains with 24–30%
AAC [35]. The test relies on the horizontal migration of cold rice
paste after cooking, cooling, and incubation using 13 × 100 mm
glass test tubes. Gel consistency is ranked as hard (27–40 mm),
medium (41–60 mm), and soft (61–100 mm) depending on the
migration of rice starch paste. Hard GC is due to high content of
long-chain amylopectin while soft GC is because of a higher pro-
portion of short-chain amylopectin. The GQNC uses high (27–
35 mm), intermediate (36–49 mm), and low (≥50 mm) scoring
system. Using this test, rice grains with softer gel consistency pro-
duce tender cooked grains which remain soft even after cooling
[36]. Because consumers tend to prefer rice grains with softer grain
consistency in Southeast Asia, this trait has been the target of most
rice breeding programs in the region [37]. Hard gel consistency
correlates with a high content of long-chain amylopectin with high
iodine affinity, found in high-AC rices [14, 38]. The “hot-water
insoluble amylose” of Bhattacharya [39] is probably long-chain
amylopectin that remains in the gelatinized starch granule.

3 Current Advanced Rice Grain Quality Tests

The three proxy tests mentioned above are insufficient to classify

the diversity of grain quality observed in rice varieties. This is
because cooking, eating, and textural quality of rice is a complex
multifactorial attribute which is best characterized by a panel of
rice consumers through sensory evaluation. This will be discussed
in a subsequent section. However, sensory evaluation is a time-­
consuming, expensive, and generally subjective method. Hence, a
combination of sensory and proxy measures is routinely employed
to infer the cooking, eating, and textural properties of rice. For
example, the eating quality of rice can be extrapolated based on
Instron hardness, gel and amylograph consistency, and final gelati-
nization of rice [40]. This highlights the need to look at grain
quality using multiple parameters at systems level [16]. It also
points out to the need to develop additional proxy measures to
exhaustively characterize the complexity of grain quality of cooked
rice kernels.
In order to gain a deeper understanding of cooking quality,
starch-based parameters need to be defined from single- to
 Improving Rice Grain Quality: State-of-the-Art and Future Prospects 25

­ ulti-­dimensional parameters. There is a need to move beyond

m
amylose content estimation to more advanced rice grain quality
tests involving starch structure, viscoelasticity, rheometry, texture,
and aroma (Table 1). All of these assays are tedious, low- to
medium-­throughput, expensive and time-consuming. These traits
need to be quantified through semi-throughput techniques to
derive new rice cultivars screened by advanced proxy measures and
validated through consumer panels. Ideally, molecular markers
associated with grain quality traits should also be developed to
facilitate multiple screening using molecular techniques (see
below). In this section, advanced state-of-the-art techniques that
can measure multiple grain quality parameters are described,
including the use of molecular markers associated with rice grain
quality.

3.1 Apparent Several analytical methods have been proposed to more accurately
Amylose Content determine the amylose content of rice grains which do not rely on
iodine binding. This includes the use of differential scanning calo-
rimetry (DSC), near-infrared spectroscopy (NIRS), and deb-
ranched starch size exclusion chromatography (SEC). The DSC
method is based on measuring the melting endotherm of the
amylose-­ lipid complex in non-glutinous rice samples [27].
Subsequent studies revealed a solid linear relationship between
amylose content and the enthalpies of melting of the amylose-lipid
complex as observed in rice flour and other major cereals and root
crops [43]. This calorimetric method is not affected by long-chain
amylopectin interference or the presence of endogenous lipids,
offering high accuracy, reproducibility, and convenience [44]. On
the other hand, the near-infrared spectroscopy (NIRS) method
relies on near-infrared reflectance (NIR) or transmittance (NIT)
spectrum of ground milled rice [45, 46]. NIRS gained popularity
not just for estimating amylose content but also in determining
other rice quality parameters such as protein content, whiteness,
transparency, and degree of milling [47]. However, this method
requires significant calibration from training population before it
can reach a significant level of accuracy. Lastly, amylose estimation
based on debranched starch structure offered a more direct mea-
surement of molecular weight distribution (MWD) of amylose
chains [48, 49]. Mechanistic information from MWD [50] was
successfully used to characterize rice kernel mutants with defects in
the starch biosynthetic pathway [15, 51, 52]. This technique was
recently employed to quantify the amylose content of a rice diver-
sity panel with superior accuracy compared to the traditional iodine
method [16]. However, this technique relies on debranching of
rice starch, hence co-elution of amylose and long-chain amylopec-
tin is possible. This issue can be indirectly addressed by isobutanol
sub-fractionation of amylose and amylopectin fractions prior to
debranching and SEC analyses [53]. Long-chain amylopectins are
Table 1
26

Summary of general routine rice grain quality tests [4, 22, 23]

Quality
parameter Measure Principle Standard method Alternative method
A. Cooking and eating quality
Apparent Apparent Determination of apparent A 100 mg milled rice sample is dispersed in 9 mL of 1 M HPLC-SEC, DSC, NIRS,
amylose amylose amylose content by NaOH. The solution is then neutralized to pH 4.5–4.8 NIR-FTR
content content iodine binding with acetic acid. Apparent amylose content is measured
by iodine colorimetry at 620 nm.
Vito M. Butardo Jr. et al.

A variation exists where the hot-­water-­insoluble amylose

content is determined by using hot water as solvent
instead of dilute alkali. Hot-water-­insoluble amylose
content is highly correlated with cooked-rice texture.
Cooking time Cooking time Estimation on the amount Rice grains are cooked in preheated water in a test tube Cooking time is indirectly
of time it takes for a immersed under boiling water bath. A few grains are measured by
grain to cook removed at regular time intervals. The grains are pressed gelatinization
between two glass slides. Cooking time is determined from temperature using DSC
the exact time at which the opaque central core disappears.
Amount of water needed to cook the rice can also be estimated
using similar technique but using a small automatic rice
cooker. Rice to water ratio (w/v) is estimated.
Grain elongation Grain elongation Determination of the ratio This test is specifically for Basmati-type rices. Raw rice None
between cooked and grains are pre-­soaked in water for 30 min. The rice
uncooked grain grains are then cooked for 10 min in boiling water. After
cooling, grain elongation ratio between the average
lengths of the cooked grain to that of the uncooked
grain is then estimated.
Grain volume Grain volume Determination of volume Exactly 20 g of rice is cooked with 50 mL water in None
expansion expansion expansion of rice grain graduated cylinder. The setup is placed in boiling water
after cooking bath and cooked. The initial and final volume of the
cooked rice is determined.
Gelatinization Alkali spreading Determination of extent of Six grains of rice are placed in rectangular plastic box Measurement of starch
temperature value or alkali disintegration of rice containing 10 mL 1.7% KOH. The extent of grain gelatinization by
digestion grain under alkali digestion is noted after 20 h. Scores range from 1 (no differential scanning
score solution effect) to 7 (all grains are completely disintegrated). calorimeter,
Commonly done in triplicates. photometric method,
alkali photometry, or
RVA pasting curve
Pasting Starch paste Determination of swelling, Rice flour sample is mixed with water. The slurry is Brabender viscograph or
properties viscosity pasting, viscosity of subjected to a controlled cycle of heating and cooling visco-­amylograph,
profiling starch paste when under continuous mixing. The starch viscosity curve is micro Viscoanalyser
cooked and how it determined where the peak, hot-paste and cold-paste
congealed when cooled viscosity is determined. Gelatinization temperature,
breakdown, setback and breakdown are then derived
from the viscogram.
B. Textural and sensory quality
Gel consistency Gel mobility test Estimation of cooked rice A 100 mg fine rice flour powder (100-mesh) is dispersed None
grain texture by mobility in a test tube with small volume of alcohol. The flour is
of starch gel after then dispersed in 2 mL dilute KOH and cooked in
cooking in dilute alkali boiling water bath. The gel dispersion is immediately
cooled in ice and then laid horizontally on a table. The
distance of the gel mobility after an hour is noted.
Normally run in duplicates.
Texture profiling Texture profiling Characterization of cooked Rice grains are cooked by boiling and three individual rice Instron hardness testing.
rice grain texture using grains parallel with each other are placed in an Parallel plate
texture analyser aluminium plate. Some methods only require single plastometer,
grain. Texture profile analysis is done using 35 mm consistometer,
aluminium cylinder probe and a 5 kg load cell. Force-­ texturometer, hardness
deformation curve is obtained using 2-cycle tester, viscoelastograph,
compression test and post-test speed of 0.5 mm/s. tensipresser, surface
Usually done in duplicates. tensiometer, Kramer
Improving Rice Grain Quality: State-of-the-Art and Future Prospects

Selected alternative methods itemized in the next column shear or texture press,
have been systematically compared. extrusion and back
extrusion, puncture test
27

(continued)
Table 1
28

(continued)

Quality
parameter Measure Principle Standard method Alternative method
Sensory Sensory Characterization of Untrained consumer panel or trained expert panel are None
evaluation evaluation of appearance, aroma, employed to score the appearance, aroma, texture, and
cooked rice texture, and taste of taste of cooked rice. Raw rice can also be subjected to
grain cooked rice sensory evaluation. A standardized attribute and
definition has been devised for descriptive sensory
analysis of cooked rice [41, 42].
Vito M. Butardo Jr. et al.

Aroma profiling Aroma Determination of aroma of The rice is cooked and scored for aroma. Raw rice grains Detection and
cooked or raw rice grains can be cracked in the mouth and scored for aroma. Rice quantification of
grains can also be placed in dilute alkali and the evolved 2-acetyle-1-­pyrroline by
aroma smelled directly. GC-MS
Detection of total volatile
metabolome by GC-MS
Rancidity test Rancidity Test of hydrolytic rancidity Ten mL of dilute methyl red and bromothymol blue are Detection of free fatty
by increased in acidity of placed in 5 g of milled rice in a test tube. The stained acids by titration or
milled rice rice grain changes with increasing acidity from yellow- colorimetry
green to yellow to orange. Freshly milled rice are
greenish while aged rice are orange to reddish. A
solution of bromothymol blue and phenol red can also
be used. Freshly milled rice will stain deep violet while
old rice will be yellow for this test.
C. Nutritional quality
Protein content Protein content Determination of protein Amount of protein in the flour is determined by semimicro NIRS
determination content by Kjeldahl or micro Kjeldahl digestion, distillation and titration of
method ammonia nitrogen. Colorimetric method is also
available.
Lipid content Crude fat and Determination of total TLC, GC, GLC determination of lipid and fatty acid Metabolomics approach
fatty acid lipids and fatty acid composition. using LC-MS or
determination composition GC-MS
Resistant starch Quantification Determination of total Flour samples are digested overnight for 16 h using None
content of total resistant starch using alpha-amylase and amyloglucosidase. The undigested
resistant enzyme digestion starch fraction is hydrolysed into glucose and quantified
starch as total resistant starch.
Nonstarch Quantification Determination of total Starch, proteins and lipids are removed using enzymatic CE, HPLC coupled with
polysaccharide of total dietary fiber after hydrolysis. Total, soluble and insoluble dietary fiber can mass spec detector
content and dietary fiber enzyme digestion then be determined. Dietary fiber composition can also
dietary fiber be done for major non-starch polysaccharides.
content
Micronutrients Elemental Atomic absorption Micronutrient identification and quantification using AAS, None
analysis spectroscopy, X-ray ICP-MS, LIBS, ICP-OES.
fluorescence imaging
Digestibility Estimation of In vitro determination of Rice grain, flour, or starch samples are digested with Time-resolved NMR
digestibility estimated glycemic enzymes that digests starch, proteins and lipids. The
kinetics scores (EGS) and amount of sugar released or starch hydrolysed are
digestibility rate quantified. EGS or k-values are then computed.
coefficient (k-value)
Improving Rice Grain Quality: State-of-the-Art and Future Prospects
29
30 Vito M. Butardo Jr. et al.

present in all high-AC cereal starches [54]. They can be separated

from amylose by butanol crystallization of amylose as a complex,
leaving amylopectin in solution [14]. The decrease in the first peak
by SEC after amylose removal is true amylose content [54].
However, adopting this protocol for routine amylose screening
will make the method more laborious and time-consuming.
Alternatively, a more accurate method that relies on direct analyses
of native (i.e., undebranched) starches may be necessary, as will be
discussed in the subsequent section.

3.2 Amylopectin Fine Quantifying the proportion of amylose and amylopectin in rice
Structure starch does not give a complete picture. The amylopectin fine
structure has to also be considered to explain functional properties
of rice as it is related to thermal, viscometric, textural, and nutri-
tional traits. Besides being the major starch fraction, the propor-
tion and fine structure of amylopectin contributes to many
properties of cooked rice. One method based on fluorophore-­
assisted carbohydrate electrophoresis (FACE) using capillary elec-
trophoresis (CE) is currently used for this purpose [55, 56]. This
method replaced an older technique that involves gel filtration
chromatography (GFC) of debranched starch samples [57]. The
chain length distribution (CLD) of amylopectin can also be quan-
tified using high-performance size-exclusion chromatography
attached to multiangle laser-light scattering (MALLS) and refrac-
tive index (RI) detectors (HPSEC-MALLS-RI) or high-­
performance anion-exchange chromatography coupled with a
pulsed amperometric detector (HPAEC-PAD). Characterizing the
CLD of amylopectin of rice grains revealed polymodal distribution
indicating several groups of B-chains (B1, B2, B3 and B4) consis-
tent with the cluster model of amylopectin structure [58, 59].
Amylopectin chain ratio (ACR) or molar ratio of Σ degree of
polymerization (DP) 6–10/Σ DP 6–24 < 0.20 of CLD of deb-
ranched amylopectin denotes intermediate/high GT and ACR of
>0.23 is low GT [29].

3.3 Starch Paste As has been mentioned above, starch paste viscosity measurement
Viscosity using Brabender Viscoamylograph (AACC International Method
61-01.01) was replaced by the Rapid Visco Analyzer (RVA,
Newport Scientific, Australia). The RVA method (AACC
International Method 61-02.01) uses a stirring and heating vis-
cometer to mimic the cooking process of rice grain by exposing a
flour-water suspension to a heat-hold-cool-hold temperature cycle
while continuously subjecting the sample to mechanical shear
stress [60, 61]. The onset of viscosity begins when amylopectin
crystals begin to melt and hydrate in excess water at elevated tem-
perature. As the suspension is heated further, viscosity increases as
a result of starch granule swelling, while amylose and small amylo-
pectin molecules leach out [62]. The pasting profiles of rice flour
 Improving Rice Grain Quality: State-of-the-Art and Future Prospects 31

are important predictors of rice eating, cooking, and processing

quality characteristics. Viscosity curves are the most useful tool
available for assessing cooking quality of rice, with the outcome of
data associated with sensory or processing attributes [63]. Aside
from gel consistency, pasting property of rice is one of the second-
ary differences that arises within varieties with the same amylose
content [64, 65]. In addition, most RVA parameters were signifi-
cantly correlated with amylose content [66, 67]. For instance, a
low setback value is associated with softness after cooking and a
high viscosity breakdown value is related to good palatability [63,
67, 68]. In addition, cohesiveness of mass, the maximum degree to
which a sample of cooked rice holds together during chewing, is
correlated with lift off and weakly correlated with break down
[69].

3.4 Starch Gel The rheological response of the system can be resolved into elastic
Rheology and a viscous component, the complete characterization of which
provides a better idea of the gelation behavior of rice flour [70].
The gelatinization and pasting properties of rice flour described
above are indicators of the ease or difficulty of cooking flour, while
rheological profiling gives a clear picture of its viscoelastic tenden-
cies [71]. Rheological parameters are increasingly correlated with
starch granule rigidity which can cause differences in the texture of
cooked rice [70]. The viscoelastic property of rice flour is com-
monly measured using an oscillating rheometer [72]. A pioneering
study using this instrument reveals that the storage modulus (G′)
of non-glutinous rice starch gels is exponentially correlated with
AAC and that it is greater than the loss modulus (G″) [73]. The G′
measures elasticity while the G″ measures viscosity. Additionally,
the same study revealed that loss tangent (tangent δ), which mea-
sures the liquid-like properties of a gel, is inversely correlated with
AAC. Mechanistically, a two-phase gelation (staling) process is
observed using this technique. The first rapid phase is due to amy-
lose, while the second slower phase is due to the recrystallization
of amylopectin [74].

3.5 Textural Profiling The textural properties of rice have been increasingly recognized as
of Cooked Rice Grains an important characteristic that can influence their acceptability
and consumption. Measurement of the texture of cooked rice grain
is traditionally conducted using Instron food tester, Texturometer,
and viscoelastograph [75–78]. Milled rice is cooked with either the
amount of water that produce the optimum texture (<75% water),
or in excess boiling water until the core is translucent (ca. 75%
water) [79]. The main concern is how to cook rice grains differing
in starch properties to the same degree that will make them inter-­
comparable. Interestingly, an international cooperative testing to
measure cooked rice grain texture revealed that hardness and stick-
32 Vito M. Butardo Jr. et al.

Table 2
Contribution of true amylose, amylopectin chains, and protein to rice quality [82]

Amylopectin

Starch property True amylose Short-chain Long-chain Protein

Crystallinity X
Gelatinization temperature X
Apparent amylose content X X
Staling during cooling X X
Slow staling on cooling (reversible) X
Gel consistency X X
RVA/Amylograph consistency/setback X X X

iness values measured by textural profiling instruments are more

accurate than sensory panel scores [78]. The use of AAC was also
validated as a highly correlated proxy measure for texture. This led
to the use of AAC estimation as the primary screening tool to clas-
sify rice based on texture, with the instrumental methods used only
for verification of selected advanced breeding lines [80]. The cur-
rent method for textural characterization involves Texture Profile
Analysis (TPA) TA.XTplus Texture Analyzer (Stable Microsystems,
UK) to measure multiple textural parameters of cooked milled rice
grains in just one experiment (see Chapter 9). It is a double-­
compression test used to describe the hardness and stickiness of
cooked milled rice grains. During TPA analysis, rice grains are
compressed twice by a texture analyzer to imitate the action of the
jaw while measuring several attributes during simulated chewing.
Limited studies have been currently conducted to validate the
relationships of rheological parameters with instrumental texture
profiles and other quality indicators. Attempts have been made to
correlate instrumental texture profiles with rice quality predictors
such as RVA measurements and amylose content [63, 80, 81], as
will be discussed below. For example, biochemical composition
such as apparent amylose content and protein content was demon-
strated to influence pasting properties to explain the differences in
cooked rice texture [80]. It is becoming clear that true amylose
content, amylopectin chain lengths, and protein content influence
the texture of cooked rice [82] (Table 2).

3.6 Aroma Profiling Rice fragrance is an eating and a sensory trait that significantly
increases the market value of rice. There are more than 100 volatile
compounds responsible for rice grain aroma but 2-acetyl-1-­
pyrroline (2-AP) is the most potent and is responsible for the
 Improving Rice Grain Quality: State-of-the-Art and Future Prospects 33

p­ andan scent in rice kernels [83–85]. It is present in jasmine and

basmati rice, two types of rice from Thailand and India ­respectively,
which are highly valued worldwide for superior aroma and grain
quality. Because 2-AP has a high odor threshold, it is traditionally
detected by smelling after reaction with 0.1 M KOH. However,
this method is harmful to the nasal cavity of the analyst upon con-
tinuous and prolonged exposure. To circumvent this problem,
methods using gas chromatography coupled with flame ionization
detector (GC-FID) or mass spectrometry (GC-­MS) were devel-
oped [86, 87]. However, not all rice breeding facilities have
GC-MS instrument due to its expensive capital investment, run-
ning and maintenance costs. Furthermore, it cannot be used to
routinely screen thousands of parental and breeding lines because
it is low-throughput. Due to these restrictions, molecular markers
related to 2-AP are currently used in rice breeding programs for
aroma (see below).
Aroma in rice grain is not just due to 2AP. Some volatile com-
pounds in rice grain are known to enhance rice grain aroma, while
others can decrease sensory quality due to off-flavors and rancidity.
Hence, comprehensive detection and comparison of aroma com-
pounds are necessary to identify other volatile compounds to
determine how they affect sensory properties. One effective
metabolomics method is a solvent-free technique using solid-phase
microextraction (SPME) which allows for simultaneous extraction
of volatile compounds during cooking of samples [88]. This tech-
nique was able to differentiate the volatile compounds present in
aromatic and non-aromatic rice grains [89]. However, jasmine-
and basmati-type aroma was not resolved using this technique
[89]. The differences in volatile profile were resolved by combin-
ing GC-MS, GC-olfactometry, and dynamic headspace system.
Volatile compounds from rice grains can be identified using odor
threshold and odor activity values to distinguish varietal differences
in rice grain aroma [90]. The complexity of rice grain aroma and
flavor can be dissected using integrated multidisciplinary approach
involving aroma profiling, flavor metabolomics, and sensory analy-
ses. These integrated approaches can be employed to further
understand the sensory properties of rice grains [91].

3.7 Sensory Sensory evaluation involves the assessment, characterization, mea-

Evaluation surement, analysis, and interpretation of the sensory properties of
food products. Sensory evaluation for rice grain quality usually
involves comparative, descriptive, or quantitative profiling of
human perception to rice grain consumption using trained sensory
panellists or untrained market consumers. Linking sensory and tex-
tural properties to quality attributes lies in the deeper understand-
ing of the physicochemical properties and the biochemical
composition of rice grain that influence the consumer perception
of cooking, eating, and sensory quality of a particular rice cultivar.
As pointed out in the previous section, routine biochemical tests
34 Vito M. Butardo Jr. et al.

are currently being used primarily in the evaluation of breeding

lines as proxy indicators of grain quality. Understanding rice grain
quality has reached an upper limit using these traditional tech-
niques. New approaches have to be used to unravel multiple facets
of rice cooking and eating quality that were not previously cap-
tured by these routine physicochemical tests. A good starting point
is the use of sensory and consumer panels to describe and quantify
various types of rice aroma and flavor. This can be done in parallel
with the exploration of other novel analytical tools to understand
and quantify traits that influence consumer acceptance and
preference.
Descriptive sensory evaluation and instrumental analyses
revealed that the aroma and flavor of rice grains are very diverse
[92]. Consumers define various classes of rice quality at the local,
regional, and international context [93–95]. Because quality pref-
erence for rice is at least as diverse as the culture and palate of the
consumers, market segmentation has to be properly identified
based on regional influence and profile of the rice varieties. It must
be pointed out, however, that there is still a huge difference on
how sensory and quality preferences are measured in the labora-
tory compared to how actual consumer preference is estimated in
target markets. In this connection, international standards for grain
quality and sensory properties should be established so that each
rice variety can be benchmarked against a set of defined criterias.
An international collaboration among rice grain quality and sen-
sory specialists need to be forged to accomplish this task. Activities
can include exhaustive textural and aroma profiling using tradi-
tional sensory and modern instrumental techniques to improve
our understanding of cooking quality. An internationally-accepted
lexicon for taste, flavor, and aroma descriptors should also be stan-
dardized [96, 97]. Because traditional sensory methods cannot be
used as routine selection tool to breed superior grain quality
­varieties due to its inherent intricacy and subjectivity, instrumental
proxy methods need to be used to bridge the gap. The results
obtained from these instrumental sensory platforms should be
highly correlated with the major traits of interest. Sensory evalua-
tion can be used to calibrate instrumental tests which can provide
quantitative values to avoid the inherent subjectivity of the sensory
evaluation process.

3.8 Molecular Advanced rice chemistry laboratories also employ molecular mark-
Markers for Rice Grain ers to classify grain quality of rice. These molecular markers are
Quality used to improve the eating, cooking, and nutritional quality of rice
[98, 99]. One limitation of this approach is that only a few molec-
ular markers have been associated with functionally validated rice
grain quality traits by far (Table 3). These molecular markers are
primarily restricted to explaining variations in rice amylose compo-
sition [104, 114], amylopectin fine structure [66, 113, 115], and
fragrance [100, 101, 116].
 Improving Rice Grain Quality: State-of-the-Art and Future Prospects 35

Allelic variations for starch synthesis-related genes (SSRGs) coding

for starch synthases, branching enzymes, and debranching enzymes
have been associated with physicochemical properties of cooked
rice grains [67, 117–121]. Of all the SSRGs highly expressed in the
rice endosperm, the Granule-Bound Starch Synthase I (GBSSI)
gene that regulates amylose synthesis is the most comprehensively
studied. Allelic variations in the structural gene for GBSSI are asso-
ciated not just with amylose content but also with other functional
properties of cooked rice grain. For instance, alleles of this gene are
also associated with gel consistency [122], starch pasting, and
digestibility in rice [123]. Another major starch synthase gene that
influences starch and functional properties is Starch Synthase IIa
(SSIIa) which determines amylopectin fine structure and gelatini-
zation property of rice [71, 123, 124]. Allelic variations in SSIIa
were also found to influence gel consistency in rice [124].
The use of molecular markers to track for grain fragrance (fgr)
in rice is made possible with the discovery of 8-bp deletion in exon
7 of a gene that codes for betaine aldehyde dehydrogenase (Badh2)
[100, 125]. This method is more cost-effective and higher-­
throughput compared to the GC-MS-based volatile profiling
method described above. This paved the way for high-throughput
screening of aromatic rice accessions by PCR amplification of the
functional mutation in Badh2 associated with presence or absence
of 2AP. However, not all aromatic rice accessions were found to
have this deletion based on screening of a rice diversity panel, hint-
ing that other alleles for this trait must be present [87]. This was
verified by subsequent studies which revealed the existence of
novel alleles in Badh2 responsible for 2AP accumulation [101,
102]. Currently, several molecular markers are needed to detect all
variant alleles present in Badh2 as listed in Table 3. Additionally,
other nonfunctional haplotypic variants of Badh2 are detected in
diverse rice germplasm collections [122, 126]. It is possible that
other yet uncharacterized mutations exist in Badh2 that were favor-
ably selected during the course of domestication of cultivated rice.
Since the establishment of GQNC in IRRI in 2004, we rou-
tinely detect allelic variants of genes related to apparent amylose
content [106], gelatinization temperature [30], and aroma by
2-AP [87, 122, 126, 127] using polymerase chain reaction (PCR)
and gel electrophoresis. These molecular markers are also employed
in routine marker-assisted breeding (MAB) programs for rice all
over the world.

3.9 Nutritional The nutritional quality of rice covers a fairly complex set of traits
Quality related to health-promoting macronutrient, micronutrient, and
phytochemicals present in the whole or milled grains [128].
Macronutrients consist of biopolymers such as carbohydrates, pro-
teins, and lipids which make up the bulk of foods primarily con-
sumed worldwide for energy homeostasis [129]. On the other
 36

Table 3
Molecular markers associated with rice grain quality
Vito M. Butardo Jr. et al.

Trait Gene/alleles PCR primers PCR conditions Reference

Fragrance from Betaine aldehyde dehydrogenase (Badh2)
2-acetyl-1-
badh2 8-bp deletion on exon 7 TTGTTTGGAGCTTGCTGATG 94 °C, 5 min Bradbury et al. [100]
pyrroline
CATAGGAGCAGCTGAAATATATACC 35× of
CTGGTAAAAAGATTATGGCTTCA  94 °C, 45 s
AGTGCTTTACAAAGTCCCGC  55 °C, 45 s
 72 °C, 45 s
72 °C, 10 min
badh2 7-bp deletion in exon 2 CCTCTGCTTCTGCCTCTGAT 94 °C, 2 min Shi et al. [101]
GATTGCGCGGAGGTACTTG 30× of
PCR = 200/207 bp  94 °C, 30 s
or  50–60 °C, 30 s
CTTCTGCCTCTGATTAGCCT  72 °C, 1 min
GCCGTGAGCCATATACACTT 72 °C, 10 min
PCR = 643/650 bp
badh2803 bp deletion between TGCTGGATGCTTTGAGTA 94 °C, 2 min Shao et al. [102]
exon 4 and 5 GTTTAGCACACCTGAAGGAAGACCA 30× of
PCR = 321/1123 bp  94 °C, 45 s
 55 °C, 45 s
 72 °C, 1 min
72 °C, 8 min
Amylose content Granule-bound starch synthase I (GBSSI)
(CT)n Repeat TTGCAGATGTTCTTCCTGATG 95 °C, 5 min Bligh et al. [103]
CTTTGTCTATCTCAAGACAC 35× of Ayres et al. [104]
 94 °C, 45 s Bergman et al. [105]
 52 °C, 45 s Roferos et al. [106]
 72 °C, 45 s Tuaño et al. [32]
72 °C, 10 min
Exon 2 23-bp duplication TGCAGAGATCTTCCACAGCA 95 °C, 5 min
GCTGGTCGTCACGCTGAG 35× of
 94 °C, 30 s
 60 °C, 30 s
 72 °C, 1 min
72 °C, 5 min
Intron 1 G/T SNP GCTTCACTTCTCTGCTTGTG 94 °C, 5 min Ayres et al. [104]
ATGATTTAACGAGAGTTGAA 35× of Tuaño et al. [32]
 94 °C, 40 s
 55 °C, 40 s
 72 °C, 1 min
72 °C, 10 min
Exon 4 Larkin and Park [107]
Mikami et al. [108]
Exon 6 A/C SNP GGTTGGAAGCATCACGAGTT 94 °C, 5 min Cai et al. [109]
TCTTCAGGTAGCTCGCCAGT 32× of Larkin and Park [110]
CCCATACTTCAAAGGAACATA  94 °C, 30 s Chen et al. [111]
 60 °C, 30 s Tuaño et al. [32]
 72 °C, 1 s
72 °C, 5 min
Exon 10 C/T SNP GCATCACCGGCATCGTC 94 °C, 5 min Larkin and Park [110]
GCTCCGGCCATGATGAGATG 35× of Dobo et al. [112]
 94 °C, 40 s Tuaño et al. [32]
Improving Rice Grain Quality: State-of-the-Art and Future Prospects

 55 °C, 40 s
 72 °C, 1 min
72 °C, 10 min
37

(continued)
Table 3
38

(continued)

Trait Gene/alleles PCR primers PCR conditions Reference

Gelatinization Starch synthase IIa (SSIIa)

temperature
SSIIa SNP2 (A/G) AGAACGACTGGAAGATGAACG 94 °C, 5 min Waters et al. [113]
GATGTCCACACCTTTCTGCC 35× of Cuevas et al. [31]
CTTGCACCGCGGCTTGCC  94 °C, 30 s Tuaño et al. [32]
 62 or 68 °C, 30 s
SSIIa SNP3 (A/G) GGGTGGGTGGGGTTCTCG
 72 °C, 1 min
Vito M. Butardo Jr. et al.

CACCATTGGTACTTGGCCTTGAC
72 °C, 7 min
GCGGGCTGAGGGACAGCA
SSIIa SNP 4 (GC/TT) GGGTGGGTGGGGTTCTCG
CACCATTGGTACTTGGCCTTGAC
GCCGCGCACCTGGAAA
Improving Rice Grain Quality: State-of-the-Art and Future Prospects 39

hand, micronutrients are small organic (e.g., vitamins) or inorganic

(e.g., minerals) molecules present in food required in trace amounts
but are essential for normal growth and metabolism [130, 131]. In
contrast, phytochemicals are naturally-occurring bioactive second-
ary metabolites responsible for organoleptic properties in plants
that also have potential health benefits in humans, but may not
necessarily be considered as essential for normal growth and
metabolism [132]. Food consumption involving balanced intake
of macronutrients, micronutrients, and phytochemicals will ensure
general health and well-being. Current efforts are underway to tai-
lor the nutritional properties of rice grains due to its potential as a
nutrient delivery system to bring optimal health to half of the
global population [128].
Just like the other grain quality traits discussed so far, the nutri-
tional quality of rice grain is also primarily influenced by starch
since it is the major macronutrient in the milled rice grain, com-
posed of around 80–90% of grain weight. Interestingly, starch is
not digested at equal rate in the human gastrointestinal tract. The
gold standard in estimating the starch digestibility of starch-­
containing food is by measuring its glycemic impact. It is assayed
under clinical settings by determining the blood glucose response
of human volunteers upon consumption of a food containing 50 g
available carbohydrates compared to a standard solution contain-
ing 50 g glucose [133, 134]. The glycemic response (GR) is
reported as glycemic index (GI) based on a pioneering study by
Jenkins et al. in 1981 [135] which was validated by several labora-
tories in 2003 [136]. The concept of glycemic load (GL) was sub-
sequently introduced in 1997 to account for the observation that
the proportion of available carbohydrates in the food can have a
profound impact on overall glycemia [137]. Unfortunately,
determining the glycemic impact of food cannot be routinely
­
employed in screening for low GI in rice breeding programs
because it is low-­throughput and requires expensive clinical assays.
Consequently, the nutritional fractions of starch can be estimated
in vitro by quantifying total sugar, total starch, rapidly digestible
starch (RDS), slowly digestible starch (SDS), and resistant starch
(RS) [138, 139]. GI can also be estimated in vitro using starch
hydrolysis [140], estimated glycemic score (EGS) [141], and
hydrolysis index (HI) values [15, 123]. In addition, the rate and
extent of starch digestibility can also be estimated by amylolysis
using log of slope (LOS) plot for first-order kinetics [142]. The
in vitro methods routinely employed in GQNC to estimate cooked
grain digestibility are presented in Chapter 13.
Aside from starch, the other macronutrients in rice are storage
proteins, storage lipids, and nonstarch polysaccharides (NSPs).
Although they are present as minor components in the rice grain,
these biopolymers play a major role in influencing nutritional qual-
ity, textural and sensory traits, and functional properties [128]. For
40 Vito M. Butardo Jr. et al.

instance, the average storage protein composition of milled rice

grain is just around 7% but it is a major source of protein in devel-
oping countries. Rice storage protein is gaining popularity world-
wide because it is hypoallergenic and has superior amino acid
composition [143]. The Kjeldahl Method [144] is the
internationally-­accepted protocol widely used for estimating total
protein content (AACCI Method 46-16.01), with modifications
to accommodate smaller sample sizes (AACCI Method 46-13.01)
[12]. Individual amino acids can also be quantified after acid
hydrolysis using pre-column derivatization using a fluorescent
derivatizing reagent followed by HPLC separation [145, 146].
Similar to storage proteins, the storage lipids in rice are just less
than 1% in milled rice grain but they significantly influence nutri-
tional and functional properties [120]. The compositional analyses
of storage lipids in rice are interesting, especially from a nutritional
and grain quality perspective [147]. For instance, minor lipids in
rice grains are found to influence pasting properties and to confer
protection against chronic ailments including cardiovascular dis-
eases and cancer [148]. Crude fat in rice grains is routinely ana-
lyzed using a standard method (AACCI Method 30-10.01).
Classical method employs thin-layer and gas chromatography to
determine lipid composition in rice grains [149]. A standard
method is routinely available to determine total, saturated, unsatu-
rated, and monounsaturated fats after acid hydrolysis (AACCI
Method 58-19.01). The proportion of nonstarch (free or bound)
and starch lipids (i.e., amylose-complexed) can also be determined
in milled rice. The fatty acid composition of the bran layer can also
be analyzed using gas-liquid chromatography (GLC) [150]. More
recently, a liquid chromatography method coupled to a mass spec
detector (LC-MS) was developed to characterize the phospholipid
content of rice [151] and to determine variation in composition
among cultivars [152].
NSPs constitute the dietary fiber in rice grains which are con-
centrated in the bran layer. Although they are just present in trace
amounts in milled grains, NSPs are interesting from a nutritional
point of view because they are compositionally unique in rice com-
pared to other cereals [135]. The total dietary fiber (TDF) in the
grain can be assayed after enzymatic removal of starch, proteins,
and lipids (AACCI Method 32-05.01). A rapid gravimetric method
is also available using parallel determinations of water-soluble and
neutral detergent-insoluble fractions (AACCI Method 32-06.01).
The total, soluble and insoluble dietary fiber can also be estimated
using a single assay (AACCI Method 32-07.01). These composi-
tional assays should be supplemented with structural determina-
tions to identify the diversity, size, and shape of NSPs in rice grain.
Micronutrients and phytochemicals in rice grains are also pri-
marily concentrated in the rice bran [153]. Polishing to produce
milled rice grains removes most of the micronutrients in rice which
 Improving Rice Grain Quality: State-of-the-Art and Future Prospects 41

could otherwise be used to mitigate the impact of micronutrient

deficiency, a form of hidden hunger that continues to plague
underdeveloped and developing countries worldwide.
Micronutrient deficiency can lead to growth stunting characterized
by growth retardation, developmental delay, impairment in the
immune system, reduction in cognitive function, and increased risk
of metabolic disorders such as obesity and hypertension [154].
Efforts are underway to increase the micronutrient density and
composition in rice endosperm to circumvent the problem of
micronutrient deficiency in rice-consuming countries [155, 156].
Hence, innovative methods are required to simultaneously identify
and quantify micronutrients in milled rice grains. This is usually
accomplished using atomic absorption spectrophotometry (AAS),
inductively coupled plasma–mass spectrometry (ICP-MS), laser-­
induced breakdown spectroscopy (LIBS), and inductively coupled
plasma–optical emission spectrometry (ICP-OES) [157, 158]. In
GQNC, we use ICP-OES to simultaneously quantify 12 elements:
Na, K, Ca, Mg, Fe, Mn, Zn, Cu, Al, P, S, and Mo (Chapter 14). In
contrast, there is no routine method for the quantification and
characterization of bioactive phytochemicals in rice grain. Readers
are referred to original research articles for analytical procedures
[159–166].

4 Emerging Rice Grain Quality Tests

Most of the emerging state-of-the-art rice grain quality tests involve

the use of nondestructive, multi-parameter assays that are highly
correlated with traditional methods. Most of these tests have not
been substantially validated using large sample population of rice
grains but they offer significant improvement in sample through-
put and accuracy. However, most of these methods would require
significant capital investments.

4.1 Grain The accurate and repeatable separation of highly branched polysac-
Structural Traits charides such as rice starch in SEC remains challenging even when
coupled with multiple detectors [167]. A range of instrumental
techniques are currently employed to analyze several levels of
starch structural organization. The main challenge in this area is to
characterize starch structure in its native conformation. Toward
this goal, combining flow field-flow fractionation (FFF) with
multi-angle laser light scattering (MALLS) and differential refrac-
tometer index (DRI) were successful in fractionating starch in
aqueous conditions [168]. More recently, two separation methods
using asymmetrical flow field flow fractionation (A4F) and hydro-
dynamic and size-exclusion chromatography (HDC-SEC) were
used to determine the molecular size and mass distributions of
native starches including rice [169]. Similar results were also
42 Vito M. Butardo Jr. et al.

accomplished with A4F in tandem with MALLS and refractive

index (RI) detectors [170]. In addition, free solution capillary
electrophoresis (CE) can provide an alternative to the conventional
debranched SEC technique because of its ability to determine
native starch structure, composion, and degree of branching [171].
In terms of higher granular organization, a distinct nonlinear spec-
tral shift in hydrated starches was detected as starch molecular
order increases using a combination of DSC, 13C nuclear magnetic
resonance (NMR), X-ray diffraction (XRD), and Fourier transform
infrared spectroscopy (FTIR) techniques [172].

4.2 Grain Spectroscopic techniques such as NIRS, Fourier transform infrared

Cooking Traits spectroscopy (FTIR), and Raman spectroscopy show promising
potential in characterizing the cooking quality of rice grains.
Several spectroscopic techniques, mostly using NIRS, were
employed to characterize apparent amylose [45, 46, 173], mois-
ture [174], protein [175], and fat contents [176]; gelatinization
[174], pasting [63, 177], and textural properties [178, 179]; as
well as solid loss, volumetric expansion, and water uptake of rice
during cooking [180].

4.3 Grain Chemometric method employing various parameters from starch

Textural Traits structure and rice grain composition highlighted the importance
of amylose-to-amylopectin ratio (AAR) and crude protein (CP) in
influencing the texture of rice grains [181]. A combination of TPA
and dynamic rheological measurement revealed that consistency
coefficient (K*) and loss tangent (tan δ) are appropriate proxy
measures for rice grain hardness and stickiness, respectively [182].
Predictive model associating textural properties of rice with the
physicochemical parameters of rice grains have also been done in
the past. One such model revealed that AAC, blue value (BV), and
moisture content (MC) significantly influence the texture of rice
[183].

4.4 Grain As mentioned above, some sensory traits such as hardness and
Sensory Traits stickiness are better measured instrumentally using TPA. However,
not all sensory traits have instrumental correlations. Hence, one of
the main challenges for sensory profiling of rice grain is to identify
instrumental methods that are well correlated with sensory traits.
Toward this goal, a single-compression TPA test coupled with
novel multivariate data analysis successfully predicted rice sensory
texture for cooked grains [184]. In the same study, seven sensory
attributes (cohesion, adhesion, hardness, cohesiveness of mass,
roughness of mass, toothpull, and toothpack) were predicted using
the novel step-wise data analysis method. Sensory preference and
taste scores were predicted by combining taste analyzer with rou-
tine grain quality data and utilizing fuzzy logic algorithm [185].
Metabolome profiling coupled with descriptive sensory evaluation
 Improving Rice Grain Quality: State-of-the-Art and Future Prospects 43

identified metabolite biomarkers related to excellent rice flavor

quality that can differentiate jasmine-type rice grains [186]. Visible
and shortwave near-infrared spectral data from NIRS were used to
predict sensory quality of rice [187]. Modeling for sensory prop-
erty of indica rice was successful using support vector machine
(SVM) [188]. Lastly, the use of electronic nose (e-nose) and elec-
tronic tongue (e-tongue) to respectively detect volatile and non-
volatile compounds in food and beverages are gaining popularity
due to advancements in sensor and separation technologies [189].
These advanced sensor and separation technology can potentially
be used in rice grain sensory evaluation in the future. In addition,
the oral processing of food is also receiving considerable attention
to describe the pleasure of taste, texture, and flavor during con-
sumption [190–192]. Toward this end, the rheology and tribology
of food texture sensation are increasingly becoming important
tools in food consumption and sensory evaluation [193, 194].
Techniques from sensory science perspective can be combined
with concepts in material science to capture the dynamic aspects of
oral processing [195]. However, no systematic study to date has
been conducted to characterize the tribology of rice grain con-
sumption. This is an important area for future study considering
that chewing behavior varies more among consumers than rice
types [196]. Systematic research on oral processing of cooked rice
grain is also necessary in altering grain composition to make them
healthier without compromising sensory quality including texture
and mouthfeel. Lastly, limited studies have also been done on
short-term and long-term satiety indices of cooked milled and
brown rice (such as satiety quotient for hunger, fullness, desire to
eat and prospective consumption; overall satiety index; and 2-h
postmeal rice intake) [41, 197].

4.5 Grain In vitro carbohydrate digestion methods are available but substan-
Digestibility Traits tial improvements and validation are needed [42, 198]. The kinet-
ics of in vitro digestibility of starch has been tracked by time-resolved
NMR [199]. Free solution CE also offers the potential to track
release of sugars and short oligosaccharides in real time by simple
UV detection during in vitro amylolysis [200].

5 Future Prospects

Rice cereal chemists have continued to develop innovative proxy

methods to estimate grain quality traits to reflect constantly
evolving and rapidly expanding consumer preference. In this last
section, future prospects for rice grain quality assessment is dis-
cussed by considering what methods are routinely used in the
past to the present and by taking into account available emerging
technologies.
44 Vito M. Butardo Jr. et al.

5.1 Accurate Actual sensory evaluation will only be a critical component in rice
Assessment breeding in characterizing parental lines and prior to releasing the
of Consumer newly-developed rice cultivar. It is therefore necessary to integrate
Preference instrumental proxy measures for sensory evaluation during the
early stages of the breeding pipeline. The extent to which correla-
tions can be made between instrumental and panel-based sensory
profiling can be validated using different samples capturing regional
preferences of various market quality segments. Toward this goal,
it would be interesting to merge methods in rheology, rheometry,
and textural analyses with taste and flavor metabolomics as well as
systems genetics [201, 202].
Instrumental and panel-based characterization of sensory attri-
butes in the private rice sector is becoming more crucial because it
is market and demand-driven. If there is no repeat demand for the
newly-released rice cultivar, the variety will not succeed in penetrat-
ing the target market. In this connection, quality and sensory pro-
filing of the rice mega varieties using instrumental and sensory tests
are important to understand the reasons why these lines continue
to be widely adopted by farmers in major rice growing regions and
preferred by majority of rice consumers despite the availability of
newer rice varieties. Market research and market intelligence will
also be crucial for preference mapping and clustering of consumer
demands for various quality and sensory attributes.

5.2 Including Rice is a diverse species with normal apparent amylose range from
Nutritional Quality 0 to 30% [16]. We therefore expect diversity not only in starch
in the Rice structure but also in rice grain digestibility. Rice has an important
Improvement Agenda role to play to mitigate the impact of non-communicable diseases.
Available germplasm can be screened for resistant starch, amylose
content, digestibility, and other health-promoting properties.
Since clinical evaluation of rice digestibility is difficult, an accurate
method of in vitro estimation of digestibility should be developed
and validated. In vitro hydrolysis index [15, 123] can be used for
screening but its suitability should be further validated in vivo.
Cooking process that can alter digestibility should also be investi-
gated. This should also be coupled with industrial processes to
modify rice digestibility such as parboiling and processing of high
amylose-based products. In connection with this, regional specifi-
cations and demands should also be considered. More importantly,
some of the cooking quality traits can also affect nutritional traits.
These tradeoffs need to be tackled to identify the key determinants
of rice grain cooking quality while ensuring health benefits.

5.3 Associating A deeper understanding of the genetic bases of rice grain quality
Additional Molecular attributes can ultimately help breeding programs predict the qual-
Markers to Rice Grain ity of their newly developed breeding lines at a faster rate through
Quality marker-assisted breeding (MAB) and genomic selection (GS). Rice
grain quality traits are controlled by major and minor quantitative
 Improving Rice Grain Quality: State-of-the-Art and Future Prospects 45

trait loci (QTLs), implying that the environmental and genetic

mechanisms underlying quality traits are complex [195]. It is also
influenced by biochemical composition of grain, storage, and pro-
cessing parameters. In this era where next-generation sequencing
(NGS) is commonly employed to generate thousands of rice
genome sequences [203], phenotyping these accessions and per-
forming genome-wide association studies to identify key regula-
tory regions of grain quality and nutritional traits will be of added
value. In this context, the development of robust and accurate
grain quality and nutritional phenotyping methods elaborated
above should keep pace with the rapid advancements in genomics
and genome-wide association tools freely available to characterize
rice diversity population.
Associating molecular markers to grain quality traits can facili-
tate predictive classification of rice at the genotype level and it can
assist in the molecular breeding of rice cultivars targeting for supe-
rior grain quality in higher yielding background. These markers
can aide in rice breeding by marker-assisted selection (MAS).
However, employing robust genomic selection for rice grain qual-
ity is still in its infancy. This is because a molecular understanding
of how various physical and biochemical properties are related to
processing, cooking, and sensory quality is generally lacking.
Development of additional functional diagnostic markers related
to rice grain quality is therefore imperative to breed for rice culti-
vars with novel combinations of traits while ensuring higher accept-
ability by consumers. Modern phenotyping methods such as
understanding comprehensive starch structure influence on cook-
ing quality and extracting the value of visco-elastic properties and
texture in selecting premium cooking quality classes during pre-­
breeding are crucial to accurately detect the genotypic basis of the
specific trait of interest. These novel traits can then be easily tracked
during the course of introgression and subsequently pyramided
into locally-adapted cultivars using a suite of functional markers.
To achieve this, a collaborative team of breeders, food scientists,
cereal chemists, and molecular geneticists need to work on a holis-
tic approach to reflect on multi-dimension traits to distinguish
low-, medium-, and premium-quality segments. The use of
genotyping-­ by-sequencing (GBS), targeted resequencing, and
next-generation sequencing (NGS) coupled with genome-wide
association studies (GWAS) and other systems genetics tools are
expected to significantly increase the identification and utilization
of single-nucleotide polymorphisms (SNPs) associated with rice
grain quality and nutrition.

5.4 Integrating Grain The success of hybrid rice breeding pipelines is normally gauged
Quality and Nutrition based on yield potential, as well as tolerances and resistances to
in Hybrid Rice various biotic and abiotic stresses. As rice consumers become
Technology increasingly particular about the quality of the rice they are eating
and how rice contributes to human health, rice hybrid breeders
46 Vito M. Butardo Jr. et al.

need to look at how to improve the quality in a more holistic way.

As a first step, initial quality indicators of cooking quality should be
linked with texture and visco-elastic properties to capture distinct
cooking grain classes. Subsequently, the genetic basis of the varia-
tion of cooking and eating quality traits should be established
through genome-wide association studies (GWAS) by studying
both diverse set of rice core collection and pre-breeding materials.
Overall emphasis has to be given to holistic understanding of vital
molecular-physiological mechanisms of grain quality traits through
gene discovery via QTL cloning and structural-functional genom-
ics strategies and utilizing this information in the hybrid rice breed-
ing sector to develop superior quality rice lines. There is an urgent
need to improve grain quality preferences in the hybrid rice sector
because many programs currently use parental materials with diver-
gent grain quality traits. This leads to the production of hybrid rice
varieties that segregate for grain quality traits.
In summary, more holistic strategies are needed to incorporate
value-added rice grain quality traits by multi-QTL pyramiding
approaches both in the inbred and hybrid rice breeding pipelines.
Toward this goal, a recent study has demonstrated that high yield-
ing elite rice lines can be pyramided with multiple complex traits to
ensure superior grain quality [204]. The systematic design of high-­
yielding and superior-quality elite rice lines in the future will rely
not only on the rapid advancements in genetics and genomics
technologies but also on the current and emerging phenotyping
methods for rice grain quality.

Acknowledgments

This work has been supported under the CGIAR thematic area
Global Rice Agri-Food System CRP, RICE, Stress-Tolerant Rice
for Africa and South Asia (STRASA) Phase III, and Australian
Centre for International Agricultural Research (Project ID
CIM/2016/046) funding.

References
1. International Rice Research Institute (1985) 3. Khush GS, Juliano BO (1985) Breeding for
Rice grain quality and marketing. Papers pre- high-yielding rices of excellent cooking and
sented at the 1985 International Rice eating qualities. In: Rice grain quality and
Research Conference, Los Banos, Laguna. marketing. Papers presented at the
International Rice Research Institute, Manila, International Rice 1985 Research Conference.
Philippines, p 74 International Rice Research Institute, Manila,
2. International Rice Research Institute (1979) Philippines, pp 61–69
Proceedings of the workshop on chemical 4. Bergman CJ, Bhattacharya KR, Ohtsubo K
aspects of Rice Grain Quality, IRRI, 1978. (2004) Rice end-use quality analysis. In:
International Rice Research Institute, Los Champagne ET (ed) Rice chemistry and tech-
Banos, Laguna, Philippines, p 390 nology, 3rd edn. AACC International, St.
Paul, MN, pp 415–472
 Improving Rice Grain Quality: State-of-the-Art and Future Prospects 47

5. Williams VR, Wu W-T, Tsai HY, Bates HG Indrasari SD, Ramli A, Ahmad R, Dipti SS,
(1958) Rice starch, varietal differences in Xie L, Lang NT, Singh P, Toro DC, Tavasoli
amylose content of rice starch. J Agric Food F, Mestres C (2009) Addressing the dilem-
Chem 6(1):47–48 mas of measuring amylose in rice. Cereal
6. Kumar I, Khush GS (1986) Genetics of amy- Chem 86(5):492–498
lose content in rice (Oryza-sativa-L). J Genet 18. Juliano BO, Tuaño APP, Monteroso DN,
65(1–2):1–11. https://doi.org/10.1007/ Aoki N, Mestres C, Duldulao JBA, Bergonio
Bf02923530 KB (2012) Replacement of acetate with
7. Juliano BO (1971) A simplified assay for ammonium buffer to determine apparent
milled-rice amylose. Cereal Sci Today amylose content of milled rice. Cereal Foods
16(10):334–338, 340, 360 World 57(1):14–19
8. Perez CM, Juliano BO (1978) Modification 19. Avaro MRA, Tong L, Yoshida T (2009) A
of the simplified amylose test for milled rice. simple and low-cost method to classify amy-
Starch 30(12):424–426 lose content of rice using a standard color
9. Juliano BO, Perez CM, Blakeney AB, Castillo chart. Plant Prod Sci 12(1):97–99
T, Kongseree N, Laignelet B, Lapis ET, Murty 20. Khoomtong A, Noomhorm A (2015)
VVS, Paule CM, Webb BD (1981) International Development of a simple portable amylose
cooperative testing on the amylose content of content meter for rapid determination of
milled rice. Starch 33(5):157–162 amylose content in milled rice. Food
10. Juliano BO, Oñate LU, del Mundo AM Bioprocess Technol 8(9):1938–1946
(1972) Amylose and protein contents of 21. Ambardekar AA, Siebenmorgen TJ, Pereira T
milled rice as eating quality factors. Phil Agric (2011) Colorimetric method for rapidly pre-
56(1–2):44–47 dicting rice amylose content. Cereal Chem
11. Juliano BO, Villareal C (1993) Grain quality 88(6):560–563
evaluation of world rices. International Rice 22. Tuaño APP, Perez LM, Padolina TF, Juliano
Research Institute, Los Baños, Philippines BO (2015) Survey of grain quality of
12. AACC International (2009) Approved meth- Philippine farmers’ specialty rices. Phil Agric
ods of analysis. AACC Inter’l, St. Paul, MN Scientist 98:446–456
13. International Organization for 23. Juliano BO (2007) Rice chemistry and qual-
Standardization (2007) ISO 6647-2:2007– ity. Philippine Rice Research Institute,
Rice—Determination of amylose content— Munoz, Nueva Ecija
Part 2: Routine methods. International 24. Bhattacharya KR (2011) Analysis of rice qual-
Organization for Standardization, Geneva ity. In: Bhattacharya KR (ed) Rice quality.
14. Takeda Y, Hizukuri S, Juliano BO (1987) Woodhead Publishing Ltd, Cambridge, UK,
Structures of rice amylopectins with low and pp 431–530
high affinities for iodine. Carbohydr Res 25. Little RR, Hilder GB, Dawson EH (1958)
168:79–88 Differential effect of dilute alkali on 25 variet-
15. Butardo VM Jr, Fitzgerald MA, Bird AR, ies of milled white rice. Cereal Chem
Gidley MJ, Flanagan BM, Larroque O, 35:111–126
Resurreccion AP, Laidlaw HK, Jobling SA, 26. Bhattacharya KR, Sowbhagya CM (1972) An
Morell MK, Rahman S (2011) Impact of improved alkali reaction test for rice quality.
down-regulation of starch branching enzyme J Food Technol 7:323–331
IIb in rice by artificial microRNA- and hairpin 27. Biliaderis CG, Page CM, Maurice TJ, Juliano
RNA-mediated RNA silencing. J Exp Bot BO (1986) Thermal characterization of rice
62(14):4927–4941 starches: a polymeric approach to phase tran-
16. Butardo VM Jr, Anacleto R, Parween S, sition of granular starch. J Agric Food Chem
Samson I, de Guzman K, Alhambra CM, 34:6–24
Misra G, Sreenivasulu N (2017) Systems 28. Normand FL, Marshall WE (1988) Thermal-­
genetics identifies a novel regulatory domain properties of whole grain milled rice and
of amylose synthesis. Plant Physiol milled rice flour studied by differential scan-
173(1):887–906 ning calorimetry. Cereal Foods World
17. Fitzgerald MA, Bergman CJ, Resurreccion 33(8):663–663
AP, Moller J, Jimenez R, Reinke RF, Martin 29. Nakamura Y, Sakurai A, Inaba Y, Kimura K,
M, Blanco P, Molina F, Chen M-H, Kuri V, Iwasawa N, Nagamine T (2002) The fine
Romero MV, Habibi F, Umemoto T, Jongdee structure of amylopectin in endosperm from
S, Graterol E, Reddy KR, Bassinello PZ, Asian cultivated rice can be largely classified
Sivakami R, Rani NS, Das S, Wang Y-J, into two classes. Starch 54(3–4):117–131
48 Vito M. Butardo Jr. et al.

30. Umemoto T, Aoki N, Lin H, Nakamura Y, 42. Woolnough JW, Monro JA, Brennan CS, Bird
Inouchi N, Sato Y, Yano M, Hirabayashi H, AR (2008) Simulating human carbohydrate
Maruyama S (2004) Natural variation in rice digestion in vitro: a review of methods and
starch synthase IIa affects enzyme and starch the need for standardisation. Int J Food Sci
properties. Funct Plant Biol 31:671–684 Technol 43(12):2245–2256
31. Cuevas RP, Daygon VD, Corpuz HM, Reinke 43. Sievert D, Holm J (1993) Determination of
RF, Waters DLE, Fitzgerald MA (2010) amylose by differential scanning calorimetry.
Melting the secrets of gelatinisation tempera- Starch 45(4):136–139
ture in rice. Funct Plant Biol 37:439–447 44. Mestres C, Matencio F, Pons B, Yajid M,
32. Tuaño APP, Aoki N, Fujita N, Oitome NF, Fliedel G (1996) A rapid method for the
Merca FE, Juliano BO (2014) Grain and determination of amylose content by using
starch properties of waxy and low-apparent differential-scanning calorimetry. Starch
amylose Philippine rices and of NSIC Rc222. 48(1):2–6
Phil Agric Sci 97:329–339 45. Delwiche SR, Bean MM, Miller RE, Webb
33. Juliano BO, Perez CM, Alyoshin EP, BD, Williams PC (1995) Apparent amylose
Romanov VB, Bean MM, Nishita AB, content of milled rice by near-­infrared reflec-
Blakeney AB, Welsh LA, Delgado L, El Baya tance spectroscopy. Cereal Chem
AW, Fossati G, Kongseree N, Mendes FP, 72(2):182–187
Brilhante S, Suzuki H, Tada M, Webb BD 46. Villareal CP, De La Cruz NM, Juliano BO
(1985) Cooperative test on amylography of (1994) Rice amylose analysis by near-infrared
milled-rice flour for pasting viscosity and transmittance spectroscopy. Cereal Chem
starch gelatinization temperature. Starch 71(3):292–296
37(2):40–50 47. Delwiche SR, McKenzie KS, Webb BD
34. Dang JMC, Bason ML (2014) AACCI (1996) Quality characteristics in rice by near-­
Approved Methods Technical Committee infrared reflectance analysis of whole-grain
report: Collaborative study on a method for milled samples. Cereal Chem 73(2):257–263
determining the gelatinization temperature of 48. Castro JV, Ward RM, Gilbert RG, Fitzgerald
milled rice flour using the rapid visco analyser. MA (2005) Measurement of the molecular
Cereal Foods World 59(1):31–34 weight distribution of debranched starch.
35. Cagampang GB, Perez CM, Juliano BO Biomacromolecules 6(4):2260–2270
(1973) A gel consistency test for eating qual- 49. Ward RM, Gao Q, de Bruyn H, Gilbert RG,
ity in rice. J Sci Food Agric 24:1589–1594 Fitzgerald MA (2006) Improved methods for
36. Juliano BO (1979) The chemical basis of rice the structural analysis of the amylose-rich
grain quality. In: Workshop on chemical fraction from rice flour. Biomacromolecules
aspects of rice grain quality 1978. International 7(3):866–876
Rice Research Institute, Los Baños, Laguna, 50. Castro JV, Dumas C, Chiou H, Fitzgerald
pp 69–90 MA, Gilbert RG (2005) Mechanistic infor-
37. Khush GS, Paule CM, de la Cruz N (1979) mation from analysis of molecular weight dis-
Rice grain quality evaluation and improve- tributions of starch. Biomacromolecules
ment at IRRI. In: Workshop on chemical 6(4):2248–2259
aspects of rice grain quality, Los Baños, 1978. 51. Butardo V Jr, Daygon VD, Colgrave ML,
International Rice Research Institute, Los Campbell PM, Resurreccion AP, Cuevas RP,
Baños, Laguna, pp 22–31 Jobling SA, Tetlow IJ, Rahman S, Morell
38. Horibata T, Nakamoto M, Fuwa H, Inouchi MK, Fitzgerald MA (2012) Biomolecular
N (2004) Structural and physicochemical analyses of starch and starch granule proteins
characteristics of endosperm starches from in the high-amylose rice mutant Goami 2.
rice cultivars recently bred in Japan. J Appl J Agric Food Chem 60(46):11576–11585
Glycosci 51:303–313 52. Dhital S, Butardo VM Jr, Jobling SA, Gidley
39. Bhattacharya KR (2009) Physicochemical MJ (2015) Rice starch granule amylolysis—
basis of eating quality of rice. Cereal Foods differentiating effects of particle size, mor-
World 54:18–28 phology, thermal properties and crystalline
40. Juliano BO, Perez CM (1976) Gel consis- polymorph. Carbohydr Polym 115:305–316
tency tests for eating quality of milled rices. 53. Jane J-L, Chen J-F (1992) Effect of amylose
Cereal Foods World 21(8):422–422 molecular size and amylopectin branch chain
41. Felix AD, Tuaño APP, Mallillin AC, Trinidad length on paste properties of starch. Cereal
TP, Juliano BO (2016) Short-term satiety of Chem 69(1):60–65
brown rice and milled rice. Phil J Crop Sci 54. Hanashiro I, Ohta K, Takeda C, Mizukami
41:52–59 H, Takeda Y (2005) Structural characteriza-
 Improving Rice Grain Quality: State-of-the-Art and Future Prospects 49

tion of long unit-chains amylopectin. J Appl other physicochemical properties in rice

Glycosci 52:233–237 (Oryza sativa L.). Theor Appl Genet
55. O’Shea MG, Samuel MS, Konik CM, Morell 113(7):1171–1183
MK (1998) Fluorophore-assisted carbohy- 67. Bao J-S, Shen S, Sun M, Corke H (2006)
drate electrophoresis (FACE) of oligosaccha- Analysis of genotypic diversity in the starch
rides: efficiency of labelling and physicochemical properties of nonwaxy rice:
high-resolution separation. Carbohydr Res apparent amylose content, pasting viscosity
307(1–2):1–12 and gel texture. Starch 58:259–267
56. Morell MK, Samuel MS, O’Shea MG (1998) 68. Suwannaporn P, Pitiphunpong S,
Analysis of starch structure using fluorophore-­ Champangern S (2007) Classification of rice
assisted carbohydrate electrophoresis. amylose content by discriminant analysis of
Electrophoresis 19(15):2603–2611 physicochemical properties. Starch
57. Biliaderis CG, Grant DR, Vose JR (1981) 59(3–4):171–177
Structural characterization of legume starches. 69. Kurasawa H, Takei K, Ogawa S, Okabe T,
I. Studies on amylose, amylopectin, and Kanauti Y, Igaue I, Hayakawa T (1972) Some
β-limit dextrins. Cereal Chem physicochemical properties of starch from
58(6):496–502 non-glutinous paddy rice cultured in niigata
58. Hizukuri S (1986) Polymodal distribution of prefecture. 2. Correlation analysis between
the chain lengths of amylopectins, and its sig- eating quality, rheological property and amy-
nificance. Carbohydr Res 147:342–347 lose content of starch. Agric Biol Chem
59. Ball SG, Guan H-P, James MG, Myers AM, 36(10):1809
Keeling PL, Mouille G, Buleon A, Colonna P, 70. Reddy KR, Subramanian R, Ali SZ,
Preiss J (1996) From glycogen to amylopec- Bhattacharya KR (1994) Viscoelastic proper-
tin: a model for the biogenesis of the plant ties of rice-flour pastes and their relationship
starch granule. Cell 86(3):349–352 to amylose content and rice quality. Cereal
60. Juliano BO (1996) Rice quality screening Chem 71(6):548–552
with the Rapid Visco Analyser. In: Walker CE, 71. Lii C-Y, Tsai M-L, Tseng K-H (1996) Effect
Hazelton JL (eds) Applications of the Rapid of amylose content on the rheological prop-
Visco Analyser, Proc AACC RVA symposium, erty of rice starch. Cereal Chem
Nashville, 1994. Newport Scientific, 73(4):415–420
Warriewood, NSW, pp 19–24 72. Biliaderis CG, Zawistowski J (1990)
61. Blakeney A (1996) Rice. In: Henry RJ, Viscoelastic behavior of aging starch gels:
Kettlewell PS (eds) Cereal grain quality, 1st effects of concentration, temperature, and
edn. Chapman Hall, London, UK, pp 55–76 starch hydrolysates on network properties.
62. Batey IL, Curtin BM (2000) Effects on past- Cereal Chem 67:240–246
ing viscosity of starch and flour from different 73. Biliaderis CG, Juliano BO (1993) Thermal
operating conditions for the Rapid Visco and mechanical properties of concentrated
Analyser. Cereal Chem 77(6):754–760 rice starch gels of varying composition. Food
63. Champagne ET, Bett KL, Vinyard BT, Chem 48(3):243–250
McClung AM, Barton FE, Moldenhauer KA, 74. Biliaderis CG, Tonogai JR (1991) Influence
Linscombe S, McKenzie KS (1999) of lipids on the thermal and mechanical-­
Correlation between cooked rice texture and properties of concentrated starch gels. J Agric
Rapid Visco Analyser measurements. Cereal Food Chem 39(5):833–840
Chem 76(5):764–771 75. Blakeney A (1979) Instron measurement of
64. Allahgholipour M, Ali AJ, Alinia F, Nagamine cooked rice texture. In: Proceedings of the
T, Kojiima Y (2006) Relationship between workshop on chemical aspects of rice grain
rice grain amylose and pasting properties for quality 1978. International Rice Research
breeding better quality rice varieties. Plant Institute, Los Banos, Laguna, Philippines,
Breed 125:357–362 pp 343–353
65. Yan C, Li X, Zhang R, Sui JM, Liang GH, 76. Suzuki H, Ikehashi H, Kushibuchi K (1979)
Shen XP, Gu SL, Gu MH (2005) Performance Rice grain quality evaluation in Japan. In:
and inheritance of rice starch RVA profile Proceedings of the workshop on chemical
characteristics. Rice Sci 12:39–47 aspects of rice grain quality. International Rice
66. Bao J-S, Corke H, Sun M (2006) Nucleotide Research Institute, Los Banos, Laguna,
diversity in starch synthase IIa and validation Philippines, pp 123–127
of single nucleotide polymorphisms in rela- 77. Laignelet B, Feillet P (1979) Use of the visco-
tion to starch gelatinization temperature and elastograph for measuring the texture of
50 Vito M. Butardo Jr. et al.

cooked rice. In: Chemical aspects of rice grain 90. Yang DS, Shewfelt RL, Lee KS, Kays SJ
quality 1978. International Rice Research (2008) Comparison of odor-active com-
Institute, Los Banos, Laguna, pp 355–361 pounds from six distinctly different rice flavor
78. Juliano BO, Perez CM, Barber S, Blakeney types. J Agric Food Chem 56(8):2780–2787
AB, Iwasaki TA, Shibuya N, Keneaster KK, 91. Calingacion M, Fang L, Quiatchon-Baeza L,
Chung S, Laignelet B, Launay B, del Mundo Mumm R, Riedel A, Hall RD, Fitzgerald M
AM, Suzuki H, Shiki JI, Tsuji S, Yama JT, (2015) Delving deeper into technological
Tatsumi K, Webb BD (1981) International innovations to understand differences in rice
cooperative comparison of instrument meth- quality. Rice 8(1):43
ods for cooked rice texture. J Texture Stud 92. Champagne ET (2008) Rice aroma and fla-
12(1):17–38 vor: a literature review. Cereal Chem
79. Juliano BO, Perez CM, Alyoshin EP, 85(4):445–454
Romanov VB, Blakeney AB, Welsh LA, 93. Calingacion M, Laborte A, Nelson A,
Choudhury NH, Delgado LL, Iwasaki T, Resurreccion A, Concepcion JC, Daygon
Shibuya N, Mossman AP, Siwi B, Damardjati VD, Mumm R, Reinke R, Dipti S, Bassinello
DS, Suzuki H, Kimura S (1984) International PZ, Manful J, Sophany S, Lara KC, Bao J, Xie
cooperative test on texture of cooked rice. L, Loaiza K, El-hissewy A, Gayin J, Sharma
J Texture Stud 15:357–376 N, Rajeswari S, Manonmani S, Rani NS, Kota
80. Champagne ET, Bett-Garber KL, McClung S, Indrasari SD, Habibi F, Hosseini M,
AM, Bergman CJ (2004) Sensory characteris- Tavasoli F, Suzuki K, Umemoto T,
tics of diverse rice cultivars as influenced by Boualaphanh C, Lee HH, Hung YP, Ramli A,
genetic and environmental factors. Cereal Aung PP, Ahmad R, Wattoo JI, Bandonill E,
Chem 81(2):237–243 Romero M, Brites CM, Hafeel R, Lur H-S,
81. Perez CM, Pascual CG, Juliano BO (1979) Cheaupun K, Jongdee S, Blanco P, Bryant R,
Eating quality indicators for waxy rices. Food Thi Lang N, Hall RD, Fitzgerald M (2014)
Chem 4(3):179–184 Diversity of global rice markets and the sci-
82. Juliano BO (2010) Grain quality of Philippine ence required for consumer-targeted rice
rice. Philippine Rice Research Institute, breeding. PLoS One 9(1):e85106
Muñoz, Nueva Ecija 94. Unnevehr LJ, Juliano BO, Perez CM,
83. Buttery RG, Ling LC, Juliano BO, Turnbaugh Marciano E (1985) Consumer demand for
JG (1983) Cooked rice aroma and 2-acetyl- grain quality in Thailand, Indonesia and the
1-pyrroline. J AgriC Food Chem Philippines. IRRI Res Paper Ser 116:20
31(4):823–826 95. Unnevehr LJ, Duff B, Juliano BO (1992)
84. Buttery RG, Ling LC, Juliano BO (1982) Consumer demand for rice grain quality:
2-Acetyl-1-pyrroline: an important aroma introduction and major findings. In:
component of cooked rice. Chem Ind Unnevehr LJ, Duff B, Juliano BO (eds)
23:958–959 Consumer demand for rice grain quality.
International Rice Research Institute, Manila,
85. Buttery RG, Ling LC, Mon TR (1986) Philippines, pp 5–20
Quantitative analysis of 2-acetyl-1-pyrroline
in rice. J AgriC Food Chem 34(1):112–114 96. Limpawattana M, Yang DS, Kays SJ, Shewfelt
RL (2008) Relating sensory descriptors to
86. Chen S, Wu J, Yang Y, Shi W, Xu M (2006) volatile components in flavor of specialty rice
The fgr gene responsible for rice fragrance types. J Food Sci 73(9):456–461
was restricted within 69 kb. Plant Sci
171(4):505–514 97. Limpawattana M, Shewfelt RL (2010) Flavor
lexicon for sensory descriptive profiling of dif-
87. Fitzgerald MA, Hamilton NRS, Calingacion ferent rice types. J Food Sci 75(4):S199–S205
MN, Verhoeven HA, Butardo VM (2008) Is
there a second fragrance gene in rice? Plant 98. Lau WCP, Rafii MY, Ismail MR, Puteh A,
Biotechnol J 6(4):416–423 Latif MA, Ramli A (2015) Review of func-
tional markers for improving cooking, eating,
88. Zeng Z, Zhang H, Chen JY, Zhang T, and the nutritional qualities of rice. Front
Matsunaga R (2007) Direct extraction of vol- Plant Sci 6:832
atiles of rice during cooking using solid-phase
microextraction. Cereal Chem 99. Bao J (2014) Genes and QTLs for rice grain
J 84(5):423–427 quality improvement. In: Yan W, Bao J (eds)
Rice - germplasm, genetics and improvement.
89. Bryant RJ, McClung AM (2011) Volatile IntechOpen, Croatia
profiles of aromatic and non-aromatic rice
cultivars using SPME/GC-MS. Food Chem 100. Bradbury LMT, Henry RJ, Jin Q, Reinke RF,
124(2):501–513 Waters DLE (2005) A perfect marker for fra-
 Improving Rice Grain Quality: State-of-the-Art and Future Prospects 51

grance genotyping in rice. Mol Breed national rice germplasm collection. J Cereal
16(4):279–283 Sci 47:536–545
101. Shi W, Yang Y, Chen S, Xu M (2008) 112. Dobo M, Ayres N, Walker G, Park WD
Discovery of a new fragrance allele and the (2010) Polymorphism in the GBSS gene
development of functional markers for the affects amylose content in US and European
breeding of fragrant rice varieties. Mol Breed rice germplasm. J Cereal Sci 52:450–456
22(2):185–192 113. Waters DLE, Henry RJ, Reinke RF, Fitzgerald
102. Shao GN, Tang A, Tang SQ, Luo J, Jiao GA, MA (2006) Gelatinization temperature of
Wu JL, Hu PS (2011) A new deletion muta- rice explained by polymorphisms in starch syn-
tion of fragrant gene and the development of thase. Plant Biotechnol J 4(1):115–122
three molecular markers for fragrance in rice. 114. Mikami I, Uwatoko N, Ikeda Y, Yamaguchi J,
Plant Breed 130(2):172–176 Hirano H-Y, Suzuki Y, Sano Y (2008) Allelic
103. Bligh HFJ, Till RI, Jones CA (1995) A mic- diversification at the wx locus in landraces of
rosatellite sequence closely linked to the Waxy Asian rice. Theor Appl Genet
gene of Oryza sativa. Euphytica 116(7):979–989
86(2):83–85 115. Umemoto T, Aoki N (2005) Single-­
104. Ayres NM, McClung AM, Larkin PD, Bligh nucleotide polymorphisms in rice starch syn-
HFJ, Jones CA, Park WD (1997) thase IIa that alter starch gelatinisation and
Microsatellites and a single-nucleotide poly- starch association of the enzyme. Func Plant
morphism differentiate apparent amylose Biol 32(9):763–768
classes in an extended pedigree of US rice 116. Chen SH, Yang Y, Shi WW, Ji Q, He F, Zhang
germplasm. Theor Appl Genet ZD, Cheng ZK, Liu XN, Xu ML (2008)
94(6–7):773–781 Badh2, encoding betaine aldehyde dehydro-
105. Bergman CJ, Delgado JT, McClung AM, genase, inhibits the biosynthesis of 2-acetyl-
Fjellstrom RG (2001) An improved method 1-pyrroline, a major component in rice
for using a microsatellite in the rice waxy gene fragrance. Plant Cell 20(7):1850–1861
to determine amylose class. Cereal Chem 117. Hsu Y-C, Tseng M-C, Wu Y-P, Lin M-Y, Wei
78(3):257–260 F-J, Hwu K-K, Hsing Y-I, Lin Y-R (2014)
106. Roferos LT, Butardo V Jr, Fitzgerald MA, Genetic factors responsible for eating and
Juliano BO (2008) Association between cooking qualities of rice grains in a recombi-
alleles of the Waxy gene and traits of grain nant inbred population of an inter-subspecific
quality in Philippine Seed Board Rice cross. Mol Breed 34(2):655–673
Varieties. Phil Agric Scientist 91(3):334–337 118. Asante MD, Offei SK, Gracen V, Adu-Dapaah
107. Larkin PD, Park WD (1999) Transcript accu- H, Danquah EY, Bryant RJ, McClung AM
mulation and utilization of alternate and non-­ (2013) Starch physicochemical properties of
consensus splice sites in rice granule-bound rice accessions and their association with
starch synthase are temperature-sensitive and molecular markers. Starch 65:1–7
controlled by a single-nucleotide polymor- 119. Kharabian-Masouleh A, Waters DLE, Reinke
phism. Plant Mol Biol 40(4):719–727 RF, Ward RM, Henry RJ (2012) SNP in
108. Mikami I, Aikawa M, Hirano H-Y, Sano Y starch biosynthesis genes associated with
(1999) Altered tissue-specific expression at nutritional and functional properties of rice.
the Wx gene of the opaque mutants in rice. Sci Rep 2:557
Euphytica 105(2):91–97 120. Yu GQ, Olsen KM, Schaal BA (2011)
109. Cai XL, Wang ZY, Xing YY, Zhang J-L, Hong Molecular evolution of the endosperm starch
M-M (1998) Aberrant splicing of intron 1 synthesis pathway genes in rice (Oryza sativa
leads to the heterogeneous 5′ UTR and L.) and its wild ancestor, O-rufipogon L. Mol
decreased expression of waxy gene in rice cul- Biol Evol 28(1):659–671
tivars of intermediate amylose content. Plant 121. Tran NA, Daygon VD, Resurreccion AP,
J 14(4):459–465 Cuevas RP, Corpuz HM, Fitzgerald MA
110. Larkin PD, Park WD (2003) Association of (2011) A single nucleotide polymorphism in
waxy gene single nucleotide polymorphisms the Waxy gene explains a significant compo-
with starch characteristics in rice (Oryza sativa nent of gel consistency. Theor Appl Genet
L.). Mol Breed 12(4):335–339 123(4):519–525
111. Chen M-H, Bergman CJ, Pinson SRM, 122. Kovach MJ, Calingacion MN, Fitzgerald MA,
Fjellstrom RG (2008) Waxy gene haplotypes: McCouch SR (2009) The origin and evolution
associations with apparent amylose content of fragrance in rice (Oryza sativa L.). Proc Natl
and the effect by the environment in an inter- Acad Sci U S A 106(34):14444–14449
52 Vito M. Butardo Jr. et al.

123. Fitzgerald MA, Rahman S, Resurreccion AP, Newman HC, Jenkins AL, Goff DV (1981)
Concepcion JC, Daygon VD, Dipti SS, Kabir Glycemic index of foods: a physiological basis
KA, Klingner B, Morell MK, Bird AR (2011) for carbohydrate exchange. Am J Clin Nutr
Identification of a major genetic determinant 34(3):362–366
of glycaemic index in rice. Rice 4:66–74 136. Wolever TMS, Vorster HH, Bjorck I, Brand-­
124. Gao Z, Zeng D, Cheng F, Tian Z, Guo L, Su Miller JC, Brighenti F, Mann JI, Ramdath
Y, Yan M, Jiang H, Dong G, Huang Y, Han DD, Granfeldt Y, Holt S, Perry TL, Venter C,
B, Li J, Qian Q (2011) ALK, the key gene for Wu XM (2003) Determination of the glycae-
gelatinization temperature, is a modifier gene mic index of foods: interlaboratory study. Eur
for gel consistency in rice. J Integr Plant Biol J Clin Nutr 57(3):475–482
53(9):756–765 137. Salmeron J, Ascherio A, Rimm EB, Colditz
125. Bradbury LMT, Fitzgerald TL, Henry RJ, Jin GA, Spiegelman D, Jenkins DJ, Stampfer MJ,
Q, Waters DLE (2005) The gene for fra- Wing AL, Willett WC (1997) Dietary fiber,
grance in rice. Plant Biotechnol glycemic load, and risk of NIDDM in men.
J 3(3):363–370 Diabetes Care 20(4):545–550
126. Shao GN, Tang SQ, Chen ML, Wei XJ, He 138. Englyst HN, Kingman SM, Cummings JH
JW, Luo J, Jiao GA, Hu YC, Xie LH, Hu PS (1992) Classification and measurement of
(2013) Haplotype variation at Badh2, the nutritionally important starch fractions. Eur
gene determining fragrance in rice. Genomics J Clin Nutr 46:S33–S50
101(2):157–162 139. Englyst KN, Englyst HN, Hudson GJ, Cole
127. Bounphanousay C, Jaisil P, Sanitchin J, TJ, Cummings JH (1999) Rapidly available
Fitzgerald MA, Sackville-Hamilton NR glucose in foods: an in vitro measurement
(2008) Chemical and molecular characteriza- that reflects the glycemic response. Am J Clin
tion of fragrance in black glutinous rice from Nutr 69(3):448–454
Lao PDR. Asian J Plant Sci 7(1):1–7 140. Goñi I, Garcia-Alonso A, Saura-Calixto F
128. Butardo VM Jr, Sreenivasulu N, Jeon K (1997) A starch hydrolysis procedure to esti-
(2016) Tailoring grain storage reserves for a mate glycemic index. Nutr Res
healthier rice diet and its comparative status 17(3):427–437
with other cereals. Int Rev Cell Mol Biol 141. Frei M, Siddhuraju P, Becker K (2003)
323:31–70 Studies on the in vitro starch digestibility and
129. Carreiro AL, Dhillon J, Gordon S, Higgins the glycemic index of six different indigenous
KA, Jacobs AG, McArthur BM, Redan BW, rice cultivars from the Philippines. Food
Rivera RL, Schmidt LR, Mattes RD (2016) Chem 83(3):395–402
The macronutrients, appetite, and energy 142. Butterworth PJ, Warren FJ, Grassby T, Patel
intake. Annu Rev Nutr 36(1):73–103 H, Ellis PR (2012) Analysis of starch amyloly-
130. Shenkin A (2006) Micronutrients in health sis using plots for first-order kinetics.
and disease. Postgraduate Med Carbohydr Polym 87(3):2189–2197
J 82(971):559–567 143. Shih FF (2004) Rice proteins. In: Champagne
131. DellaPenna D (1999) Nutritional genomics: ET (ed) Rice: chemistry and technology, 3rd
manipulating plant micronutrients to improve edn. American Association of Cereal
human health. Science 285(5426):375–379 Chemistry, Inc, St. Paul, MN, pp 143–162
132. Johnson I, Williamson G (2003) 144. Kjeldahl J (1883) Neue methode zur bestim-
Phytochemical functional foods. CRC, mung des stickstoffs in organischen körpern.
Cambridge, England Fresenius' J Analyt Chem 22(1):366–382
133. Wolever TMS (2006) Glycaemic index : a 145. Cohen SA, Michaud DP (1993) Synthesis of
physiological classification of dietary carbohy- a fluorescent derivatizing reagent, 6-­aminoqu
drate. CAB International Book Archive, titles inolyl-­N-hydroxysuccinimidyl carbamate, and
from 2005, 2006, 2007. CABI Publishing, its application for the analysis of hydrolysate
Wallingford, UK amino acids via high-performance liquid
134. Trinidad TP, Mallillin AC, Encabo RR, chromatography. Anal Biochem
Sagum RS, Felix AD, Juliano BO (2013) The 211(2):279–287
effect of apparent amylose content and dietary 146. Cohen SA (2000) Amino acid analysis using
fiber on the glycemic response of different precolumn derivatization with 6-­aminoquinol
varieties of cooked milled rice and brown rice. yl-­
N-hydroxysuccinimidyl carbamate. In:
Int J Food Sci Nutr 64:89–93 Cooper C, Packer N, Williams K (eds) Amino
135. Jenkins DJ, Wolever TM, Taylor RH, Barker acid analysis protocols. Humana Press,
H, Fielden H, Baldwin JM, Bowling AC, Totowa, NJ, pp 39–47
 Improving Rice Grain Quality: State-of-the-Art and Future Prospects 53

147. Liu L, Waters DLE, Rose TJ, Bao J, King GJ solvent extracts from Japonica rice bran. Food
(2013) Phospholipids in rice: significance in Chem 117(3):538–544
grain quality and health benefits: a review. 160. Heuberger AL, Lewis MR, Chen MH, Brick
Food Chem 139(1–4):1133–1145 MA, Leach JE, Ryan EP (2010) Metabolomic
148. Godber JS, Juliano BO (2004) Rice lipids. In: and functional genomic analyses reveal vari-
Champagne ET (ed) Rice: chemistry and etal differences in bioactive compounds of
technology, 3rd edn. American Association of cooked rice. PLoS One 5(9):e12915
Cereal Chemistry, St. Paul, MN, 161. Moongngarm A, Daomukda N, Khumpika S
pp 163–190 (2012) Chemical compositions, phytochemi-
149. Maniñgat CC, Juliano BO (1980) Starch lip- cals, and antioxidant capacity of rice bran, rice
ids and their effect on rice starch properties. bran layer, and rice germ. APCBEE Proc
Starch 32(3):76–82 2:73–79
150. Lugay JC, Juliano BO (1964) Fatty acid com- 162. Niu YG, Gao BY, Slavin M, Zhang XW, Yang
position of rice lipids by gas-liquid chroma- F, Bao JS, Shi HM, Xie ZH, Yu LL (2013)
tography. J Am Oil Chem Soc Phytochemical compositions, and antioxidant
41(4):273–275 and anti-inflammatory properties of twenty-­
151. Liu L, Tong C, Bao JS, Waters DLE, Rose TJ, two red rice samples grown in Zhejiang. Lwt-­
King GJ (2014) Determination of starch lyso- Food Sci Technol 54(2):521–527
phospholipids in rice using liquid 163. Pereira-Caro G, Cros G, Yokota T, Crozier A
chromatography-­ mass spectrometry (LC-­ (2013) Phytochemical profiles of black, red,
MS). J Agric Food Chem brown, and white rice from the Camargue
62(28):6600–6607 region of France. J Agric Food Chem
152. Tong C, Liu L, Waters DLE, Rose TJ, Bao JS, 61(33):7976–7986
King GJ (2014) Genotypic variation in lyso- 164. Pereira-Caro G, Watanabe S, Crozier A,
phospholipids of milled rice. J Agric Food Fujimura T, Yokota T, Ashihara H (2013)
Chem 62(38):9353–9361 Phytochemical profile of a Japanese black-­
153. Kamal-Eldin A (2007) Micronutrients in purple rice. Food Chem 141(3):2821–2827
cereal products: their bioactivities and effects 165. Pintha K, Yodkeeree S, Pitchakarn P,
on health. In: Hamaker BR (ed) Technology Limtrakul P (2014) Anti-invasive activity
of functional cereal products. Elsevier, against Cancer cells of phytochemicals in red
Amsterdam. https://doi. Jasmine rice (Oryza sativa L.). Asian Pac
org/10.1533/9781845693886.1.86 J Cancer Prev 15(11):4601–4607
154. Branca F, Ferrari M (2002) Impact of micro- 166. Reddy CK, Kimi L, Haripriya S (2016)
nutrient deficiencies on growth: the stunting Variety difference in molecular structure,
syndrome. Ann Nutr Metab 46:8–17 functional properties, phytochemical content
155. Bashir K, Takahashi R, Nakanishi H, and antioxidant capacity of pigmented rice.
Nishizawa NK (2013) The road to micronu- J Food Meas Charact 10(3):605–613
trient biofortification of rice: progress and 167. Gaborieau M, Castignolles P (2011) Size-­
prospects. Front Plant Sci. https://doi. exclusion chromatography (SEC) of branched
org/10.3389/fpls.2013.00015 polymers and polysaccharides. Anal Bioanal
156. Welch RM, Graham RD (2004) Breeding for Chem 399(4):1413–1423
micronutrients in staple food crops from a 168. Roger P, Baud B, Colonna P (2001)
human nutrition perspective. J Exp Bot Characterization of starch polysaccharides by
55(396):353–364 flow field-flow fractionation–multi-angle laser
157. Hansen TH, Laursen KH, Persson DP, Pedas light scattering–differential refractometer
P, Husted S, Schjoerring JK (2009) Micro-­ index. J Chromatogr A 917(1):179–185
scaled high-throughput digestion of plant tis- 169. Rolland-Sabaté A, Guilois S, Jaillais B,
sue samples for multi-elemental analysis. Plant Colonna P (2011) Molecular size and mass
Methods 5:12 distributions of native starches using comple-
158. Wheal MS, Fowles TO, Palmer LT (2011) A mentary separation methods: ­ asymmetrical
cost-effective acid digestion method using flow field flow fractionation (A4F) and hydro-
closed polypropylene tubes for inductively dynamic and size exclusion chromatography
coupled plasma optical emission spectrometry (HDC-SEC). Anal Bioanal Chem
(ICP-OES) analysis of plant essential ele- 399(4):1493–1505
ments. Anal Methods 3(12):2854–2863 170. Chiaramonte E, Rhazi L, Aussenac T, White
159. Lai P, Li KY, Lu S, Chen HH (2009) DR (2012) Amylose and amylopectin in
Phytochemicals and antioxidant properties of starch by asymmetric flow field-flow fraction-
54 Vito M. Butardo Jr. et al.

ation with multi-angle light scattering and rheological testing and its relation to the fine
refractive index detection (AF4-MALS-RI). structure of rice starch. Carbohydr Polym
J Cereal Sci 56(2):457–463 146(1879–1344):253–263
171. Mantovani V, Galeotti F, Maccari F, Volpi N 183. Zhu B, Li B, Zheng XD, Xu BL, Liang S,
(2018) Recent advances in capillary electro- Kuang X, Ma MH (2010) Study on predictive
phoresis separation of monosaccharides, oli- models relating physicochemical properties to
gosaccharides, and polysaccharides. texture of cooked rice and the application in
Electrophoresis 39(1):179–189 rice blends. J Texture Stud 41(2):101–124
172. Warren FJ, Gidley MJ, Flanagan BM (2016) 184. Sesmat R, Meullenet JF (2001) Prediction of
Infrared spectroscopy as a tool to characterise rice sensory texture attributes from a single
starch ordered structure—a joint FTIR–ATR, compression test multivariate regression, and
NMR, XRD and DSC study. Carbohydr a stepwise model optimization method.
Polym 139:35–42 J Food Sci 66(1):124–131
173. Bao J-S, Cai YZ, Corke H (2001) Prediction 185. Champagne ET, Richard OA, Bett KL,
of rice starch quality parameters by near-­ Grimm CC, Vinyard BT, Webb BD, McClung
infrared reflectance spectroscopy. J Food Sci AM, Barton FE, Lyon BG, Moldenhauer KA,
66(7):936–939 Linscombe S, Mohindra R, Kohlwey DE
174. Barton FE, Himmelsbach DS, McClung AM, (1996) Quality evaluation of US medium-­
Champagne E (2002) Two-dimensional grain rice using a Japanese taste analyzer.
vibration spectroscopy of rice quality and Cereal Chem 73(2):290–294
cooking. Cereal Chem 79(1):143–147 186. Daygon VD, Prakash S, Calingacion M,
175. Himmelsbach DS, Barton FE, McClung AM, Riedel A, Ovenden B, Snell P, Mitchell J,
Champagne ET (2001) Protein and apparent Fitzgerald M (2016) Understanding the
amylose contents of milled rice by NIR-FT/ Jasmine phenotype of rice through metabo-
Raman spectroscopy. Cereal Chem 78(4): lite profiling and sensory evaluation.
488–492 Metabolomics 12(4):63. https://doi.
176. Wang HL, Wan XY, Bi JC, Wang JK, Jiang L, org/10.1007/s11306-016-0989-6
Chen LM, Zhai HQ, Wan JM (2006) 187. Lapchareonsuk R, Sirisomboon P (2015)
Quantitative analysis of fat content in rice by Sensory quality evaluation of rice using visible
near-infrared spectroscopy technique. Cereal and shortwave near-infrared spectroscopy. Int
Chem 83(4):402–406 J Food Prop 18(5):1128–1138
177. Meadows F, Barton FE (2002) Determination 188. Lu L, Zhu ZW (2014) Prediction model for
of Rapid Visco Analyser parameters in rice by eating property of indica rice. J Food Qual
near-infrared spectroscopy. Cereal Chem 37(4):274–280
79(4):563–566 189. Baldwin EA, Bai JH, Plotto A, Dea S (2011)
178. Bao JS, Wang YF, Shen Y (2007) Electronic noses and tongues: applications for
Determination of apparent amylose content, the food and pharmaceutical industries.
pasting properties and gel texture of rice Sensors 11(5):4744–4766
starch by near-infrared spectroscopy. J Sci 190. Chen JS (2009) Food oral processing - a
Food Agric 87(11):2040–2048 review. Food Hydrocoll 23(1):1–25
179. Champagne ET, Bett-Garber KL, Grimm 191. Pascua Y, Koc H, Foegeding EA (2013) Food
CC, McClung AM, Moldenhauer KA, structure: roles of mechanical properties and
Linscombe S, McKenzie KS, Barton FE oral processing in determining sensory tex-
(2001) Near-­infrared reflectance analysis for ture of soft materials. Curr Opin Colloid
prediction of cooked rice texture. Cereal Interface Sci 18(4):324–333
Chem 78(3):358–362 192. Pereira LJ, Van der Bilt A (2016) The influ-
180. Cho MJ, Morawicki R, Guerra C (2009) Ability ence of oral processing, food perception and
of near-infrared spectroscopy to predict solid social aspects on food consumption: a review.
loss, volumetric expansion, and water uptake of J Oral Rehabil 43(8):630–648
cooked rice. Cereal Chem 86(1):113–116 193. Chen JS, Stokes JR (2012) Rheology and tri-
181. Patindol J, Gu XF, Wang YJ (2010) bology: two distinctive regimes of food tex-
Chemometric analysis of cooked rice texture ture sensation. Trends Food Sci Technol
in relation to starch fine structure and leach- 25(1):4–12
ing characteristics. Starch 62(3–4):188–197 194. Prakash S, Tan DDY, Chen JS (2013)
182. Li H, Prakash S, Nicholson TM, Fitzgerald Applications of tribology in studying food
MA, Gilbert RG (2016) Instrumental mea- oral processing and texture perception. Food
surement of cooked rice texture by dynamic Res Int 54(2):1627–1635
 Improving Rice Grain Quality: State-of-the-Art and Future Prospects 55

195. Stokes JR, Boehm MW, Baier SK (2013) Oral 200. Toutounji MR, Van Leeuwen MP, Oliver JD,
processing, texture and mouthfeel: from rhe- Shrestha AK, Castignolles P, Gaborieau M
ology to tribology and beyond. Curr Opin (2015) Quantification of sugars in breakfast
Colloid Interface Sci 18(4):349–359 cereals using capillary electrophoresis.
196. Moongngarm A, Bronlund JE, Grigg N, Carbohydr Res 408:134–141
Sriwai N (2012) Chewing behavior and bolus 201. Concepcion JCT, Ouk M, Zhao D, Fitzgerald
properties as affected by different rice types. MA (2015) The need for new tools and
Int J Med Biol Sci 6:51–56 investment to improve the accuracy of select-
197. Felix AD, Trinidad TP, Tuaño APP, Juliano ing for grain quality in rice. Field Crops Res
BO (2013) Short-term satiety of cooked 182(0):60–67
Philippine rices of varying apparent amylose 202. Sreenivasulu N, Butardo VM, Misra G,
content and glycemic index. Phil Agric Sci Cuevas RP, Anacleto R, Kavi Kishor PB
96:179–186 (2015) Designing climate-resilient rice with
198. Hasjim J, Lavau GC, Gidley MJ, Gilbert RG ideal grain quality suited for high-­temperature
(2010) In vivo and in vitro starch digestion: stress. J Exp Bot 66(7):1737–1748
are current in vitro techniques adequate? 203. The 3000 Rice Genomes Project (2014) The
Biomacromolecules 11(12):3600–3608 3,000 rice genomes project. GigaScience 3:7
199. Dona AC, Pages G, Gilbert RG, Gaborieau 204. Zeng D, Tian Z, Rao Y, Dong G, Yang Y, Huang
M, Kuchel PW (2009) Kinetics of in vitro L, Leng Y, Xu J, Sun C, Zhang G, Hu J, Zhu L,
digestion of starches monitored by time-­ Gao Z, Hu X, Guo L, Xiong G, Wang Y, Li J,
resolved 1H nuclear magnetic resonance. Qian Q (2017) Rational design of high-yield and
Biomacromolecules 10(3):638–644 superior-quality rice. Nat Plants 3:17031
 Chapter 3

Morphology of Rice Seed Development and Its Influence

on Grain Quality
Paul A. Counce and Karen A. K. Moldenhauer

Abstract
Various quality attributes of rice seed are affected by the wide array of biochemical products accumulated
during the course of reproductive development and the environmental conditions which impact the grain
composition. The staging of rice plant reproductive development is needed in experiments to define phase
transitions of seed biology. The application of the nomenclature and criteria of the rice growth staging
system can facilitate recording the reproductive development by distinct stages. The meaningful progres-
sion from one stage to another in time can then be evaluated in a tabular or graphic manner. In order to
determine the developmental stages of rice in experiments, it is desirable to select a representative group
of plants and to record the development of those plants. We provide procedures for efficiently (1) observ-
ing and recording development of rice plants and (2) collecting, storing and seperating seed by develop-
mental stages. It is necessary to divide seeds into differing groups to track development from fertilization
until maturity. The earliest seeds to be fertilized on a panicle are superior grains and the latter seeds to
develop are inferior grains. In some cases, it is necessary to divide individual seeds into the aforementioned
groups and the different stages of development for various analyses. A procedure for dividing seeds into
differing stages of development is presented to more appropriately select seeds for further analysis. The
developmental record can then be statistically and graphically analyzed to better understand responses to
treatments and interactions among treatments, years, and locations.

Key words Rice, Oryza sativa L., Reproductive development, Seed, Grain, Seed development,
Growth staging system, Sampling, Timing, Inferior grains, Superior grains

1 Introduction

The rough rice kernel includes the husks or hulls and pedicel, as
well as the caryopsis (Fig. 1). “The weight distribution of the vari-
ous parts of the rice caryopsis [at maturity] is as follows: pericarp
(1–2%), seed coat and aleurone (5%), starchy endosperm (89–
91%), and embryo (2–3%)” [1]. The rice caryopsis (the rice seed or
brown rice kernel) accumulates mass rapidly during the R6 stage of
development, which lasts approximately from 5 to 15 days after
fertilization under optimum conditions for development (Fig. 2).
Starch is accumulated in the greatest concentration in the starchy

Nese Sreenivasulu (ed.), Rice Grain Quality: Methods and Protocols, Methods in Molecular Biology, vol. 1892,
https://doi.org/10.1007/978-1-4939-8914-0_3, © Springer Science+Business Media, LLC, part of Springer Nature 2019

57
58 Paul A. Counce and Karen A. K. Moldenhauer

Fig. 1 Parts of rough rice grain

Fig. 2 Rice mass accumulation by growth stage for Cocodrie and Guichao2 rice
Morphology of Rice Seed Development and Its Influence on Grain Quality 59

endosperm. Small amounts of starch are found in the subaleurone

layer and very small amounts are present in the embryo and aleu-
rone layer [2]. Starch synthesis is carried out in the plastids of these
tissues and is aggregated into highly structured starch granules [3].
From approximately 4 to 18 days after anthesis, starch accumulates
rapidly to a maximum in the endosperm tissue [4]. At maturity,
protein concentration declines gradually from the outer layer of
the caryopsis to the center of the grain in the starchy endosperm.
Functional proteins are found in the embryo, aleurone and subal-
eurone layers, starchy endosperm and maternal tissues during
development; storage proteins accumulate in these tissues as well
[5]. The quantity of storage proteins are greatest in the starchy
endosperm but concentration of protein is greater in the aleurone
layer, less in the subaleurone layer, and least in the starchy endo-
sperm [1]. Storage proteins begin to accumulate in the starchy
endosperm slightly later than starch. At maturity, lipid concentra-
tions are greatest in the embryo, followed by the aleurone layer,
the subaleurone layer, and the starchy endosperm [1]. Lipids, in
the form of lipid bodies, are reported to begin to accumulate
around 5 days after anthesis and increase in content in conjunction
with starch and storage protein accumulations but may continue to
accumulate over a somewhat longer period [6]. During develop-
ment, the biological activity of the pericarp and seed coat is impor-
tant for barley (and likely other cereals including rice) but the
synthetic activity of the seed covering maternal tissue begins to
decline prior to endosperm and embryo maturity [7].
The panicles composing the rice crop and the individual
grains within each panicle develop together beginning at differ-
ent times and proceeding at different rates. The rice staging sys-
tem was developed to improve production, research, extension,
and education communication regarding the plants and crop [8],
but it can also be adapted to differentiate the individual seeds at
different stages of development. Reproductive development con-
sists of ten growth stages based on discrete morphological criteria
(Figs. 3, 4, and 5, Table 1): panicle initiation (RO), panicle dif-
ferentiation (R1), flag leaf collar formation (R2), panicle exertion
(R3). Stages R4–R8 refer to the first grain on a panicle to reach
the following criteria: anthesis (R4), grain length and width
expansion (R5), grain depth expansion (R6), grain dry down
(R7), single-grain maturity (R8). When all grains that reach R6
have reached R8 then complete panicle maturity (R9) has been
reached. The grains change visibly during their development
both externally and internally Fig. 6). Consequently, the first
grain for a panicle to reach the criteria for R4 denotes the R4
stage. Consequently, when the panicle is at the R6 stage, the indi-
vidual grains on the panicle will frequently be at the R4, R5, and
R6 stages of development and at the R8 stage, individual grains
60 Paul A. Counce and Karen A. K. Moldenhauer

Reproductive growth stages with morphological markers.

Growth
Stage
RO R1 R2 R3
Panicle exertion
Morphological

Panicle from boot,

Marker
Panicle development Collar formation
branches have tip of panicle is
has initiated on flag leaf
formed above collar of
flag leaf
Illustration

Fig. 3 Rice growth stages R0–R3

Reproductive growth stages with morphological markers.

R4 R5 R6
One or more florets At least one caryopsis At least one caryopsis
on the main stem on the main stem on the main stem
panicle has reached panicle is elongating panicle has elongated
anthesis to the end of the hull to the end of the hull

Fig. 4 Rice growth stages R4–R6

Fig. 5 Rice growth stages R7–R9

Table 1
Rice reproductive stages of development, external visibility, description, and criteria

Rice
reproductive Externally
growth stage visible Brief description Criteria
R0 Not visible Panicle Initial formation of panicle
differentiation
R1 Not visible Panicle initiation Panicle Branches differentiate
R2 Visible Flag leaf collar Collar has formed on Flag Leaf
formation
R3 Visible Panicle exertion Panicle has emerged from flag leaf sheath
R4 Visible Anthesis Anthers and filaments exert from one floret on
panicle
R5 Visible Grain length and Elongating caryopsis visible but not extending
width expansion to the length of hull
R6 Visible Grain Caryopsis of one grain on panicle has elongated
Depth expansion to the length of the hull
R7 Visible Grain dry down One hull with a filled caryposis (length, width
and depth expanded) on the panicle is not
visibly green but is yellow over the entire hull
compared to floret color during R6
R8 Visible Single grain One hull has turned visibly brown and the luster
maturity of the hull is more flat or dull compared to
that of R7
R9 Visible Complete panicle Last grain to reach R6 on panicle has turned
maturity brown with same criteria for that kernel as for
R8
62 Paul A. Counce and Karen A. K. Moldenhauer

Fig. 6 Progressive development of the rice grain from anthesis through grain maturity

of the panicle will be at all stages from R3 through R8 although

those individual grains that have not reached R6 by this plant
growth stage are unlikely to successfully fill.
The rice seed develops from the panicle progressively over time
with the upper grain developing first and the lower grain develop-
ing later. The seed per se develops after pollination and fertilization
of the egg and the polar nuclei by the sperm. Subsequently the
fertilized egg cell develops into the embryo, while the fertilized
polar nuclei form the endosperm. The surrounding maternal tissue
including the seed covering, the lemma, and palea also develop con-
currently. The florets are exposed to light and air as the panicle
exerts (R3), the floret opens forcefully and the anther and filament
project out of the floret (R4) facilitating pollination and fertiliza-
tion. Cell division in both the embryo and the endosperm proceeds
during R4. Subsequently, the caryopsis expands visibly (R5); and
the cells in the embryo continue to develop as the endosperm cells
fill (primarily with starch) and increase in mass (R6); after the endo-
sperm cells have filled, the chlorophyll in the various maternal tis-
sues covering the seed is degraded. The lemma and palea lose
chlorophyll and these tissues turn from green to yellow (R7). As the
process continues, maternal tissue chlorophyll completes degrada-
tion and the lemma and palea, seed coat, endosperm, and embryo
dry out and the seed enters a quiescent state (R8). Usually, the crop
is harvested during the late part of the R8 stage for the crop but R9
is reached when all grains that reach R6 have reached R8.
The panicle on rice has branches off the central axis called pri-
mary branches while branches from the primary branches are
termed secondary branches. Branches arising from secondary
branches are the tertiary branches. The seeds on the panicle pro-
gressively develop from florets to fully developed seeds and a hier-
archy among the seeds of the panicle becomes progressively more
visible: the florets that develop first (usually the upper branches,
outer periphery of the branches and on the primary branches) are
followed progressively by the later florets (usually those on the
lower branches, the inner parts of the branches and at the ­secondary
 Morphology of Rice Seed Development and Its Influence on Grain Quality 63

and tertiary branches). The first florets to develop into seeds are
called superior grains and the later developing seeds are called infe-
rior grains. The superior grains on the panicle fill rapidly, are larger
and denser while the inferior grains fill slower, are smaller and less
dense [9]. Many scientists have divided the seeds into more detailed
delineations: Reports include varying levels of distinction among
various delineations of superior and inferior grains [10–12]. While
most scientists simply report the aggregate of harvested grains
without distinguishing position on the panicle, subdividing the
grains on panicle samples for supplementary analyses can supply
valuable information to aid in interpreting treatment effects.
Subdividing grains, in particular, can eliminate sources of variation
which can be critical to enzyme and starch analyses [10, 11].

1.1 Individual Grain For R4–R8, the staging system is primarily keyed to when the first
Development grains on the panicle begin to open and develop. Within the pani-
Within the Panicles cle, individual grains are at different stages of development. The
multiple stages of development of the individual grains within the
panicles during crop development (R6–R8) have distinct implica-
tions for rice grain quality [13–16], while earlier reproductive
stages have large effects on pollen viability [17, 18] and conse-
quently on rice crop grain yield [19].
During the process of grain filling storage products accumulate
in distinct phases with starch followed by storage proteins and lipid
contents. Grain filling, in terms of mass, is done in the R6 stage of
the individual grains. The R7 is a transitional stage from filling grain
to seed drying. The R8 stage is also critical to seed quality because
during this stage, the seed dries and can absorb water after drying.
When this occurs fissures can form in the rice grains. During the R8
stage fissures often form in the superior grains while the inferior
grains are still filling. Fissure formation leads to reduced head rice
yield (unbroken kernels or kernels ¾ or more of their original
length which is one of the most critical measures of grain quality).

2 Materials

2.1 Determining 1. R2 stage rice plants.

and Recording Crop, 2. White plastic tags (Kaywood Distributing Company, 4399
Panicle, and Individual Westgrove Drive, Addison, TX 75001; kaywood.com).
Seed Development
3. Waterproof marker.
4. Field book (“Rite in the Rain” All Weather Field Book 310,
614 Pacific Highway, Tacoma, WA 98424) (or alternative elec-
tronic device) for recording data.
5. Pen with stable ink after drying.
64 Paul A. Counce and Karen A. K. Moldenhauer

2.2 Collecting, 1. Rice reproductive stage plants.

Storing 2. Containers for seed of different stages of development.
and Separating
3. Table lamp or flash light.
Individual Seed
by Stages 4. Liquid nitrogen in Dewar or thermos.
of Development 5. Mortars for holding liquid nitrogen for seeds at different stages
of development.
6. Tweezers.
7. 10 or 50 mL Falconer tubes.
8. −80 ° C freezer.
9. Freezer boxes for organizing stored seeds.

3 Methods

Experimental treatments are imposed to determine the effects of

the treatments on various seeds on the entire panicle and on grain
at various stages of development. In this chapter we address meth-
ods for recording grain development, subdividing grain from the
panicle for purposes of enzyme assay and quantification, gene
expression, starch, protein, and lipid analysis.

3.1 Determining For some studies it is essential to know the timing of seed develop-
and Recording Crop, ment. The methods we present are the result of trial and error over
Panicle, and Individual several years of recording rice reproductive development. These
Seed Development methods can be modified depending on (1) the objectives of the
study, (2) the uniformity of plants in the study, and (3) the use to
be made of the staging information. The staging system utilized in
these methods is presented by Counce et al. [8].
Culms at the R2 stage are easily identified by the emergence of
the final leaf or flag leaf and the formation of the collar (Fig. 7).
The culm is marked with a plastic numbering tag and waterproof
ink (Fig. 8). The numbered tag is placed below the penultimate
leaf (second leaf from the top) to prevent dislodging or tangling
when the panicle exerts. Alternatively, the leaf itself can be num-
bered with an ultrafine point permanent marker but finding the
culm is often more difficult without the tag because the tag can be
seen more easily. For this reason, we use white tags so that the
culms can be located and notes more efficiently made. Therefore,
flags or visible markers in the field plots are useful for locating the
culms on the plants to be sampled. Subsequently, the culms are
monitored daily for the dates of R3, R4, R5, R6, R8, and R9 and
are noted for each panicle (Table 2). Determining R3 is relatively
easy and the exertion of the panicle is highly visible but the panicle
may exert through the flag leaf collar or below the flag leaf collar
(Fig. 9). Growth stage R4 is also very visible with the anthers now
present on the outside of the floret and the anthers being turgid
 Morphology of Rice Seed Development and Its Influence on Grain Quality 65

Fig. 7 R2 rice culms viewed from different angles when the leaf is definitely the final or flag leaf and after tag
attachment below the second leaf on left photo

Fig. 8 Tags for growth staging along with sheet of tags

Table 2
Excerpt from the field book for a rice field test conducted in 2015 at Stuttgart, Arkansas, U.S.A

Test C15S1 . C15S1 . C15S1 . C15S1 . C15S1 . C15S1

Plot 403 . 307 . 301 . 203 . 107 . 101
Cultivar Presidio . Cypress . LaGrue . RoyJ . Presidio . LaGrue
Rep 4 . 3 . 3 . 2 . 1 . 1
Panicle 4721 . 4735 . 4748 . 4762 . 4777 . 4790
Stage Day of the year in 2015
R3 205 . 211 . 208 . 215 . 208 . 209
R4 208 . 215 . 209 . 216 . 209 . 212
R5 210 . 216 . 212 . 219 . 212 . 216
R6 211 . 217 . 215 . 223 . 215 . 217
R7 222 . 229 . 223 . 229 . 219 . 226
R8 224 . 230 . 226 . 230 . 226 . 229
R9 238 . 258 . 253 . 257 . 239 . 254
The dates are days of the year numbering from Day 1 on January 1 and ending with Day 365 on non-leap years. The
table layout is in bold and the dates recorded for stages are in italics. The columns have the identification and plot
information for each panicle and the rows are for each respective growth stage. The columns with periods indicate
breaks in the data from the field book and the panicle numbers are continuous in the actual field book

Fig. 9 Freshly exerted panicles (rice growth stage R3). Panicle on left exerted through the collar of the flag leaf
sheath while the panicle on the right exerted through the flag leaf sheath below the collar
Morphology of Rice Seed Development and Its Influence on Grain Quality 67

and standing out just after anthesis (Fig. 10). Determining R5

requires close scrutiny of the upper florets on the panicle (ordinar-
ily) with the floret interposed between the sunlight and the
observer to allow the light to shine through the hull to view cary-
opsis elongation. Likewise R6 requires close scrutiny (normally of
the upper florets) to determine when the caryopsis has elongated
to the end of the hull (Fig. 11). When the grain begins to visibly
have both depth and width, R6 is also visible but this is usually
sometime after the start of R6. Photographs of the R7 stages are
misleading but easily visible by comparing the green color in the
R6 grains and the gradual yellowing leading to R7 (Fig. 12). R8 is
usually a phase change and different cultivars differ in the luster
and shade of brown color indicating R8 but in this case also the
comparative difference among grains (green vs. yellow) makes this
much less subjective than the verbal description suggests. Usually,
the luster of the R8 grains goes from bright to dull as well as brown
(Fig. 13). Comparing the grains within the panicle on a daily basis
is essential to accurately note the transitions from R6 through R8.
Relying on photographs or written descriptions must be aug-
mented with observation, especially during R8 determination
because the lemma and palea of rice lines differ with some being
almost white to light tan or tan at R8 while other cultivars may
turn dark brown at R8: progressive observations over days will
allow these differences to be noted. Determination of the R9 stage
requires close scrutiny of the final grains to mature on the panicle:
many unfilled grains at the bottom of the panicle will remain green
after the filled grains have matured (Fig. 14). The best way, in
many cases, is simply to press the last green florets between the
fingers to determine if the hull is actually flat. Extreme care is
needed in handling the panicles at this stage as the pedicles of the
earliest superior grains will often easily break leading to loss of
those kernels but this is unavoidable in our experience for some
panicles. If you require all of the grains to be collected for the
experiment a bag can be placed over the panicle to avoid loss. Loss
of the earliest grains during growth staging is usually only a prob-
lem if the panicles are to be collected for further analysis at R9.
The interval between making notes can, of course, be done at
varying frequencies but many stages, especially intervals R3–R4,
R4–R5, and R5–R6, are frequently less than 4 days. It is important
to clearly determine whether staging data is needed, the purpose
intended for the data, and the precision desired. While longer inter-
vals for making notes will lessen the precision of the information, the
precision needed for a particular study may be sufficient with less
frequent note taking, i.e., taking notes twice a week rather than daily.
The number of culms needed for sampling is dependent on the
plant uniformity and the potential use for the information. For
uniform field plots, we usually sample 7–10 culms per plot. For
pots in a greenhouse or in controlled climate conditions we sample
one culm per plant if one pot is the experimental unit.
68 Paul A. Counce and Karen A. K. Moldenhauer

Fig. 10 Anthers and filaments on panicle at the R4 stage of development shortly after dehiscence

In cases in which the vegetative development is monitored,

main stems are progressively monitored through reproductive
development as well. In pots with single plants per pot or with a
few plants physically separated within the pot, main steins can usu-
ally be identified after the fact even in cases in which the main stem
is not identified and monitored beginning at early vegetative devel-
opment. Identifying main stems in late vegetative and early repro-
ductive development in separated plants can normally be done by
observing the culms and surmising which is the main stem. This is
time consuming, requires some judgment and experience, and is
not certain. In dense field plots with either drill-seeded rows or
transplanted rice with more one plant per hill, determining main
stems can be nearly impossible without uprooting plants (which
would defeat the purpose intended for sequentially observing the
same undisturbed plants in the crop). Consequently in field plots,
selecting the earliest culms allows assessment of panicles most likely
to produce most of the yield. In field plots, it is normally best to
mark the earliest culms (in bordered areas within the plots) to
reach R2 for monitoring panicles. The earliest panicles generally
 Morphology of Rice Seed Development and Its Influence on Grain Quality 69

Fig. 11 Rice panicle (Growth Stage R6) with typically different stages of development at the different positions
within the panicle. Note that florets at the top of the panicle are at R6 and in some grains the caryopsis com-
pletely fills the enclosure of the hull while many of the lower grains are still at R3 (anthers visible within the
hull)

Fig. 12 Panicle just reaching growth stage R7 with several grains reaching the
critical stage of the same day. Note that grains on nearby panicles are also
reaching the criterion for R7 (loss of visible green color) and their easily visible
difference from the green R6 grains on the same panicles
70 Paul A. Counce and Karen A. K. Moldenhauer

Fig. 13 Rice panicle at Growth Stage R8

contribute a disproportionally greater part of crop yield than later

culms and plants (assuming uniform genetic material such as a crop
stand derived from pure seed). The tags should be placed below
the penultimate leaf on the culm. If the tags are placed only below
the flag leaf, the exerting panicle can become entangled in the tag.
At the completion of the study, an analysis of variance can be
applied to the staging data and, if appropriate given the analysis,
the data can be averaged within treatments to produce a tabular
(Table 3) or graphic timeline of development [20]. The intervals
can be by either calendar days, thermal time units, or both.
Combining the timeline for development (dates for R3–R9) with
environmental conditions during that development can be
extremely useful in understanding the many responses which differ
under different environments or treatments. Specifically treatment
by year and treatment by location interactions for responses can
often become clearer by thoughtful analysis of varying environ-
mental conditions at particular stages of development in different
years and locations. Moreover, even if only one stage of reproduc-
tive development is determined at one point in time for the various
treatments, projections of subsequent reproductive stages can be
predicted using calendar days or thermal time and relevant data
sets such as Counce et al. [20] used can be obtained.
 Morphology of Rice Seed Development and Its Influence on Grain Quality 71

Fig. 14 Panicle observed as R9 although a number of unfilled but green florets

are present on the panicle

Table 3
Thermal-unit accumulation (DD10 values) and duration in days for “Wells” rice from field
experiments conducted at Stuttgart, Arkansas, U.S.A. from 2007 to 2010 [20]

2007 2008 2009 2010

DD10 Days DD10 Days DD10 Days DD10 Days

R3 10.4 1.2 15.0 1.9 25.9 3.0 19.6 2.6
R4 35.5 3.8 24.4 3.5 20.6 2.4 24.7 3.3
R5 19.9 2.1 13.9 2.0 18.3 2.2 19.8 2.6
R6 57.7 6.1 56.5 8.1 58.9 7.3 46.8 6.0
R7 34.1 3.6 29.6 3.8 17.4 2.6 28.2 3.8
R8 149.6 17.2 204.4 30.7 178.4 26.3 172.8 23.5
72 Paul A. Counce and Karen A. K. Moldenhauer

3.2 Collecting, A familiarity with the progress of rice plant development in field
Storing and plots, greenhouse, or controlled climate environments is essential
Separating Individual to avoid wasted effort or even misleading data. Critical timing of
Seed by Stages of data collection is frequently determined too late to be of value.
Development Thus samples are frequently missed because the stage for sampling
is already passed. This is particularly true of sampling seeds which
are actively filling. The superior grains within the crop reach R6
within 7–12 days of the R3 for the plants. The inferior grains nor-
mally reach R6 over a much longer period. It is critical to be able
to identify which grains are selected and to sample those grains at
the rapid period of grain filling. Various enzymes involved in grain
filling increase rapidly in R6 and have declined by several folds in
its activity by R7. Moreover, the enzyme activity of critical enzymes
for grain filling is different for the same stage of development for
superior and inferior grains. Consequently, to meaningfully com-
pare treatment effects, seeds should be sampled in such a manner
as to eliminate these potential sources of variation.
For many enzyme activity determinations, fresh seed samples
can be removed from the plants, immediately placed on ice, rapidly
separated into the stage needed, quickly extracted and promptly
assayed. If the seeds are kept cold from removal from the plant
through assay, such a procedure often yields the maximum enzyme
activity. If seeds cannot be assayed within 2–4 h, the best proce-
dure is to quickly freeze them in liquid nitrogen and store them
frozen. For most enzymes, activity is rapidly reduced at tempera-
tures above freezing. The fresh panicles should be removed from
plants, placed in falconer tubers (50 mL volume often works well),
lids should be screwed on tightly, and the tubes dropped in liquid
nitrogen. On delivery to the lab, tubes can be placed in containers
and stored in a −80 °C freezer.
To separate grains into growth stages, the seeds are removed
from the panicles and inspected for individual grain growth stages
using the criteria for determining the plant growth stage (Table 1
and Fig. 6). The determination of R3 can be made for individual
grains by inspection against a suitable light source to see if the
anthers and filaments are still within the hull. R4 can be determined
if the anthers or filaments are visible outside of the hull or are absent
while the caryopsis is still not visible; R5 can be determined if the
caryopsis is elongating within the hull, R6 is determined if the cary-
opsis has elongated to the length of the hull; R7 is determined
when the hull is yellow, lacking any visible green areas such as veins
in the hull; R8 is determined when the hull has proceeded from yel-
low to the mature grain color. If the panicles are observed daily
during development: the distinctions between R7 and R8 are quite
obvious by the hue as well as the color of the hull. The divisions of
individual grains by growth stages can yield valuable information by
itself (Table 4) as well as providing the grains for further analysis.
 Morphology of Rice Seed Development and Its Influence on Grain Quality 73

Table 4
Individual grain growth stage percentages at the beginning of three
reproductive plant growth stages from field experiments conducted at
Stuttgart, Arkansas, U.S.A. in 2011

Individual grain growth stage

R3 and R4 R5 R6 R7 R8

Plant growth stage Percentage of grains per panicle

R6 74.1 11.7 14.1 0 0
R7 29.3 23.9 39.7 7.1 0
R8 15.5 24.9 33.1 15 11.6

If required, the process described above can be done with fro-

zen seeds and tweezers, working rapidly from removing the frozen
seed from panicles individually, determining growth stage and
placing seeds for each growth stage present into separate, labeled
mortars containing liquid nitrogen (see Note 1). At the end of col-
lection, seeds can be removed from the mortars and placed into
labeled falconer tubes, attaching tops to tubes quickly and freezing
those tubes (see Note 2). This is critical for both enzyme assay and
expression analysis of seed.

4 Notes

1. To avoid various forms of contamination, the seeds collected

for enzyme analysis and enzyme expression analysis should
either be collected in such a manner as to avoid skin contact
with seed or be handled with sterile gloves or tweezers. In
many field situations, with limited assistance, sterile gloves are
impractical and the best alternative is simply to remove the
panicle below the peduncle and drop the panicle into the fal-
coner tube. Usually even then a suitable instrument is required
to force fit the panicle branches into the tube. Ziploc bags can
be used for transport and storage. Fitting the panicles into
such bags is easier than for falconer tubes. On freezing, how-
ever, the plastic bags become brittle with further handling and
break. This can lead to spills, lost samples, or contamination.
2. Care should be taken to avoid introducing liquid nitrogen into
the falconer tubes for storage. Since any liquid nitrogen easily
transitions to gas, a risk of exploding the falconer tube exists.
74 Paul A. Counce and Karen A. K. Moldenhauer

References

1. Lu S, Luh BS (1991) Properties of the rice on grain weight and quality. J Sci Food Agric
caryopsis. In: Luh BS (ed) Rice. Springer, 88(5):897–903
New York, pp 389–419 13. Counce PA, Bryant RJ, Bergman CJ, Bautista
2. Champagne E, Wood DF, Juliano BO, Bechtel RC, Wang YJ, Siebenmorgen TJ, Moldenhauer
DB (2004) The rice grain and its gross compo- KAK, Meullenet JFC (2005) Rice milling qual-
sition. In: Champagne ET (ed) Rice chemistry ity, grain dimensions, and starch branching as
and technology. AACC International, St. Paul, affected by high night temperatures. Cereal
MN, pp 77–100 Chem 82(6):645–648
3. Smith AM, Denyer K, Martin C (1997) The 14. Cooper NTW, Siebenmorgen TJ, Counce PA,
synthesis of the starch granule. Annu Rev Plant Meullenet JF (2006) Explaining rice milling
Biol 48(1):67–87 quality variation using historical weather data
4. Singh R, Juliano BO (1977) Free sugars in analysis. Cereal Chem 83(4):447–450
relation to starch accumulation in developing 15. Ambardekar AA, Siebenmorgen TJ, Counce
rice grain. Plant Physiol 59(3):417–421 PA, Lanning SB, Mauromoustakos A (2011)
5. Yamagata H, Sugimoto T, Tanaka K, Kasai Z Impact of field-scale nighttime air temperatures
(1982) Biosynthesis of storage proteins in during kernel development on rice milling
developing rice seeds. Plant Physiol quality. Field Crops Res 122(3):179–185
70(4):1094–1100 16. Lanning SB, Siebenmorgen TJ, Counce PA,
6. Ichihara K, Kobayashi N, Saito K (2003) Lipid Ambardekar AA, Mauromoustakos A (2011)
synthesis and acyl-CoA synthetase in develop- Extreme nighttime air temperatures in 2010
ing rice seeds. Lipids 38(8):881–884 impact rice chalkiness and milling quality. Field
7. Sreenivasulu N, Radchuk V, Strickert M, Crops Res 124(1):132–136
Miersch O, Weschke W, Wobus U (2006) Gene 17. Coast O, Ellis RH, Murdoch AJ, Quiñones C,
expression patterns reveal tissue-specific signal- Jagadish KS (2015) High night temperature
ing networks controlling programmed cell induces contrasting responses for spikelet fertil-
death and ABA-regulated maturation in devel- ity, spikelet tissue temperature, flowering char-
oping barley seeds. Plant J 47(2):310–327 acteristics and grain quality in rice. Func Plant
8. Counce PA, Keisling TC, Mitchell AJ (2000) A Biol 42(2):149–161
uniform, objective, and adaptive system for 18. Krishnan P, Ramakrishnan B, Reddy KR,
expressing rice development. Crop Sci Reddy V (2011) Chapter three-High-tempera-
40(2):436–443 ture effects on rice growth, yield, and grain
9. Mohapatra P, Patel R, Sahu S (1993) Time of quality. In: Sparks DL (ed) Advances in agron-
flowering affects grain quality and spikelet par- omy, vol 111. Academic, Burlington,
titioning within the rice panicle. Func Plant pp 87–206
Biol 20(2):231–241 19. Peng S, Huang J, Sheehy JE, Laza RC,
10. Cheng F, Zhang Q, Zhu H, Zhao N, Wang F, Visperas RM, Zhong X, Centeno GS, Khush
Chen K, Zhang G (2007) The difference in GS, Cassman KG (2004) Rice yields decline
amylose content within a panicle as affected by with higher night temperature from global
the panicle morphology of rice cultivars. warming. Proc Natl Acad Sci U S A
J Cereal Sci 46(1):49–57 101(27):9971
11. Umemoto T, Nakamura Y, Ishikura N (1994) 20. Counce P, Siebenmorgen T, Ambardekar A
Effect of grain location on the panicle on activi- (2015) Rice reproductive development stage
ties involved in starch synthesis in rice endo- thermal time and calendar day intervals for six
sperm. Phytochemistry 36(4):843–847 US rice cultivars in the Grand Prairie, Arkansas,
over 4 years. Ann Appl Biol 167(2):262–276.
12. Wang F, Cheng F, Zhang G (2008) Impact of https://doi.org/10.1111/aab.12226
cultivar variation in grain density of rice panicle
 Chapter 4

Novel Imaging Techniques to Analyze Panicle Architecture

Erstelle Pasion, Roinand Aguila, Nese Sreenivasulu, and Roslen Anacleto

Abstract
Panicle architecture is known to directly influence grain yield in rice, and thus is an important trait for rice
varietal improvement. However, spike branching consequences trigger variation in number of superior and
inferior grains and thus affect grain quality. The genetics behind the length of both primary and secondary
branches were studied resulting in the identification of cloned genes. Extending this knowledge to include
other physiological parameters of panicle architecture is not yet well studied, and it requires high-­
throughput imaging techniques that are accurate. In this chapter we put the spotlight on Panicle Trait
Phenotyping Tool (P-TRAP), a freely available platform independent software to analyze the panicle archi-
tecture of rice, as one of such methods that can be used to generate a comprehensive and reproducible
panicle architecture data and identify superior breeding lines. P-TRAP measures 15 panicle structure and
nine spikelet traits. These quantitative traits can be used in genome-wide association studies to understand
their genetic basis.

Key words Panicle architecture, Grain yield, Rice breeding, Grain quality

1 Introduction

One of the major challenges in post-green revolution rice breeding

is to maximize the yield of stress-resilient semi-dwarf rice plants to
ensure global food security. If the rice-consuming world was to
grow by another billion people in 2030, then breeding has to pro-
duce high-yielding superior varieties with good grain quality fea-
tures to catch up with the estimated additional requirement of 65
million tons of milled rice by then [1]. New plant type was concep-
tualized in the mid-1980s as a strategy for genetic improvement to
increase the yield potential in rice [2]. Varietal improvement
addressed wide adaptation, optimal growth duration, resistance to
biotic and abiotic stresses, and necessary modifications to the com-
ponents of plant architecture, particularly panicle structure, to
increase the number and density of grains that the panicle can pro-
duce [3, 4].
Genetic improvement of panicle architecture is correlated with
yield and grain quality traits [5]. It directly affects the number of

Nese Sreenivasulu (ed.), Rice Grain Quality: Methods and Protocols, Methods in Molecular Biology, vol. 1892,
https://doi.org/10.1007/978-1-4939-8914-0_4, © Springer Science+Business Media, LLC, part of Springer Nature 2019

75
76 Erstelle Pasion et al.

grains per panicle and therefore the final rice yield [6]. The num-
ber of grains per panicle is further determined by the number of
spikelets per panicle that is mainly determined by the primary and
secondary branching as well as the seed setting rate of the spikelets
[7]. New plant type studies conducted in the 1990s however have
shown that large sink size results in poor grain filling and thus
negatively impact grain quality. From lax panicles of wild-type vari-
eties, new cultivars were developed with compact panicles bearing
more spikelets [8]. The increased number of spikelets per panicle
for most cultivars is mainly attributed to extra grains located at the
secondary rachis branches where spikelets are inefficiently filled
[9]. Greater variation in terms of spikelet development and grain
filling was also observed among rachis branches within a compact
panicle type compared to loose panicle type [9]. Moreover, higher
percentage of chalkiness was found in grains at the upper rachis
branches and primary rachis branches compared to those grains
located at the bottom rachis branches and secondary rachis
branches [10]. Initial reports also showed that inferior grains have
lower palatability, higher protein content, lower amylose content,
and less weight compared to superior grains [11, 12]. Because of
the large variation among grains within a panicle, the overall grain
yield and quality is differentially impacted [13, 14]. Negative cor-
relations are also usually observed between the length of panicle
and number of panicles per plant [5], as well as between the size of
panicles and overall size of spikelets per plant [15]. These tradeoffs
between assimilates and sinks compromise both the quantity and
quality of grains within a panicle [5], factors contributing to poor
grain filling have been identified [16] and addressed throughout
IRRI’s recent breeding history.
A reliable method to measure panicle architecture and a few
derived traits are presented in this chapter. Analyzing panicle archi-
tecture traits is performed using 2D image-based panicle pheno-
typing software. A prototype developed by Duan et al. [17]
automatically evaluates grain yield-related traits such as total num-
ber of spikelets, number of filled spikelets, grain length, grain
width, and 1000-grain weight. This method was only intended for
one rice plant at a time. Smart-PL system based on dual-camera
imaging was developed [18] to measure only the rice panicle length
including the length from the first rachis node to the panicle neck.
Additional 2D-image-based panicle phenotyping software includes
PASTAR, P-TRAP, and PANorama. Panicle TRAits Phenotyping
Tool (P-TRAP) is a free software written in Java with a graphical
user interface [6] to capture images of panicles taken using a cam-
era at a fixed distance, lighting, shutter speed, ISO, and minimum
camera resolution of 1024×768 pixels, where images can be batch
loaded into the software as a “project.” From then on, users can
preprocess the images within P-TRAP by cropping out non-­panicle
areas without altering the resolution of the images. P-TRAP can be
 Novel Imaging Techniques to Analyze Panicle Architecture 77

calibrated to the precise cm/pixel ratio (the supported scale of

P-TRAP) to accurately measure the panicle and grain parameters.
The novelty of P-TRAP that is not in other software of similar
function is that it can simultaneously measure panicle structure
parameters and quantify grain related traits such as the spikelet
number per panicle and several spikelet dimensional traits such as
length, width, circularity, ellipticity, perimeter, area, compactness,
and aspect ratio. Unlike P-TRAP that can measure numerous traits
of spikelets whether or not the spikelets are attached to rachis
branches, SmartGrain can only determine the shape and size of
spikelets threshed out from the panicles [19]. PASTAR (Panicle
Structure Analyzer for Rice) [20] is similar to P-TRAP in a way
that it can measure the number of spikelets and length and number
of rachis branches, but it is not available for public use and it does
not measure spikelet dimensional traits. PANorama [5] is faster
and more accurate in panicle traits analysis than P-TRAP version
20130213220. However, PANorama measures only attributes of
rachis and rachis branches but not spikelet-related traits, so in that
sense P-TRAP provides more value.
An integration of an image analysis method and a 5-point cali-
bration model was done [21] in order to rapidly estimate the num-
ber of spikelets per panicle. Although this approach eliminates the
need for high-resolution images, it provides measurement for only
one panicle trait. Another method focusing on only one trait is
number of panicles per plant during heading is based on multi-­
angle imaging developed [22]. A simple and noninvasive technique
was developed [23] based on X-ray computed tomography for rice
panicle post-anthesis development analysis. This method ultimately
aims to provide precise prediction of the optimal crop harvest time
and to facilitate selection in breeding for higher grain yield. One
more recent method for phenotyping rice panicles is the Panicle-­
SEG [24]. This software however focuses on panicle segmentation
and generates data about heading period and panicle development
based on maturity test and changes in area and color of panicles in
the field. No definite measurements about rachis branches and
spikelets can be measured by Panicle-SEG.
Although new methods are available, P-TRAP was still used in
the panicle architecture analysis performed in order to consider
both panicle structural variation and spikelet traits. Recent genome-­
wide association studies also used P-TRAP in generating rice pani-
cle phenotype data [25].

2 Materials Used in Panicle Architecture Measurement

1. Nikon D90 DSLR camera—chosen mainly because of its com-

patibility with DCamCapture—the software for tethered
78 Erstelle Pasion et al.

shooting. Any camera make and model may be used provided

that it works with the software chosen for tethered shooting.
2. USB Cable connector—Type A to Type mini-B cable (Nikon
D90 uses a mini-B USB connector).
3. MS Windows XP (or higher) for compatibility to both the
image acquisition and panicle architecture analysis software.
4. DCamCapture—free software for tethered shooting down-
loadable from http://www.bernd-peretzke.de/index.php/
dcamcapture-en). Other software for tethered shooting that is
compatible with the DSLR camera may also be used.
5. GIMP—open source software for faster image editing that can
be freely downloaded from http://www.gimp.org/
downloads.
6. Panicle trait phenotyping (P-TRAP) tool—downloadable
from http://bioinfo.mpl.ird.fr/index.php/resources?id=102.
7. Caliper (optional)—to be used in generating a reference image
for P-TRAP.
8. White 11” × 16” mounting sheet or board.
9. Putty-like adhesive (e.g., Blu-Tack)—to be used to attach the
panicles to the mounting sheet.
10. Rice panicles ~24 days after fertilization stored individually
in labeled containers (see Note 1).

3 Procedure

1. Trim the panicle 1 cm below the first primary branch node.

2. Securely mount the trimmed panicle onto the white mounting
sheet (e.g., A3 heavy bond paper) (see Note 2) using
enough inconspicuously placed glue tacks. Make sure that the
glue tacks are small enough to be undetectable during imag-
ing. Mount the panicles such that the branches (primary, sec-
ondary, etc.) are as straight as possible without any grains from
different branches (or parts of different branches) touching
each other, and that no panicle part extends beyond the bor-
der of the mounting sheet (Fig. 1a). Consecutive primary
rachis branches should also be mounted alternately on oppo-
site sides to attain maximum distance from neighboring rachis
branches. Higher order branches should be well distinguished
from the rachis branches from which they originated in order
to attain an accurate count of the number of spikelets per
rachis branch.
3. Mount the Nikon D90 camera on a tabletop tripod and tether
it to the computer using the specified USB cable. Choose a
 Novel Imaging Techniques to Analyze Panicle Architecture 79

Fig. 1 Image acquisition. (a) The panicle mounted on a white board with sufficient margins. (b) The image
acquisition setup where the camera is positioned at an optimal and consistent distance to the magnetic board.
Constant diffused office lighting was used when capturing the images. This image also shows the Microsoft
Windows workstation tethered to the camera via USB cable and the image acquisition software (DCamCapture)
already open. (c) How the viewfinder should appear when capturing images. Different camera models use
different viewfinder focus points

fixed ISO, shutter speed, and image resolution (see Note 3)

making sure that this will be the same for all images taken.
4. Open the DCamCapture software and make sure that the
computer detects the camera. Take a test shot, if necessary.
Decide on a naming convention of the image files in the
“Filename” setting (see Note 4).
5. Position a clipboard some distance opposite the camera enough
to frame the whole mounting sheet. This clipboard shall be
used to securely hold the mounted panicle when taking images
(Fig. 1b).
6. Clip a mounted panicle onto the clipboard and then manipu-
late the positions and orientations of the camera and the clip-
board so that the edges of the mounting sheet are framed as
80 Erstelle Pasion et al.

tightly as possible in the camera’s view finder (Fig. 1c).

Maintain this framing for all images.
7. By now, the imaging setup is ready—camera system and the
position and distance of the clipboard. Take images of all the
panicles to be analyzed. Ensure that shadowing is absent dur-
ing image capture. Complete the imaging first before doing
any analysis to minimize batch effect (see Note 5).
8. Using the same imaging setup, take a reference image of an
object of known size (e.g., the caliper mentioned in the list of
materials) (see Note 6).
9. Open the P-TRAP software.
10. Create a new project (File>New Project) for this analysis and
then set it as the main project (Run>Set Main Project).
11. Highlight the project’s name at the project navigation window
(by clicking on it) and then import the images of the panicles
to this project (by clicking the toolbar labeled “+”). Verify that
the names of the images appear when you expand the images
node in the project’s tree view (see Note 7).
12. Using the reference image (whose name should appear in the
list of images), calibrate the scaling factor of P-TRAP. The
goal is to set the correct “Real Units/Pixels” calibration ratio.
Refer to the P-TRAP manual for calibration details. Use the
reference image obtained in a previous step (see Note 8).
13. Open one of the images (by double-clicking its name in the
list of images) and then set the “grain color” and “background
color” using the context menu (right-click on a spikelet and
on the background, respectively). Please see the P-TRAP
manual on how to do color selection. In cases where color is
not uniform within the panicle, all the possible colors of spike-
lets may be selected as “grain color” in order to ensure all
spikelets are detected by P-TRAP (see Notes 9, 10).
14. Detect the structure of the panicle and edit manually to ensure
correctness. Refer to the P-TRAP manual for more details on
structure detection (see Note 11).
15. Detect the grains and edit the wrongly indicated number of
grains in the grain clusters. Also, mark the grains that were not
detected. Refer to the P-TRAP manual on how to edit the
number of spikelets initially indicated (see Note 11).
16. Collect the data for both the panicle structure and the grains.
P-TRAP generates two files for the panicle analysis results:
MainTraits.csv and GrainsTraits.csv, respectively. MainTraits.
csv has the following columns: file_name, PA_length, PA_
diameter, Node_nb, SA_nb, SA_average, SA_int, TA_nb, TA_
average, TA_int, QA_nb, Sp_nb. GrainsTraits.csv will have the
following columns: file_name, Sp_nb, Sp_length, Sp_width,
 Novel Imaging Techniques to Analyze Panicle Architecture 81

Table 1
List of (a) panicle structure traits and (b) spikelet traits

Variable name Trait measured

(a) Panicle structure traits
pa_length Length of the rachis
pa_diameter Diameter of the rachis
sa_po Position of first primary branch
sa_nb Number of primary branches
sa_length Length of primary branch
node_nb Number of rachis nodes
sa_int Length of intervals between primary branches
ta_nb Number of secondary branches
ta_length Length of secondary branches
ta_po Position of first secondary branch
ta_int Length of intervals between secondary branches
qa_nb Number of tertiary branches
qa_length Length of tertiary branches
qa_po Position of first tertiary branch
qa_int Length of intervals between tertiary branches
(b) Spikelet traits
sp_nb Number of spikelets
sp_length Length of the spikelets
sp_width Width of the spikelets
sp_area Area of the spikelets
sp_perimeter Perimeter of the spikelets
sp_circularity Circularity of the spikelets
sp_compactness Compactness of the spikelets
sp_ellipticity Ellipticity of the spikelets
sp_ar Aspect ratio of the spikelets
Please see Al-Tam et al. [6] for the complete details

Sp_area, Sp_primeter, Sp_circularity, Sp_compactness, Sp_

ellipticity, Sp_AR. Please see Table 1a, b for the list of traits
with short descriptions.
17. Merge both files using the “file_name” column, and then save
into a new file. This new file shall contain the final result.
82 Erstelle Pasion et al.

Fig. 2 Hierarchical clustering on 19 traits shows five clusters with approximate representation from each
group. The lone accession that is unique from the rest of the panicles analyzed differentiates because of its
low values in spikelet ellipticity, compactness, and circularity (see Fig. 4). The second smallest cluster (II)
shows a combination of spikelet density and unique branching. The rest of the clusters differentiate because
of unique combinations of branching and spikelet density, as well as other traits

18. Panicle structure and spikelet related traits measured by

P-TRAP are listed in Table 1a, b.
19. An optional but useful step in getting a sense of the data is to
do unsupervised statistics such as hierarchical clustering
(Fig. 2), principal components analysis (PCA) (Fig. 3a), all-­
pairs correlation (Fig. 3b), and heatmap generation (Fig. 4).
The optimal methods for distance matrix and clustering calcu-
lations in generating the dendrograms are “Euclidean dis-
tance” and “complete linkage” methods, respectively.
Complete linkage ensures good balance between maximum
variation between clusters and maximum similarity within a
cluster. For the PCA, generate the biplot to see the spatial
arrangements in both the sample and variable spaces. The bip-
lot is very useful in making initial inferences about correlations
among variables as well as detecting inherent population
 Novel Imaging Techniques to Analyze Panicle Architecture 83

Fig. 3 (a) Plot of the first two principal components shows spatial clustering of the variables that suggests their
correlations. Variability in the dataset is mostly explained by the spikelet traits where circularity and compact-
ness are the highest positive contributors along the first principal component (PC1), and spikelet length, perim-
eter and area are negative contributors for PC1. The number of spikelets, nodes, primary and secondary rachis
branches are highly correlated and negatively influence the second principal component. These variable cor-
relations are shown in the all-pairs correlation plot (b), where the color gradient in the scale bar shows the
strength and direction of correlation, while the size of the circles represents the correlation likelihood (p value).
Blank cells indicate no correlation between the traits

structure. All-pairs correlation among variables is a way of

quantifying what is being shown in the biplot’s variable space.
Hierarchical clustering, PCA, and heatmap can be generated
using R. It is important that the distance and clustering func-
tions used in generating both the dendrograms and the heat-
map (row and column dendrograms) should be the same:
Euclidean and complete linkage, respectively—as mentioned
earlier. For PCA, use the prcomp function instead of princomp
(both coming from the stats package) because the calculation
is done “by a singular value decomposition of the (centered
and possibly scaled) data matrix, not by using eigen on the
covariance matrix” [17], which is considered more numeri-
cally accurate (see Notes 12, 13).

4 Notes

1. Twenty-four days after fertilization is when the spikelets are of

the final size and shape and will not easily shatter when
harvested until after a month after mounting.
Fig. 4 Heatmap shows both trait and sample clustering. Trait clusters are within these sets: 1. {panicle diameter,
spikelet ellipticity, spikelet compactness, spikelet circularity}, 2. {number of spikelets, number of secondary branches,
number of tertiary branches}, 3. {spikelet width, spikelet length, spikelet perimeter, spikelet aspect ratio}, 4. {Length of
intervals of secondary branches}, 5. {rachis length, number of nodes, number of primary branches}, and 6. {length of
primary branch, length of secondary branch, length of intervals of primary branches, spikelet aspect ratio}
 Novel Imaging Techniques to Analyze Panicle Architecture 85

2. P-TRAP allows either white or dark colored mounting sheets

or board. Just ensure that you match the setting of P-TRAP to
whichever you used using the toggle switch displayed in the
toolbar labeled by an asterisk. Note, however, that a black
background reduces accuracy in measuring the number of
spikelets per rachis branch. If this trait is to be considered, a
white background should be used instead.
3. For efficiency, a DSLR camera was used and set to 2144 × 1424
pixels resolution. Also, better images may be captured with
studio lighting.
4. When taking images of the panicles, it is suggested that you
keep a spreadsheet that matches the filename of the panicle’s
image and its sample ID to prevent confusion later when
matching the panicle’s image to the data generated.
5. When analyzing multiple panicles, it is more efficient to com-
plete the imaging of all panicles before proceeding with the
analysis. It also minimizes batch effects due to variations in the
conditions when the images are taken such as camera setting,
consistency of the technician, possible variations in lighting,
variations in the distance of the mounted panicle to the cam-
era, and other possible confounding factors.
6. When framing the mounted panicle, align as tightly as possible
the margins of the mounting paper (e.g., A3 heavy bond
paper) with the margins of the view finder. When that is
achieved, maintain the position of the camera system and the
distance toward the panicle when taking subsequent images.
7. Note to keep an open thumbnail view of captured images to
ensure that framing of the images is consistent. This can be
done through an open Windows Explorer in thumbnail view.
8. This scaling factor affects the accuracy of the panicle structure
and grain dimension measurements.
9. All sheets used in mounting the panicles should be of the same
type (dimensions and color).
10. Ensure consistent framing, distance of the mounted panicle
to the camera, camera setting for image size and resolution,
aperture, film speed, and lighting.
11. In measuring both the panicle and the spikelet traits, of utmost
importance is the consistency by which the determinant spots
in the panicle are marked. Figure 5 shows an image of the
panicle and the resulting graph decomposition. More details
about how P-TRAP’s algorithm processes this decomposed
image are discussed in the paper.
12. Generating the PCA biplot (Fig. 3a) shows the positive correla-
tion and inverse correlation among the variables. Spikelet traits
show the highest contributors to data variability, particularly
spikelet compactness and circularity. These traits are inversely
86 Erstelle Pasion et al.

Fig. 5 Graphical decomposition of the panicle images requires precise placement of the start/end nodes of the
rachis, the start nodes of the primary, secondary (and tertiary, if present) branches, and the terminal nodes.
(a) Shows these nodes indicated in the image, and (b) shows the resulting graphical decomposition of the
panicle’s image

correlated to spikelet aspect ratio (calculated as the ratio of

spikelet length and width), and to spikelet length, perimeter,
and area. Data also shows causal relationship between the inter-
vals between primary branches and the average length of sec-
ondary branches. Another interesting cluster of variables are the
numbers of nodes, spikelets, branches (although the relative
contribution of tertiary branches is minimal) that explains the
highest variability along the PC2 axis. Intuitive inverse correla-
tion is shown for the panicle length and diameter. Traits related
to milled grains such as chalkiness, dimensions (length and
width), and shape (ratio of length and width) do not contribute
much to data variability although they show intuitive direct
correlations to their counterpart spikelet traits.
13. Panicle architecture analysis can be further improved by
generating derived traits based on the measurements from
 Novel Imaging Techniques to Analyze Panicle Architecture 87

P-TRAP analysis. Density-related traits may be calculated by

getting the ratio of a count trait over the length of another
trait. For instance, the density of primary rachis branches may
be computed by getting the ratio between the number of pri-
mary rachis branches over the length of the rachis. In the same
way, the average density of secondary rachis branches may be
derived by calculating the mean of all the ratios between the
number of secondary rachis branches over the length of a pri-
mary rachis branch. P-TRAP results can also be used to parti-
tion the panicle into top, middle, or bottom gradients and
determine the mean profile for each gradient. To achieve this,
the total rachis length must be divided into three partitions
and the sum of each internode length starting from the first
rachis node until the last one must be calculated. Considering
the quotient representing one-third of the length of rachis, the
positions of the primary rachis branches can be determined
based on the range within which internode they are attached
to. Also, panicle shape categories may be created by compar-
ing the means of primary rachis branches’ lengths for each
panicle gradient. For instance, if primary rachis branches at the
lower gradient have a mean length longer than the middle
gradient, and the mean length at the middle is also longer than
that at the top, then the panicle shape can be described as
pyramidal. The inverse of this mean length comparisons results
in an upright cone-like shape panicle. If the mean length of
primary rachis branches is longest at the middle gradient, the
panicle can be described as diamond-like. Otherwise, panicle
shape may be categorized as irregular.

Acknowledgment

This work has been supported under the CGIAR thematic area
Global Rice Agri-Food System CRP, RICE, Stress-Tolerant Rice
for Africa and South Asia (STRASA) Phase III, and Australian
Centre for International Agricultural Research (Project ID
CIM/2016/046) funding.

References

1. Zeigler RS (2014) Toward food security in 5. Crowell S, Falcão AX, Shah A, Wilson Z,
2050: when I think about poverty. CSA News Greenberg AJ, McCouch SR (2014) High-­
59:8 resolution inflorescence phenotyping using
2. Vergara BS (1987) Raising the yield potential a novel image-analysis pipeline, PANorama.
of rice. Trans Nat Acad Sci Tech 9:397–413 Plant Physio 166:479–495
3. Khush GS (2001) Green revolution: the way 6. FM AL-T, Adam H, Anjos A, Lorieux M,
forward. Nat Rev Genet 2:815–822 Larmande P, Ghesquière A, Jouannic S,
4. Khush GS (1995) Breaking the yield frontier of Shahbazkia HR (2013) P-TRAP: a panicle trait
rice. Geo J 35:329–332 phenotyping tool. BMC Plant Biol 13:122
88 Erstelle Pasion et al.

7. Xing Y, Zhang Q (2010) Genetic and molec- challenge of increasing rice yield potential.
ular bases of rice yield. An Rev Plant Biol Crop Sci 39:1552–1559
61:421–442 17. Duan L, Yang W, Huang C, Liu Q (2011)
8. Zhu Z, Tan L, Fu Y, Liu F, Cai H, Xie D, A novel machine-vision-based facility for the
Wu F, Jianzhong F, Wu J, Matsumoto T, Sun automatic evaluation of yield-related traits in
C (2013) Genetic control of inflorescence rice. Plant Methods 7:44
architecture during rice domestication. Nat 18. Huang C, Yang W, Duan L, Jiang N, Chen G,
Commun 4:2200 Xiong L, Qian L (2013) Rice panicle length
9. Mohapatra PK, Panigrahi R, Turner NC measuring system based on dual-camera imag-
(2011) Physiology of spikelet development on ing. Comp Elect Agri 98:158–165
the rice panicle: is manipulation of apical dom- 19. Tanabata T, Shibaya T, Hori K, Ebana K,
inance crucial for grain yield improvement? Yano M (2012) SmartGrain: high through-
Adv Agro 10:333–359 put phenotyping software for measuring seed
10. Zhang X, Alim N, Lin Z, Liu Z, Li G, Wang shape through image analysis. Plant Physiol
Q, Wang S, Ding Y (2013) Analysis of varia- 160:1871–1880
tions in white-belly and white-core rice kernels 20. Ikeda M, Hirose Y, Shibata Y, Yamamura T,
within a panicle and the effect of panicle type. Komura T, Doi K, Ahikari M, Matsuoka M,
J Integ Agr Adv Online Pub:1–11 Kitano H (2010) Analysis of rice panicle traits
11. Matsue Y, Odahara K, Hiramatsu M (1995) and detection of QTLs using an image analyz-
Differences in amylose content, amylo- ing method. Breed Sci 60:55–64
graphic characteristics and storage proteins 21. Zhao S, Gu J, Zhao Y, Hassan M, Li Y, Ding W
of grains on primary and secondary rachis- (2013) A method for estimating spikelet num-
branches in rice. Japan J Crop Sci 64(3): ber per panicle: integrating image analysis and
601–1606 a 5-point calibration model. Sci Rep 5:16241
12. Umemoto T, Nakamura Y, Ishikura N (1994) 22. Duan L, Huang C, Chen G, Xiong L, Liu Q,
Effect of grain location on the panicle on activ- Yang W (2015) Determination of rice panicle
ities involved in starch synthesis in rice endo- numbers during heading by multi-angle imag-
sperm. Phytochemistry 36:843–847 ing. The Crop J 3:211–219
13. Iwasaki Y, Mae T, Makino A, Ohira K, Ojima 23. Jhala T (2015) X-ray computed tomography
K (1992) Nitrogen accumulation in the infe- to study rice (Oryza sativa L.) panicle develop-
rior spikelet of rice ear during ripening. Soil Sci ment. J Expt’l Bot 66(21):6819–6825
Plant Nutri 38:517–525 24. Xiong X, Duan L, Liu L, Tu H, Yang P, Wu
14. Liu ZH, Cheng FM, Cheng WD, Zhang GP D, Chen G, Xiong L, Yang W, Liu Q (2017)
(2005) Positional variations in phytic acid and Panicle-SEG: a robust image segmentation
protein content within a panicle of japonica method for rice panicles in the field based on
rice. J Cereal Sci 41:297–303 deep learning and superpixel optimization.
15. Ohsumi A, Takai T, Ida M, Yamamoto T, Plant Methods 13:104
Arai-Sanoh YM, Ando T, Kondo M (2011) 25. Rebolledo MC, Peña AL, Duitama J, Cruz
Evaluation of yield performance in rice near-­ DF, Dingkuhn M, Grenier C, Tohme J (2016)
isogenic lines with increased spikelet number. Combining image analysis, genome wide asso-
Field Crops Res 120:68–75 ciation studies and different field trials to reveal
16. Peng S, Cassman KG, Virmani SS, Sheehy J, stable genetic regions related to panicle archi-
Khush GS (1999) Yield potential trends of tecture and the number of spikelets per panicle
tropical rice since the release of IR8 and the in rice. Front Plant Sci 7:1384
 Chapter 5

Measuring Head Rice Recovery in Rice

Jennine Rose Lapis, Rosa Paula O. Cuevas, Nese Sreenivasulu,
and Lilia Molina

Abstract
Head rice recovery (HRR) is a milling quality attribute that is highly influential toward the market price
of rice. It is defined as the proportion of paddy rice that retains 75% of its length after milling. For a new
rice variety to be accepted and adopted by farmers, the new variety’s HRR should satisfy consumer
requirements of at least 55% or above. Hence, HRR is a crucial attribute by which new varieties are
selected for release. Although the amount of head rice recovered depends on the genetic background of
a rice variety, HRR is also highly affected by postharvest processing conditions that the variety goes
through. To determine the maximum HRR, therefore, one must ensure that the processing conditions
are as optimal as possible. This book chapter outlines how paddy rice is processed into head rice and how
HRR is measured. It also proposes an improved laboratory-scale means for postharvest drying to mini-
mize head rice losses.

Key words Quality evaluation, Dehulling, Milling, Head rice recovery, Drying

1 Introduction

Rice varietal improvement programs aim to develop high-yielding

rice varieties. In order to encourage farmers to adopt these
improved varieties, the milling quality of these varieties has to be,
at least, at par with the varieties that these farmers cultivate regu-
larly. One of the indicators of milling quality is HRR [1, 2], the
amount of rice grains that retain at least three-quarters (75%) of
their original length after being polished, relative to the amount of
paddy rice [3]. Although this attribute can be explained partially
by genetics, farming conditions during the grain-filling stage and
postharvest conditions also affect HRR [4, 5]. These postharvest
factors include moisture content (MC), drying conditions of
freshly harvested paddy rice, presence of foreign matter in the sam-
ple lot, and grain dimensions. The critical MC values of rice

Nese Sreenivasulu (ed.), Rice Grain Quality: Methods and Protocols, Methods in Molecular Biology, vol. 1892,
https://doi.org/10.1007/978-1-4939-8914-0_5, © Springer Science+Business Media, LLC, part of Springer Nature 2019

89
90 Jennine Rose Lapis et al.

varieties (i.e., MC below the critical MC leads to grain fissuring)

vary, but for screening of varietal differences in grain cracking, 14%
MC was deemed to be the optimum level [6]. Foreign matter like
small stones or soil, and other plant material, on the other hand,
are removed prior to milling [7]; too much of these foreign materi-
als can potentially reduce throughput. Long and slender rice grains
tend to have greater breakage compared to short and bold rice
grains [8].
Although varieties that meet yield expectations, which fail to
satisfy the HRR criterion, risk low adoption by farmers because
HRR is directly linked to market price of rice set by millers, and
therefore projected farmer income may vary. Farmer adoption sur-
veys [9] indicate that modern improved varieties have not seen
widespread use, with HRR being an attribute of concern. For
instance, rice varieties developed by IRRI’s breeding programs
were reported to average 45% HRR [10]. It appears, therefore,
that there is still room for improving HRR. A two-pronged
approach is employed in efforts to increase HRR: (1) decrease the
grain’s vulnerability to breakage through genetic improvement
and (2) improve postharvest conditions to further reduce grain
breakage. Chalkiness is recognized as an additional factor in
reduced HRR [4] and the potential of eliminating chalkiness in
grains through genetic approaches is being explored [8]. On the
other hand, drying rough rice is considered the most critical post-
harvest process that affects milling quality and HRR. Several dry-
ing experiments were conducted by Siebenmorgen et al. to
understand the response of rice grains to various drying and tem-
pering environments using the glass transition principle and to find
ways on how grain fissuring can be minimized [11–14].
The milling process may also contribute to HRR. There are
different approaches to mill rice: (1) the traditional method that
uses mortar and pestle, (2) the single-stage mill in which the husk
and the bran are removed in one pass, and (3) the multistage mill-
ing that has different processing steps for dehulling, milling, and
separating head rice from the broken grains [15]. In IRRI’s Grain
Quality and Nutrition Service Laboratory (GQNSL), grains
undergo multistage milling.
This book chapter describes the method used in the GQNSL
for processing rice from paddy grain to head rice and how HRR is
measured. This book chapter also presents a case study exploring
various means of drying rice samples, at the laboratory scale, that
can possibly contribute to the improvement of HRR measurement
through reduced grain breakage and increased throughput. This
chapter does not discuss the genetic approach to improving HRR
nor does it suggest large-scale modifications that may decrease
grain breakage at the industrial level.
 Measuring Head Rice Recovery in Rice 91

2 Materials

1. Drying room with controlled temperature (50 ± 5 °C) and

Relative Humidity (63 ± 5%) maintained using an automatic
heater and humidifier.
2. Grain moisture tester.
3. Metallic drying racks.
4. Nylon drying net bags.
5. Re-sealable plastic bags used for the tempering stage.
6. Dehuller (Satake, Model THU-35A, Satake Corp., Hiroshima,
Japan, Fig. 1).
7. Cleaning brush.
8. Plastic bin for rice husk disposal.
9. Top-loading balance (Model PGW 2502i, Adam Equipment,
United Kingdom).
10. Barcode scanner (Model LS2208-SR20007R-UR, Symbol
Technologies, Mexico, Fig. 2).
11. Mill (Grainman, Model 60-220-60-DT, Grain Machinery
Mfg. Corp., Miami, Florida, USA, Fig. 3).
12. Collecting tray for milled rice.
13. Plastic bin for rice bran disposal.
14. Spatula for Grainman mill head scraping/cleanup.

Fig. 1 Satake blower/dehuller

92 Jennine Rose Lapis et al.

Fig. 2 Barcode scanner

Fig. 3 Grainman mill

15. Head rice separator (Satake rice machine Type TRG with three
different trays, Satake Engineering Co., Ltd., Japan, or Rice
sizing device, Figs. 4 and 5).
16. Collecting trays for broken and head rice grains.
 Measuring Head Rice Recovery in Rice 93

Fig. 4 Rice sizing device

Fig. 5 Trays (short grain, medium grain, long grain)

3 Methods

1. Drying.
(a) Determine the moisture content (MC) of the freshly har-
vested paddy rice sample using a grain moisture tester (see
Note 1).
(b) Weigh 200 g paddy rice samples and pack in a net drying
bag. Arrange samples across drying racks and allow drying
inside the controlled drying room (24 h) or until 12–14%
MC is obtained.
2. Dehulling.
(a) Get the moisture content of the paddy rice sample using
the grain moisture tester. Clean the paddy rice grain by
­blowing out the unfilled grains and other impurities in the
samples using the Satake dehuller (see Note 2).
(b) Weigh 125.00 ± 1.00 g cleaned paddy rice grains.
94 Jennine Rose Lapis et al.

(c) Dehull the paddy rice grain sample using the Satake dehuller
(see Note 3).
(d) Collect and weigh the unpolished rice grain using the top-­
loading balance.
3. Milling.
(a) Remove the bran of the unpolished rice grain using the
Grainman mill for 45 s (see Note 4).
(b) Collect and weigh the polished rice grain using the top-­
loading balance.
4. Determining the head rice recovery.
(a) 
Remove broken rice grain (<75% of the original grain
length after milling) by passing the grains through the
vibrating rice sizing device (see Note 5).
(b) Collect and weigh the head rice grain using the top-­loading
balance.
5. Data processing.
The proportions of brown, milled, and head rice from paddy
rice are calculated based on the equations below (see Note 6).

weight of brown rice

% Brown rice = ×100
weight of paddy rice ( g )

weight of milled rice

% Milled rice = ×100
weight of paddy rice ( g )

weight of head rice ( g )

% Head rice = ×100
weight of paddy rice ( g )

4 Applying a Drying Protocol to Improve HRR: A Case Study

In support of IRRI’s pursuit on breeding rice with better HRR, a

drying protocol that incorporates a tempering step [14] was tested.
This additional step to drying supposedly minimizes grain fissuring
by preventing abrupt changes in the moisture gradient within the
rice kernel. It involved the application of (1) heat while monitoring
RH to dry freshly harvested paddy rice and of (2) tempering con-
ditions as a post-drying step.
In this study, four varieties grown at IRRI in the 2016 wet
season (Table 1) were obtained right after harvest and were sub-
jected to two laboratory-scale postharvest drying conditions. These
varieties were chosen to represent different classes of grain size and
shape. NSIC Rc392 is a saline-tolerant variety that was included in
the participatory varietal selection (PVS) in 2015 and was favored
 Measuring Head Rice Recovery in Rice 95

Table 1
Description of grain type of four IRRI varieties obtained from IRRI’s plant
breeding division

Variety Description
NSIC Rc 392 (IRRI 185) Long, slender
NSIC Rc 348 (IRRI 175) Long, bold
NSIC Rc 400(IRRI 186) Medium, intermediate
NSIC Rc 334 (IRRI 171) Extra-long, slender

in central Philippines. All other varieties will be introduced in the

PVS in the near future.
For samples that were dried in controlled conditions, the
freshly harvested samples were threshed and stored in re-sealable
plastic bags at 4 °C for 1 week. Grains (200 g) were then trans-
ferred into nylon drying bags and were allowed to equilibrate at
room temperature overnight. Initial MC was measured using a
moisture meter. Then, the samples were placed inside the con-
trolled drying room, in which the temperature range applied was
40–52 °C while RH was monitored (range was 47–63%). When
the grains reached the MC range of 12–14%, the drying bags were
placed in re-sealable plastic bags for tempering (1 h, 50 ± 5 °C).
Samples were removed from the re-sealable bags and then allowed
to cool at room temperature until 12–14% moisture content was
reached. Samples were refrigerated (4 °C, in resealable plastic bags)
until ready for HRR determination.
For samples dried in unmonitored conditions, freshly har-
vested samples were placed in nylon drying bags and placed in the
glasshouse for sun-drying. When the target moisture content range
(12–14%) was reached, the samples were hand-threshed, and were
stored at 4 °C in resealable plastic bags until ready for HRR
determination.
Head rice recovery was determined as described in Subheading
3, in triplicate.
Results (Table 2) suggest that samples subjected to controlled
drying conditions had higher HRR compared to those dried in
uncontrolled conditions in the glasshouse. Among the varieties
tested, IRRI 171 (extra-long and slender grains) appeared to be
most susceptible to fissuring among the samples tested under
uncontrolled sun-drying, but showed improved HRR when dried
under controlled drying condition (Table 2). However, it is noted
that only IRRI 185 reached an HRR of 60% under controlled dry-
ing conditions. The HRRs of the rest of the samples were around
50% when drying was controlled.
96 Jennine Rose Lapis et al.

Table 2
Comparison of HRR (%) of four IRRI rice varieties subjected to two drying
conditions

HRR (%)

Sample Drying room (controlled) Glasshouse (unmonitored)

IRRI 185 60.5 ± 0.9 55.0 ± 0.5
IRRI 175 52.1 ± 5.1 43.8 ± 0.4
IRRI 186 51.1 ± 0.6 45.7 ± 0.1
IRRI 171 50.6 ± 2.3 31.6 ± 1.5

The developed controlled drying room protocol, therefore, is

recommended as a routine drying method for IRRI varieties
requiring HRR analysis

5 Notes

1. Freshly harvested paddy rice will be used for controlled post-

harvest drying of paddy samples for HRR analysis. Fresh MC is
determined using grain moisture tester by placing an ample
amount of the paddy grain to be measured in the sampling tray.
The paddy grain must have been presorted to remove unripe or
degenerated grains. Insert the sampling tray in the testing
chamber at the side of the tester. Rotate the crushing handle
counterclockwise until it reaches its limit. Wait until % MC
appears on the screen. Moisture content measurements using a
moisture meter are based on the conductivity of the paddy rice;
which, in turn, is relative to the paddy grain’s moisture content
[16]. This method is less accurate than the primary (oven-­
drying) method but produces results within seconds or minutes
and is usually used in postharvest management and for trade.
2. For removing the semi-filled and the unfilled grains, the roller
distance of the Satake dehuller is adjusted to keep rollers apart.
This allows the passage of paddy samples for blowing of unfilled
or semi-filled grains without dehulling them. To do this, the
machine is switched on, paddy samples are poured on the feed
hopper, and the sample passage is controlled to prevent clog-
ging of paddy grains. The filled grains fall into a large container
for collection and the semi-filled or unfilled grains will fall into
a small container (for discarding). Turn the machine on and off
to ensure that no grains are left inside the dehuller. Open the
 Measuring Head Rice Recovery in Rice 97

trap door in the hopper, use a brush or vacuum to clean the

inside surface of the dehuller. Clean the machine after each
sample.
3. The same Satake dehuller machine is used for dehulling. To do
this, the same procedure for cleaning (see Note 2) is followed,
except that the roller distance is adjusted to keep rollers at a
distance (i.e., slightly more than half the length of the grain)
that allows passage of paddy samples to achieve the optimum
huller efficiency at 75–80%, at minimum breakage [3]. Allow
two sample passes to ensure maximum paddy grain dehulling.
Collect the brown rice samples from the large container. Return
samples into their appropriate container (properly labeled enve-
lopes). Discard the undehulled samples.
4. The milling process using Grainman mill removes the outer
layer (bran) from brown rice, resulting in a white or polished
rice fraction, also known as milled rice. To operate the machine,
switch the Grainman mill on and load the brown rice samples
into the mill. Lock the milling chamber by pushing the locking
handle upward, push the cover downward and hook the weight
lever over the cover. Set the timer to 45 s and press the white
button to start milling. Once the milling process is complete,
remove the weight lever. Clean the milling chamber using a
brush to remove the bran on the outside surface of the milling
chamber. Place a tray/collection pan under the milling cham-
ber. Gently lift the cover off from the milling chamber. Unlock
the milling chamber by pulling the lock handle downward.
Turn the milling chamber upside-down in order for the samples
to fall into the collection pan. Use spatulas and/or brushes to
remove the milled rice off the screen and all sides of the roller,
ensuring that all rice grains from the milling chamber are
removed. Transfer the milled rice collected on the tray/collec-
tion pan into its properly labeled envelope. Clean the mill thor-
oughly after each use. Use a brush to scrub all accessible parts
of the mill (i.e., cover, roller, and chamber). Make sure to
remove all the debris on and around the milling chamber, as
well as in the collection pans. Once all samples are milled, hold
the toggle switch to turn off the mill.
5. The rice sizing device sorts the milled rice to separate the bro-
ken grains from head rice. Pass the milled rice sample onto the
vibrating or oscillating sieve of the head rice separator. Collect
the grains with lengths at least 75% of the unbroken grain (i.e.,
those grains that do not fit the wells of the sizing device) into a
pan. Manually remove broken milled grain from the collected
sample.
6. Other terms for “brown rice” and “milled rice” are “unpolished
rice” and “polished rice,” respectively.
98 Jennine Rose Lapis et al.

Acknowledgments

The authors thank Marnol Santos, Edgar Amoloza, Anna Carriza

Basilio, Dennis Villegas, Teodoro Atienza, Leah Villanueva, for the
controlled drying room setup and optimization, and assistance in
processing and collecting data for the samples used in the case
study at GQNSL. Fresh paddy samples for the drying experiment
were donated by Dr. Georgina Vergara and Rafael Julian Panerio
while Jun Correa allowed us to avail the rice threshing service at
Zeigler Experiment Station in IRRI. This work has been supported
under the CGIAR thematic area Global Rice Agri-Food System
CRP, RICE, Stress-Tolerant Rice for Africa and South Asia
(STRASA) Phase III, and Australian Centre for International
Agricultural Research (Project ID CIM/2016/046) funding.

References

1. Zhou L, Liang S, Ponce K, Marundon S, Ye G, preference for rice traits: insights from farm
Zhao X (2015) Factors affecting head rice yield surveys in Central Luzon, Philippines, 1966–
and chalkiness in indica rice. Field Crops Res 2012. PLoS ONE 10(8):e0136562
172:1–10 10. Anacleto R, Cuevas RP, Jimenez R, Llorente
2. Patindol J, Wang YJ (2002) Fine structures of C, Nissila E, Henry RJ, Sreenivasulu N (2015)
starches from long-grain rice cultivars with differ- Prospects of breeding high-quality rice using
ent functionality. Cereal Chem 79(3):465–469 post-genomic tools. Theor Appl Genet
3. Juliano BO (2007) Rice chemistry and quality. 128:1449–1466
Philippine Rice Research Institute, Munoz, 11. Cnossen AG, Siebenmorgen TJ, Yang W,
Nueva Ecija, p 402 Bautista RC (2001) An application of glass
4. Zhao X, Fitzgerald MA (2013) Climate transition temperature to explain rice kernel fis-
change: implications for the yield of edible rice. sure occurrence during the drying process.
PLoS One 8(6):e66218 Drying Technol 19(8):1661–1682
5. Pearce MD, Marks BP, Meullenet J-FC (2001) 12. Cnossen AG, Siebenmorgen TJ (2000) The
Effects of postharvest parameters on functional glass transition temperature concept in rice
changes during rough rice storage. Cereal drying and tempering: effect on milling quality.
Chem 78(3):354–357 Trans ASAE 43(6):1661–1667
6. Juliano BO, Perez CM (1993) Critical mois- 13. Cnossen AG, Siebenmorgen TJ, Yang W
ture content for fissures in rough rice. Cereal (2002) The glass transition temperature con-
Chem 70(5):613–615 cept in rice drying and tempering: effect on
7. Bhattacharya KR, Ali SZ (2015) An introduc- drying rate. Trans ASAE 45(3):759–766
tion to rice-grain technology, vol 288. CRC 14. Ondier GO, Siebenmorgen TJ,
Press, Boca Raton, FL Mauromoustakos A (2012) Drying characteris-
8. Sreenivasulu N, Butardo VM Jr, Misra G, tics and milling quality of rough rice dried in a
Cuevas RP, Anacleto R, Kavi Kishor PB (2015) single pass incorporating glass transition prin-
Designing climate-resilient rice with ideal grain ciples. Drying Technol 30:1821–1830
quality suited for high-temperature stress. 15. Dhankar P (2014) Rice milling. IOSR J Eng
J Exp Bot 66(7):1737–1748 4(5):34–42
9. Laborte AG, Paguirigan NC, Moya PF, Nelson 16. Janier JB, Maidin MB (2011) Paddy moisture
A, Sparks AH, Gregorio GB (2015) Farmers’ content detector. J Appl Sci 11:1476–1478
 Chapter 6

Measurement of Rice Grain Dimensions and Chalkiness,

and Rice Grain Elongation Using Image Analysis
Marnol V. Santos, Rosa Paula O. Cuevas, Nese Sreenivasulu,
and Lilia Molina

Abstract
Measurements of rice grain dimensions, percent grain chalkiness, and grain elongation used to be tedious
and slow due to the manual nature of measurements (e.g., use of calipers to measure grains one at a time)
and the subjective nature of scoring based on visual inspection (i.e., chalkiness). Recent developments in
imaging technologies have enabled more high-throughput means for measuring physical traits (i.e., grain
dimensions and chalkiness) in raw grains and grain elongation by comparing ratio between raw versus
cooked rice. The digital images of rice grains are captured through computer scanning and analyzed using
software that can calculate area and pixel value statistics of user-defined parameters. The improvements in
throughput made possible by the use of imaging technologies will allow faster quality grading of rice vari-
eties. Market quality is usually defined based on the rice grain physical traits (grain size and shape), degree
of chalkiness, and the ability of rice to elongate on cooking. In this chapter, the routine methods to mea-
sure the physical traits of rice and grain elongation using image analysis are described.

Key words Physical traits, Chalkiness, Size and shape, Grain elongation, Image analyses

1 Introduction

Grain quality can be subdivided into search, experience, and cre-

dence attributes [1]. Search attributes are those intrinsic properties
that consumers can evaluate and base their purchase decisions on,
which includes physical traits for rice grains such as length, width,
and degree of chalkiness. Ensuring these intrinsic properties are
important market criteria for valuation of milled rice into distinct
classes based on shape (shape is expressed as the ratio of grain
length to width): slender (>3.1), medium (2.1–3.0), and bold
(≤2.0). Traditionally, grain dimensions are determined per grain
using calipers or micrometers [2–6], and mean grain dimensions
are then reported. However, manual measurements of these prop-
erties are slow and tedious. The advent of digital imaging systems
has increased the throughput of measurement, with the imaging

Nese Sreenivasulu (ed.), Rice Grain Quality: Methods and Protocols, Methods in Molecular Biology, vol. 1892,
https://doi.org/10.1007/978-1-4939-8914-0_6, © Springer Science+Business Media, LLC, part of Springer Nature 2019

99
100 Marnol V. Santos et al.

systems calibrated using, or compared against, manual measuring

techniques using calibrated instruments [7, 8]. Flatbed scanning
systems, such as the SeedCount SC5000 Rice Analyser, operate by
capturing images of individual seeds within a sample lot using the
reflectance mode. Algorithms are used to estimate grain length and
width, and calculate the sample lots’s mean and its variation [9,
10]. At IRRI, grain length is categorized as short (<5.50 mm),
medium or intermediate (5.51–6.60 mm), long (6.61–7.50 mm),
and very long (>7.51 mm) [11].
Chalkiness, an attribute that affects the visual quality of the raw
rice grain, can also be classified as a search attribute [1] as the per-
cent of opacity (“chalkiness”) and translucency of the rice grain
factor into purchase decisions of consumers; i.e., in most markets,
the higher the degree of chalkiness, the lower is its acceptability in
the market [11]. Chalkiness and milling quality (e.g., head rice
recovery, HRR) are negatively correlated; i.e., HRR decreases with
increased chalkiness [12]. Chalkiness is often expressed as the pro-
portion of opaque areas relative to transluscent areas in a single
milled rice grain [13]. Typically, visual observation and scoring
based on a scale is used to evaluate the percentage of chalkiness in
the grain (reviewed in [14]). However, evaluation of chalkiness in
rice by image analysis is more accurate than by human visual inspec-
tion [15]. Image analysis also enables an increase in efficiency of
measuring chalkiness.
Grain elongation is another visual attribute evaluated in cooked
rice rather than in uncooked grains. It is an important attribute for
consumers in India and Pakistan, while it is an attribute also
observed in sadri varieties (Iran) and in japonica rice [16]. The
metrics used to describe this attribute are the grain elongation ratio
(the ratio of mean length between cooked grain and raw grain [17]
and the grain elongation index (the ratio of mean length-to-width
ratios between cooked grain and raw grain [16]. Although grains
can be measured one by one directly or through the use of a photo
enlarger, digital capture of images of individual cooked and raw
grains via an ordinary flatbed scanner and measurement of grain
length and width through the ImageJ analysis software and its
user-written plug-ins [18] has significantly improved efficiency in
characterizing grain elongation.
This book chapter describes the routine sample scanning and
image analysis procedures for grain dimensions, chalkiness, and
grain elongation measurements being conducted at IRRI’s Grain
Quality and Nutrition Services Laboratory (GQNSL).

2 Materials

2.1 Sample 1. Paddy rice grains.

Preparation 2. Drying oven, Model UFE800, Memmert GmbH + Co. KG,
Germany.
 Measurement of Rice Grain Dimensions and Chalkiness, and Rice Grain Elongation… 101

3. Dehuller, Satake, Model THU-35A, Satake Corp., Hiroshima,

Japan.
4. Mill, depends on the amount of sample to be polished:
(a) For samples 70–100 g: Grainman Model 60-220-60-DT
and 60-230-60-24 T (Grain Machinery Mfg. Corp,
Miami, FL, USA).
(b) For samples 10–15 g: Kett mill (Kett Electric Laboratory,
Tokyo, Japan).
5. Head rice separator (Satake rice machine Type TRG, Satake
Engineering Co., Ltd., Japan; or Rice sizing device).

2.2 Measuring 1. SeedCount SC5000 Image Analyzer, Next Instruments,

Physical Traits in Raw Australia.
Grain 2. SC-5000 Image Analysis System Software Version 2.6.3.
3. Image Analyzer sample tray (there are different trays for grains
of different lengths and shapes, Note 1).
4. Computer running Microsoft Windows Operating System XP
or 2007 as scanner controller.
5. Reference rice image dimension template (printed).
6. Reference color chart template (printed).
7. Sample scoop.
8. Sample tweezers/forceps.
9. Soft-bristle paintbrush.

2.3 Measuring Grain 1. HP Scanjet 8200, Hewlett Packard (for image capture).
Elongation 2. Computer running Microsoft Windows Operating System
Xp/2007 as scanner controller.
3. ImageJ image analysis software, ImageJ 1.47, downloaded
from http://imagej.nih.gov/ij
4. 50 mL test tube (for cooking rice sample).
5. Water bath (Temperature set point range: 98.0–100.0 °C).
6. Mild solvent and lint-free tissue for scanner surface cleanup.
7. Reference rice image dimension template (printed).

3 Methods

3.1 Sample 1. Remove the hull from paddy rice grains using a dehuller.
Preparation 2. Using a mill suitable for the amount of sample, mill the dehu-
lled grains into white polished rice (see Note 1).
3. Place the milled rice into bar-code labeled envelopes.
102 Marnol V. Santos et al.

3.2 Measuring This procedure is specific for the operations conducted at the
Physical Traits in Raw GQNSL. Please refer to the SeedCount SC5000 user’s guide to
Grain customize the settings for other laboratories.
1. Switch on computer and SeedCount unit.
2. Open SeedCount program icon.
3. As quality control, scan the reference rice image dimension
template (Fig. 1) before analysis and check if the values are
within the calibration standard verification (CSV) limits
(Table 1).
4. If the dimension limits are exceeded, clean the scanner surface
or restart the scanner and the computer. If these do not help
correct the dimension readings, call personnel trained at trou-
bleshooting the equipment (either in-house staff or from the
supplier) to recalibrate the system’s dimension readings.
5. For measurement of the degree of chalkiness (see Notes 2–4),
presort and label the samples according to the color of milled
rice using the reference color chart template (Fig. 2).
6. Using the designated sample scoop, pour at least 100 milled
grains onto the tray (medium or long grain tray, depending on
the size of the grains, Note 1). Make sure that the grains are
scattered evenly on the wider wells (bottom part of the tray).
All grains must be in the wells.
7. Load the sample tray into the SeedCount unit.

Fig. 1 The SeedCount reference rice image dimension template (not presented to scale), provided by Next
Instruments. The dimensions in the template have been validated using a calibrated Vernier caliper prior to
use. The template is scanned on the SeedCount unit on a daily basis or as needed to confirm system accuracy
and stability
 Measurement of Rice Grain Dimensions and Chalkiness, and Rice Grain Elongation… 103

Table 1
Calibration standard verification (CSV) limits for rice dimensions
measured using the SeedCount SC5000

Dimension limits

Image dimension template1. Length (mm) Width (mm)

CSV1 7.3 ± 0.1 3.4 ± 0.1
CSV2 6.1 ± 0.1 2.8 ± 0.1
CSV3 8.0 ± 0.1 4.0 ± 0.1

Fig. 2 Reference color chart template, developed in-house. Before scanning, the grain color is compared with
the color chart for appropriate adjustment of the reflectance brightness threshold. Color chart values of a, b,
c, and d correspond to seed count’s reflectance slider position values of 121, 122, 123, and 124,
respectively

8. In the SeedCount software window, the Module must be set to

“White Long Grain Rice” and the Standard must be “ISO.”
9. Match the grain color with the color chart template (Fig. 2).
Adjust the reflectance brightness threshold slider position from
121 to 124 depending on the corresponding color of the grain
on Fig. 2 (see Note 5).
10. Select “Scan and Analyze.”
11. Enter approximate weight of the sample in grams.
12. Wait for the results (grain length, grain width, and % chalki-
ness) to appear.
13. Select “‘Save Data.”
14. In the Sample ID window, enter the filename and sample ID or
scan the barcode of the sample.
15. Click OK and save.
16. Remove the tray from the SeedCount unit.
17. Return all grain samples into the sample envelope.
104 Marnol V. Santos et al.

18. For succeeding samples, repeat Subheading 3.2, steps 5–17.

19. Rescan the reference image dimension template after the anal-
ysis. Check the limits as stated on Subheading 3.2, step 3.
20. At the end of the analysis, close the SeedCount window.

3.3 Measuring Grain 1. Switch on the computer and the image scanner (see Note 6).
Elongation 2. Place at least ten whole (unbroken) raw grains directly on the
3.3.1 Capturing scanner surface for image capture. Ensure that the grains do
Grain Images not touch each other.
3. While leaving the scanner cover open, scan the grains and save
the scanned image file.
4. Capture images of unbroken cooked rice grains (Subheading
3.3.2, steps 1–7) following Subheading 3.3.1, steps 1–3.

3.3.2 Cooking 1. In a 50 mL test tube, place at least 20 raw unbroken grain

Rice Grains (milled rice) samples and add water (10 ± 2 mL).
2. Soak the grains in water for 1 h at room temperature
(25 ± 5 °C).
3. Heat the test tube (10 ± 1 min) in a boiling water bath.
4. Cool the test tube (5 ± 1 min) in a water bath (20–30 °C)
which has a water level at least 1 cm higher than the water
height in the test tube.
5. Remove/drain excess water in the test tube.
6. Place the cooked rice on a Petri dish and blot dry using filter
paper or tissue.
7. Using tweezers or forceps, place unbroken cooked grains on
the scanner surface for image scanning.

3.3.3 Image Capture The methods written below are specific for the operations con-
and Processing Using ducted at the GQNSL. Please refer to the ImageJ user’s guide to
ImageJ customise the procedure for other laboratories.
1. Click the ImageJ image analysis software. The system will stan-
dardize the pixel-to-mm ratio automatically every time the sys-
tem starts using the preset or stored reference Vernier caliper
image (see Note 7).
2. Click on the “Plugin” Menu and select the appropriate cus-
tomized plugin module, then browse and open for the scanned
image for measurement.
3. The selected plugin module will measure the lengths and the
widths of all objects in the selected file. And will give the fol-
lowing images and result list:
Measurement of Rice Grain Dimensions and Chalkiness, and Rice Grain Elongation… 105

(a) 
Processed equivalent 8-bit black and white, negative
image.
(b) Outlined images with assigned number per object.
(c) Result list of length and width measurement.
4. The analyst may copy the results list onto a spreadsheet file for
further data control check (see Note 8), and for statistical
analyses.
5. Selecting the day’s scanned reference rice image CSV template
(Fig. 3), compare the results with the standards’ dimension
limits (Table 2).
6. If the dimension limits set are exceeded, clean the scanner sur-
face or restart the scanner and the computer; if the values are
still outside the dimension limits, call for technical support (in-­
house trained staff or from the supplier) to recalibrate the sys-
tem’s dimension readings.
7. For the sample measurement, record the average length, the
average width, and the length-to-width ratio of both the raw
and the cooked rice grains. Save the file.
8. Report the grain elongation ratio [17] and the grain elonga-
tion index [16].

Fig. 3 ImageJ reference rice image dimension template (not shown to scale),
developed in-house. Using a certified Vernier caliper, the printed template images
dimensions were measured manually and the dimensions were set as the refer-
ence values for length and width. The template is scanned on a daily basis or as
needed to confirm system accuracy and stability
106 Marnol V. Santos et al.

Table 2
Dimension limits for three image dimension templates used for ImageJ

Dimension limits

Image dimension template1. Length (mm) Width (mm)

CSV1 4.1 ± 0.1 2.0 ± 0.1
CSV2 8.9 ± 0.1 4.1 ± 0.1
CSV3 14.7 ± 0.1 6.0 ± 0.1

4 Notes

1. Under-milled and colored rice grains may affect the measure-

ment for % chalkiness. Make sure that grains to be scanned are
well milled or polished before analyzing in the SeedCount
SC5000. The method is validated for grains with maximum
length of 6.5 mm and maximum width of 3.9 mm for medium
tray; and maximum length of 8.8 mm and maximum width of
2.5 mm for the long tray.
2. Using SeedCount in reflectance mode, the % chalkiness of the
complete sample is determined as the average of the % chalki-
ness of the individual grains. There is a need to adjust the
reflectance brightness threshold (using the reference color
chart template) because the default setting provides inaccurate
chalkiness values for grains of different shades or translucen-
cies. Adjusting the reflectance brightness threshold does not
affect measurements for grain length and for grain width.
3. Chalkiness measurement by the SeedCount SC5000 was cali-
brated using values obtained from processing raw rice grain
images using the ImageJ software. Figure 4 is an example of
how chalkiness measurements were obtained using ImageJ.
4. The opacity of waxy grain is neither determined nor validated
using the color chart and is not covered by the machine
capability.
5. Grain sample scanned with a darker color shade than that of
the color chart option D (Fig. 2) has been provided improper
% chalkiness values for rice samples.
6. Always clean the scanner surface by using any mild solvent
such as ethyl or isopropyl alcohol, or water and lint-free tissue
prior to and after use.
7. Scan the reference rice image dimension template at least
twice, as the initial and the last sample for the day’s quality
control verification run. Then using the ImageJ software, com-
pare results with the standards’ set values.
 Measurement of Rice Grain Dimensions and Chalkiness, and Rice Grain Elongation… 107

Fig. 4 Different degrees of chalkiness in grains, with chalkiness of the complete sample reported as the aver-
age of % chalkiness of the individual grains. These images were captured using a scanner (Subheading 3.3).
There are ten grains per image, positioned into two rows with five grains each row (I). ImageJ generates grain
images in which the chalky portions are blackened or shaded (II). The % chalkiness is then based on the area
pixel of the chalky area in relation to the whole grain (III). Sets (a), (b), and (c) illustrate samples whose calcu-
lated degrees of chalkiness are 32%, 47%, and 70%, respectively

8. The resulting value for the scanned working standards should

be compared to the set value and should be within ±0.1 mm of
the validated record or from the control chart, where available.
If these requirements are met, start processing the sample
images; otherwise, recalibrate the pixel-to-mm value based on
manual measurements using a Vernier caliper.

Acknowledgments

The authors thank Teodoro Atienza (who operates the SeedCount

SC5000), and Leah Villanueva (who measures grain elongation met-
rics) for providing technical assistance during method development
and optimization of the image analysis protocols. This work has been
supported under the CGIAR thematic area Global Rice Agri-Food
System CRP, RICE, Stress-Tolerant Rice for Africa and South Asia
(STRASA) Phase III, and Australian Centre for International
Agricultural Research (Project ID CIM/2016/046) funding.
108 Marnol V. Santos et al.

References
1. Rutsaert P, Demont M, Verbeke W (2013) 9. Armstrong B, Aldred G, Armstrong T,
Consumer preferences for rice in Africa. In: Blakeney AB, Lewin L (2005) Measuring rice
Wopereis MCS, Johnson D, Ahmadi N, Tollens grain dimensions with an image analyser. In:
E, Jalloh A (eds) Realizing Africa’s rice prom- Blanchard C, Truong H, Allen H, Blakeney
ise. CAB International, Wallingford, AB, O’Brien L (eds) Cereals 2005: 55th
pp 293–301 Australian cereal chemistry conference, Sydney,
2. Fan C, Xing Y, Mao H, Lu T, Han B, Xu C, Li Australia pp 326–329
X, Zhang Q (2006) GS3, a major QTL for 10. Tikapunya T, Fox G, Furtado A, Henry RJ
grain length and weight and minor QTL for (2016) Grain physical characteristic of the
grain width and thickness in rice, encodes a Australian wild rices. Plant Genet Res
putative transmembrane protein. Theor Appl 15(5):1–12
Genet 112(6):1164–1171 11. Graham R (2002) A proposal for IRRI to
3. Li M-m XL, J-f R, G-l C, Yu L-q, He H-h, L-z establish a Grain Quality and Nutrition
H, H-j K (2010) Identification of quantitative Research Center. In: IRRI discussion paper
trait loci for grain traits in japonica rice. Agr Sci series no. 44, International Rice Research
China 9(7):929–936 Institute, Los Banos p 15
4. Xie X, Song M-H, Jin F, Ahn S-N, Suh J-P, 12. Zhao X, Fitzgerald MA (2013) Climate change:
Hwang H-G, McCouch SR (2006) Fine map- implications for the yield of edible rice. PLoS
ping of a grain weight quantitative trait locus One 8(6):e66218
on rice chromosome 8 using near-isogenic lines 13. Lisle AJ, Martin M, Fitzgerald MA (2000)
derived from a cross between Oryza sativa and Chalky and translucent rice grains differ in
Oryza rufipogon. Theor Appl Genet starch composition and structure and cooking
113(5):885–894 properties. Cereal Chem 77(5):627–632
5. Govindaraj P, Vinod K, Arumugachamy S, 14. Raju GN, Manjunath N, Srinivas T (1991)
Maheswaran M (2009) Analysing genetic con- Grain chalkiness in cereals. Trop Sci
trol of cooked grain traits and gelatinization 31:407–415
temperature in a double haploid population of 15. Yoshioka Y, Iwata H, Tabata M, Ninomiya S,
rice by quantitative trait loci mapping. Ohsawa R (2007) Chalkiness in rice: potential
Euphytica 166(2):165–176 for evaluation with image analysis. Crop Sci
6. Fan C, Yu S, Wang C, Xing Y (2009) A causal 47(5):2113–2120
C-A mutation in the second exon of GS3 highly 16. Juliano BO, Perez CM (1984) Results of a col-
associated with rice grain length and validated laborative test on the measurement of grain
as a functional marker. Theor Appl Genet elongation of milled rice during cooking.
118(3):465–472 J Cereal Sci 2(4):281–292
7. Koutroubas SD, Mazzini F, Pons B, Ntanos 17. Dela Cruz NM, Khush GS (2002) Rice grain
DA (2004) Grain quality variation and rela- quality evaluation procedures. In: Singh RK,
tionships with morpho-physiological traits in Singh US, Khush GS (eds) Aromatic rices.
rice (Oryza sativa L.) genetic resources in Mohan Primlani, New Delhi, pp 15–28
Europe. Field Crops Res 86(2–3):115–130
18. Schneider CA, Rasband WS, Eliceiri KW
8. van Dalen G (2005) Characterisation of rice (2012) NIH image to ImageJ: 25 years of
using flatbed scanning and image analysis. In: image analysis. Nat Methods 9(7):671–675
Riley AP (ed) Food policy, control and research.
Nova Biomed, pp 149–186
 Chapter 7

Method Development of Near-Infrared Spectroscopy

Approaches for Nondestructive and Rapid Estimation
of Total Protein in Brown Rice Flour
Rosario Jimenez, Lilia Molina, Iman Zarei, Jennine Rose Lapis,
Ruben Chavez, Rosa Paula O. Cuevas, and Nese Sreenivasulu

Abstract
Rice varietal development and improvement programs are constantly seeking means to shorten the breeding
cycle in order to deliver new, consumer-acceptable rice varieties to farmers and to consumers. Advances in
molecular biology technologies have enabled breeders to use high-throughput genotyping to screen
breeding lines. However, current phenotyping technologies, particularly for rice cooking and eating prop-
erties, have yet to match the efficiency of genotyping methodologies. A high-throughput and cost-­effective
phenotyping suite is essential because without phenotype, the value of genotypic information cannot be
maximized. In this book chapter, we explore the application of near-infrared spectroscopy (NIRS), a high-­
throughput and nondestructive approach in characterizing rice grains, primarily describing method devel-
opment and validation, instrument calibration, upgrading, and maintenance. We then focus on estimating
protein content (PC) in brown rice as a case study because (1) PC is an attribute that contributes to the
cooking behavior and the eating properties of cooked rice; and (2) proteins contain chemical bonds that
can easily be detected by NIRS.

Key words Protein content, Near-infrared spectroscopy (NIRS)

1 Introduction

Measurement of indicators of rice cooking and processing charac-

teristics is considered important in varietal improvement programs
and thus has always been an essential component in selecting mate-
rials in rice breeding programs [1].The most commonly quantified
metrics for predicting grain quality include amylose content (AC),
gelatinization temperature (GT), and gel consistency (GC) [2–4].
The methodologies that have been developed for determining
these attributes are relatively high-throughput (e.g., differential
scanning calorimetry, alkali spreading test, iodine colorimetry, and
gel consistency test) but are often labor- and resource-intensive.

Nese Sreenivasulu (ed.), Rice Grain Quality: Methods and Protocols, Methods in Molecular Biology, vol. 1892,
https://doi.org/10.1007/978-1-4939-8914-0_7, © Springer Science+Business Media, LLC, part of Springer Nature 2019

109
110 Rosario Jimenez et al.

Because current improvements in rice breeding programs attempt

to shorten delivery times of new rice varieties, efforts have been
made to increase efficiency of breeding line selection through the
use of high-throughput genotyping approaches [5–8]. However,
current routine grain quality phenotyping methodologies, albeit
being high-throughput, are neither fast nor efficient enough to
match the rapid pace of genotyping pipelines and the increasingly
efficient modern breeding approaches. It is, therefore, important
to develop easy-to-implement, high-throughput, multi-trait, and
cost-effective grain quality phenotyping tools to ensure the timely
delivery of new rice varieties with desirable quality attributes.
A cost-effective, high-throughput, and nondestructive method
with the potential to screen for phenotypic variability in rice is
NIRS. In this technique, the wavelengths within the NIR range
(λ = 750–2500 nm) are absorbed differently by XHn functional
groups (O–H, N–H, C–H), due to specific vibrations of the chem-
ical bonds of the atoms in these functional groups [9]. These
chemical bonds (like in O–H, N–H, and C–H) differ in strength,
leading to differences in the amounts of energy required to change
the chemical bonds’ normal vibration patterns (e.g., transitioning
from a ground state to the second excited state). These energy dif-
ferences can then be visualized as spectra of varying absorptions
bands (e.g., overtones and combination bands) at different wave-
lengths [10, 11]. The absorbed energy can then be associated with
the concentration of the functional groups [12].
Since the 1970s, NIR instrumentation has been used for a
wide variety of applications in the grain industry as well as in proxi-
mate analysis of other foods and feeds, and in on-line process con-
trol in pharmaceutical manufacturing [11, 13–16]. However,
NIRS was slow to gain wide acceptance as an analytical tool [17]
because it is an indirect method whose accuracy is influenced by
the sum of errors of both the reference method and the NIR cali-
bration system, and by interferents or other variables not necessar-
ily related to the analyte of interest (e.g., particle size distribution
and moisture content [18]). With the advent of more sophisticated
NIR instrumentation, more powerful microprocessors, advanced
chemometrics software, and computational algorithms, more
robust and reliable NIRS prediction models for various analytes of
interest have become possible. These prediction models are typi-
cally regressions taking on the general form:
Y = a + b1 X1 + b2 X 2 + … + bk X k ,

where Y is the response variable; X1, X2,…,Xk are k independent

variables such that each variable is a combination of spectral values;
b1, b2, …, bk are the regression coefficients; and a is the intercept.
Chemometrics is used to determine the regression coefficients
­during prediction model development using known X and Y val-
ues, and is used to predict Y given a set of Xk and bk [16].
Method Development of Near-Infrared Spectroscopy Approaches for Nondestructive… 111

The improved application of NIRS is evident in the increasing

number of NIRS publications for predicting AC and other quality
components in whole-grain brown and milled rice [19–23], rice
flour [24–26], as well as in individual milled and bulk milled ker-
nels [27]. The best standard error of prediction (SEP) for AC by
NIRS obtained in earlier studies made by Villareal [19], Delwiche
[20, 21], and Bao [24] ranged from 1.0 to 1.6%. Besides AC, a
more comprehensive NIR prediction of rice starch quality param-
eters was reported by Bao [24] for GT, GC, pasting parameters,
and gel texture. NIRS was found reasonably accurate in predicting
pasting parameters of setback (SB) (SEP = 13.6 RVU, R2 = 0.92),
and breakdown (BD) (SEP = 10.2 RVU, R2 = 0.88); and gelatini-
zation peak temperature (Tp) (SEP = 1.33 8C, R2 = 0.89). Although
the various studies reported SEP and/or standard error of cross-­
validation (SECV) that were reasonably acceptable for the range of
samples analyzed, most of the calibration models developed were
restricted to specific types of rice varieties that do not cover the
entire compositional range. Rice samples used in these studies were
sometimes specific to the region where they were grown or were
selected from commercially released rice varieties and advanced
breeding lines. The accuracy of NIR global prediction models for
rice with widely diverse physical and compositional traits still has to
be validated.
Protein content (PC), although not routinely measured during
early selection breeding, likewise plays a substantial role in affect-
ing the cooking behavior of rice and the texture of cooked rice
because of protein’s ability to imbibe water during the cooking
process [28]. Although protein contributes to the nutritional value
of rice, high-protein rice may be unacceptable to some consumers
because it increases hardness, reduces the stickiness, and affects
milled rice color [29]. Much of the variability in the PC of rice is
believed to be caused by environmental factors [30] or by G × E
interactions more than by general heritability [31]. Moreover, a
negative correlation between PC and grain yield was reported in
earlier studies done at IRRI [30]. Therefore, a method that can
estimate PC in a high-throughput manner is important. The NIRS
method for protein estimation has been established with good pre-
cision for years. In fact, the first commercial NIRS instrument was
designed to analyze PC in grain and the success and importance of
the technique was demonstrated in the determination of protein in
wheat [32–34], using partial least squares (PLS), reported that the
calibration models obtained from near-infrared transmission
(NIRT) spectroscopy had an SEP of 0.24 for brown rice protein
content and 0.22 for milled rice PC. The derivative method for
NIR calibration to determine both PC and AC in milled rice flour,
resulting in a SECV of 0.23% for PC using four factors and an
SECV of 1.0% for AC using ten factors [35].
112 Rosario Jimenez et al.

While NIRS is viewed as a fast and simple method, the com-

plexity lies in the development and maintenance of reliable and
robust calibration models for the analytes or traits of interest. Most
commercial near-infrared spectrophotometers can now be pur-
chased with built-in calibrations. Some employ more complex but
efficient methods such as Artificial Neural Networks (ANN) for the
development of global NIR calibration models on large databases
[36]. However, since instrument manufacturers have limited access
to all possible variations in the types of samples the client may
encounter, pre-calibrated instruments are often not satisfactory or
reliable enough to be applied to a particular client’s samples. This
could be due to a number of reasons, mainly:
1. The range in composition of the samples used in the calibra-
tion does not cover all possible variation;
2. The sampling technique or presentation used for NIR spectra
collection is not applicable to the client’s limited amounts of
rice grains available for analysis;
3. The reference data is unreliable or the reference method is
highly susceptible to artifacts, and therefore problematic;
4. The calibrations provided with the instrument need to be vali-
dated and updated with client’s samples or entirely new cali-
bration models need to be developed that fit the specific
purpose of the assay.
This book chapter aims to present a general guideline for the
development, calibration, validation, as well as the upgrading and
maintenance of NIRS procedures. Figure 1 outlines the general
workflow for an NIRS analytical procedure. This workflow is
divided into two main phases: the initial calibration development
phase and the subsequent management phase of the NIRS proce-
dure lifecycle which includes constant validation and monitoring,
and updating and maintenance of the NIRS library or database.
Phase 1 involves initial NIRS calibration development for the
analyte or trait of interest using selected and representative sam-
ples. The spectral pattern of a set of diverse samples are first
explored by scanning all samples in the spectrophotometer using
the vis-NIR wavelength range (400–2500 nm) and interpreted by
the use of chemometric techniques [37]. A standard and well-­
established reference method is required to develop a reliable cali-
bration model. Ideally, the reference analysis should be performed
at the same time or immediately after NIR scanning. Information
on the performance (linearity, accuracy, and precision) of the refer-
ence method should be known since it will influence the perfor-
mance of the NIR prediction model that is calibrated against it. A
calibration model is developed by relating spectral data to labora-
tory values using multivariate statistical algorithms and then vali-
dated by testing the model with an independent set of samples
(usually around 10%, but not found in the calibration set) [18].
 Method Development of Near-Infrared Spectroscopy Approaches for Nondestructive… 113

Phase 1
Initial Calibration
samples NIRCalibraon
Development

NIR Scanning
(400–2500 nm wavelength)

Wet
chemistry by
Reference Method

Calibration Validation set

set (*.cal)
(*.cal) SD of validation set

Calibration Perform Report

Development Validation Acceptability
SEP(C), (*.lod) SEP of
(*.eqa) RPD=SD/SEP
Calibration

Phase 2
Routine NIR Analysis

Outliers NIR NIR Scanning

(400–2500 nm Test Samples
GH>3, NH>0.6
Check wetchem
Analysis
wavelength)
Using *.eqa

Validation
(select 10% of
test samples
for wet chem.)

Report Analysis
with validation
data

Fig. 1 General workflow for method development, validation, and routine analysis by NIRS

Phase 2 involves calibration maintenance and updating while

performing routine assay of succeeding batches of the same type of
test samples using the prediction model developed in Phase 1. The
development and implementation of a NIRS calibration is dynamic
114 Rosario Jimenez et al.

and iterative, meaning the NIRS library may be constantly updated

with new data, usually with outliers detected among new test samples
during routine analysis. Validation samples having analyte content
outside of the calibration range may appear as outliers when tested by
the NIRS procedure. Any suspected outlier should first be investi-
gated, whether spectral or reference outlier, before inclusion in the
library. Validation samples used to verify the standard error of predic-
tion (SEP) for every batch of analysis, provide added information,
and hence can also be used to expand the existing NIRS library.
When reporting NIR analysis data, validation statistics for the assur-
ance of robustness of the method should be provided. Changes (both
planned and unplanned), which might affect the performance of an
NIRS procedure, may necessitate re-­calibration and/or re-validation
of the NIRS model to demonstrate continued model suitability. All
changes should be validated accordingly, appropriately documented
and recorded according to valid change management protocols.
To illustrate this, the application of these guidelines for NIRS
screening of PC in brown rice flour in a diverse set of rice varieties
is reported.

2 Materials

2.1 Rice Samples In developing the calibration and the validation sets, it is important
to use rice samples coming from a diverse collection of materials
(see Note 1).

2.2 Vis-NIRS The Grain Quality and Nutrition Service Laboratory (GQNSL) is
Spectrophotometer equipped with the NIR Systems 6500 (Perstorp Analytical, Silver
Spring, MD, U.S.A.), a visible/near-infrared scanning spectropho-
tometer (see Note 2) able to collect reflectance readings and calcu-
late for log (1/reflectance) at 2 nm intervals within the
400–2500 nm wavelength range. Other NIR spectrophotometer
models can be used as well; however, this book chapter presents
settings that may be specific for the NIR Systems 6500 as opti-
mized for rice samples.

2.3 Sample Module Spinning sample cup (for reflectance mode), 36 mm diameter
(depth = 9 mm) black ring cups with quartz windows on one side
and disposable white backs on the other side, fitted with black
inserts with 12 mm diameter windows.

2.4 Standardized The standardized “Check Cell” is provided with the instrument to
Reference (Check Cell) evaluate and keep track of overall instrument performance over time.

2.5 Sample 1. Dehuller.

Preparation Devices 2. Retsch MM400 mixer mill (Retsch GmbH, Retsch-Allee 1-5,
42781 Haan, Germany) for fine-grinding (up to ~5 μm) small
amounts of grains.
 Method Development of Near-Infrared Spectroscopy Approaches for Nondestructive… 115

2.6 Chemometrics The chemometrics software available with the NIRSystem6500 is

Software WinISI v1.02a (Infrasoft International, LLC) although other
third-party multivariate analysis software (e.g., Unscrambler X
10.3, CAMO Software, Oslo, Norway) may also be utilized for
this purpose (see Note 3).

2.7 Apparatus 1. Skalar San++ Continuous Flow Analyzer unit consisting of

and Materials sampler, manifold, proportioning pump, heating bath, color-
for Micro-Kjeldahl N imeter (with a 660 nm filter).
Analysis (Reference 2. Analytical balance, 1000 mg capacity.
Method) 3. Aluminum block digestor with automatic temperature
control.
4. Digestion tubes (2.5× 20 cm) calibrated to 20.000 ± 0.100 g
of the same solution matrix as the calibration standard
(K2SO4-H2SO4).
5. Dispensers, 10 and 50 mL capacities.
6. Vortex mixer.
7. Calibrated scoops (for catalyst mixture).
8. Polyethylene bottles.
9. Volumetric flasks.
10. Funnel.
11. Autosampler cups.
12. Concentrated sulfuric acid (Analytical Grade, 18 M).
13. Ammonium sulfate (AR grade).
14. Potassium sulfate (AR grade).
15. Ammonium chloride (AR grade).
16. Anhydrous Potassium sulfate/Selenium mixture, 100 g:2 g
(see Note 4).
17. Calibration standards (0, 10, 30, 50, 70, 90, 120, 150 mg
N/kg from 1000 ppm N ammonium sulfate).
18. Sodium hydroxide solution, 20% (Stock).
19. Sodium potassium tartrate solution, 20% (Stock).
20. 0.5 M Buffer Solution (Stock) (see Note 5).
21. Working Buffer, a solution made up of 20% sodium potassium
tartrate (250 mL), 0.5 M stock buffer solution (200 mL), and
20% sodium hydroxide (250 mL); the volume is then made up
with deionized water to 1 L.
22. Sulfuric acid (0.135 M)/sodium chloride (1.71 M) solution.
23. Sodium salicylate (0.94 M)/sodium nitroprusside
(0.00026 M) solution.
24. Sodium hypochlorite solution, 0.525%.
116 Rosario Jimenez et al.

25. Internal rice reference standard, 99G (rice flour).

26. 30% Brij35 (AR grade).

3 Method Development

3.1 Sample This section describes method development for brown rice samples.
Preparation However, the approach should be applicable to milled rice as well.
1. Dehull the paddy rice samples using the rice dehuller to obtain
brown rice.
2. Grind and homogenize the brown rice grains (2–3 g) using the
Retsch MM400 mixer mill to get very fine flour (see Note 6).
3. Prior to NIR scanning and wet chemical analysis, the samples are
oven-dried (80 °C, 30 min), and then kept in a desiccator cabi-
net until ready to be loaded into NIR sample cups or weighed
for chemical analysis.

3.2 NIRS Scanning Since NIRS is a nondestructive method, scanning is usually done
of Brown Rice Flour prior to destructive wet chemical analysis, especially for samples
Samples with limited available grains. The method described here is specific
for the NIRSystems 6500. For more details, refer to the manufac-
turer’s manual.
1. Power on and warm up the (NIRSystems6500) spectropho-
tometer (1 h).
2. Perform the following instrument diagnostics according to
instructions from the manufacturer [18]:
(a) Instrument Response refers to the absolute reflectance from
the ceramic reference tile (supplied by the manufacturer).The
maximum value for absolute reflectance in the NIR Systems
6500 is 65,535. The auto-gain function of the instrument
allows the reflectance at the visible range to be above 50,000
and that of the NIR range to be above 35,000.The dark
period values should be less than 1600 (see Note 7).
(b) Wavelength Accuracy is measured using internal standards:
didymium (for the visible range) and polystyrene (for the
NIR). In aligning the wavelengths, the error should be within
±0.30 nm; i.e., any peak found within the 0.30 nm margin of
the wavelength is acceptable. The reported wavelength loca-
tion and suggested or nominal locations for the wavelengths
should not have day-to-day deviations of more than 0.10 nm.
Wavelength accuracy can be tested by comparing the current
K and Phi values, and its associated error value. There is no
need to change the K and Phi values if the current error does
not exceed twice the suggested error (see Note 8).
Method Development of Near-Infrared Spectroscopy Approaches for Nondestructive… 117

(c) NIR Repeatability indicates the magnitude of deviation

(noise) in optical data at each wavelength. Testing for
repeatability requires the use of the ceramic tile, scanned
as a reference, as an unknown sample, and then as a refer-
ence. Repeatability is measured using the statistic average
root mean square corrected for bias (RMS(C)), calculated
in the WinISI software. The RMS(C) of five scans should
be <20 in a room with stable temperature (see Note 9).
(d) Scan “Check cell” before measurements are made on test
samples to evaluate overall day to day performance of the
instrument (see Note 10). This step may also be included
within the sample sequence. The spectra of the “Check
cell” are stored in the Chk.NIR file (see Note 11).
3. Record details in the NIR sample logbook.
(a) List the sample codes beside the corresponding sample
cup numbers. Sample cups are numbered 1–100 on the
disposable backs or covers.
(b) List down other relevant information such as the sample
designation, the source of sample, the harvest year and
season, the date of scanning, and the filename where the
NIR spectral data will be stored. Also indicate the insert
used (see Note 12).
4. Pack the flour into sample cups
Standard small ring cups (IH-0307, i.d. = 36 mm,
depth = 9 mm) have to be uniformly and completely filled
with flour samples. However, if not enough sample is avail-
able, the ring cups can be fitted with black ring inserts to
reduce the volume of the cups that have to be filled with
smaller amounts of sample (see Note 12). It is important to
decide from the start what type or size of cups to use based on
expected amount of samples available for the entire application
or intended use of the procedure.
(a) Arrange the disposable white back covers on piles accord-
ing to number (1–100, 1 on top).
(b) Clean the sample cup and the insert using a soft camel hair
brush and/or a small vacuum cleaner. Additional cleaning
may be done by wiping the quartz glass with lint-free
tissue.
(c) Mix the flour sample in the envelope thoroughly using the
stainless steel spatula.
(d) Scoop portions from the envelope with the spatula and
pour into the sample cup (with or without the insert).
Overfill the cup a little to ensure that there are no empty
air spaces inside.
118 Rosario Jimenez et al.

Fig. 2 NIRSystems 6500

(e) Use the white disposable back to cover the cup. Make sure
to use the correct sample cup numbers corresponding to
the sample codes (refer to the NIR sample logbook).
(f) Lightly push the cover to secure it and to compress the
sample so that there will be no loose or empty spaces
inside.
(g) Vacuum or wipe again the outer quartz window of the
sample cup to ensure it is clean or free of dust.
5. Collect NIR spectra of samples
(a) Load the sample cup into the Autocup magazine con-
nected to the Spinning Sample Cup module (for reflec-
tance mode) of the NIRSystems 6500 (Fig. 2).
(b) Create a file where all sample spectra are to be stored, for
example, “BRF-DS15.NIR” for brown rice flour sampled
from 2015 dry season field plots. “NIR” is the file exten-
sion for files containing spectra.
(c) Scan the sample over the entire visible-NIR range
(λ = 400–2500 nm) in 2 nm increments. Each sample
spectrum recorded is the average of 32 scans.
(d) Check the sample cups for any air spaces after scanning.
Note the sample number and repeat step 4.
 Method Development of Near-Infrared Spectroscopy Approaches for Nondestructive… 119

6. Cleaning sample cups

(a) Collect the sample and return to the designated envelope.
(b) Clean the sample cup, white disposable cover, and insert
(if used) by using a vacuum, soft camel hair brush, and
lint-­free tissue.

3.3 Reference The GQNSL performs protein content determination by Kjeldahl

Method (Wet Chemical method developed and modified based on the procedure reported
Analysis) by Yoshida et al. [38].
1. Convert protein N to NH4+ by chemical digestion
(a) Weigh 50.0 ± 0.1 mg of oven-dried sample in a 2.5 × 20 cm
pre-calibrated digestion tube.
(b) Add 1 scoop (approximately 1.0–1.3 g) of potassium
­sulfate/selenium mixture into the digestion tube.
(c) Add 2.0 mL concentrated sulfuric acid from a calibrated
dispenser and mix using a Vortex mixer to completely wet
the brown rice flour material.
(d) Place the tube in the digestion block preheated to
380 ± 10 °C to digest the sample (1 h), shaking the tube
after 30 min to ensure complete digestion (to wash down
flour that may cling to the side of the tube into the diges-
tion mixture).
(e) Remove the tube from the digestor and let it stand to cool
at room temperature before adding deionized water to
final volume.
(f) Transfer the diluted digest into a 20 mL test tube and stand
overnight to allow for sedimentation before analysis.
2. Measure the NH4+ concentration based on its colorimetric
reaction (an emerald green complex) with sodium salicylate in
the presence of sodium hypochlorite and nitroprusside (modi-
fied Berthelot reaction [39], see Note 13).
(a) In the software of the Skalar San++, create the “Sample
Sequence Run,” which includes:
i. Primer, the standard with the highest concentration
(150 ppm).
ii. Wash Solution (deionized water).
iii. Calibration curve standards: blank, 10 ppm, 30 ppm,
50 ppm, 70 ppm, 90 ppm, 120 ppm, 150 ppm
(NH4)2SO4.
iv. Unknown samples in batches of ten.
v. The 90-ppm standard, treated as quality check sam-
ple, is analyzed for every ten unknown samples.
vi. Calibration standards as unknowns.
vii. Wash solution.
120 Rosario Jimenez et al.

Fig. 3 Skalar San++ Continuous Flow Analyzer

(b) Prepare the reagents

i. Sodium salicylate/sodium nitroprusside solution
(see Note 14).
ii. Sodium hypochlorite solution (see Note 15).
(c) Pour the wash solution, the standards, and the diluted
sample digests into autosampler cups and arrange accord-
ing to the “Sample Sequence Run” order.
(d) Start the sequence run of the autoanalyzer (Fig. 3) with
the following settings:
i. Absorbance is measured at λ = 660 nm;
ii. Sample flowrate is set at 0.16 mL/min;
iii. Sampler settings at 30 s (sample time; the time allowed
for the sample to pass through the chemistry mani-
fold), 60 s (wash time) and 6 s (air time);
iv. Peak detection based on peak height and using a
Calibration I Order ISO 8466.
(e) Construct a standard curve based on the absorbance val-
ues of the calibration standards.
(f) Calculate the % Kjeldahl N based on the standard curve,
which is based on the Beer-Lambert’s Law [40].
(g) Protein content (%) is calculated by multiplying % Kjeldahl
N by 5.95 [40].
(h) The expanded uncertainty for percent Kjeldahl N is ±0.072
with coverage factor = 2, which gives a level of confidence
of approximately 95%.
 Method Development of Near-Infrared Spectroscopy Approaches for Nondestructive… 121

3.4 Initial NIRS 1. Sample Selection for the Calibration Set

Calibration There are no hard and fast rules to determine the optimum
Development number of selected samples to include in the initial or “starter”
calibration set. The calibration set should be composed of
samples that are sufficient to provide all of the likely sources of
variance (Infrasoft International 1995). A NIRS method
development procedure can make use of large sets of data
already available, such as from a database in a rice quality eval-
uation service laboratory or from a screening experiment.
There are three different approaches to select the calibration
set. Choose one approach depending on the availability of
samples and reference values:
(a) Selection from a large data pool based on reference data
A large database can provide valuable source of all kinds
of variation. Taking into consideration the range and dis-
tribution of reference data, samples with uniform distribu-
tion across the expected range may be assembled to
compose the calibration set. A large number of samples
with nonuniform distribution in the calibration set may,
however, lead to poor prediction models due to increased
risk of involving noise in the calibration and bias toward
the mean or toward the range with dense population [41].
Representative and evenly distributed samples should be
selected to cover the entire range of component of
interest.
(b) Selection from the spectra files without available data from
the reference method
It is possible that spectra of a large set of samples are
available but reference data and the range of variability are
still unknown. Collection of NIR spectra of all samples
prior to wet chemical analysis proves to be a more simple
and cost-saving approach since only samples with selected
representative spectra that are unique are included in the
calibration set and are sent to the laboratory for wet chem-
ical analysis.
(c) Selection from a large sample spectra file with reference
data
When there is no prior knowledge about the diversity
and composition of all available samples, NIR scanning
and wet analysis are performed simultaneously on all sam-
ples even before selection is done. This is usually the case
when a laboratory starts to explore the applicability of an
NIR method as an alternative for routine wet chemical
analysis. Although the process may be labor-intensive at
the start, this is considered as the best approach since the
SELECT algorithm in WinISI fine-tunes the scores to
maximize the differences among the spectra as well as the
reference value.
122 Rosario Jimenez et al.

2. Processing of NIRS spectra from the calibration set

(a) Select the wavelength segment of the spectra to be used
for creating scores
For the protein calibration, the factory default segment
or the normal recommended wavelength for the instru-
ment in the NIR range (1108–2492, 8) is used. This
means that the wavelengths used for creating scores by
multivariate statistical algorithm are every 8 nm from 1100
to 2500 nm (NIR region) where the first (1100 nm) and
last (2500 nm) wavelengths are not included.
(b) Mathematical treatment of spectral data
The “SNV and Detrend” option is applied on all spectra
for scatter correction. The SNV (Standard Normal Variate)
function scales each spectrum to have a standard deviation
of 1.0 to help reduce particle size effects. “Detrend”
removes the linear and quadratic curvature of each
spectrum.
A good exploratory math treatment is a (1, 4, 4, 1) where:
1 = first derivative
4 = gap over which the derivative is calculated
4 = smoothing of points
1 = no second smoothing
An example of “raw” spectra of 3148 samples is shown
in Fig. 4a and the resulting normalized, or mathematically
treated, spectra in Fig. 4b. Reducing the particle size
effects through normalizing the data set has visibly reduced
the noise (Fig. 4b).
(c) Create the score file from the spectra file, done using the
WinISI software
A “score file” is created wherein the mathematically
treated spectral data are first reduced to independent
sources of variation and converted to scores by principal
component analysis (PCA). This option is important in
understanding the structure and distribution of a popula-
tion of samples based on spectra; no reference values
are used. However, it places a lot of emphasis on particle
size, moisture [18], and other variables that influence
the spectral pattern more than the sample chemical
composition.
(d) Use SELECT algorithm for pattern recognition
The method uses an algorithm called SELECT in the
chemometrics software. This algorithm identifies patterns
in a group of normalized or math-treated spectra that
 Method Development of Near-Infrared Spectroscopy Approaches for Nondestructive… 123

Fig. 4 NIR spectra of 3148 brown rice flour samples showing the average spectrum (in yellow) for (a) raw data
and (b) normalized data

c­ ontribute most to the variation among the spectra. Each

sample score is ranked by its distance from the center of a
spectra file (global H or GH) and by its distance from
every other sample score in the file (neighborhood H or
NH). Redundant samples or samples whose scores are
close to selected neighbors (NH value) are excluded in the
calibration set. The cutoff NH is set according to the spec-
tral range of the data (0.6–1.0 for monochromators col-
lecting data at λ = 1100–2498 nm). If too many samples
are automatically selected by the SELECT algorithm in
WinISI based on the cutoff NH value, the number can be
reduced to whatever is deemed enough by WinISI to rep-
resent the entire population.
124 Rosario Jimenez et al.

On the other hand, the SELECT function uses partial

least squares (PLS) to identify patterns in a group of spec-
tra that contribute most to the variation within spectra-
reference data.
3. Calibration development in the WinISI software
The following steps represent a good place to start to make
an initial calibration:
(a) Create the calibration file
Use the *.NIR file or the file containing the spectra of
selected calibration samples (from step 1). Encode or import
the reference data into the file, taking care to properly match
the reference data (PC by Kjeldahl N, Subheading 3.3) with
the correct sample number or code (Refer to logbook
­prepared in Subheading 3.2, step 3). The resulting spectra-
reference data file is automatically saved as a *.CAL file.
(b) Wavelength selection
There are two wavelength segments that can be consid-
ered for the calibration:
i. Visible range: 408–1092, 8
ii. NIR range: 1108–2492, 8
(c) Mathematical pretreatment of spectral data
The same scatter correction and math treatment method
applied to create a score file from the spectra file should be
used (as in step 2b).
(d) Open/Create the required files
i. Input files
● Calibration file: Select the file containing the
spectra and reference data of samples in calibra-
tion set from step 3a (*.CAL).
● Repeatability file: Select the file containing spectra
of the “Check C ell” repeatedly scanned on differ-
ent days and at specific intervals during routine
analysis of test samples (e.g., “Chk.NIR” file from
Subheading 3.2, step 2d).
ii. Output files
● Equation file: Create the resulting filename where
the calibration equation or the prediction model
will be saved. This file usually has the same name as
calibration file (sample set) but with the default
extension (*.EQA).
● Loadings file: Create the resulting filename that
will contain the loadings or the patterns that
explain the spectral variation in the calibration file.
This file usually has the same name as the calibra-
tion file but with the default extension (*.LOD).
Method Development of Near-Infrared Spectroscopy Approaches for Nondestructive… 125

iii. Choose and perform the regression method

The recommended “Regression Method” is modified PLS,
which reduces the spectral data to a few combinations of
absorptions that account for both the spectral information
and reference values. The modified PLS procedure algorithm
[42] copes more effectively with non-analyte interference and
provides an estimate of the interference spectrum that can be
used to correct the measured spectrum before determination
of analyte concentration.
4. Sample Selection for the Validation Set
Validation of calibration models is normally performed on
an independent set of samples or samples not found in the
calibration set. However, in the course of calibration develop-
ment, simultaneous validation is also performed using internal
samples within the calibration set (cross-validation).
(a) External or independent validation set
To demonstrate linearity, it is likewise required that
samples in the validation set are distributed across the
specified range of interest. From among the redundant
samples excluded from the calibration set in step 1, around
10% of the number of samples in calibration set may be
chosen either randomly (with uniform distribution across
the protein content range) or by using the SELECT algo-
rithm on the score file of external samples.
(b) Internal validation set
The internal validation set is composed of samples in
the calibration set where each sample is cross-validated
against the resulting calibration model. The algorithm
usually performs this automatically during the calibration
procedure to determine the optimum number of terms (or
factors) that will result in the least possible SECV.
5. Validation
NIRS validation presents evidence for the accuracy and
robustness of the calibration model in predicting the compo-
sition of samples for which the calibration was made. The
range should be confirmed by use of a suitable validation set
(step 2) which matches in extent the proposed range for the
intended use of the procedure. The NIRS predicted analyte
concentration of samples in the validation set should be com-
pared with reference values.
(a) Create the validation file
Use the spectra file (*.NIR) of the validation set con-
taining the selected independent samples or samples not
included in the calibration set (selected from step 2a).
Send the samples for laboratory analysis if reference data
are still unknown (or if samples are selected from a score
126 Rosario Jimenez et al.

file). Enter or import the reference data into the spectra file
(*.NIR), taking care to properly match the reference data
(PC by Kjeldahl N method) with the correct sample num-
ber or code. The resulting file containing the spectra and
reference data is likewise saved with the default extension
(*.CAL) but use another character in the file name to
differentiate it from the calibration set (ex. *v.CAL).
(b) Use the output calibration equation file (*.EQA devel-
oped from step 3dii) to predict the PC analyte concentra-
tion of samples in the validation set (*v.CAL).
(c) Compare NIR predicted and reference values of samples
in validation file and report validation statistics (SEP, R2,
range and standard deviation of analyte concentration).
(d) Rate the performance of the calibration model by the ratio
of the standard deviation of analyte concentration in the
validation set to the SEP.

4 Application of the NIRS Method Development Pipeline for Protein Content

Estimation in Brown Rice

A diverse set of rice samples (N = 3148) belonging to indica,

tropical and temperate japonica, and aus varietal groups were
grown and harvested at IRRI’s experiment station during the 2013
dry season. These samples were selected based on the diversity of
their cooking and texture characteristics (e.g., AC, GT, GC) as
well as their physical characteristics (e.g., chalkiness, grain length,
grain width, grain shape). The paddy grains were dehulled to
obtain brown rice and were subjected to quantification of PC by
Kjeldahl N method. The brown rice was ground using the Retsch
MM400 mixer mill to get very fine flour (2–3 g). The flour was
oven-dried (80 °C, 30 min) then packed into the NIRS sample
cups. NIR spectra were obtained using the NIRSystems 6500,
scanning over the entire visible-NIR range (λ = 400–2500 nm in
2 nm increments).
Results showed that the range of PC was from 5.36 to 13.98%
(μ = 9.50%). The frequency distribution of PC, more or less, fol-
lows the bell shape of a normal distribution and shows that majority
of the samples have PC at around 8.1–9.0% (Fig. 5), consistent with
reported literature [34].
Determination of the calibration set was then conducted using
the three approaches specified (Subheading 3.4, step 1).
In the first approach (selection based on reference data), 10%
of the population was randomly selected from within each bin in
the frequency distribution (Fig. 5) such that each bin, except for
the tails, was equally represented (Fig. 6).
Method Development of Near-Infrared Spectroscopy Approaches for Nondestructive… 127

Fig. 5 Histogram of PC in a diverse set of 3148 brown rice samples. The blue
shaded portions represent the number of samples selected from each range for
the calibration set

Fig. 6 Uniformly distributed samples selected from each PC range bin for the
calibration set

For the population of 3148 samples, around 300 samples were

selected for the calibration set such that the second to the eighth
range bins were represented by 40 samples; the first and last bins
were each represented by 10 samples since there were fewer samples
in those bins (Fig. 6). The NIR spectra of these samples were
obtained and subjected to further mathematical treatment. The
spectra-reference data sets were used to create a calibration file.
128 Rosario Jimenez et al.

Fig. 7 Three-dimensional scores plot of 3148 sample spectra (blue) generated by

PCA showing distances from the center of the spectral hypersphere and the 301
representative samples (green) selected by WinISI for the calibration set

In the second approach (selection based on NIR spectra), a

score file was generated by WinISI using all available spectra (which
have already undergone mathematical treatment) through
PCA. WinISI was then used to select 10% of the population. The
number of samples in the calibration set (nCS), selected by WinISI
was 301 (Fig. 7). The PCs of the selected samples were then
obtained. The spectra-reference data sets were then combined in a
calibration file.
In the third approach (selection based on both spectra and
reference data), the NIR spectra are mathematically treated and an
input file containing spectra and reference data (N = 3148) was
created in WinISI. The samples that went into the calibration file
were then selected based on modified PLS on the entire popula-
tion. This resulted in a three-dimensional scores plot that visual-
ized sample scores and the samples (nCS = 307) that were selected
for inclusion in the calibration set (Fig. 8).
Calibration models were developed for each approach, the
number of terms (i.e., wavelengths used) for each equation is
found in Table 1. The coefficient of determination (R2) for the
calibration models ranged from 0.951 to 0.970, indicating that the
models were highly linear.
The standard error of calibration (SEC) ranged from 0.303 to
0.373. As the number of wavelengths (terms) used in the regres-
sion model increased, the SEC and the R2 decreased. On the other
hand, samples composing ~10% of the calibration set were
­randomly selected to be part of the independent validation set
(Table 1). For the first approach, the samples were selected ran-
domly based on the PC frequency distribution (Fig. 6). For the
Method Development of Near-Infrared Spectroscopy Approaches for Nondestructive… 129

Fig. 8 Three-dimensional scores plot of 3148 spectra-reference data (blue) gen-

erated by PLS showing distances from the center of the spectral hypersphere
and the 307 samples (green) selected by WinISI for the calibration set to repre-
sent the whole population

second and third approaches, the samples were selected by

WinISI. The spectral data of the independent validation set were
fed into each of the calibration models to predict PC. Results
showed that the R2 ranged from 0.919 to 0.977, indicating that
the models predicted PC with high accuracy; while the SEP ranged
from 0.312 to 0.376.
To select the best calibration model for predicting PC rou-
tinely, the balance between SEC and SEP must be achieved (or at
least be within 20% of each other). The data (Table 1) shows that
the SEC and the SEP ranges overlap and are well within 20% of
each other, hence it was difficult to select the best calibration
model. Therefore, the RPD values were reviewed. The highest
RPD values were from the first approach where calibration set
defined based on reference data (Table 1). According to Baillères
et al. [43] and Agelet and Hurburgh [44], at this RPD, the cali-
bration model can be used for quality control purposes. The
other models were good enough for screening purposes, accord-
ing to the standards previously reported [44, 43]. Hence, for
predicting PC by NIR, it is recommended to use the first calibra-
tion model (based on the calibration set with uniformly distrib-
uted PC reference data) (see Note 16).
It is recommended that the calibration models developed for
brown rice in the population (N = 3148) be used for routine PC
prediction and to expand the NIRS by including outlier samples
(both spectral and reference data) to the calibration model.
Table 1
130

Validation of calibration models developed for protein in brown rice flour

Calibration sample set Calibration models developed Validation test with independent samples

No. of No. of Rating/

Selection method nCS Range Mean Filename terms R 2 SEC SECV samples R2 SEP Range SD RPD acceptability
1. Uniform 300 5.36– 9.50 C123s-1. 8 0.970 0.373 0.401 30 0.977 0.356 5.64– 2.35 6.6 Good
distribution 13.98 eqa 13.86
Rosario Jimenez et al.

2848 0.919 0.359 5.64– 1.24 3.45 Fair

13.86
2a. Spectra scores 301 5.72– 9.03 C123s-2. 9 0.954 0.338 0.366 33 0.938 0.376 5.60– 1.46 3.88 Fair
(PCA) 13.74 eqa 11.66
 Sample
cutoff = 300
2b. Spectra scores 1245 5.36– 9.03 C123s-22. 10 0.951 0.323 0.334 1903 0.94 0.321 5.70– 1.305 4.06 Fair
(PCA) 13.98 eqa 13.74
 NH cutoff = 0.6
3a. Spectra 307 5.70– 9.12 C123s-3. 11 0.967 0.303 0.355 30 0.966 0.312 5.74– 1.63 5.22 Good
scores- 13.98 eqa 13.68
Reference data
 Sample
cutoff = 300
3b. Spectra 299 5.73– 9.08 C123s-32. 11 0.959 0.322 0.369 2849 0.937 0.343 5.35– 1.34 3.91 Fair
scores- 13.51 eqa 13.98
Reference data
 NH cutoff = 0.6

Where: SEC = Standard Error of Calibration, SECV = Standard Error of Cross-validation, SEP = Standard Error of Prediction, an indicator of overfitting of the model such that
the minimum SEP value indicates the maximum number of terms that can be put in the model without overfitting, SD = Standard Deviation of the range of PC (data based on
the reference method) in the validation set, RPD = Ratio of Performance to Deviation, the ratio of standard deviation of the PC data in the validation set to the SEP (SD/SEP).
Ratings of RPD are as follows: Excellent (>10); Very good (7–10); Good (5–7); Fair (3–5); Not satisfactory (<3)
Note: Range and SD are derived from data derived from the reference method
 Method Development of Near-Infrared Spectroscopy Approaches for Nondestructive… 131

5 Notes

1. The diversity and selection of samples used to build the data-

base is crucial to the robustness and reliability of the calibra-
tion or the prediction model for the substance or trait being
analyzed [45]. The samples ideally cover the whole range of
known variations in concentration, rice genotype, particle size
of ground material, growing season, location, and other
expected and unexpected sources of variation. Subsets may
also be maintained for rice samples prepared and analyzed in
different forms (e.g., whole grain vs. flour, or brown vs. milled)
and those that used other NIRS data collection mode (e.g.,
reflectance vs. transmittance mode).
2. The NIR spectrophotometer is a means for collecting and
measuring the intensity of the transmitted or reflected light by
the sample in the near-infrared region (λ ~ 780–2500 nm) that
are mainly due to overtone and combination bands primarily
of O–H, C–H, N–H, and C–O groups (Infrasoft International
1995). Essentially, NIRS takes a “fingerprint” or spectral pat-
tern of a sample based on its organic constituents. Most NIR
instruments also include the visible region (λ = 400–780 nm)
that permit visible color difference analysis.
3. An appropriate chemometrics procedure is required to be able
to extract meaningful information from the NIR spectra.
Analyte concentrations are not directly determined from the
NIR spectra because the spectra at the NIR range feature
overtone and represent combination bands primarily of O–H,
C–H, N–H, and C–O groups [10, 11]. Thus, an equation or
a mathematical model relating spectral data (treated as the
independent variables) to wet chemistry-based laboratory val-
ues (the response variables) is necessary. This calibration pro-
cess normally uses multivariate statistical procedures and
computer algorithms on a large number of sample spectra with
their ­corresponding reference data to arrive at an appropriate
calibration model for predicting analyte concentrations. It is
important for an analyst who uses NIR to understand the per-
formance and suitability of the procedure. The calibration
model then needs to be tested or validated either on indepen-
dent samples or cross-validated on randomly selected samples
from the calibration set.
4. Mix well each time before use as the mixture tends to be het-
erogeneous. Use this mixture within 1 year.
5. Dissolve 134 g of sodium phosphate dibasic heptahydrate in
deionized water (800 mL). Add sodium hydroxide pellets
(20 g), and then add deionized water to make 1 L volume.
Mix thoroughly. Use this solution within 3 months.
132 Rosario Jimenez et al.

6. Store the flour in properly labeled coin envelopes at −20 °C

until ready for wet chemical analysis and NIR scanning.
7. These values may vary from one sample attachment module to
another but these values are under the control of the instru-
ment manufacturer. If the instrument response is not within
the specified range, it may be time to replace lamp.
8. If current error is equal to or more than twice the suggested
error, change K and Phi values. The K and the Phi values are
coefficients for linearizing the spectra and for aligning the
wavelengths.
9. If the average RMS(C) is not within the specified range, room
conditions (i.e., temperature and humidity) may not be stable,
record and monitor room conditions to establish their specific
ranges where the average RMS(C) is less than 20.
10. The current analysis is compared to previous values stored in
the ISISTDSF.CHK file. The “T” value is computed based on
how far the current values are from the mean of the previous
values. If check cell analysis fails (e.g., T > 3), an error message
will appear to warn that one or more analyses have changed.
This could be due to one of the following:
(a) a module has been changed;
(b) a repair has been done;
(c) the check cell may be out of date.
If it is due to the any of the first two, there is a need to re-­
standardize the NIRS. If check cell is out of date, replace the
current check cell with a new one.
11. The Chk.NIR file is helpful in the calibration procedure to
minimize the effects of variation in temperature and instru-
ment optics and to make the calibration model insensitive to
these changes during analysis.
12. Different sizes of inserts can be used if the amount of flour
sample is not enough to fill the standard sample cup. For flour
samples
(a) >2 g: use sample cup with no insert
(b) >500 mg–2 g: use sample cup with size 1 insert (18 mm
hole, SI1)
(c) >200–500 mg: use sample cup with size 2 insert* (12 mm
hole, SI2)
(d) >40–200 mg: use sample cup with size 3 insert (6 mm
hole, SI3)
*This is the recommended size of insert to use for scanning
of brown rice flour prepared from 2 to 3 g paddy rice.
Separate files have to be created for samples that were charac-
terized using samples cups with different inserts. The amount
 Method Development of Near-Infrared Spectroscopy Approaches for Nondestructive… 133

of sample available and the size of the ­“window” of the insert

will affect the resulting NIR ­spectrum. Therefore separate
calibrations are performed for samples characterized in samples
cups equipped with different inserts.
13. For the NIRS calibration for protein, the accepted and recog-
nized Kjeldahl method for the wet chemical analysis of protein
in brown rice flour samples is reported by Bietz [46].
14. Dissolve of sodium salicylate (150 g) in about 600 mL of
deionized water and add 0.3 g of sodium nitroprusside to the
solution. Filter the solution through a fast filter paper into a
1 L volumetric flask and make the volume to 1 L. Add about
1 mL of 30% Brij-35 and mix thoroughly. Use this solution
within 1 month.
15. Dilute 10.0 mL of any commercial bleach solution (containing
at least 5.25% available chlorine) to 100 mL with deionized
water. Add two drops of 30% Brij-35 and mix thoroughly.
Prepare fresh daily.
16. A large number of samples with nonuniform distribution
(i.e., the PC bins are not equally represented) in the calibra-
tion set may lead to poor prediction models due to increased
risk of involving noise in the calibration and bias toward the
mean or toward the range with dense population [41].
Representative and evenly-distributed samples should be
selected to cover the entire range of component of interest,
such as PC in this case.

Acknowledgments

The authors thank the following GQNSL staff for the technical
assistance in generating the protein data: Jose Rosales, Edgar
Amoloza, Leonel Borebor, Anna Carissa Basilio, and Ruben
Chavez. This work has been supported under the CGIAR thematic
area Global Rice Agri-Food System CRP, RICE, Stress-Tolerant
Rice for Africa and South Asia (STRASA) Phase III, and Australian
Centre for International Agricultural Research (Project ID
CIM/2016/046) funding.

References
1. Cuevas RP, Pede VO, McKinley J, Velarde O, Melting the secrets of gelatinisation tempera-
Demont M (2016) Rice grain quality and con- ture in rice. Func Plant Biol 37:439–447
sumer preferences: a case study of two rural 3. Cagampang GB, Perez CM, Juliano BO
towns in the Philippines. PLoS One (1973) A gel consistency test for eating quality
11(3):e0150345 of rice. J Sci Food Agric 24:1589–1594
2. Cuevas RP, Daygon VD, Corpuz HM, Reinke 4. Juliano BO, Perez CM, Resurreccion AP
RF, Waters DLE, Fitzgerald MA (2010) (2009) Apparent amylose content and gelatini-
134 Rosario Jimenez et al.

zation temperature types of Philippine rice 17. Martens H, Martens M (2001) What makes
accessions in the IRRI Gene Bank. Phil Agric NIR data so information-rich? In: Martens H,
Sci 92(1):106–109 Martens M (eds) Multivariate analysis of qual-
5. Tanaka J, Hayashi T, Iwata H (2016) A practi- ity: an introduction. Wiley, Chichester,
cal, rapid generation-advancement system for pp 415–417
rice breeding using simplified biotron breeding 18. Infrasoft International (1995) NIRS 2 Version
system. Breed Sci 66(4):542–551 3.10 Routine Operation and Calibration
6. Janwan M, Sreewongchai T, Sripichitt P (2013) Development Software for Near Infrared
Rice breeding for high yield by advanced single Instruments, State College PA, USA
seed descent method of selection. J Plant Sci 19. Villareal CP, Dela Cruz NM, Juliano BO
8(1):24–30 (1994) Rice amylose analysis by near-infared
7. Spindel J, Begum H, Akdemir D, Virk P, transmittance spectroscopy. Cereal Chem
Collard BCY, Redoña E, Atlin G, Jannink J-L, 71(3):292–296
McCouch SR (2015) Genomic selection and 20. Delwiche SR, Bean MM, Miller RE, Webb
association mapping in rice (Oryza sativa): BD, Williams PC (1995) Apparent amylose
effect of trait genetic architecture, training content of milled rice by near-infrared reflec-
population composition, marker number and tance spectroscopy. Cereal Chem 72(2):
statistical model on accuracy of rice genomic 182–187
selection in elite, tropical rice breeding lines. 21. Delwiche SR, McKenzie KS, Webb BD (1996)
PLoS Genet 11(2):e1004982 Quality characteristics in rice by near-infrared
8. Varshney RK, Terauchi R, McCouch SR reflectance analysis of whole-grain milled sam-
(2014) Harvesting the promising fruits of ples. Cereal Chem 73(2):257–263
genomics: applying genome sequencing tech- 22. Rahim HA, Ibrahim S (2013) Using near-­
nologies to crop breeding. PLoS Biol infrared spectroscopy to investigate the amy-
12(6):e1001883 lose content in rice. Appl Mech Mater
9. Rahim HA, Ibrahim S (2013) Near infrared 239–240:163–166
spectroscopy measurement: the assessement of 23. Lee H-S, Choi Y-M, Lee Y-Y, Ma K-H, Gwag
amylose content in rice grain. Sensors J-G, Lee JR, Yoon Y-T, Cho Y-G, Lee S-Y
Transducers 156:251–258 (2014) Selecting high amylose rice germplasm
10. Bokobza L (1998) Near infrared spectroscopy. combined with NIR spectroscopy at the RDA
J Near Infrared Spectrosc 6:3–17 genebank conserved. Plant Breed Biotechnol
11. Pasquini C (2003) Near infrared spectroscopy: 2(4):380–385
fundamentals, practical aspects and analytical 24. Bao J-S, Cai YZ, Corke H (2001) Prediction of
applications. J Brazilian Chem Soc 14(2): rice starch quality parameters by near-­infrared
198–219 reflectance spectroscopy. J Food Sci 66(7):
12. Agelet LE, Hurburgh CR Jr (2014) Limitations 936–939
and current applications of Near Infrared 25. Shu Q, Wu D, Xia Y, Gao M, McClung AM
Spectroscopy for single seed analysis. Talanta (1999) Apparent amylose content of rice by
121:288–299 near infrared reflectance analysis of ground
13. Roggo Y, Chalus P, Maurer L, Lema-Martinez milled samples. Chinese J Rice Sci 13(3):
C, Edmond A, Jent N (2007) A review of near 189–192
infrared spectroscopy and chemometrics in 26. Wu JG, Shi CH (2007) Calibration model
pharmaceutical technologies. J Pharma Biomed optimization for rice cooking characteristics by
Anal 44(3):683–700 near infrared reflectance spectroscopy (NIRS).
14. Luypaert J, Massart DL, Heyden YV (2007) Food Chem 103(3):1054–1061
Near-infrared spectroscopy applications in 27. Rash J, Meullenet J-FC (2010) Apparent amy-
pharmaceutical analysis. Talanta 72(3): lose content prediction using near infrared
865–883 spectroscopy of individual and bulk rice ker-
15. Butz P, Hofmann C, Tauscher B (2005) Recent nels. AAES Res Ser:312–321
developments in noninvasive techniques for 28. Martin M, Fitzgerald MA (2002) Proteins in
fresh fruit and vegetable internal quality analy- rice grains influence cooking properties.
sis. J Food Sci 70:R131–R141 J Cereal Sci 36(3):285–294
16. Givens DI, De Boever JL, Deaville ER (1997) 29. Juliano BO (2003) Rice chemistry and quality.
The principles, practices and some future appli- Philippine Rice Research Institute, Munoz,
cations of near infrared spectroscopy for pre- p 492
dicting the nutritive value of foods for animals 30. Gomez KA (1979) Effect of environment on
and humans. Nutr Res Rev 10:83–114 protein and amylose content of rice. In:
 Method Development of Near-Infrared Spectroscopy Approaches for Nondestructive… 135

Workshop on chemical aspects of rice grain 38. Yoshida S, Forno DA, Cock JH, Gomez KA
quality. Rice Grain Res, pp 59–68 (1976) Laboratory manual for physiological
31. Shi C, Zhu J, Yang X, Yu Y, Wu J (1999) studies of rice, 3rd edn. International Rice
Genetic analysis for protein content in indica Research Institute, Los Banos
rice. Euphytica 107:135–140 39. Searle PL (1984) The berthelot or indophe-
32. Osborne BG (2006) Applications of near infra- nol reaction and its use in the analytical
red spectroscopy in quality screening of early- chemistry of nitrogen. A review. Analyst
generation material in cereal breeding 109:549–568
programmes. J Near Infrared Spectrosc 14: 40. Holtzhauer M (2006) Basic methods for the
93–101 biochemical lab, 1st edn. Springer, Berlin
33. Williams PC (1975) Application of near infra- 41. Cao N (2013) Calibration optimization and
red reflectance spectroscopy to analysis of efficiency in near infrared spectroscopy. Iowa
cereal grains and oilseeds. Cereal Chem 52: State University, Ames, p 184
561–576 42. Shenk JS, Westerhaus MO (1991) Population
34. Kawamura S, Natsuga M, Itoh K (1999) structuring of near infrared spectra and modi-
Determination of undried rough rice constitu- fied partial least squares regression. Crop Sci
ent content using near-infrared transmission 31(6):1548–1555
spectroscopy. Trans Am Soc Agric Eng 42(3): 43. Baillères H, Davrieux F, Ham-Pichavant F
813–818 (2002) Near infrared analysis as a tool for rapid
35. Sohn M, Barton FE, McClung AM, screening of some major wood characteristics in a
Champagne ET (2004) Near-infrared spec- eucalyptus breeding program. Ann For Sci
troscopy for determination of protein and amy- 59:479–490
lose in rice flour through use of derivatives. 44. Agelet LE, Hurburgh CR Jr (2010) A tutorial
Cereal Chem 81(3):341–344 on near infrared spectroscopy and its calibra-
36. Nørgaard L, Lagerholm M, Westerhaus MO tion. Crit Rev Anal Chem 40:246–260
(2013) Artificial neural networks and near 45. Kays SE, Barton FE II, Windham WR (2000)
infrared spectroscopy: a case study on protein Predicting protein content by near infrared
content in whole wheat grain. FOSS, Hillerød, reflectance spectroscopy in diverse cereal food
Denmark products. J Near Infrared Spectrosc 8:35–43
37. Wang H, Peng J, Xie C, Bao Y, He Y (2015) 46. Bietz JA (1974) Micro-Kjeldahl analysis by an
Fruit quality evaluation using spectroscopy improved automated ammonia determination
technology: a review. Sensors 15: following manual digestion. Anal Chem 46:
11889–11927 1617–1618
 Chapter 8

Multi-Dimensional Cooking Quality Classification Using

Routine Quality Evaluation Methods
Lilia Molina, Rosario Jimenez, Nese Sreenivasulu,
and Rosa Paula O. Cuevas

Abstract
A battery of assays to characterize the cooking and eating attributes of rice have been in routine use for
several decades. The classification system to group rice varieties into different quality types are often based
on cooking and eating attributes defined based on amylose content, rather than being considered a set of
attributes contributing to an overall quality type based on multi-dimensional approach. In this chapter, the
methods developed to measure the cooking quality attributes of rice are described. Instead of considering
each attribute on its own, the authors employ multidimensional data generated from the estimation of
amylose content, gel consistency, gelatinization temperature, Rapid Visco-Analyzer parameters to classify
rice into distinct cooking quality ideotypes. If used universally, such an approach can improve prediction
of cooking quality classifications of rice varieties in the breeding programs.

Key words Quality evaluation, Amylose content, Gelatinization temperature, Rapid Visco-Analyzer,
Multivariate analyses

1 Introduction

Rice grain quality evaluation programs have been dependent

largely on three indicators: amylose content (AC), gel consistency
(GC), and gelatinization temperature (GT). Eventually, the Rapid
Visco-Analyzer (RVA) also became a routine part of grain quality
analyses [1] as it can provide information on samples’ cooking
behaviors in conditions simulating real-life processing conditions
[2]. Through these indicators, rice varieties could be grouped into
distinct cooking quality classes on a per-trait approach. This
approach toward classification has always been based on ranges for
each indicator: AC classes, for example, are as follows: waxy (0–2%),
very low (3–9%), low (10–19%), intermediate (20–25%), and high
(> 25%); GC of cooked rice pastes, on the other hand, can be used
to group high-AC varieties into hard (gel length < 40 mm),
medium (gel length 41–60 mm), and soft (gel length > 61 mm);

Nese Sreenivasulu (ed.), Rice Grain Quality: Methods and Protocols, Methods in Molecular Biology, vol. 1892,
https://doi.org/10.1007/978-1-4939-8914-0_8, © Springer Science+Business Media, LLC, part of Springer Nature 2019

137
138 Lilia Molina et al.

rice samples may have high (>74 °C), intermediate (70–74 °C), or
low (<70 °C) GT [3–7]. Moreover, the different classes of AC,
GT, and GC have been associated with different alleles. These
alleles can potentially be developed further into molecular markers
that can be used for routine marker-assisted screening for predict-
ing cooking quality [8, 9]. Just using established AC, GT, and GC
classes, it is possible to define 45 quality classes, assuming that each
AC-GT combination has three classes of GC. However, consider-
ing that GC is primarily designed as a secondary screening tool for
high-AC samples, there are possibly 21 quality classes. RVA param-
eters [10] are not even considered yet; if RVA metrics are added to
the classification, the number of quality classes will inevitably
increase into impractical numbers, with many of these quality
classes not represented by samples. The per-trait approach, there-
fore, is not appropriate because (1) it paints a rather disjointed
picture of cooking quality; (2) it makes classification of rice samples
and development of product profiles complex; and (3) the advent
of post-genomics tools [11] and the growing dependence of breed-
ing programs on genotyping technologies [12, 13] requires a deep
understanding of overall cooking quality, particularly at the pheno-
typic level.
An alternative approach is to characterize cooking quality attri-
butes using all available quantitative values. In this multivariate
approach, the traditional quality classes, based on individual attri-
butes, are eschewed; rather, the metrics that predict the cooking
quality of a sample are considered simultaneously. Through this
means, it is predicted that defining quality classes will become
more grounded in statistics and quality classes that reflect upon
real sample behavior will be formulated.
In this book chapter, methods routinely used to evaluate grain
quality of cooked rice are described. For each method, appropriate
rice reference materials and intermediate checks are incorporated.
The reference materials are traceable to SI units of measurement or
to certified reference materials, against which internal reference
samples and working standards are checked. Meanwhile, to dem-
onstrate the value of the holistic approach of characterizing rice
varieties using all available metrics simultaneously, a case study
involving a multivariate approach has been presented to distinguish
cooking quality ideotypes.

2 Materials

2.1 Sample 1. Paddy rice grains.

Preparation 2. Drying oven.
3. Dehuller: Satake, Model THU-35A, Satake Corp., Hiroshima,
Japan.
 Multi-Dimensional Cooking Quality Classification Using Routine Quality Evaluation… 139

4. Mill (depends on sample size): Grainman Model 60-220-60-­

DT and 60-230-60-24 T (Grain Machinery Mfg. Corp,
Miami, FL, USA) or Kett mill (Kett Electric Laboratory,
Tokyo, Japan).
5. Grinder: Udy cyclone mill, Model 3010-019 with sieve mesh
size 60 or 0.25 mm (Udy Corporation, Fort Collins, CO, USA).

2.2 Apparent 1. Rice flour (prepared as per Subheading 3.1).

Amylose Content (AC) 2. Milled rice reference materials (flour of IR65, mixed IR65/IR24,
Determination IR24, IR64, and IR8; preparation from Subheading 3.1).
3. Spatula.
4. Weighing paper.
5. Analytical balance (0.1 mg accuracy): Mettler Toledo Model
ML 204/02, Switzerland.
6. Volumetric flask (capacities: 100 mL and 1 L).
7. Aluminum foil.
8. Magnetic stirrer: Cole Parmer Model 4810, USA.
9. Plastic carboy container, 10 L.
10. Pipettes (capacities: 5 mL, 10 mL).
11. Test tubes, 20 mL.
12. Water bath 95°C: Blue M Model MW-1140C-1, USA.
13. Skalar San++ System Segmented Flow Analyser (SFA): Model
3000 (Skalar Analytical B.V. 4800 De Breda, The Netherlands).
It consists of an autosampler, an amylose chemistry unit (man-
ifold, proportioning pump, and colorimeter (with 620 nm
filter)), an interface, and a computer for data processing
­
(see Note 1, Fig. 1).

Fig. 1 The San++ SFA set-up with three components: (a) autosampler; (b) amylose chemistry unit; (c) interface,
and (d) computer for data processing
140 Lilia Molina et al.

14. Deionized water.

15. Sodium hydroxide (NaOH), AR Grade.
16. Glacial acetic acid (CH3COOH), AR Grade.
17. Potassium iodide (KI), AR Grade.
18. Iodine (I2), AR Grade.
19. Ethanol (EtOH), technical grade.

2.3 Measurement 1. Rice flour (prepared as per Subheading 3.1).

of Gelatinization 2. In-house quality control checks (e.g., flour of IR 36, and IR
Temperature 42, prepared as per Subheading 3.1).
by Differential
3. Deionized water.
Scanning Calorimetry
4. Differential scanning calorimeter (DSC): TA Instruments
Q100, USA (see Note 2).
5. TA Instruments Universal Analysis software.
6. Analytical balance: Mettler Toledo Model AG 204, Switzerland.
7. Encapsulating press: TA Instruments.
8. Hermetic seal pre-forming tool/crimping die: TA Instruments,
USA.
9. Pipettor, 10 μL.
10. Spatula.
11. Hermetic aluminum pans: TA Instruments, USA (see Note 3).
12. Hermetic aluminum lids: TA Instruments, USA (see Note 4).
13. Nitrogen gas, HP and industrial grade.
14. Indium standard.

2.4 Measurement 1. Rice flour (prepared as per Subheading 3.1).

of Gel Consistency 2. In-house reference materials (flour of IR42, IR32, and IR48,
prepared as per Subheading 3.1).
3. Deionized water.
4. Top-loading balance: Mettler Toledo Model AG 2014,
Switzerland.
5. Adjustable-volume dispenser bottle.
6. Culture tubes (13-mm OD × 100-mm length).
7. Vortex mixer: Cole Parmer Model 4721-00, USA.
8. Hot-water bath: Julabo Model SW23, Germany.
9. Cold-water bath: Julabo Model 33, Germany.
10. Leveled reading board lined with millimeter graphing paper.
11. Thymol blue (0.025% in 95% Ethanol).
12. 0.2 M KOH.
 Multi-Dimensional Cooking Quality Classification Using Routine Quality Evaluation… 141

2.5 Characterization 1. Rice flour (from Subheading 3.1).

of Rice Viscosity 2. In-house reference material (IR32 flour, prepared as per
Profiles Using RVA Subheading 3.1).
3. Deionized water.
4. Top-loading balance: Mettler Toledo Model PB1502-S,
Switzerland.
5. Adjustable-volume dispenser bottle.
6. Pasteur pipette.
7. Rapid Visco-Analyzer Series 4 (RVA-4), Newport Scientific,
USA.
8. RVA-4 canisters and paddles.
9. Thermocline for Windows software, Analysis Tool for RVA,
Newport Scientific v. 2.6.
10. Cooling water bath: Julabo Model F12, Germany.
11. Viscosity Reference Standard, Product Code S2000S, Paragon
Scientific Ltd., UK.

3 Methods

3.1 Flour Preparation 1. Gradually air-dry the paddy grains in the laboratory to bring
down moisture content to 14%.
2. Remove the hull from paddy rice grains using a dehuller.
3. Mill the unpolished rice to remove the bran (see Note 5).
4. Select a test portion of grains (20–50 grains) (see Note 6).
5. Grind the rice grains to flour using the Udy mill to pass a
0.25-mm mesh.
6. Allow the rice flour to equilibrate to room temperature before
the assays.

3.2 Determination This procedure is based on published ISO methods [14, 15]. It has
of Apparent Amylose been modified and adapted for routine operations using the Skalar
Content Via Iodine San++ SFA system.
Colorimetry

3.2.1 Preparation 1. 1 N NaOH. Weigh NaOH pellets (400 g) into a plastic carboy
of Reagents containing about 9 L deionized water and mix using a mag-
netic stirrer for 5 min. Remove the magnetic bar and make up
the volume with deionized water to 10 L.
2. 1 N CH3COOH. In a 1-L volumetric flask, add 57.2 mL
(60 g) of glacial CH3COOH to 500 mL deionized water.
Make up the volume to 1-L using deionized water.
3. (2%:0.2%) KI-I2 stock solution. Dissolve KI (20 g) in 500 mL
deionized water, by stirring, in a 1-L volumetric flask. Add I2
142 Lilia Molina et al.

(2 g) and mix until it is completely dissolved. Make up the

volume to 1L with deionized water (see Note 7).
4. KI-I2 + CH3COOH solution. Mix 1 N CH3COOH (100 mL)
with KI-I2 stock solution (300 mL) in a 1L volumetric flask.
Make up the volume to 1-L with deionized water.
5. 0.09 N NaOH. Add of 1 N NaOH (9 mL) to deionized water
and make up the volume to 100mL with deionized water. Use
a 100-mL volumetric flask.

3.2.2 Gelatinization This method is applicable for both sample and reference flour
of Rice Flour materials. See Note 8 for special instructions.
1. Turn the water bath on and heat to 95 °C.
2. Add flour (100 ± 0.5 mg) in a 100-mL volumetric flask.
3. Add 95% ethanol (1 mL) into the flask.
4. Add of 1 N NaOH (9 mL) into the flask.
5. Heat the flask for 10 min in the water bath to gelatinize the
starch.
6. Remove the flask from heat and cool at room temperature for
at least 2 h (or up to 16 h or overnight).
7. Make up the volume to 100 mL with deionized water.

3.2.3 Preparing 1. Selected IRRI-released rice varieties are used as reference

a Standard Curve Using materials in constructing the standard curve for AC determi-
Rice Flour Standards nation: IR 65 (waxy), IR 65/24 (very low AC), IR 24 (low
AC), IR 64 (intermediate AC), and IR 8 (high AC). The AC
of these samples were determined following the ISO 6647-1:
2015 Reference method [14]. The Skalar San++ Segmented
Flow Analyzer (wavelength = 620 nm) was used to determine
the absorbance of the reference materials (see Subheading
3.2.4).

3.2.4 Iodine Absorbance Detailed instructions on how to operate the SFA can be found in
Measurements of Samples the instrument’s manual.
by SFA
1. Set up the SFA (Fig. 1) and warm up the colorimeter for at
least 30 min. Use the following conditions:
(a) Filter wavelength: 620 nm.
(b) Sample injection rate: 0.42 mL/min.
(c) I2-CH3COOH injection rate: 0.60 mL/min.
(d) Flow cell: 10 mm.
2. Dip all the reagent tubings in deionized water.
3. Set up the pump in the amylose chemistry unit.
4. Place assigned reagent tubings into their corresponding
reagents:
 Multi-Dimensional Cooking Quality Classification Using Routine Quality Evaluation… 143

(a) KI-I2-CH3COOH.
(b) Deionized water.
(c) 0.09 N NaOH.
5. Let the system run for 15 min and make sure that bubbles are
discretely segmented.
6. Monitor the baseline until a good reagent baseline is achieved.
7. Pour samples (from the overnight standing, Subheading
3.2.2) and internal reference materials into individual sample
cups and arrange them as per the pre-prepared sample sequence
on the autosampler.
8. Start sample injections into the SFA.
9. After the sequence run, flush the system with deionized water
for 15 min.
10. Switch off the interface, the autosampler, and the chemistry
unit.

3.2.5 Data Processing The Flow Access Software automatically calculates the ACs of the
samples based on their absorbance values (indicated by peak
height) and on the standard curve generated based on the refer-
ence materials.
1. Using the Flow Access Software of the Skalar San++ Analyzer,
in the data processing window, click on the file for
processing.
2. Click on the “Realtime” tab to view the chart. In this screen,
manually correct the baseline as necessary. Click on “Results”
tab to see corrected peak height values and ACs.
3. Click on “Options” and “Export” to “Excel format” and save
the file.

3.3 Measurement This method is applicable for both sample and reference flour
of Gelatinization materials. In preparing the samples for DSC, use metal forceps to
Temperature handle the pans and the lids. These items may be too small for
by Differential hands to handle. Also, the forceps are used for handling to avoid
Scanning Calorimetry contaminating the pans and the lids. Instructions on operating the
sealing equipment can be found in the manual provided by the
3.3.1 Rice Sample
manufacturer.
Preparation
1. Weigh flour (4.0 ± 0.1 mg) onto the center of the aluminum
pan.
2. Add Milli-Q-filtered water (8 μL) onto the flour (see Note 9).
3. Put the lid on the pan with the concave side up.
4. Seal the pan and lid combination hermetically using the forming
tool and the encapsulating press.
5. Place the hermetically sealed pans onto the DSC autosampler.
Add the sample identifier (name or code) into the DSC’s
144 Lilia Molina et al.

software as soon as the sample is positioned in the autosampler.

A typical sequence run will have the following sequence:
Pan No. 1 = Indium Standard, use as calibrant.
Pan No. 2–3 = Reference Materials (IR 36 and IR 42 rice
flour).
Pan No. 4 onwards = samples.
6. Allow the first sample to equilibrate for 30 min (to allow the
water to be fully absorbed by the flour) before starting the
experiment.

3.3.2 Differential Instructions on how to use the software are found in the instru-
Scanning Calorimetry ment’s manual. The DSC is kept switched on to maintain the tem-
perature within the DSC cell above ambient temperature. It is
NOT switched off, unless it will not be used for an extended length
of time. Ensure that the instrument is calibrated prior to sample
analyses (see Note 10). The method used is based on TA- Q100
Differential Scanning Calorimetry Manual- TA Instruments [16].
1. Allow N2 gas to flow through the system.
2. Turn on the refrigerated cooling system (RCS).
3. Set up sequence runs with the following conditions using
the TA Thermal Advantage Instrument Control software
(see Note 11):
For indium only:
Sample size: 3.000–5.000 mg.
Heating Ramp Rate: 10.0 °C/min.
Final Temperature: 180.0 °C.
Load and Unload Temperature: 25.0–35.0 °C.
For rice flour samples only:
Sample size: 4.0 mg.
Heating Ramp Rate: 10 °C/min.
Final Temperature: 120.0 °C.
Load and Unload Temperature: 25.0–35.0 °C
4. Start the sequence run.
5. At the end of the sequence run, turn off the RCS and wait
until the flange temperature has returned to ambient.
6. Stop the flow of the N2 gas.

3.3.3 Data Processing Instructions on how to use the software for data processing are
found in the instrument’s manual.
1. Load the TA Universal Analysis software.
2. To view the results, open the sample file and set the y-axis to
heat flow and the x-axis to temperature.
 Multi-Dimensional Cooking Quality Classification Using Routine Quality Evaluation… 145

3. Set the left and right baselines of the melting curve and deter-
mine, using the software, the following parameters: onset
temperature (To), peak temperature (Tp), and enthalpy (ΔH).

3.4 Measurement This method is based on Cagampang et al. [3]. See Note 12 for
of Gel Consistency limitations to its application.

3.4.1 Preparation 1. Thymol blue (0.025% in 95% ethanol): Dissolve thymol blue
of Reagents indicator (250 ± 10 mg) in 95% ethanol in a 1-L volumetric
flask. Mix then add ethanol to make up the volume to 1 L.
2. 0.2 M KOH: Dissolve KOH (12.76 ± 0.01 g, ACS, JT Baker,
87.7%) in deionized water in a 1-L volumetric flask. Mix and
add deionized water to make up the volume to 1L.

3.4.2 Procedure 1. Turn water bath ON and heat to ≥90 °C. Turn ON cooling
bath to ≤10 °C.
2. Weigh flour of the sample (100.0 ± 0.1 mg) into a culture
tube and place the tube in a test tube rack. Each test tube rack
is a sample batch. Prepare duplicate batch.
3. Weigh flour of (100.0 ± 0.1 mg) in a culture tube for each of
the reference materials. Assign three different reference mate-
rial samples for each batch. Prepare the reference tubes in
duplicate.
4. Add the 0.025% thymol blue solution (200 ± 8 μL) to each of
the tubes.
5. Add the 0.2 M KOH (2.0 ± 0.1 mL) to each of the tubes and
mix thoroughly.
6. Place an empty steel test tube rack (with side handles) in the
boiling water bath.
7. Place the freshly mixed samples (from one batch) in the steel
test tube rack (in the boiling water bath) and cover the tubes
with glass marbles. The water should not reach half the height
of the culture tubes to prevent the boiling water from spilling
into the samples.
8. Heat the tubes for 8 min. Immediately start the timer as soon
as tubes (i.e., usually a set of four tubes) are in the boiling
water bath.
9. Occasionally lift the rack out of the water to prevent the
contents of the tubes from boiling over.
10. Allow the samples to cool at room temperature for 5 min.
Immediately start the timer once the first tube or set of tubes
are placed at room temperature.
11. Cool the tubes further in an ice-water bath (15 min), again
starting the timer as soon as the samples or set of samples are
placed in the ice-water bath.
146 Lilia Molina et al.

12. Lay the tubes horizontally on the leveled reading board lined
with millimeter graphing paper and allow the gel to solidify
(1 h).
13. Measure the length of the blue gel in the horizontally posi-
tioned tubes, using the millimeter graphing paper as basis.

3.4.3 Quality Control Included in the analysis are three different reference materials with
each sample batch. The reference materials are IR42, IR32, and
IR48.

3.5 Viscosity This method is based on the AACC Method 61-02 [17].
Profiling by RVA Instructions on how to use the software and the instrument are in
the instrument’s manual. See Note 13 for instructions about sam-
ple handling.
1. Switch on the water bath (set at 10 °C) and the RVA-4.
2. Load the Thermocline software for Windows (TCW.exe).
3. Use the following conditions for the RVA run:
(a) Temperature ramp up range: 50–95 °C.
(b) Temperature ramp down range: 95–50 °C.
(c) Temperature ramp rate: 11.8 °C/min.
(d) Holding time at 95 °C: 2 min 30 s.
(e) Rotational speed during RVA run: 160 rpm.
4. Weigh a test portion of rice flour (3.00 ± 0.01 g) into a canis-
ter and add deionized water (25 ± 0.01 g). This applies for
samples and for reference materials. The reference materials
must be included in every sample batch.
5. Place a paddle in the canister and use it to suspend the samples
and to disperse lumps.
6. Place the canister onto the sample platform under the tower of
the RVA and insert the paddle on top of the coupling.
7. Lower the tower and start the analysis.
8. Remove the canister and the paddle from the tower after the
run has completed.
9. Data processing: When a test run has been completed, the
TCW software will display the results of the analysis based on
­predefined formulas. The results include the following param-
eters: Peak1 (PV), Trough1 (TV), Breakdown (BD), Final
Viscosity (FV), Setback (SB), Peak Time (Pk_Time), Pasting
temperature (PT), and Retrogradation (also known as lift-off,
LO).

3.5.1 Quality Control Aside from the inclusion of in-house reference material per sample
batch, the calibration of the RVA unit is checked at least once per
month using a reference standard oil of known viscosity.
 Multi-Dimensional Cooking Quality Classification Using Routine Quality Evaluation… 147

4 Applying a Multi-Dimensional Approach in Characterizing Rice Samples:

A Case Study

Ten indica rice lines were grown and harvested in IRRI in the dry
season of 2014. Samples were dehulled, milled, and then ground
to flour. The ground samples were characterized based on the
methodologies described for AC (by iodine colorimetry), GT (by
DSC), GC, and RVA attributes. In total, there are 10 parameters
that were used to characterize the rice lines. Pasting temperature
was excluded from the parameters in the analyses because it is also
an indicator of cooking time, like GT. Instead of putting the rice
lines into quality classes per attribute, the quantitative nature of the
dataset was viewed holistically. This allows categorization of rice
samples beyond the classic system, possibly into more comprehen-
sive quality classes because all available phenotypic data points were
considered. Because the parameters used in this study had different
ranges, the data was pre-processed through scaling; i.e., the data
set was scaled so that for each attribute, the mean = 0 and the
SD = 1. The data was then plotted into spider plots. The shapes of
these plots were then compared.
Results showed that the ten indica rice lines represented the
five AC classes, the low- and the intermediate-GT classes, while the
two high-AC samples belonged to the soft GC class (Table 1). The
rest of the samples had soft GC as well.
The RVA plot points PV, TV, and FV suggest that the samples’
properties are in a continuum, rather than classifiable into discrete
quality classes; however, the derived attributes (SB, LO, and BD)
have indications of groupings. Samples 1–3 had low BD, high LO,
and positive SB. Samples 4–10, however, have negative SB. These
samples can further be grouped based on BD and LO: Samples
4–6 had higher BD and LO than Samples 7–10. Also, it is noted
that Samples 1–6 attained PV at around 5–6 min while Samples
7–10 attained PV at around 4 min.
Plotting scaled values for all these attributes shows that the
indica rice lines could be clustered into three groups based on the
similarities of the spider plots (Fig. 2). The advantage of the spider
plot over other graph types is that spider plots allow simultaneous
multivariate presentation in two dimensions and comparison of the
samples in the dataset, such as in this case study.
It is important to consider the overall shape of the plots and
not just the values of individual attributes, especially in attempts to
understand cooking and eating quality beyond the typical quality
classes. This can then be used as a first step in developing predic-
tion models for cooking and eating quality using quantitative
parameters.
148 Lilia Molina et al.

Table 1
Cooking and eating quality metrics for 10 indica rice lines

AC GT GC PV Peak Time TV FV BD LO SB
Designation (%) (°C) (mm) (cP) (min) (cP) (cP) (cP) (cP) (cP)
Sample 1 26.3 68.68 84 2628 6.1 2088 3894 540 1806 1266
Sample 2 22.9 68.73 70 2727 6.1 2133 4234 594 2101 1507
Sample 3 26.4 68.85 70 3708 6.2 2903 4750 805 1847 1042
Sample 4 9.8 70.25 70 3689 5.3 1417 2076 2272 659 −1613
Sample 5 11.6 70.54 76 3911 5.9 2064 3189 1847 1125 −722
Sample 6 14.1 71.03 84 3756 5.9 1736 2718 2020 982 −1038
Sample 7 2.7 69.97 90 2487 3.7 1217 1616 1270 399 −871
Sample 8 1.3 70.17 100 1911 3.7 981 1200 930 219 −711
Sample 9 3.3 70.25 81 2405 4.0 1472 2087 933 615 −318
Sample 10 1.3 70.33 100 2482 3.9 1326 1733 1156 407 −749

a b c
AC AC
AC 4 4
4
GC 2 PV GC PV
GC PV 2
2
0 0
0
-2 -2
-2 GT TV GT TV
GT TV
-4 -4
-4
-6 -6
-6

LO BD LO BD LO BD

Pk_Time FV Pk_Time FV Pk_Time FV

SB SB SB

Fig. 2 Spider plots indicating the differences among the 10 indica lines categorized into three classes (a, b, c)
based on 11 quantitative grain quality attributes (AC amylose content, PV peak viscosity, TV trough viscosity,
BD breakdown, FV final viscosity, SB setback, Pk_Time time of PV, LO lift-off, GT gelatinization temperature,
and GC gel consistency). Each plot represents one rice line

5 Notes

1. The Skalar San++ SFA Analyzer has been set–up with Amylose
Module 1 and 2 for amylose determination in rice flour by
iodine method.
2. The TA Q100 DSC is composed of a calorimeter and a
RCS. Other DSC models can be used. The method defined
here is specifically targeted for using the Q100 and its
accessories.
Multi-Dimensional Cooking Quality Classification Using Routine Quality Evaluation… 149

3. TA hermetic aluminum pans: P/N 900793.901.

4. TA hermetic aluminum lids: P/N 900794.901.
5. The 20–50 grains selected for grinding have to be well-milled
(i.e., no bran left on the grain surface) and must not be
discolored.
6. The type of mill to use to polish the unpolished samples
depends on the amount of sample available. For small amounts
(<70 g), the Kett mill is typically used. However, for larger
sample amounts (i.e., ≥70 g), the Grainman mill can be used.
The type of mill used has the potential to change the results of
the tests due to the differences in the degree of milling.
7. Iodine (I2) is light-sensitive. Cover the flask used for preparing
the KI-I2 solution with aluminum foil during preparation of
the solution. Dissolution via stirring may take a few
minutes.
8. For steps 3–4, while adding the solutions, tilt the volumetric
flask and slowly put the solutions in such that the solutions do
not fall directly into the flour and cause it to splatter inside
the flask.
9. Make sure that the flour that gets displaced during the process
is scraped back into the middle of the pan. Also, ensure that
the water has been absorbed by the flour and that there is no
water on the side of the pan. Gentle agitation may be neces-
sary to put all displaced material in the middle of the pan.
10. Calibration of the DSC can be conducted each time the N2
tank is replaced or at a pre-set frequency. Refer to the User’s
Manual for the instructions on how to conduct DSC
calibrations.
11. The conditions of DSC runs may be modified to fit one’s
research objectives. Changes in temperature ranges and ramp
rates may affect endotherm formation and must be taken into
account when comparing sample profiles across different
experiments. The Q100 offers a “modulated DSC” option, a
setting in which the temperature ramp is sinusoidal instead of
linear, but still following the ramp rate set for the run.
12. The gel consistency test differentiates varieties of high- and
intermediate-AC in terms of tenderness of cooked rice. It clas-
sifies high-amylose rice into three categories based on gel
length.
13. Right after the sample has been suspended in water, it must be
analyzed in the RVA-4. Therefore, only add water when the
sample is about to be analyzed. Otherwise, keep the canister
covered to minimize changes in moisture content of the flour,
which is hygroscopic.
150 Lilia Molina et al.

Acknowledgments

The authors thank Dennis Villegas, Teodoro Atienza, Leah

Villanueva, Jennine Rose Lapis, Reah Gonzales, Eric Jhon Cruz,
Marnol Santos, Ruben Chavez, and Mitzi Alodia Asih for assis-
tance in processing and collecting data for the samples used in the
case study and in the validation and optimization of the method-
ologies in GQNSL. This work has been supported under the
CGIAR thematic area Global Rice Agri-Food System CRP, RICE,
Stress-Tolerant Rice for Africa and South Asia (STRASA) Phase III
funding.

References

1. Blakeney AB, Lewin L, Reinke RF (2001) 10. Fitzgerald MA, Martin M, Ward RM, Park
Quality rice for North Asia. Rural indus- WD, Shead HJ (2003) Viscosity of rice flour:
tries research and development corporation. a rheological and biological study. J Agri Food
RIRDC, Kingston, ACT, p 34 Chem 51:2295–2299
2. Deffenbaugh LB, Walker CE (1989) 11. Anacleto R, Cuevas RP, Jimenez R, Llorente
Comparison of starch pasting properties in C, Nissila E, Henry RJ, Sreenivasulu N
the brabender viscoamylograph and the rapid (2015) Prospects of breeding high-quality rice
visco-analyser. Cereal Chem 66:493–499 using post-genomic tools. Theor Appl Genet
3. Cagampang GB, Perez CM, Juliano BO 128:1449–1466
(1973) A gel consistency test for eat- 12. Collard BCY, Septiningsih EM, Das SR,
ing quality in rice. J Sci of Food and Agr Carandang J, Pamplona AM, Sanchez DL, Kato
24:1589–1594 Y, Ye G, Reddy JN, Singh US, Iftekharuddaula
4. Kumar I, Khush GS, Juliano BO (1987) KM, Venuprasad R, Vera Cruz CN, Mackill
Genetic analysis of waxy locus in rice DJ, Ismail AM (2013) Developing new flood-­
(Oryza sativa L.). Theor Appl Genet 73: tolerant varieties at the International Rice
481–488 Research Institute (IRRI). SABRAO J Breed
5. Graham R (2002) A proposal for IRRI to Genet 45:42–56
establish a grain quality and nutrition research 13. Gregorio GB, Islam MR, Vergara GV,
center. IRRI Discussion Paper Series No. 44. Thirumeni S (2013) Recent advances in rice
International Rice Research Institute, Los science to design salinity and other abiotic
Banos, p 15 stress tolerant rice varieties. SABRAO J Breed
6. Cuevas RP, Daygon VD, Corpuz HM, Reinke Genet 45:31–41
RF, Waters DLE, Fitzgerald MA (2010) 14. International Organization for Standardization
Melting the secrets of gelatinisation tempera- (2015) ISO 6647-1: 2015-Rice—
ture in rice. Funct Plant Biol 37:439–447 Determination of amylose content—Part 1:
7. Cuevas RP, Fitzgerald MA (2012) Genetic Reference method p 1–4
diversity of rice grain quality. In: Caliskan M 15. International Organization for Standardization
(ed) Genetic diversity in plants. InTech, Rijeka, (2015) ISO 6647-2: 2015–Rice—
pp 285–310 Determination of amylose content—Part 2:
8. Tuaño AP, Perez LM, Padolina TF, Juliano Routine methods p 1–4
BO (2015) Survey of grain quality of 16. TA Instruments (2003) Differential scanning
Philippine farmers’ specialty rices. Phil Agric calorimeter q series getting started guide. TA
Sci 98:446–456 Instruments –Waters LLC, New Castle
9. Zhao X, Zhou L, Ponce K, Ye G (2015) The 17. American Association of Cereal Chemists Inc
usefulness of known genes/QTLs for grain (2000) Approved methods of the American
quality traits in an indica population of diverse Association of Cereal Chemists. American
breeding lines tested using association analysis. Association of Cereal Chemists International,
Rice 8:29 St. Paul, MN
 Chapter 9

Characterization of Mechanical Texture Attributes

of Cooked Milled Rice by Texture Profile Analyses
and Unraveling Viscoelasticity Properties Through
Rheometry
Rosa Paula O. Cuevas, Pawan S. Takhar, and Nese Sreenivasulu

Abstract
The mechanical attributes of cooked rice grains reflected by textural characteristics capture consumers
preferences. Two of these attributes such as hardness and stickiness are typically indicated in grain quality
evaluation programs by the amylose content of rice. However, the association of amylose content with two
other textural attributes such as cohesiveness and springiness remains unknown. Hence, texture profile
analyses play a role in quantifying these mechanical parameters of texture. Rheometry on the other hand
can be utilized to characterize both viscous and elastic properties of rice during cooking in a water-rice
proportion closer to what consumers typically use for cooking. In this chapter, methods for texture profil-
ing and rheometry are presented to capture inferences on cooking quality modeling. Data extracted from
rice texture and viscoelastic properties that go beyond what amylose content can predict, has been deci-
phered through mathematical modeling, which can help predict cooking quality of new rice breeding lines
to improve textural and cooking quality specifications.

Key words Rice, Rheometry, Texture, Texture profile analyses, Viscoelastic properties

1 Introduction

In rice varietal development the cooked grain’s eating quality of

breeding lines is typically assessed by measuring amylose content,
gel consistency, and gelatinization temperature [1]. These three
tests are used extensively all along breeding line selection processes.
Amylose content determination by iodine colorimetry has long
been used routinely as an indicator of hardness and stickiness [2].
These two mechanical attributes of textural characteristics are
reported to heavily influence Asian consumer preferences [3, 4].
Mechanical textural attributes are those that describe deforma-
tion of food caused by biting and chewing forces [5]. However,
other mechanical textural attributes may not be modeled to a­ mylose

Nese Sreenivasulu (ed.), Rice Grain Quality: Methods and Protocols, Methods in Molecular Biology, vol. 1892,
https://doi.org/10.1007/978-1-4939-8914-0_9, © Springer Science+Business Media, LLC, part of Springer Nature 2019

151
152 Rosa Paula O. Cuevas et al.

content. For instance, premium and second-best rice varieties often

of the same amylose content class were reported to be distinguish-
able by a trained group of descriptive sensory profilers [6], implying
that amylose content is not the sole determinant of rice eating qual-
ity. Other components of the rice grain, such as amylose versus
amylopectin proportions, protein [3, 7], non-starch polysaccha-
rides [4, 8, 9], and lipids [10], are reported to contribute to visco-
elastic and textural properties of cooked grains. Thus, by relying
heavily on amylose content and its supplementary test gel consis-
tency [11], there is a probability that current screening methods
provide incomplete information about a rice variety’s textural qual-
ity when cooked. Such incomplete information poses a challenge in
breeding programs to capture the premium cooking quality (e.g.,
[12, 13]). Texture is multi-faceted trait and must be understood by
considering different attributes such as hardness, adhesiveness,
cohesiveness, and springiness. Expanding phenotyping capacity by
measuring other facets of mechanical texture may prove useful in
further understanding rice eating quality and their contributions to
the overall sensory experience to support breeding for improved
cooking quality in rice. Two ways of doing so are by characterizing
the viscoelastic properties of cooked rice flour by rheometry (static
testing), and by measuring compression-­dependent attributes of
cooked rice grains by texture profile analyses (TPA).
TPA is a two-compression-cycle test often performed with a
texture analyzer. It provides a sense of the chewing behavior of
food products. Hence, it is often called a “two-bite test.” The TPA
procedure provides information about four parameters [14, 15]:
●● Hardness (HRD): The maximum force reached during the first
compression (positive value).
●● Adhesiveness (ADH): It indicates the force required to separate
the probe from the base plate after the sample has been
compressed.
●● Cohesiveness (COH): It indicates the capacity of a material to
withstand two consecutive deformations.
●● Springiness (SPR): It measures how much a material has
returned to its original shape after it has been previously com-
pressed and allowed to recover before the next compression.
Rheology is the study of the deformation properties of materi-
als upon application of stress [16]. Rheometry can provide insight
into the textural properties of food materials, which exhibit both
viscous and elastic properties [17]. This is particularly important in
product development for food purposes since consumer satisfaction
is strongly linked with food texture. The viscoelastic properties can
be measured using static or dynamic testing. The static testing
involves measuring stress-relaxation or creep compliance functions
 Rice Rheology 153

by applying constant-strain or -stress to a sample. During dynamic

testing, the samples are oscillated sinusoidally at a constant
frequency and often the temperature of the samples increased dur-
ing cooking [16–18]. Some of these viscoelastic properties are:
●● Storage modulus (G′): A measure of a material’s elastic (i.e.,
solid-like, rigid) behavior under shear stress. It is an indication
of how much energy is stored in a sample as the components
return to their initial configurations.
●● Loss modulus (G″): A measure of a material’s viscous (i.e.,
liquid-­like) behavior under shear stress. It indicates mechanical
energy lost as heat as the structure (e.g., molecular) of the
material is damaged, allowing it to flow.
●● Phase (loss) angle δ (°): The phase shift between the oscillatory
deformation and the material’s response. The angle range is
0°< δ < 90°.
●● Tan (δ). The ratio of G″ to G′. Its value decreases with an
increase in solid-like, elastic behavior.
●● Storage compliance (J′): Like G′, J′ is also a measure of energy
storage in a material. It is the ratio of strain in phase with the
applied sinusoidal stress to the stress [19].
●● Loss compliance (J″): It is a measure of a material’s energy dis-
sipation ability or its liquid like behavior. It is the ratio of strain,
which is 90° out of phase with the applied sinusoidal stress to
the stress.
●● Stress Relaxation Function [G(t)]: Obtained by compressing a
sample at a constant strain, while recording the decay in stress
with time. G(t) is the ratio of stress decay with constant applied
strain.
●● Creep Compliance Function [J(t)]: Obtained by compressing a
sample at a constant stress, while recording the change in strain
with time. J(t) is the ratio of strain as a function of time with
constant applied stress.
The static and dynamic testing provides complementary infor-
mation on viscoelastic characteristics of biopolymers at different
time scales. The static testing measures the viscoelastic characteris-
tics of biopolymers at long-time scales are generally performed
using texture analyzer. The dynamic tests are performed using a
rheometer for gels or liquid foods, and dynamic mechanical ana-
lyzer for solid foods. Some newer dynamic testing equipment can
perform static testing as well.
The dynamic properties are usually more sensitive in detecting
transitions such as glass-transition, gelatinization, gelation, etc.
The static properties are easier to correlate with textural attributes
154 Rosa Paula O. Cuevas et al.

such as stiffness, hardness, sogginess etc., and can be directly

utilized in mathematical models describing mechanical texture
­
change during imbibition of water into rice during cooking. To
obtain complete understanding of polymers behavior in a material,
the data from tests performed at a wide range of time scales need
to be combined [19].
In this chapter, the methodologies for rheometry and texture
profile analyses are described as phenotyping tools to understand
the mechanical attributes of cooked rice and case studies are pre-
sented. Here we demonstrate the value of texture profiles and rhe-
ometric information, which add value as selection tools in rice
breeding pipelines to capture premium cooking properties.
Furthermore, cooking quality modeling will be inferred with the
previously mentioned phenotyping measures. Since it may be
tedious to perform all complementary rheological measurements;
often the mathematical relations converting data from one type of
test to the other type can be utilized. As an example, case study 2
has been presented, where the methodology for converting stor-
age, and loss moduli to the corresponding compliance moduli or
vice-versa; and the stress relaxation function to the creep compli-
ance function or vice-versa has been demonstrated.

2 Materials

2.1 Texture Profile 1. Uncooked polished (milled) rice grains (see Note 1).
Analyses (TPA) 2. Spatula (metal).
3. 50 mL glass test tubes.
4. Marbles or foil (to seal the test tubes during cooking).
5. Water bath set at 50 °C (see Note 2).
6. Water bath set at 100 °C (see Note 3).
7. Milli-Q water.
8. Texture Analyzer (TA-XT Plus, Stable Micro Systems) fitted
with a 5 kg load cell and equipped with a 35 mm diameter
cylinder flat-end probe.
9. 2 kg weight for calibration.

2.2 Rheometry 1. Rice flour.

2. Milli-Q water.
3. Rheometer (Advanced Rheometer 2000, TA Instruments)
equipped with a peltier plate system, and a stainless steel paral-
lel plate geometry (diameter = 40 mm). A water circulation
system is used to control the cooling rate of the rheometer.
4. Beaker.
 Rice Rheology 155

5. Metal spatula.
6. Balance.

3 Methods

The objective in performing TPA (a two-compression test) is to

quantify four mechanical attributes of texture: hardness, adhesive-
ness (or stickiness), cohesiveness, and springiness. It is important
that the cooked rice grains chosen for TPA are not broken or split
and are of the same height.

3.1 Texture Profile In this procedure, three setups per sample (i.e., three test tubes)
Analyses (TPA) are prepared. Hence, there are three cooking replicates.
3.1.1 Rice Sample 1. Milled rice grains that are not broken (25 grains) are placed in
Preparation one test tube.
2. Wash grains thrice with water. The wash water is decanted.
3. Put water with 1:2 (w/v) ratio with rice into the tube.
4. Seal the test tube with a marble or foil, and tie with rubber
band.
5. Place the test tube in the 100 °C water bath and heat for
15 min. During this cooking process the water has been
absorbed by the grains.
6. Transfer the test tube to the 50 °C water bath. Keep it there
until ready to conduct TPA on the cooked grains.

3.1.2 TPA For each cooking replicate, three measurements are performed.
Hence, nine cooked rice grains are selected for TPA per test tube.
In total, nine measurements are performed for each sample (3
measurements per tube × 3 tubes). The Texture Analyzer is cali-
brated once daily before samples are tested (see Note 4). For
instructions on how to use the software controlling the Texture
Analyzer, refer to the instrument manual.
1. Calibrate the Texture Analyzer as per instructions in the
instrument manual.
2. Set the Texture Analyzer parameters to the following values
(see Note 5):
(a) Pretest speed is set to 1 mm s−1.
(b) Test speed and post-test speed are both set to 5 mm s−1.
(c) Target mode is set to Strain.
(d) Strain is set to 90% (this means that the cooked grains are
compressed to 10% of their heights; the software detects
156 Rosa Paula O. Cuevas et al.

the initial heights of the grains and automatically calculates

from these values).
(e) Time is set to 5 s.
(f) Trigger type is set to Autoforce.
(g) Trigger force is set to 3 g (this is the force that is being
recognized by the instrument to indicate that probe has
made contact with the samples).
(h) Tare mode is set to Auto.
3. Select three cooked rice grains of the similar heights (from the
same tube kept at 50 °C) and place them in parallel at the cen-
ter of the base plate such that the grains will all be compressed
in their entirety by the probe.
4. Using the Exponent Lite software to control the instrument,
compress the grains twice. The output graph is a profile com-
posed of two positive peaks and two negative peaks (Fig. 1).
5. Clean the probe and the base plate with a moist cloth. Make
sure that residues from the compressed rice grains have been
thoroughly removed. And then repeat steps 3 and 4.
6. Data preprocessing can be conducted using the Exponent Lite
software. Export the data as .csv files, as per the software man-
ufacturer’s instructions. The pre-processing will provide hard-
ness (HRD, Force 2) and adhesiveness (ADH) values along
with other parameters.
7. To calculate cohesiveness (COH) and springiness (SPR), use
the following formulae (this can be performed as an additional
macro in the software or on a spreadsheet after the data has
been exported) based on Fig. 1:
A4:6
COH = (1)
A1:3
t 4:5
SPR = (2)
t 1:2

3.2 Rheometry In this procedure, three replicates are tested. Hence, at least 3 g of
flour is needed per sample. Each flour-water suspension has to be
3.2.1 Rice Sample
prepared just before the rheometry test as to avoid settling of the
Preparation
flour to the bottom of the beaker.
1. Put 1 g flour in a glass beaker.
2. Add 2 g water.
3. Stir the mixture until flour is thoroughly wetted and sus-
pended; i.e., there are no clumps of flour at the bottom of the
beaker or streaks of dry flour on the side of the beaker.
 1 2 3 4 5 6

2.5
2.0
1.5
Force(kg)
1.0
0.5
0.0
0 10 20 30 40 50 60
Time (s)

Fig. 1 An example of a TPA profile for cooked rice. Six dashed numbered lines divide the profile into different regions. These lines indicate the boundaries of the
different components of the TPA profile. These features of the curve are: (F2) peak force during the first compression (hardness); (A1:3) area of work during the first
compression; (A1:3) area of work during the first compression; (A3:4) are of work during the upstroke after the first compression (adhesiveness); (A4:6) area of work
during the second compression; (t1:2) travel time on the downstroke of the first compression; (t4:5) travel time on the downstroke of the second compression. These
variables are used in calculating the derived values, cohesiveness, and springiness
Rice Rheology
157
158 Rosa Paula O. Cuevas et al.

3.2.2 Rheometry For instructions on how to use the software, please refer to the
instrument’s manual. A rheometry procedure (a “run”) consists of
a heating ramp and a cooling ramp (see Note 6).
1. Start up the Advanced Rheometer 2000 according to the
instrument manufacturer’s instructions. This includes turning
on the air compressor, the circulating water bath, the rheom-
eter, and the computer.
2. Install the parallel plate geometry.
3. Set the run parameters as follows:
(a) Conditioning set to 35 °C.
(b) 
Temperature ramp 1 start temperature set to current
(35 °C).
(c) Temperature ramp 1 end temperature set to 95 °C.
(d) Temperature ramp 2 start temperature set to 95 °C.
(e) Temperature ramp 2 end temperature set to 35 °C.
(f) Ramp rate for both temperature ramps set to 4 °C min−1.
(g) Strain for both temperature ramps set to 1%.
(h) Single frequency set at 1 Hz.
4. After mixing the rice flour-water suspension, pour it onto the
Peltier plate.
5. Lower the geometry up to the geometry gap using the TA
Advantage 2003 software (see Note 7).
6. Remove excess sample sandwiched between the geometry and
the Peltier plate (Fig. 2a).
7. Place water on the geometry and cover the system with the
solvent trap to saturate the air within and prevent the sample
from drying out (Fig. 2b).
8. Using the TA Advantage 2003 software, supply the necessary
information about the sample and the geometry, and then
start the run.
9. After each run, clean the geometry and the Peltier plate using
a moist cloth. Do not scrape or make deep scratches while
cleaning the geometry and the Peltier plate.
10. Repeat steps 1–6 for each replicate of each sample.
11. For each run, values for storage moduli (G′), loss moduli (G″),
and tan (δ), among other parameters are recorded by the soft-
ware during the temperature ramps. Export these into a .csv
file as needed. The graph of the viscoelastic profiles is in Fig. 3.
12. Data processing steps and correlation analysis have to be
outlined.
 Rice Rheology 159

Fig. 2 Proper sample preparation for rheometry using a parallel plate geometry and a peltier plate. (a) How an
adequate amount of sample looks like when viewed from the side; (b) Installation of a solvent trap to prevent
evaporation of moisture from the sample during heating

3.3 Case Study 1: The relationships among the routine grain quality indicators
Correlations of TPA (i.e., amylose content, gel consistency, and gelatinization tempera-
and Rheometry ture) and more direct measurements of mechanical textural attri-
Properties butes (TPA) and viscoelastic properties (rheometry) were studied
from Diverse Lines using a subset of diverse collection of indica rice varieties (n = 61).
The contrasting lines depict extreme differences in TPA and rheo-
metric properties (Fig. 4). Characterizing the diversity panels doc-
umented with huge phenotyping variation for TPA and rheometric
properties might be useful to define cooking quality characteristic
features.
The correlation results for the diversity samples indicate that
elevated amylose content is positively correlated with hardness and
also negatively correlated with adhesiveness (Fig. 5). These results
agree with previously published literature (e.g., [20]). Amylose
content is also correlated with springiness and cohesiveness, though
these correlations are weaker than with adhesiveness and with hard-
 160

5000
15000
90

4000
80
Rosa Paula O. Cuevas et al.

3000
70

G’(Pa)
Temp(°c)
60

2000
5000 G”(Pa) 10000
50

1000
40

0
0

0 5 10 15 20 25 30

Time (min)

Fig. 3 A profile of rheometric properties obtained froma 1:2 (w/v) flour-to-water slurry was heated from the 35 °C to 95 °C. The black profile corresponds to the
storage modulus (G′); the red profile is the loss modulus (G″): and the blue profile denotes the temperature throughout the experiment
 Rice Rheology 161

Fig. 4 A comparison of TPA profiles and rheometric properties of contrasting samples in the indica diversity
collection. (a) TPA profiles of a hard variety (red) and a soft variety (black); (b) storage moduli (G′) of two rice
varieties, IR36 (solid) and IR65 (dashed). The blie profile corresponds to the temperature

ness. Amylose content is also positively correlated with storage and

loss modulus [20, 21]; however, this trend was not previously
observed in slurries of high concentration and with inherent amy-
lose [22]. These trends indicate that AC remains to be an interest-
ing parameter in the determination of the cooking and eating
quality of the rice grain. Consumers may negatively associate hard-
ness with adhesiveness. However, Fig. 5 suggests that the relation-
ship may not be strong. This indicates that there may be varieties
162 Rosa Paula O. Cuevas et al.

Fig. 5 Correlation matrix of routinely measured grain quality properties (AC amy-
lose content, GT gelatinization temperature, GC gel consistency, PC protein con-
tent) with TPA parameters (HRD hardness, COH cohesiveness, SPR springiness,
ADH adhesiveness) and rheometric attributes (G′ maximum storage modulus, G″
loss modulus at maximum G′, Tan d tan (δ) at maximum G′)

that are soft but are not necessarily sticky, and vice versa. Adhesiveness
appears to be negatively correlated with the storage modulus, which
is quite intuitive, because materials that are sticky tend to exhibit
attenuated elastic properties. It is rather expected that stickier mate-
rials exhibit enhanced flow properties; however Fig. 5 indicates that
the adhesiveness is negatively correlated with the loss modulus.
Clearly, starchy materials such as rice-water slurries are complex
materials and thus it remains important to capture not simply vis-
cosity properties but also elastic properties.
Protein content negatively correlated with TPA and rheomet-
ric properties with moderate significance. Most notable among
these correlations are protein content correlation with cohesive-
ness and springiness, two textural properties of the cooked rice
grain that have not gained as much attention as hardness. Despite
 Rice Rheology 163

the weaker contributions of protein to texture, it still warrants a

closer look since varieties with similar amylose contents may have
differences in textures (i.e., [6]) with proteins as a discriminating
factor. It is also interesting to note that interactions between amy-
lose versus amylopectin and amylose to protein ratio may also con-
tribute to unique textural and rheological behaviors that make
indica varieties distinctive among each other.
These new insights gleaned from diversity panel of 61 samples
show deeper understanding of cooking and eating quality. Although
amylose content is an important attribute and is still worth explor-
ing, other measures such as rheology and texture profile analyses
may comprehend better knowledge of cooking and eating quality,
and provide breeders with new and efficient tools that will allow
them to expand selection criteria to distinguish distinct cooking
quality classes.

3.4 Case Study 2: It is straightforward to convert the storage and loss moduli to the
Conversion of Storage corresponding compliance properties using the complex func-
(G′) and Loss Moduli tions (involving the use of imaginary number, i) [19]. The stor-
(G″) to Storage (J′) age and loss moduli can be denoted using the complex modulus
and Loss Compliance function (G*):
(J″)
G ∗ = G ′ + iG ′′ (3)

Similarly J′ and J″ can be denoted using the complex compli-

ance function (J*):
J ∗ = J ′ − iJ ′′ (4)

The conversion relation between the two complex functions is

[19]:
1
G∗ = (5)
J∗

or
1 J ′ + iJ ′′
G ′ + iG ′′ = = 2 (6)
J ′ − iJ ′′ J ′ + J ′′2

By relating the real and imaginary components on left and

right hand sides of (Eq. 6) we get
J′
G′ =
J ′ + J ′′2
2

(7)
J ′′
G ′′ = 2
J ′ + J ′′2
164 Rosa Paula O. Cuevas et al.

3.5 Conversion The stress relaxation data collected using a texture analyzer can be
of Stress Relaxation described using a mathematical relation. Often the generalized
Function to Creep Maxwell relation is used to describe the stress relaxation data [17].
Compliance Function The stress relaxation function (G(t) is related to stress (σ(t)) in the
material by:
t
dε (τ )
σ (t ) = ∫ G (t − τ ) dτ (8)
0
dτ

where ε(t)is the strain in the material. The above equation repre-
sents a convolution. Assuming that the food is free from stress and
strain before the start of an experiment, the Laplace transformation
of (Eq. 8) gives
σ ( s ) = sG ( s ) ε ( s ) (9)

where the bars over a function denote the corresponding Laplace

transformed function, and s denotes the frequency in Laplace
space.
The creep compliance function (J(t)) is related to strain using
the following relation:
t
dσ (τ )
ε (t ) = ∫ J (t − τ ) dτ (10)
0
dτ

Laplace transformation of (Eq. 10) gives

ε ( s ) = sJ ( s ) σ ( s ) (11)

By substituting ε ( s ) from (Eq. 11) in (Eq. 9) and simplifying,

we get
1
J (s ) = (12)
s 2G ( s )

The inverse Laplace transformation of (Eq. 12) would give the

creep compliance function, J(t). Similarly, if the creep compliance
function (J(t)) were measured using a mechanical test, it can be
converted to the corresponding stress relaxation function (G(t)) by
first performing Laplace transformation of J(t), and then using
(Eq. 12) to obtain (G(s)) and finally performing the inverse Laplace
transformation of G(s) to obtain G(t).

4 Notes

1. Milled, or polished, rice are eaten by most rice consumers.

In preparing milled rice, paddy rice grains are dehulled and
then milled. The milling process removes the outer layers of
 Rice Rheology 165

the rice grain [23]. The methodologies outlined in this chap-

ter can also be applied to brown (unpolished) rice bran. The
rheological and textural profiling data will, expectedly, be dif-
ferent because the layers removed during polishing may affect
transfer of heat and/or water absorption into the rice grains
during cooking [24].
2. The water bath (50 °C) will keep the cooked rice grains from
retrograding. Retrogradation is the development of firmness
of the cooked grain brought about by the formation of net-
works of amylose molecules [25]. With the temperature set
below gelatinization temperatures (based on a survey of highly
diverse samples reported by [26]), rice grains will not get
overcooked. It is important to keep the water bath covered to
keep the air saturated with water vapor; this will prevent the rice
grains from becoming dehydrated before they are subjected
to TPA.
3. The water bath (100 °C) is used to cook rice grains in the test
tubes. If a water bath is not available, a sand bath set at 100 °C
or a beaker with water being heated to 100 °C on a hot plate
can be used instead.
4. To calibrate the Texture Analyzer, follow the instrument
manufacturer’s instructions before compressing any samples.
With the 5 kg load cell, use the 2 kg weight.
5. The settings of the TPA compression test can be pre-­
programmed in a Project. See the instrument’s manual for
instructions on how to do this. Alternatively, there already is a
template for TPA in the Texture Analyzer. Parameters may be
customized as well.
6. The procedure and the geometry can be further customized
by changing the temperature ramp rates, the geometry (e.g.,
use of the starch cell or of the steel cone geometry), the oscil-
lation stress, among other parameters. These changes should
depend on the objective of the study.
7. The software currently being used to operate the Advanced
Rheometer 2000 is the TA Advantage Software 2003
(v.4.0.0). Other rheometer brands may have other proprie-
tary software while more recent models of the Advanced
Rheometer from TA Instruments may have later versions of
the software.

Acknowledgments

We thank Cyril John Domingo and Lucena Samadio for assistance

in obtaining the data used in this chapter. The authors acknowl-
edge the Office of International Programs in College of ACES at
166 Rosa Paula O. Cuevas et al.

University of Illinois, Urbana-Champaign for helping initiate this

collaborative work. This work has been supported under the
CGIAR thematic area Global Rice Agri-Food System CRP, RICE,
Stress-Tolerant Rice for Africa and South Asia (STRASA) Phase III
funding.

References
1. Cuevas RP, Fitzgerald MA (2012) Genetic 11. Cagampang GB, Perez CM, Juliano BO
diversity of rice grain quality. In: Genetic diver- (1973) A gel consistency test for eating quality
sity in plants. InTech, Rijeka, pp 285–310 in rice. J Sci Food Agri 24:1589–1594
2. Juliano BO, Perez CM, Barber S, Blakeney AB, 12. Unnevehr LJ (1986) Consumer demand for
Iwasaki TA, Shibuya N, Keneaster KK, Chung rice grain quality and returns to research for
S, Laignelet B, Launay B, del Mundo AM, quality improvement in Southeast Asia. Am
Suzuki H, Shiki JI, Tsuji S, Yama JT, Tatsumi J Agri Econ 68:634–641
K, Webb BD (1981) International cooperative 13. Vandeplas A, Minten B (2015) Food quality in
comparison of instrument methods for cooked domestic markets of developing economies: a
rice texture. J Texture Stud 12:17–38 comparative study of two countries. Agri Econ
3. Zhou Z, Robards K, Helliwell S, Blanchard C 14. Chen L, Opara UL (2013) Texture measure-
(2002) Ageing of stored rice: changes in chem- ment approaches in fresh and processed foods:
ical and physical attributes. J Cereal Sci a review. Food Res Int 51:823–835
35:65–78 15. Champagne ET, Lyon BG, Min BK, Vinyard
4. Mestres C, Ribeyre F, Pons B, Fallet V, BT, Bett KL, Barton FE, Webb BD, McClung
Matencio F (2011) Sensory texture of cooked AM, Moldenhauer KA, Linscombe S,
rice is rather linked to chemical than to physical McKenzie KS, Kohlwey DE (1998) Effects of
characteristics of raw grain. J Cereal Sci postharvest processing on texture profile analy-
53:81–89 sis of cooked rice. Cereal Chem 75:181–186
5. Szczesniak AS, Brandt MA, Friedman HH 16. Schramm G (2000) A practical approach to
(1963) Development of standard rating scales Rheology and Rheometry. Gebrueder HAAKE
for mechanical parameters of texture and cor- GmbH, Dieselstrasse, p 291
relation between the objective and the sensory 17. Steffe JF (1996) Rheological methods in food
methods of texture evaluation. J Food Sci process engineering. Freeman Press, East
28:397–403 Lansing, p 418
6. Champagne ET, Bett-Garber KL, Fitzgerald 18. Mezger TG (2006) The Rheology handbook:
MA, Grimm CC, Lea J, Ohtsubo KI, Jongdee for users of rotational and oscillatory rheometers,
S, Xie L, Bassinello P, Resurreccion AP, Ahmad vol 299. Vincentz Network GmbH, Hannover
R, Habibi F, Reinke RF (2010) Important sen-
sory properties differentiating premium rice 19. Ferry JD (1980) Viscoeslastic properties of
varieties. Rice 3:270–281 polymers. John Wiley & Sons, New York, p 672
7. Martin M, Fitzgerald MA (2002) Proteins in 20. Biliaderis CG, Juliano BO (1993) Thermal and
rice grains influence cooking properties. mechanical properties of concentrated rice
J Cereal Sci 36:285–294 starch gels of varying composition. Food Chem
48:243–250
8. Das M, Gupta S, Kapoor V, Banerjee R, Bal S
(2008) Enzymatic polishing of rice–a new pro- 21. Li JY, Yeh AI (2001) Relationships between
cessing technology. Food Sci Tech 41: thermal, rheological characteristics and swell-
2079–2084 ing power for various starches. J Food Eng
50:141–148
9. Shibuya N, Iwasaki T (1984) Effect of cell wall
degrading enzymes on the cooking properties 22. Lii CY, Tsai ML, Tseng KH (1996) Effect of
of milled rice and the texture of cooked rice. amylose content on the rheological property of
J Jap Soc Food Sci Tech 31:656–660 rice starch. Cereal Chem 73:415–420
10. Biliaderis CG, Tonogai JR (1991) Influence of 23. Bryant RJ, Dorsch JA, Peterson KL, Rutger
lipids on the thermal and mechanical proper- JN, Raboy V (2005) Phosphorus and mineral
ties of concentrated starch gels. J Agri Food concentrations in whole grain and milled low
Chem 39:833–840 phytic acid (lpa) 1-1 rice. Cereal Chem
82:517–522
 Rice Rheology 167

24. Juliano BO, Bechtel DB (1985) The rice grain 26. Cuevas RP, Daygon VD, Corpuz HM,
and its gross composition in rice: chemistry and Reinke RF, Waters DLE, Fitzgerald MA
technology. Am Assoc Cereal Chem:17–57 (2010) Melting the secrets of gelatinisation
25. Tang MC, Copeland L (2007) Investigation of temperature in rice. Func Plant Biol 37:
starch retrogradation using atomic force 439–447
microscopy. Carb Pol 70:1–7
 Chapter 10

Characterizing Starch Molecular Structure of Rice

Cheng Li, Hongyan Li, and Robert G. Gilbert

Abstract
A better understanding of the nutritional properties of rice starch is important because of the rapid rise of
diet-related health complications, particularly obesity, type 2 diabetes, and colorectal cancers. Rice starch
that is slowly digested to glucose, and where significant quantities of starch which reach the lower gut
(“resistant starch”), can mitigate, and also delay the onset of, these diseases. These digestibility properties
depend to some extent on starch molecular structure. The characterization of this structure is therefore
significant for understanding and developing healthier slower digestible rice. In this chapter, a series of
techniques used for characterizing starch structure are reviewed and the procedure for preparing rice starch
samples with minimum degradation for characterizing starch chain length distribution (CLD) and overall
molecular structure is given. Some methods for choosing or developing plants showing desirable structural
characteristics are briefly summarized.

Key words Rice quality, Starch structure, Size-exclusion chromatography, Chain-length distribution,
Digestibility

1 Introduction

Starch digestibility is one of the important qualities for rice. High

levels of rapidly digestible starch (RDS), defined here as the amount
of starch digested in the first 20 min as measured by the method of
Englyst [1], have been associated with increasing incidence of met-
abolic syndrome, including obesity and diabetes. On the other
hand, rice varieties containing larger amounts of slowly digestible
starch (SDS) and enzyme-resistant starch (RS), which are starches
digested between 20 and 120 min and that not digested after
120 min, respectively, are desirable to prevent and alleviate these
complications [2–4].
Starch is a complex branched homopolymer of glucose, having
a broad distribution of molecular weights and sizes, and with two
major components: amylose and amylopectin. Amylose (weight-­
average molar mass~105–106 Da) is a slightly branched polymer
with a small number of long branches, while amylopectin (weight-­
average molar mass ~107–109 Da) is a highly branched polymer

Nese Sreenivasulu (ed.), Rice Grain Quality: Methods and Protocols, Methods in Molecular Biology, vol. 1892,
https://doi.org/10.1007/978-1-4939-8914-0_10, © Springer Science+Business Media, LLC, part of Springer Nature 2019

169
170 Cheng Li et al.

containing a vast number of short chains. Some starches also

contain an intermediate component, which consist of highly
branched molecules with smaller molecular weights than amylo-
pectin, but similar to that of amylose [5–7]. All these starch com-
ponents have polymer backbones that are linked by α-(1⟶4)
glycosidic bonds; the branch points are formed by α-(1⟶6) gly-
cosidic bonds. Different plant species have different amylose and
amylopectin contents. The normal amylose content of starch is
approximately 15–30%, whereas waxy starch is essentially devoid
of amylose [8]. Amylose contents as high as 50% have been
found in high-amylose starches [6, 9]; note in this context that
the value of “amylose content” depends on the method of mea-
surement [10].
Starch structure can be divided into multiple structural levels
[11]. Individual starch chains (branches) form the first level. The
second level consists of fully branched individual amylose and
amylopectin molecules without any chain alignment. Level 3
describes the conformation of starch molecules, encompassing the
following features in the native unprocessed grain: starch chains
aggregate, entwining into a helical structure; the helices aggregate
to form crystallites; and finally the crystallites form alternating
amorphous and crystalline lamellae. Higher structural levels of
native starch granules comprise growth rings, individual granules,
and association with non-starch components [12].
The first two levels of structure can be significant determinants
of both the higher-level structures and a number of structure-­
property relations for starch-containing substances [13, 14]. For
instance, the first level structure (the starch chain-length distribu-
tion, the CLD) influences starch digestibility: the ratio of short and
long chains in amylopectin has been found to be related to the
amount of SDS [15]. This has two mechanistic reasons. First, the
higher proportion of longer amylopectin chains leads to a greater
amount of retrogradation (the partial crystallization of these lon-
ger chains after some branches have been removed during the
digestion process). Second, the reduction in digestion rate of the
higher proportion of shorter chains can be attributed to an increase
in the number of branch points which inhibit enzymatic hydrolysis
[16]. Both native uncooked and cooked rice starch granules con-
taining a greater amount of chains with degree of polymerization
(DP) in the range 6–12 have increased starch digestions rates,
whereas the opposite trend is observed for longer amylopectin
chains [17, 18]. Furthermore, higher amounts of amylose and
extra-long amylopectin chains can greatly decrease the digestibility
of raw starch granules and gelatinized starch [19].
Different techniques have been used to characterize starch
molecular structure. Size-exclusion chromatography (SEC),
fluorophore-­ assisted carbohydrate electrophoresis (FACE), and
 Characterizing Starch Molecular Structure of Rice 171

high-performance anion-exchange chromatography (HPAEC) are

used to characterize level 1 structure, giving the CLD of starch
molecules. These techniques use enzymatically debranched starch
as the analyte. FACE, which separates molecules based on mass-to-­
charge ratio, is currently the best technique to characterize short
starch branches with a DP up to ~180 [20, 21]. The first part of
the experimental procedure for FACE involves labeling the reduc-
ing end of the short starch branches with a fluorophore (typically
8-amino-1, 3, 6-pyrenetrisulfonic acid, APTS); separation is then
performed using capillary electrophoresis with fluorescence detec-
tion. The labeling has been proven to be independent of starch
chain size, at least up to DP ~135 [22]. The size distribution
derived from FACE is the number distribution (relative number)
of chains of DP X, Nde(X). HPAEC, usually with pulsed ampero-
metric detection (PAD), also provides the number CLD of starch.
Both FACE and HPAEC can only give information on shorter,
largely amylopectin, chains; HPAEC also suffers from a mass bias
which is very laborious to correct. While SEC is able to character-
ize the larger amylose chains, it suffers from problems such as band
broadening, which is the effect where molecules of a given size
“leak” into a range of elution volumes, instead of all eluting at the
same elution volume. SEC can also be used to characterize the
second level of starch structure, with multiple-detection, the num-
ber, weight-average molecular weight, and z-average radius of
gyration distributions [20]. Unfortunately, there is unavoidable
shear scission of the amylopectin molecules in current SEC systems
and therefore it cannot give accurate molecular size characteriza-
tion of the amylopectin component [23]. Multi-angle laser light
scattering (MALLS) without size separation can measure the
weight-average molecular weight and average radius of gyration of
starch; with careful sample handling, there would be no shear scis-
sion (in the absence of size separation). However, neither dynamic
nor static light scattering is able to give reliable size distributions
without size separation [11].
Field-flow fractionation (FFF) is a range of related techniques
that has attracted considerable interest recently in regards to the
characterization of high molecular weight macromolecules. It sep-
arates macromolecules without a stationary phase, so shear scission
is minimal, which gives this technique the potential to reveal the
full size distribution of undegraded native starch. However, with-
out major technical advances, it is not possible to obtain the size
distribution of fully dissolved and undegraded native starch using
FFF techniques, due to problems with obtaining adequate signal
to noise ratio for starch in an optimal eluent [11].
It is essential to be aware that (contrary to what is sometimes
stated incorrectly) SEC separates by molecular size, not by molec-
ular weight. For complex branched polymers such as starch,
172 Cheng Li et al.

there is no relation between molecular weight and molecular size:

­molecules of the same size can have different branching structures
and hence different molecular weights. The size separation param-
eter with SEC is the so-called hydrodynamic volume Vh or the
corresponding hydrodynamic radius Rh. Vh is defined [24] as the
volume of a hypothetical impenetrable sphere displaying in a hydro-
dynamic field the same frictional effect as an actual polymer mole-
cule. For SEC, Vh is a complicated quantity with the dimensions of
molecular size, proportional to the product of the weight-­average
intrinsic viscosity and the number-average molecular weight [25]:
V h = 2 5 η (V h )  M n (V h ) / N A , where NA is Avogadro’s constant.
w

To obtain SEC data which can be reproduced, it is essential to

calibrate with appropriate standards, and either report the results
in terms of molecular size or give the calibration curve; data
reported only in terms of elution time or volume are irreproduc-
ible, as this varies from day-to-day and with the SEC setup, and it
is scientifically inadmissable to publish irreproducible data. Starch
CLD methods involve one more step, debranching, which removes
the branching points of starch with isoamylase [21, 26].

2 Materials

We now give a typical procedure for obtaining the SEC distribu-

tion of debranched starch.

2.1 Rice Nipponbare rice grains.

2.2 Buffer Tricine buffer (250 mM, pH 7.5)

and Solutions
1. Add 44.80 g tricine to 1 L deionized water.
2. Prepare 1 M NaOH by dissolving 4 g NaOH in 100 mL
deionized water.
3. Adjust the pH of 1). To 7.5 with ~20 mL of 2).
4. Store the tricine buffer at 4°C.
Sodium bisulfite solution (0.45% w/w)
1. Add 0.45 g sodium bisulfite to 100 mL deionized water.
2. Store at room temperature (RT).
Dimethyl sulfoxide (DMSO)/LiBr
1. Add 13.75 g LiBr to 2.5 L pure DMSO with dissolved LiBr
(0.5% w/w).
2. Store at RT.
Acetate buffer (0.1 M, pH ~3.5)
 Characterizing Starch Molecular Structure of Rice 173

1. Acetic acid (0.1 M)—diluted 1.15 mL of 17.4 M acetic acid

with deionized water to 200 mL.
2. Sodium acetate (0.1 M)—dissolve 1.64 g of anhydrous sodium
acetate (or 2.72 g of sodium acetate trihydrate) in deionized
water and dilute to 200 mL.
3. Adjust the pH of acetic acid solution (0.1 M) with sodium
acetate solution (0.1 M) to pH 3.5 (the ratio is 18.9:1).
4. Store acetate buffer at 4°C. It expires in 6 months.
Sodium azide solution (0.04 g/mL)
1. Add 8 g of sodium azide to 200 mL deionized water.
2. Store at RT.
Protease from Streptomyces griseus (Type XIV, ≥3.5 units/mg
solid, powder, Sigma-Aldrich P5147).
1. Store protease in −20 °C Freezer.
Isoamylase from Pseudomonas sp. (Lot 130104b, Megazyme).
1. Store Isoamylase at 4 °C.
Pullulan standards (peak molecular weight ranging from 342 to
2.35 × 106, Polymer Standard Services, Mainz, Germany)
1. Dissolve 1 mg of pullulan standards in 1 mL DMSO/LiBr
(0.5%, w/w) solution in the thermomixer at 80 °C at 350 rpm
for 3 h.
2. Store at RT.

2.3 Equipment Hydrophilic Teflon membrane filter

Millipore, Billerica, MA, USA.
Cryogrinder
Freezer/Mill 6850 SPEX, Metuchen, NJ, USA.
Thermomixer
Thermomixer Comfort, Hamburg, Germany.
Centrifuge
Avanti® J-E, Beckman Coulter, USA.
Freeze-drier
BenchTop Pro with Omnitronics, VirTis SP Scientific, USA.
SEC
Agilent 1100 Series SEC, Agilent Technologies, Waldbronn, Germany
1. Separation columns (GRAM precolumn, GRAM30, and 1000
analytical columns, Polymer Standard Services, PSS, Mainz,
Germany).
2. Refractive index detector (RID, 235RID-10A, Shimadzu,
Kyoto, Japan).
174 Cheng Li et al.

PSS WinGPC Unity software

Polymer Standard Services, Mainz, Germany.

3 Methods

3.1 Sample An essential preliminary for characterizing starch structure using

Preparation SEC is that the sample be prepared in a molecularly dispersed
(unaggregated) form, without loss (or at least without loss that is
biased in any structural property such as size) and without degra-
dation. This is particularly hard for native starch, as distinct from
modified or processed starch: the presence of non-starch material
(non-starch polysaccharides, cell-wall material, proteins, and lip-
ids) and high crystallinity complicates nondestructive extraction
and dissolution. Moreover, the isolation of starch from a natural
source (e.g., a cereal grain) requires grinding, which can easily
induce molecular degradation [26, 27]. Dissolving previously
extracted or degraded/processed starch is a simpler task to imple-
ment without further degradation, because extraction (and/or
degradation and/or processing) is highly likely to reduce molecu-
lar size, and partially or wholly remove non-starch materials.
Another difficulty in dispersing starch is its propensity to hydrogen-­
bond with itself.
Water-based systems (without any other hydrogen-bonding
solvent) can be readily implemented for the starch extraction/
dissolution. This however normally involves using aggressive
conditions, for example, high temperature, microwave heating,
pressure, and high or low pH, although other techniques involve
initial dissolution in DMSO followed by redispersion in a water-
based solvent, and/or dispersion using a microwave oven [28].
However, thus far, there is no firm evidence that any such tech-
niques avoid significant degradation, re-association and/or loss
[29, 30].
Methods have been described [21, 26, 31] for ensuring that
dissolution is complete and with minimum degradation using a
non-aqueous hydrogen-bonding solvent, specifically dimethyl
sulfoxide (DMSO) with small amounts of a hydrogen-bond dis-
rupter, for example LiBr. The methodology needs some degree
of optimization for each type of grain. A typical procedure uses a
combination of protease, sodium bisulfite, DMSO with 0.5%
w/w LiBr, and ethanol solutions to completely dissolve starch
and remove non-starch components. For this method, different
characterization procedures have been applied to test for loss and
degradation of starch, and where the final amounts of non-starch
components are minimized. For obtaining the first-level starch
structure, isoamylase is used to remove the branching points of
starch [21, 26].
 Characterizing Starch Molecular Structure of Rice 175

3.1.1 Cryogenic Milling 1. Pour enough liquid nitrogen into the cryogrinder (see Note 1).
of Rice Grains 2. Put rice grain samples in the plastic vials containing a grinding
bead.
3. Lock the plastic vials in position in the cryogrinder.
4. Close the lid of the cryogrinder.
5. Freeze the rice grain samples in liquid nitrogen for 2 min.
6. Pulverize samples for 5 min with grinding speed at 10 s−1.
7. Open the lid and remove the samples.
8. Transfer the rice flour to another tube.
9. Store samples at RT until use.

3.1.2 Starch Extraction 1. Transfer 7–9 mg flour (equivalent to ~4 mg dry starch) to a

(For Pure Starch, Skip 2 mL microcentrifuge (Eppendorf) tube.
Steps 1 and 3–6) 2. Mix the flour with 0.5 mL protease in tricine buffer (2.5
Units/mL) using a vortex mixer at 37 °C for 30 min.
3. Centrifuge at 4000 × g for 10 min and discard the
supernatant.
(steps 4 and 5 are recommended only for grains containing
proteins with high disulfide bonds)
4. Mix the precipitate with 0.5 mL sodium bisulfite solution (0.45%)
using a vortex mixer, and incubate at 37 °C for 30 min.
5. Centrifuge at 4000× g for 10 min and discard the
supernatant.
6. Suspend the precipitate in 1.5 mL DMSO/LiBr solution
(0.5% w/w) and shake it gently by hand to make a homoge-
nous mixture (see Note 2).
7. Heat in the thermomixer at 80 °C 350 rpm for 24 h, inverting
occasionally by hand to ensure a homogenous mixture.
8. Mix well, centrifuge at 4000 × g for 10 min, and collect the
supernatant in a 15 mL centrifuge tube (this step is to remove
insoluble materials, mostly non-starch polysaccharides and
lignin).
9. Precipitate the starch in the supernatant using 10 mL absolute
ethanol.
10. Centrifuge at 4000 × g for 10 min and discard the supernatant
(this step is to remove non-starch soluble materials, mostly
lipids and any residual proteins).
11. Add another 10 mL absolute ethanol and shake gently.
12. Centrifuge at 4000 × g for 10 min and discard the supernatant
(this step is to remove residual DMSO/LiBr).
13. Trap the last drop of ethanol by inverting the tube on a paper
towel.
176 Cheng Li et al.

(For molecular analysis of fully branched starch, after the pro-

cedure of starch extraction above, dissolve the samples in DMSO/
LiBr with a concentration of 2 mg/mL in a thermomixer at 80 °C
with shaking for 2 h. Centrifuge the samples at 4000 × g for 10 min
and transfer the supernatant to SEC vials to store until SEC analysis
(see Note 3)).

3.1.3 Debranching (Also 1. Disperse the precipitate in 0.9 mL of warm or hot deionized
Prepare a Blank water and heat in a boiling water bath for 15 min.
Without Starch) 2. Cool the dispersion to room temperature and add 5 μL of
sodium azide solution (0.04 g/mL) as preservative, 0.1 mL of
0.1 M acetate buffer (pH ~3.5).
3. Transfer 2.5 μL isoamylase solution to the starch dispersion.
4. Vortex the whole mixture and incubate in a water bath at
37 °C for 3 h.
5. Neutralize the mixture to pH ~7 (measured using pH paper)
with 0.1 M NaOH (about 0.1 mL).
6. Transfer the mixture to a 2 mL microcentrifuge (Eppendorf)
tube.
7. Heat in the thermomixer at 80 °C 350 rpm for 1 h.
8. Lift up the lid of the microcentrifuge tube and cover with a
punctured cap (two holes).
9. Freeze the starch dispersion in liquid nitrogen until reduced
or no bubbling is observed.
10. Freeze-dry the starch overnight (dried samples may be stored
in desiccator until use; best to tightly seal the microcentrifuge
tube with parafilm).
11. Dissolve the dry starch in 1.0 mL DMSO solution containing
LiBr (0.5% w/w) in the same microcentrifuge tube in the
thermomixer at 80 °C 350 rpm for 24 h and invert it by hand
occasionally to ensure complete dissolution.
12. Mix well, centrifuge at 4000 × g for 10 min (see Note 4).
13. Transfer the supernatant in a SEC vial.
14. Store it at room temperature until SEC analysis.

3.2 SEC Separation Separation. The main component of the SEC setup is the separa-
and Detection tion column, which contains a stationary phase of porous polymer
packing material. Polymer molecules in the mobile phase with a
large hydrodynamic volume cannot penetrate into the pores and
will therefore elute out of the column first, while smaller polymers
take a longer time, as they have access to the extra volume inside
the pores. The different detectors give complementary informa-
tion on the whole distribution of polymer molecules. The analytes
must not interact enthalpically with the stationary phase in order
to have the proper size-separation regime.
 Characterizing Starch Molecular Structure of Rice 177

There are technical problems with SEC, particularly band

broadening and shear scission. Band broadening is the effect where
molecules with the same Vh elute at a finite range of elution vol-
umes. Although this broadening is quite small for good modern
columns, it can significantly affect the distribution shape [32] and
hide some important features, such as shoulders or subsidiary max-
ima that could represent different starch components and enzyme
sets. Although some methods have been suggested to correct the
band-broadening problem [33–37], these are not yet readily avail-
able to be implemented for starch systems. Another major problem
with SEC is shear scission for large branched molecules like amylo-
pectin. Shear forces inside the column will break large molecules
into two roughly equal size parts [20, 23]. Some efforts have been
made to reduce the amount of shear scission, such as changing elu-
tion conditions, e.g., flow rate [23, 29]. While shear scission is
reduced with a slower flow rate, the effect cannot be completely
removed satisfactorily (as shown by dimensionless-number analysis
[23]), because the lower the flow rate, the greater the band-­
broadening. There is also the limitation of finding an appropriate
solvent system with sufficiently low viscosity. Despite these limita-
tions, the degree of shear scission for different polymer molecules
in the same SEC setup and same run would vary only slightly,
which allows for some meaningful comparison between the amylo-
pectin components of starches.
Contamination from those non-starch components in the sam-
ple which have not been removed in the dissolution procedure
could result in elution in a region which overlaps that of the starch
in the sample. Further, contaminants such as lipids and phospho-
lipids can cause starch molecules to form helical complexes and
entangle with one another, which can also result in artifacts.
Further, it is essential to check that there are no artifacts in the
apparent distribution because of column exclusion limits: for
example, if different samples all show a rapid drop to the baseline
at the same size, it is highly likely that this is an artifact arising
because all molecules larger than a certain size elute at the same
time, irrespective of their true molecular size, because all larger
sizes are too big to enter the pores in the column.
Detection. The objective is to detect the amount of starch present
in a given elution slice, with the signal dependent only on a chosen
property of the starch (e.g., distribution of the number or weight
of chains) in that elution slice.
The common detectors for SEC are viscometric, differential
refractive index (DRI), and multiple-angle laser light scattering
(MALLS), which give, respectively, the number (N(Vh)) and
weight (w(Vh)) distributions, plus weight-average molecular weight
Mw(Vh) and z-average radius of gyration Rg,z(Vh) [20, 38]. The
more types of detectors, the more information about starch struc-
178 Cheng Li et al.

ture can be obtained by size separation, although one-dimensional

methods are always limited to single moments of an infinite-­
dimensional distribution [38]. The SEC weight distribution w(log
Vh) is defined by w(logVh)d(logVh) being the weight of chains in
the volume increment log Vh to logVh + d(logVh). This definition
is because this quantity is the signal given by an ideal SEC with no
band broadening and with a linear calibration curve [39]. For lin-
ear polymers, DRI detection is able to give both number and
weight distributions, as there is a trivial relationship between N(Vh)
and w(log Vh), given by w(logM) = M2 N(M), where M is the
molecular weight [39]. However, this relationship is not valid for
complex branched polymers.
A MALLS detector used with a DRI detector enables one to
obtain Mw(Vh), and thus enables the X axis to be put in terms of Mw
rather than Vh or Rh. However, this is the weight-average molecu-
lar weight in that elution slice, not the actual molecular weight. No
SEC device can give a molecular weight distribution for a branched
polymer, because the size separation is by size, not molecular
weight. (In principle, mass-spectrometric techniques could give a
true molecular weight distribution, but with problems of fragmen-
tation and molar-mass-dependent signal intensity, this is impracti-
cal except for special cases such as relatively low molecular weights
using MALDI or electrospray MS.)
For debranched starch, as with any linear polymer, there is a
unique relationship between molecular size (Vh or Rh) and molec-
ular weight. In principle, the MALLS detector could give absolute
molecular weight for this, but the problem is that the light-­
scattering signal is weak with small molecules such as debranched
amylopectin chains. Instead, the principle of universal calibration is
invoked, which is the assumption that separation is only a function
of Rh and is independent of polymer composition. One then cali-
brates with essentially monodisperse standards of known Rh, as
directly measurable from the intrinsic viscosity of an individual
monodisperse sample. Knowing Rh for a given polymer, one can
then find the corresponding molecular weight using the Mark-­
Houwink equation: Vh = 2/5 KM1 + α/NA. Here K and α are the
Mark-Houwink parameters, which depend on the polymer, solvent
and temperature, and can be obtained from independent viscom-
etry measurements (and are often available in the literature). Note
that the Mark-Houwink equation is an empirical one, and may not
work well under certain circumstances, such as quite low molecular
weights.
Here is a typical procedure.
1. DMSO/LiBr (0.5% w/w) solution after filtration through a
0.45 μm hydrophilic Teflon membrane filter is used as mobile
phase (the Mark—Houwink parameters for this eluent at
80 °C are K = 2.424 × 10−4 dL g−1 and α = 0.68).
 Characterizing Starch Molecular Structure of Rice 179

2. Separation columns are PSS GRAM precolumn, GRAM30,

and 1000 analytical columns.
3. Only a refractive index detector is used.
4. The separation columns are held at 80 °C and the detector is
set at 45 °C.
5. The flow rate is set at 0.5 mL min−1.
6. A series of pullulan standards with molecular weights ranging
from 342 to 2.35 × 106 is used for calibration.

3.3 Data Processing Data processing. Data processing is needed to convert the raw sig-
and Reporting nals from one or more detectors (these signals depend on the par-
ticular setup of the separation and detection system) into one or
more distribution functions (e.g., a number or weight distribu-
tion). The distribution functions are expressed as Vh or Rh distribu-
tions, which depend only on the sample and not on the separation/
detection system. Note that w(log Vh) and w(log Rh) are the same,
to within an arbitrary normalization constant.
Data reporting. The objective for data reporting is to present the
distribution data along with sufficient information that the experi-
ment is reproducible. It should contain a complete description of
the starting material and the dissolution technique. A full descrip-
tion of the separation setup should also be provided, including for
example flow rate, material used for the column, and so on. The
calibration information should also be reported, which can be
deposited as supplementary information. It is strongly preferable
to report results in terms of Rh rather than Vh, as scientists have a
“feel” for the size (diameter) or a molecular entity rather than for
its volume.
1. The SEC chromatograms are analyzed using appropriate soft-
ware, such as Astra (Wyatt) or WinGPC (PSS).
2. A calibration curve is made to relate the elution time to Vh and
thus Rh, with Vh = 4/3 π Rh3.
3. The size distribution w(log Rh) is calculated from the detector
signal (see Fig. 1 for an example).
The normalization of any distribution, including those from
SEC or FACE, is arbitrary, and can be chosen for convenience to
bring out whatever feature is being considered. For example, to look
at amylopectin CLDs from different samples, the CLD can conve-
niently be normalized to the overall maximum (which will almost
always be the amylopectin component in a CLD). For looking at
amylose features, it is convenient to normalize to the maximum in
the amylose component of the CLD. Under other circumstances,
it can be convenient to normalize to the amount of starch present
in the sample, e.g., when comparing changes in the debranched
distributions as enzymatic digestion proceeds [40].
180 Cheng Li et al.

Nipponbare
1.2
Amylopectin Amylose

w(log Vh) [arb. unit]

0.8

0.4

0
1 10 100 1000 10000
DP

Fig. 1 A typical example of starch CLD from Nipponbare rice grain, obtained with
SEC. The amylopectin and amylose chain ranges are shown, with the dividing
line set at the minimum in w(log Vh) (which is close to DP 100)

The CLD can be expressed in terms of the weight or number

distributions, related as given above: w(logM) = M2 N(M).
Molecular weight for a linear homopolymer can also be expressed
in terms of the degree of polymerization X, where for linear starch,
one has M = 162.2X + 18.0, 162.2 being the molecular weight of
the anhydroglucose monomeric unit and 18.0 that of the addi-
tional water in the reducing end group.
For reasons given in detail elsewhere [41, 42], it is strongly
preferable to plot the number CLD, Nde(X), as log Nde(X), which
voids many artifacts of normalization and brings out features at
higher DPs which are essentially invisible in a non-logarithmic plot
(as unfortunately is often done in the literature). On the other
hand, an amylose CLD is best plotted as w(logX) against logX.

4 Notes

1. Make sure the cryogrinder is used in a well-ventilated area as a

large amount of liquid nitrogen is consumed by the cryogrinder.
2. The grinding step might take a few days, depending on the
nature and the amount of grain in the sample. For less than 30
samples, the time for starch extraction will take an afternoon
and the debranching procedure a whole day. For more than 30
samples, it is better to divide the samples into different sets
and do the starch extraction and debranching accordingly.
The freeze-dry procedure for starch samples normally takes
overnight followed by another 24 h to dissolve the dried
samples in DMSO/LiBr (0.5% w/w) solution. Finally, the
separation time for a single starch sample using SEC is ~1 h.
 Characterizing Starch Molecular Structure of Rice 181

3. This protocol also applies to measuring fully branched starch

molecular structure (except of course that the debranching
procedure is omitted, and usually the columns will be
different).
4. Carrying out the centrifugation step must be done just before
transferring the starch solution into the SEC vial. This step is
to remove the impurities in the solution which could block the
SEC columns from blocking, and this step also reduces the
height of contamination peaks in the chromatograms.

4.1 FACE Many of the practical procedures shown above are also used in
FACE measurements, especially the starch isolation, dissolution,
and debranching steps. The most important difference is the need
to label with a charged fluorophore. This technology was devel-
oped in elegant work by Morell and coworkers over some years
[22, 43]. Recently, some improvements in these techniques have
been developed by Wu et al. [21]. Figure 2 gives an example of a
FACE CLD. This was obtained with a PA-800 Plus System and
monitored with a solid-state laser-induced fluorescence (LIF)
detector with an argon-ion laser for excitation (Beckman-Coulter,
Brea, CA, USA), a 50 μm diameter N-CHO coated capillary with
effective capillary separation length of 40 cm, and the Beckman
carbohydrate separation buffer as the separating medium. The sep-
aration used an applied voltage of 30 kV (current ~14 mA) at
25 °C over 90 min of total separation time. The areas under the
peaks give the number CLD directly, as this device gives baseline
separation between each DP. This setup enables CLDs to be mea-
sured as high as DP ~170, which is much higher than the range in
earlier types of FACE equipment. A reminder that FACE gives the
number CLD Nde(X) directly, whereas SEC gives the SEC weight
distribution with w(logX) = X2 Nde(X).

FULLY BRANCHED
4.2 The techniques used above, but without the debranching step,
DISTRIBUTIONS give the fully branched distributions. As stated above, it is essential
to be aware that SEC characterization cannot avoid shear scission
of the larger amylopectin molecules. The example given in Fig. 3
shows a typical result (recalculated from data in [44]). It shows the
smaller amylose molecules, the larger amylopectin ones, and also
the region where shear scission of the amylopectin component is
significant. The extraneous features for smaller sizes are non-starch
impurities.

USING THIS
4.3 Although not the main topic of this chapter, we now give a brief
METHODOLOGY overview showing how this methodology can be used to develop
IN THE DEVELOPMENT improved plant varieties. The simplest example is to choose or
OF NUTRITIONALLY breed varieties where the CLD is known to be nutritionally advan-
IMPROVED STARCHES tageous, e.g., with longer amylopectin branches [45]. Another is
182 Cheng Li et al.

Sorghum
1

0.1

log10Nde(X )
0.01

0.001

0.0001
0 40 80 120 160 200
DP

Fig. 2 A typical example of starch CLD from sorghum grain, obtained with FACE. It
gives the number CLD; the characterization range is as high as DP ~170.
Although FACE yields Nde(X) with resolution of individual X values, for visual ease
the data are presented as a continuous line

Fig. 3 SEC weight distribution w(log Rh) (arbitrary units) for a rice starch (data
recalculated from Vh distribution given in [44]). Also shown is the size value
above which shear scission occurs, and two peaks at low size which are artifacts
from non-starch impurities

a transgenic method developed by the present authors [46]. This

is based on fundamental theory for the biosynthetic processes
controlling the CLD [47, 48]; from this, it is predicted that a
slight change in the minimum chain-length requirements for the
function of starch branching enzyme would have a significant
effect on the CLD. Such a change could be brought about by
changing a single residue on the enzyme branching site, and can
be implemented by standard transgenic techniques in molecular
biology. At each step of the process, it was important to monitor,
 Characterizing Starch Molecular Structure of Rice 183

using the techniques given in this chapter, the CLDs in vitro pro-
duced by the modified enzymes. This led to selection of sites
which could give reduced activity of the enzyme, as required for
this goal. The same technology could be used to develop plants
producing starches with slower digestibility and hence improved
nutritional performance.

Acknowledgments

Robert G. Gilbert greatly appreciates the support of the Chinese

Government’s 1000 Talents program of the State Foreign Experts
Bureau.

References
1. Englyst HN, Kingman SM, Cummings JH chemical properties from a generation-means
(1992) Classification and measurement of analysis study using amylomaize v and vii par-
nutritionally important starch fractions. Eur ents. J Agric Food Chem 58:5633–5639
J Clin Nutr 46(Suppl 2):S33–S50 10. Vilaplana F, Hasjim J, Gilbert RG (2012)
2. Byrnes SE, Miller JCB, Denyer GS (1995) Amylose content in starches: toward optimal
Amylopectin starch promotes the development definition and validating experimental meth-
of insulin-resistance in rats. J Nutr ods. Carbohydr Polym 88:103–111
125:1430–1437 11. Gilbert RG (2011) Size-separation character-
3. Johnston KL, Thomas EL, Bell JD, Frost GS, ization of starch and glycogen for biosynthesis-­
Robertson MD (2010) Resistant starch structure-­property relationships. Anal Bioanal
improves insulin sensitivity in metabolic syn- Chem 399:1425–1438
drome. Diabet Med 27:391–397 12. Gilbert RG, Witt T, Hasjim J (2013) What is
4. Lehmann U, Robin F (2007) Slowly digestible being learned about starch properties from
starch, its structure and health implications: a multiple-level characterization. Cereal Chem
review. Trends Food Sci Technol 18:346–355 90:312–325
5. Baba T, Arai Y (1984) Structural features of 13. Witt T, Gilbert RG (2014) Causal relations
amylomaize starch. 3. Structural characteriza- between structural features of amylopectin, a
tion of amylopectin and intermediate material semicrystalline hyperbranched polymer.
in amylomaize starch granules. Agric Biol Biomacromolecules 15:2501–2511
Chem 48:1763–1775 14. Witt T, Doutch J, Gilbert EP, Gilbert RG
6. Li L, Jiang HX, Campbell M, Blanco M, Jane (2012) Relations between molecular, crystal-
JL (2008) Characterization of maize amylose-­ line, and lamellar structures of amylopectin.
extender (ae) mutant starches. Part i: relation- Biomacromolecules 13:4273–4282
ship between resistant starch contents and 15. Zhang G, Ao Z, Hamaker BR (2008)
molecular structures. Carbohydr Polym Nutritional property of endosperm starches
74:396–404 from maize mutants: a parabolic relationship
7. Vilaplana F, Gilbert RG (2010) Two-­ between slowly digestible starch and amylopec-
dimensional size/branch length distributions tin fine structure. J Agric Food Chem
of a branched polymer. Macromolecules 56:4686–4694
43:7321–7329 16. Zhang G, Sofyan M, Hamaker BR (2008)
8. Nakamura T, Yamamori M, Hirano H, Hidaka Slowly digestible state of starch: mechanism of
S, Nagamine T (1995) Production of waxy slow digestion property of gelatinized maize
(amylose-free) wheats. Mol Gen Genet starch. J Agric Food Chem 56:4695–4702
248:253–259 17. Srichuwong S, Sunarti TC, Mishima T, Isono
9. Jiang HX, Jane LL, Acevedo D, Green A, Shinn N, Hisamatsu M (2005) Starches from differ-
G, Schrenker D, Srichuwong S, Campbell M, ent botanical sources i: contribution of amylo-
Wu YS (2010) Variations in starch physico- pectin fine structure to thermal properties and
184 Cheng Li et al.

enzyme digestibility. Carbohydr Polym 60: heating for analysis by size-exclusion chroma-
529–538 tography. J Agric Food Chem 54:9664–9669
18. Benmoussa M, Moldenhauer KA, Hamaker BR 31. Schmitz S, Dona AC, Castignolles P, Gilbert
(2007) Rice amylopectin fine structure vari- RG, Gaborieau M (2009) Quantification of the
ability affects starch digestion properties. extent of starch dissolution in dimethylsulfox-
J Agric Food Chem 55:1475–1479 ide by 1h nmr spectroscopy. Macromol Biosci
19. Zhu LJ, Liu QQ, Wilson JD, Gu MH, Shi YC 9:506–514
(2011) Digestibility and physicochemical 32. van Berkel KY, Russell GT, Gilbert RG (2005)
properties of rice (oryza sativa l.) flours and Molecular weight distributions and chain-­
starches differing in amylose content. stopping events in the free-radical polymeriza-
Carbohydr Polym 86:1751–1759 tion of methyl methacrylate. Macromolecules
20. Vilaplana F, Gilbert RG (2010) 38:3214–3224
Characterization of branched polysaccharides 33. Schnoll-Bitai I, Vega J, Mader C (2007)
using multiple-­detection size separation tech- Estimation of the band broadening parameters
niques. J Sep Sci 33:3537–3554 in single detection size-exclusion
21. Wu AC, Li E, Gilbert RG (2014) Exploring ­chromatography: a comparative study of vari-
extraction/dissolution procedures for analysis ous column combinations. Anal Chim Acta
of starch chain-length distributions. Carbohydr 604:9–17
Polym 114:36–42 34. Gruendling T, Guilhaus M, Barner-Kowollik C
22. O'Shea MG, Samuel MS, Konik CM, Morell (2008) Quantitative lc-ms of polymers: deter-
MK (1998) Fluorophore-assisted carbohydrate mining accurate molecular weight distributions
electrophoresis (face) of oligosaccharides: effi- by combined size exclusion chromatography
ciency of labeling and high-resolution separa- and electrospray mass spectrometry with maxi-
tion. Carbohydr Res 307:1–12 mum entropy data processing. Anal Chem
23. Cave RA, Seabrook SA, Gidley MJ, Gilbert RG 80:6915–6927
(2009) Characterization of starch by size-­ 35. Gruendling T, Guilhaus M, Barner-Kowollik
exclusion chromatography: the limitations C (2009) Fast and accurate determination of
imposed by shear scission. Biomacromolecules absolute individual molecular weight distri-
10:2245–2253 butions from mixtures of polymers via size
24. Jones RG, Kahovec J, Stepto R, Wilks ES, Hess exclusion chromatography-electrospray ion-
M, Kitayama T, Metanomski WV (2009) ization mass spectrometry. Macromolecules
Compendium of polymer terminology and 42:6366–6374
nomenclature. Iupac recommendations 2008. 36. Castro JV, van Berkel KY, Russell GT, Gilbert
Royal Soc Chem, Cambridge RG (2005) General solution to the band-­
25. Kostanski LK, Keller DM, Hamielec AE (2004) broadening problem in polymer molecular
Size-exclusion chromatography-a review of weight distributions. Aust J Chem 58:178–181
calibration methodologies. J Biochem Biophys 37. Konkolewicz D, Taylor JW II, Castignolles P,
Methods 58:159–186 Gray-Weale AA, Gilbert RG (2007) Towards a
26. Syahariza ZA, Li E, Hasjim J (2010) Extraction more general solution to the band-­broadening
and dissolution of starch from rice and sor- problem in size separation of polymers.
ghum grains for accurate structural analysis. Macromolecules 40:3477–3487
Carbohydr Polym 82:14–20 38. Gray-Weale A, Gilbert RG (2009) General
27. Hasjim J, Jane JL (2009) Production of resis- description of the structure of branched poly-
tant starch by extrusion cooking of acid-­ mers. J Polymer Sci Polymer Chem
modified normal-maize starch. J Food Sci 74: 47:3914–3930
C556–C562 39. Clay PA, Gilbert RG (1995) Molecular-weight
28. Bello-Perez LA, Roger P, Baud B, Colonna P distributions in free-radical polymerizations.1.
(1998) Macromolecular features of starches Model development and implications for data
determined by aqueous high-performance size interpretation. Macromolecules 28:552–569
exclusion chromatography. J Cereal Sci 27: 40. Witt T, Gidley MJ, Gilbert RG (2010) Starch
267–278 digestion mechanistic information from the
29. You SG, Lim ST (2000) Molecular characteriza- time evolution of molecular size distributions.
tion of corn starch using an aqueous hpsec-malls- J Agric Food Chem 58:8444–8452
ri system under various dissolution and analytical 41. Castro JV, Dumas C, Chiou H, Fitzgerald MA,
conditions. Cereal Chem 77:303–308 Gilbert RG (2005) Mechanistic information
30. Kim HS, Huber KC, Higley JS (2006) Alkaline from analysis of molecular weight distributions
dissolution of starch facilitated by microwave of starch. Biomacromolecules 6:2248–2259
 Characterizing Starch Molecular Structure of Rice 185

42. Wu AC, Witt T, Gilbert RG (2013) altering its chain length, branch density, and
Characterization methods for starch-based crystalline structure. J Agric Food Chem
materials: state of the art and perspectives. Aust 55:4540–4547
J Chem 66:1550–1563 46. Li C, Wu AC, Go RM, Malouf J, Turner MS,
43. O'Shea MG, Morell MK (1996) High resolu- Malde AK, Mark AE, Gilbert RG (2015) The
tion slab gel electrophoresis of 8-amino-­1,3, characterization of modified starch branching
6-pyrenetrisulfonic acid (apts) tagged oligosac- enzymes: toward the control of starch chain-­
charides using a DNA sequencer. length distributions. PLoS One 10:e0125507
Electrophoresis 17:681–686 47. Wu AC, Morell MK, Gilbert RG (2013) A
44. Gray-Weale A, Cave RA, Gilbert RG (2009) parameterized model of amylopectin synthesis
Extracting physically useful information from provides key insights into the synthesis of gran-
multiple-detection size-separation data for ular starch. PLoS One 8:e65768
starch. Biomacromolecules 10:2708–2713 48. Wu AC, Gilbert RG (2010) Molecular weight
45. Ao ZH, Simsek S, Zhang GY, Venkatachalam distributions of starch branches reveal genetic
M, Reuhs BL, Hamaker BR (2007) Starch constraints on biosynthesis. Biomacromolecules
with a slow digestion property produced by 11:3539–3547
 Chapter 11

Rice Grain Quality Benchmarking Through Profiling

of Volatiles and Metabolites in Grains Using Gas
Chromatography Mass Spectrometry
Cindy Llorente, Rosario Jimenez, Jackie, Yariv Brotman,
Alisdair R. Fernie, and Nese Sreenivasulu

Abstract
Gas chromatograph coupled with mass spectrometer is widely used to profile volatiles and metabolites
from the homogenized rice flour obtained from mature grains. Rice grains consist of central endosperm
which stores majorly starch and, in addition, accumulate various storage proteins as storage reserves. The
outer nutritious aleurone layer stores lipids, sugar alcohols, volatiles, antioxidants, vitamins, and various
micronutrients. Once paddy sample is dehulled, milled, and ground cryogenically, the brown rice flour is
subjected to extraction of primary metabolites and volatiles using an appropriate extraction method. In
metabolite profiling of the liquid extract obtained from the rice sample, mixture is initially subjected to
methoxyamination then silylation before being subjected to untargeted metabolite profiling. Peaks
obtained are processed for noise reduction and specific signal selection. Volatile compounds are initially
extracted using a solid phase adsorbent prior to analysis. All these compounds, metabolites, and volatiles
are detected in the mass selective detector by fragmentation at 70 eV ionization energy and the resultant
mass spectrum compared with a built-in library of compounds. Data mined from the gas chromatography
mass spectrometry analysis are then subjected to post-processing statistical analysis.

Key words Rice, Aroma, Metabolites, Metabolite profile analyses

1 Introduction

Chemical composition of seeds comprises a wide array of storage

products including starch, storage proteins, lipids, various primary
metabolism intermediates, and secondary metabolites carrying
human health benefits [1, 2]. Various metabolite profiling tech-
niques such as gas chromatography mass spectrometry (GC-MS),
liquid chromatography mass spectrometry (LC-MS), capillary
electrophoresis mass spectrometry (CE-MS), and chromatography
coupled nuclear magnetic resonance (NMR) technologies are
available with various degrees of accuracy and sensitivity [3].
Combining the information obtained from several of these

Nese Sreenivasulu (ed.), Rice Grain Quality: Methods and Protocols, Methods in Molecular Biology, vol. 1892,
https://doi.org/10.1007/978-1-4939-8914-0_11, © Springer Science+Business Media, LLC, part of Springer Nature 2019

187
188 Cindy Llorente et al.

t­echnologies, it is possible to cover the global metabolome index

with thousands of metabolites. Among these platforms, the GC-MS
is one of the key technologies applied routinely to profile several
hundred metabolites that helped reveal the metabolic signature
index of different plant species which possess diverse seed storage
products [4–6]. Over the last decade, the GC-MS platform is the
most widely used due to superior chromatographic resolution, bet-
ter reproducibility, repeatability of mass fragmentation through
electron impact ionization, as well as its low operational and main-
tenance cost [7]. These attributes make GC-MS reliable in terms
of characterization of numerous compounds in different biological
systems, especially at various stages of their development.
Rice grains contain up to 90% starch and 6–8% proteins prefer-
entially in endosperm, while the outer nutritious layer bran and
endosperm contain lipids, antioxidants, dietary fiber, and other
micronutrients [8]. Thus, rice remains an important cereal crop
providing a valuable source of energy for human consumption.
GC-MS profiling technique has been instrumental in acquiring
biological information for the identification of varietal differences
in terms of rice grain quality [9], starch quality [10] and for distin-
guishing aroma through volatile profiling [11–13]. The measured
759 metabolic signatures investigated from the grains of 85 lines
from backcrossed inbred population were investigated for associa-
tion mapping using metabolome quantitative trait loci analysis [14,
15]. Additional application includes the identification of genotypic
and phenotypic differences of germinating rice seeds [16].
Elucidation of seed storage composition of varietal differences
[17], drought implications on seed storage composition [18], and
application of metabolomic techniques as screening method for
predicting rice quality traits were reported in the past [19–21].
Metabolomic studies have also helped in the investigation of the
molecular basis of rice quality [22] as well as the identification of
metabolic variation between transgenic lines and its wild type,
mapping populations created from breeding lines due to higher
accumulation of sucrose, mannitol, and glutamic acid [23]. The
GC-MS platform was also shown to be useful in elucidating the
type of mutation in rice as part of food safety assessments [24].
Metabolites in rice grains are widely studied using the GC-MS
platform for separation, identification and quantification [7, 25].
By facilitating extraction procedures specific to compounds of
interest, the GC-MS aids not only in the identification of more
than 100 primary metabolites, but also to unravel the volatile sig-
natures of the sample being analyzed [11, 26, 27]. In this chapter,
we describe the GC-MS protocols for volatiles and metabolites
found in ground rice grains as performed at the Grain Quality and
Nutrition Center, IRRI. The rice paddy is dehulled, milled, and
ground with liquid nitrogen using a cryomill prior to extraction
 Rice Grain Quality Benchmarking Through Profiling of Volatiles and Metabolites… 189

Fig. 1 Workflow of GC-MS in volatile and metabolite profiling in rice grains

and GC-MS analysis. Resulting chromatogram is initially processed

for noise reduction and peak alignment before subjecting to statis-
tical analysis (Fig. 1).

2 Materials

2.1 Reagents 1. Methyl tert-butyl ether:Methanol solution (5:1) with

(0.3 mg/mL) decanoic acid as internal standard (see Note 1).
2. Methanol:H2O (1:3).
3. Methoxyamine hydrochloride.
4. Pyridine.
5. N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA) with 1%
trimethylchlorosilane (TMCS).
6. C4—C24 Even Carbon Saturated FAMEs (1000 μg/mL each
component in hexane, analytical standard).

2.2 Equipment 1. Analytical balance (10 mg–220 g ± 0.1 mg capacity).

2. Dehuller (8–10 g capacity).
3. Miller (8–10 g capacity).
4. Ball grinder (equipped with grinding jar with 50 g capacity for
handling cryo-cooled samples).
5. Liquid nitrogen container with dispenser.
6. Cold water bath (−10 °C temperature).
190 Cindy Llorente et al.

7. Temperature-controlled vortex mixer.

8. Sonicator.
9. Bench top microcentrifuge.
10. Speed vacuum dryer.
11. 11.0 mm capper for 2 mL glass vials.
12. 2 mL glass vials.
13. Glass inserts.
14. Eppendorf® Safe-Lock microcentrifuge tubes (1.5 mL and
2.0 mL capacity).
15. Agilent 6890 GC 5975 MS or its equivalent.

3 Methods

3.1 Sample Handling of rice samples is meticulous and, prior to extraction,

Preparation for Both handling is cryogenic.
Volatile and Metabolite
1. Weigh around 2 g of the rice paddy and clean for any
Profiling impurities.
2. Immediately dehull and place in a grinding jar which is pre-
cooled with liquid nitrogen.
3. Pour a generous amount of liquid nitrogen to a grinding jar
with the grinding ball to freeze the sample (see Note 2).
Repeat for 2–3 times.
4. Mill the frozen rice sample (see Note 3).
5. Grind milled samples using the ball grinder for 1 min.
6. Place samples in labeled Ziploc bags or microfuge tubes and
immediately store in −80 °C freezer prior to analysis.

3.2 Extraction 1. Place 1 g of rice flour sample in 10 mL GC vial.

of Compounds 2. Seal immediately using the suitable crimp cap and septum (see
3.2.1 Volatiles Note 4).
3. Allow the vials to stand for 30 min to let all the flour settles at
the bottom of the vial. This will prevent contamination of the
SPME fiber with rice flour.
4. Place the vials in the GC sample tray.
5. From the GC method, heat the rice flour with agitation to
allow for the release of the volatiles in the headspace of the
sample vial.
6. Insert the SPME fiber to the headspace of the vial for the
absorption of the volatiles.
7. Desorb the fiber in the injector and allow for the volatiles pass
through the GC column for separation and to the mass spec-
trometer for fragmentation.
 Rice Grain Quality Benchmarking Through Profiling of Volatiles and Metabolites… 191

3.2.2 Metabolites 1. Weigh 100 mg of rice flour sample (see Note 5) per sample in
2.0 mL Eppendorf® Safe-Lock microcentrifuge tubes.
2. Add sample with precooled with MTBE:MeOH containing
decanoic acid (0.3 mg/mL) as an internal standard.
3. Place tubes in ice bath while finishing a batch of samples.
4. Vortex mix tubes for 10 min.
5. Sonicate mixture for 15 min making sure that the water bath
will not heat up.
6. Add sample mixture with 500 μL MeOH:H2O solution.
7. Centrifuge samples at 12,000 × g for 5 min.
8. Aliquot 1 mL of the polar supernatant into properly labeled
1.5 mL Eppendorf® Safe-Lock microcentrifuge tubes.
9. Dry the extracts overnight in a speed vacuum dryer or its
equivalent.
10. Once extracts are dried, remove vials from speed vacuum dryer
(see Note 6). Dried extracts are temporarily stored in
dessicator.

3.3 Derivatization To improve the volatility of some compounds, analytes of particu-

of Compounds (See lar interest must undergo methoxyamination prior to derivatiza-
Note 7) tion where the polar functional groups are modified to decrease
their polarity and hence they can be separated in the GC column.
1. Cool the vials to room temperature.
2. Add dried extract with 10 μL methoxyamine hydrochloride
(40 mg in 1 mL pyridine), and mix for 90 m in 30 °C dry
bath.
3. After the reaction, derivatize the samples using 90 μL BSTFA
with 1%TMCS (with 20 μL FAMES in 1 mL BSTFA).
4. Heat the mixture with shaking for 30 min at 37 °C to ensure
that derivatization is complete.
5. 60 μL of derivatized mixture is then transferred to 200 μL
glass inserts inside 2 mL microfuge tubes.
6. Centrifuge samples at 12,000 × g for 30 s to settle down any
solid particles into the bottom of the glass insert. This will also
avoid clogging the syringe of the GC.
7. Once done, transfer glass inserts into properly labeled 2 mL
GC vials and samples are analyzed in the GCMS.
The method described in this procedure employs silylation to
alter the functionality of the targeted functional groups with the use
of N, O-bis(trimethylsilyl)trifluoroacetamide or BSTFA with 1% tri-
methylchlorosilane (TMCS). With BSTFA, the active hydrogens of
the compounds containing –SH, –OH, –NH, and –COOH are
replaced with the trimethylsilyl (TMS) group. This process is cata-
192 Cindy Llorente et al.

lyzed by trimethylchlorosilane (TMCS) which is already mixed with

BSTFA at 1%. Pyridine is used as a solvent which contributes to the
reactivity of the solvent by accepting the protons (H+) during
derivatization process. There had been many published methods
using different reagents to derivatize compounds of different classes
such as MSTFA w/ 1% TMCS [25] in rice and in biological sample
extracts [28]; BSTFA w/ 1% TMCS [23] in transgenic rice; MSTFA
[24] in rice; hydroxylamine hydrochloride and hexamethylsily-
limidazole and trifluoroaceteic acid [29] for sugars in carrots. A
comprehensive discussion on the derivatization of sugars [30] pro-
vides a selection for the preparation of procedures for the analysis of
carbohydrates which may be useful in rice research.

3.4 GC-MS Run (See For every batch of run in the GC a reagent blank and a set of stan-
Note 8) dards in a quality control (QC) mix should be included. Both
blank and QC mix should also undergo similar extraction and
derivatization procedure together with the sample.
The variability in samples can arise from multiple sources
including physiological differences and variability from the ana-
lytical method itself. Measuring metabolites using mass spectrom-
etry techniques to explore natural variation from the diversity
panel requires appropriate care about homogenous representation
of tissue samples, including enough biological and technical repli-
cations. In addition, analytical variation caused by suboptimal per-
formance of the chosen apparatus and instrument drift over time
are additional major issues in large-scale metabolomics studies,
which requires further attention. Batch-to-batch variation is
another technical source of variation arising from the sum of both
manual and robotic samples handling. The presence of batch-to-­
batch variation makes it difficult to integrate data from indepen-
dent batches of samples. This issue is particularly problematic
when dealing with a large number of samples such is the case when
analyzing structured plant populations. To counter this, several
normalization methods have been developed to minimize non-
biological variation. For example, normalizations by a single or
multiple internal or external standard compounds [7, 31] were
considered. Similarly, isotope-labeled internal standard approaches
[32] were established to monitor analytical error. While there is
no single best way to conduct metabolomic studies, there are a
number of pitfalls and known problems that need to be carefully
avoided. Detailed guidelines and practice and normalization pro-
tocols [33] have been published previously for this purpose. As
the number of samples in the data set increases there is a corre-
sponding time-dependent variation in the metabolite data.
Removing platform-specific sources of variability such as system-
Rice Grain Quality Benchmarking Through Profiling of Volatiles and Metabolites… 193

atic errors is one of the top priorities in metabolomics data pre-

processing. However, metabolite diversity leads to different
responses to variations at given experimental conditions, making
normalization a very demanding task. The Quality Control (QC)
samples are of key importance, and these are best prepared by
pooling equal volumes of material from all of the biological sam-
ples to be analyzed [34, 35]. Alternatively, a chemically defined
mixture of authenticated reference compounds that mimics the
metabolic composition of the investigated biological material can
be employed. These synthetic mixtures are then subjected to the
same sample extraction, subject to instrumental analyses (ideally
distributed across the analytical run), and data processing, thus
providing quality checks for technical and analytical error, and
quantitative calibration to eliminate batch effects for the final pro-
cessed data. This normalization is a crucial step for minimizing the
batch-to-batch data variability across extended periods. As such,
this is a crucial requirement for large-scale phenotyping, which
facilitate inter-batch data integration.
Depending on the objectives of the study, a single or a triple
quadrupole MS can be employed to carry out the profiling for
both volatiles and metabolites. Single-quadrupole MS is very use-
ful in giving a full untargeted scan of the compounds while in triple
quadrupole the resulting transitions from the precursor ion to the
product ion become the fingerprint of the targeted analytes. In a
single quadrupole, compounds are first fragmented in the ioniza-
tion chamber using electron impact ionization set at 70 eV to pro-
duce fragment ions. These fragments are then separated by mass in
the quadrupole, allowing the ions to be detected at slightly differ-
ing time intervals. The resultant mass spectrum can be indexed
against the built-in library of compounds commercially available
for the identification of analytes.
In the triple quadrupole MS system, there is a second quadru-
pole or collision cell that further fragments the fragment ions gen-
erated from the first quadrupole, also known as the precursor ions.
Prior to the collision induced dissociation, the precursor ions are
first selected based on their selectivity and sensitivity to produce
the product ions. The product ions then pass to the third quadru-
pole for another mass filtering process. Instead of detecting a spe-
cific ion, the transition from a precursor ion to product ion is
recorded. This type of detection allows us to properly identify and
quantify trace-level compounds in a complex matrix such as that of
rice (Fig. 2). Matrix compounds might coincide in having the same
precursor ions as the target analytes, but the chances of having the
same precursor-to-product transition are rare [36, 37].
194 Cindy Llorente et al.

Fig. 2 Total ion chromatogram of a rice sample analyzed for volatiles using (a) Dynamic Headspace Sampling
and (b) SPME for sampling. Both insets show corresponding selected ion monitoring profiles of 2-acetyl-­1-
pyrroline or 2-AP (chromatogram obtained from Shimadzu GCMS-TQ8040)

Application Box: GCMSMS Analysis of Volatiles and Metabolites in Rice

using Triple Quadrupole GC [38]
Targeted metabolites need to be accurately quantified to use
them as biomarkers to discriminate differences in traits of
diverse rice species. Profiling with a GC-MS/MS (GC triple
quadrupole MS) was proven to be an effective tool for complex
matrices following 2 phases during profiling. Phase I requires
the discovery of the compounds that are present in the sample
using the full scan mode. Phase II involves the analysis of
selected secondary fragmentation in the multiple reaction mon-
itoring (MRM) mode. Once these MRM transitions are estab-
lished, they can be used to create targeted screening methods
that will scan the whole sample for the identified transitions of
each of the specific analytes. This type of workflow allows the

(continued)
 Rice Grain Quality Benchmarking Through Profiling of Volatiles and Metabolites… 195

researcher to identify and quantify metabolites in a sample

­without the worry of varying matrix. Metabolomic workflow
showing the use of full scan and MRM measurements for dis-
covery and targeted analyses respectively were described earlier
[38]. Different systems will have different ways of creating the
method for the targeted phase of metabolite profiling using the
GC-MS/MS. Generally, compounds of interest, particularly
those that may be considered potential biomarkers are chosen
for the precursor ions study based on the fragment’s intensity
and selectivity. From the target compound, product ions are
selected and transitions from the precursor ions to the product
ions are optimized. That is, the collision energy that will pro-
duce the maximum ion intensity of the compound will then be
used as the suitable collision energy of the compound for the
actual analysis of the samples.
In metabolites analysis, common fragments with m/z
ratio such as 147 and 73 may not be a good precursor ion as
these fragments of trimethylsilyl (TMS) group are common
in all the derivatized compounds. Once the transition param-
eters are identified for each compound, this information is
then used by the GC-MS/MS system to identify and quantify
the compounds (see Fig. 3).
Aside from volatiles, rice metabolites have already been
investigated showing the capacity of the GC triple quadru-
pole MS system to identify useful mass fragment transitions
of compounds. This can be used as a guide in selecting transi-
tions for the analytes of interest.

(x100,000)
(x1,000) 2.00
82.00>67.05
111.00>83.10
5.0 82.00>41.05
111.00>69.00 1.75

4.0 1.50

1.25
3.0
1.00

2.0 0.75

0.50
1.0
0.25

8.0 8.5 9.0 9.5 3.50 3.75 4.00 4.25

Fig. 3 Multiple reaction monitoring chromatograms showing peaks corresponding to transition ions of (a) 2-AP
and (b) hexanal generated from a GC-MS/MS analysis (chromatogram obtained from Shimadzu GCMS-TQ8040)
196 Cindy Llorente et al.

4 Post-Run Data Analysis

Raw data obtained after every GC run can be analyzed one-at-a-

time or as a batch. Processing data in batches will ensure the
analyst that similar data treatment methods areapplied to all the
chromatograms as suggested by Fig. 4. Several chromatography
data systems are available like LabSolutions for Shimadzu and
ChemStation for Agilent. Generally, integration parameters are
set up and applied automatically to most of the chromatograms.
It is possible that since the samples can be largely varied in terms
of sample matrices, they would have slight differences due to
retention time shifts in the chromatogram. While the built-in
library provides the spectrum of the desired compound, the value
of running the actual standard compound and comparing the
analyte’s mass spectra to that of the standard from pooled librar-
ies around the world [39] is still important for the correct identi-
fication of the compound. There are additional softwares available
for deconvolution, baseline correction, and peak alignment that
could be useful during the data mining process such as AMDIS
[40]. The TargetSearch Package is a preprocessing package from
Bioconductor for searching and identifying metabolites using
corrected retention time indices [41]. The identification process
requires the use of retention index m ­ arkers or standards to obtain
aligned peaks and to identify outliers all throughout the sample
runs (Fig. 4).
Moreover, several data pretreatment methods down the pipe-
line such as centering, scaling, and transformations are in place to
further improve the biological interpretability of the biological
data set [42]. After the preprocessing, multivariate statistical analy-
sis is often used to initially describe and explain the profiles obtained
in the analysis.

Fig. 4 Workflow for GCMS post run data analysis using TargetSearch
 Rice Grain Quality Benchmarking Through Profiling of Volatiles and Metabolites… 197

5 Notes

1. Prepare the extraction solution for the whole set of samples for
analysis to minimize day effects.
2. Ultra-cooled samples and environment during extraction will
ensure that any enzymes present are quenched and any meta-
bolic process will no longer occur.
3. If the interest of the researcher includes the bran, there is no
need to mill or polish the rice sample. Cryo-grinding of the
sample must be immediately done.
4. Make sure that the septum used for the vials is suitable for
SPME analysis to avoid breakage of the SPME fiber.
5. For mature grains, and depending on the sensitivity of the
equipment, amount of sample recommended for extraction is
300 mg while for developing grains and germinating seeds,
smaller amount (20 mg) is desired.
6. Moisture will hydrolyze the derivatizing reagents rendering
the latter to lose its integrity for sample derivatization. Make
sure that there is no water or moisture adhering on the sur-
faces of the vials, glass inserts, and pipette tips and all through-
out the process especially right after freeze-drying. If the
extracts will not be derivatized immediately, store vials in a
dessicator. Store the derivatizing reagents in a cool, dry place
when not in use.
7. Derivatize only extracts that will be analyzed in the GC on the
same day.
8. Make sure that the GC-MS has no leak or moisture in the
system. Perform auto-tune and replace filaments, septa, liners
as needed.

Acknowledgments

This work has been supported under the CGIAR thematic area
Global Rice Agri-Food System CRP, RICE, Stress-Tolerant Rice
for Africa and South Asia (STRASA) Phase III funding.

References

1. Sreenivasulu N, Wobus U (2013) Seed-­ omics trilogy. Nat Rev Mol Cell Biol
development programs: a systems biology-­ 13(4):263–269
based comparison between dicots and 4. Zhen S, Dong K, Deng X, Zhou J, Xu X, Han
monocots. Annu Rev Plant Biol 64:189–217 C, Zhang W, Xu Y, Wang Z, Yan Y (2016)
2. Weckwerth W (2003) Metabolomics in systems Dynamic metabolome profiling reveals signifi-
biology. Annu Rev Plant Biol 54:669–689 cant metabolic changes during grain develop-
3. Patti GJ, Yanes O, Siuzdak G (2012) ment of bread wheat (triticum aestivum L.).
Innovation: metabolomics: the apogee of the J Sci Food Agric 96(11):3731–3740
198 Cindy Llorente et al.

5. Q-q Y, C-q Z, M-l C, D-s Z, J-z C, Wang Q, A 90-day toxicology study of high-amylose
Li Q-f, H-x Y, Gu M-h, SS-m S, Q-q L (2016) transgenic rice grain in Sprague–Dawley rats.
Biofortification of rice with the essential amino Food Chem Toxic 49(12):3112–3118
acid lysine: molecular characterization, nutri- 16. Vu NT, Shimada H, Kakuta Y, Nakashima T,
tional evaluation, and field performance. J Exp Ida H, Omori T, Nishi A, Satoh H, Kimura M
Bot 67:4285–4296 (2008) Biochemical and crystallographic char-
6. Tian JS, Liu CC, Xiang H, Zheng XF, Peng GJ, acterization of the starch branching enzyme I
Zhang X, Du GH, Qin XM (2015) Investigation (BEI) from Oryza sativa L. Biosci Biotech
on the antidepressant effect of sea buckthorn Biochem 72(11):2858–2866
seed oil through the GC-MS-based metabolo- 17. C-y L, Lai VMF, Shen M-C (2004) Changes in
mics approach coupled with multivariate analy- retrogradation properties of rice starches with
sis. Food Funct 6(11):3585–3592 amylose content and molecular properties.
7. Lisec J, Schauer N, Kopka J, Willmitzer L, Cereal Chem 81(3):392–398
Fernie AR (2006) Gas chromatography mass 18. Calingacion M, Fang L, Quiatchon-Baeza L,
spectrometry–based metabolite profiling in Mumm R, Riedel A, Hall RD, Fitzgerald M
plants. Nat Prot 1:387–396 (2015) Delving deeper into technological
8. Butardo VM Jr, Sreenivasulu N (2016) innovations to understand differences in rice
Tailoring grain storage reserves for a healthier quality. Rice 8:43
rice diet and its comparative status with other 19. Annison G, Illman RJ, Topping DL (2003)
cereals. Int Rev Cell Mol Biol 323:31–70 Acetylated, propionylated or butyrylated
9. Calingacion MN, Boualaphanh C, Daygon starches raise large bowel short-chain fatty
VD, Anacleto R, Sackville-Hamilton NR, Biais acids preferentially when fed to rats. J Nutr
B, Deborde C, Maucourt M, Moing A, Mumm 133(11):3523–3528
R, de Vos RCH, Erban A, Kopka J, Hansen 20. Champagne ET, Bett-Garber KL, Grimm CC,
TH, Laursen KH, Schjoerring JK, Hall RD, McClung AM, Moldenhauer KA, Linscombe
Fitzgerald MA (2012) A genomics and multi-­ S, McKenzie KS, Barton FE (2001) Near-
platform metabolomics approach to identify infrared reflectance analysis for prediction of
new traits of rice quality in traditional and cooked rice texture. Cereal Chem
improved varieties. Metabolomics 8:771–783 78(3):358–362
10. Kusano M, Fukushima A, Fujita N, Okazaki Y, 21. Laguerre M, Mestres C, Davrieux F, Ringuet J,
Kobayashi M, Oitome NF, Ebana K, Saito K Boulanger R (2007) Rapid discrimination of
(2012) Deciphering starch quality of rice kernels scented rice by solid-phase microextraction,
using metabolite profiling and pedigree network mass spectrometry, and mutivariate analysis
analysis. Mol Plant 5:442–451 used as a mass sensor. J Agric Food Chem
11. Bryant RJ, McClung AM (2011) Volatile pro- 55:1077–1083
files of aromatic and non-aromatic rice cultivars 22. Redestig H, Kusano M, Ebana K, Kobayashi
using SPME/GC-MS. Food Chem M, Oikawa A, Okazaki Y, Matsuda F, Arita M,
124(2):501–513 Fujita N, Saito K (2011) Exploring molecular
12. Zeng Y-X, Hu C-Y, Lu Y-G, Li J-Q, Liu X-D backgrounds of quality traits in rice by predic-
(2009) Abnormalities occurring during female tive models based on high-coverage metabolo-
gametophyte development result in the diver- mics. BMC Syst Biol 5:176
sity of abnormal embryo sacs and leads to 23. Zhou J, Ma C, Xu H, Yuan K, Lu X, Zhu Z,
abnormal fertilization in indica/japonica Wu Y, Xu G (2009) Metabolic profiling of
hybrids in rice. J Integr Plant Biol 51(1):3–12 transgenic rice with cryIAc and sck genes: an
13. Kim JK, Park S-Y, Lim S-H, Yeo Y, Cho HS, evaluation of unintended effects at metabolic
Ha S-H (2013) Comparative metabolic profil- level by using GC-FID and GC-MS. J
ing of pigmented rice (Oryza sativa L.) culti- Chromatogr B: Analyt Technol Biomed Life
vars reveals primary metabolites are correlated Sci 877:725–732
with secondary metabolites. J Cereal Sci 24. Frank T, Meuleye BS, Miller A, Shu Q-Y,
57:14–20 Engel K-H (2007) Metabolite profiling of two
14. Matsuda F, Okazaki Y, Oikawa A, Kusano M, low phytic acid (lpa) rice mutants. J Agric Food
Nakabayashi R, Kikuchi J, J-i Y, Ebana K, Yano Chem 55:11011–11019
M, Saito K (2012) Dissection of genotype– 25. Heuberger AL, Lewis MR, Chen M-H,
phenotype associations in rice grains using Brick MA, Leach JE, Ryan EP (2010)
metabolome quantitative trait loci analysis. The Metabolomic and functional genomic analyses
Plant J 70:624–636 reveal varietal differences in bioactive
15. Zhou XH, Dong Y, Xiao X, Wang Y, Xu Y, Xu ­compounds of cooked rice. PLoS One 5(9):
B, Shi WD, Zhang Y, Zhu LJ, Liu QQ (2011) e12915
 Rice Grain Quality Benchmarking Through Profiling of Volatiles and Metabolites… 199

26. Yang DS, Shewfelt RL, Lee K-S, Kays SJ 35. Zhao X, Zhang G, Wang Y, Zhang F, Wang W,
(2008) Comparison of odor-active compounds Zhang W, Fu B, Xu J, Li Z (2016) Metabolic
from six distinctly different rice flavor types. profiling and physiological analysis of a novel
J Agric Food Chem 56(8):2780–2787 Rice introgression line with broad leaf size.
27. Verhoeven HA, Jonker H, RCH DV, Hall RD PLoS One 10(12):e0145646
(2012) Solid phase micro-extraction GC–MS 36. Li J, Dong F, Xu J, Liu X, Li Y, Shan W,
analysis of natural volatile components in Zheng Y (2011) Enantioselective determina-
melon and Rice. In: Hardy NW, Hall RD (eds) tion of triazole fungicide simeconazole in veg-
Plant metabolomics: methods and protocols. etables, fruits, and cereals using modified
Humana Press, Totowa, NJ, pp 85–99 QuEChERS (quick, easy, cheap, effective,
28. Agilent Technologies, Inc (2013) Agilent fiehn rugged and safe) coupled to gas chromatogra-
GC/MS metabolomics RTL library a complete phy/tandem mass spectrometry. Anal Chim
GC/MS solution for metabolite identification. Acta 702(1):127–135
Agilent Technologies 37. Tranchida PQ, Zoccali M, Schipilliti L,
29. Soria AC, Sanz ML, Villamiel M (2009) Sciarrone D, Dugo P, Mondello L (2013)
Determination of minor carbohydrates in car- Solid-phase microextraction with fast GC com-
rot (Daucus carota L.) by GC-MS. Food Chem bined with a high-speed triple quadrupole mass
114:758–762 spectrometer for targeted and untargeted food
30. Ruiz-Matute AI, Hernández-Hernández O, analysis. J Sep Sci 36(13):2145–2150
Rodríguez-Sánchez S, Sanz ML, Martínez-­ 38. Matthews J, Gemme S, Huebschmann H-J,
Castro I (2011) Derivatization of carbohy- Llorente C, Jimenez R, Sreenivasulu N (2015)
drates for GC and GC–MS analyses. Metabolomics of rice genotypes using GC-MS/
J Chromatogr B Analyt Technol Biomed Life MS. ThermoScientific Application Note 1–9
Sci 879:1226–1240 39. Schauer N, Steinhauser D, Strelkov S,
31. Lou Q, Ma C, Wen W, Zhou J, Chen L, Feng Schomburg D, Allison G, Moritz T, Lundgren
F, Xu X, Lu X, Luo L, Mei H, Xu G (2011) K, Roessner-Tunali U, Forbes MG, Willmitzer
Profiling and association mapping of grain L, Fernie AR, Kopka J (2005) GC–MS libraries
metabolites in a subset of the core collection of for the rapid identification of metabolites in
Chinese rice Germplasm (Oryza sativa L.). complex biological samples. FEBS Lett
J Agric Food Chem 59:9257–9264 579(6):1332–1337
32. Chang WT, Lin DL, Liu RH (2001) Isotopic 40. Agilent Technologies, Inc (2005) Automated
analogs as internal standards for quantitative mass spectral deconvolution and identification
analyses by GC/MS—evaluation of cross-­ system AMDIS 2.64, in Released December,
contribution to ions designated for the analyte NIST
and the isotopic internal standard. Forensic Sci 41. Cuadros-Inostroza Á, Caldana C, Redestig H,
Int 121(3):174–182 Kusano M, Lisec J, Peña-Cortés H, Willmitzer
33. Chen M, Rao RSP, Zhang Y, Zhong CX, Thelen L, Hannah MA (2009) TargetSearch-a biocon-
JJ (2014) A modified data normalization ductor package for the efficient preprocessing
method for GC-MS-based metabolomics to of GC-MS metabolite profiling data. BMC
minimize batch variation. Springerplus 3:439 Bioinformatics 10:428
34. Winnike JH, Wei X, Knagge KJ, Colman SD, 42. van den Berg RA, Hoefsloot HCJ, Westerhuis
Gregory SG, Zhang X (2015) Comparison of JA, Smilde AK, van der Werf MJ (2006)
GC-MS and GC×GC-MS in the analysis of Centering, scaling, and transformations:
human serum samples for biomarker discovery. improving the biological information content
J Proteome Res 14(4):1810–1817 of metabolomics data. BMC Genomics 7:142
 Chapter 12

Re-sequencing Resources to Improve Starch and Grain

Quality in Rice
Gopala Krishnan Subbaiyan, Ardashir K. Masouleh, Agnelo Furtado,
Daniel L. E. Waters, and Robert J. Henry

Abstract
Next-generation sequencing can identify differences in the rice genome that explain the genetic basis of
grain quality variation. Differences in rice grain quality are mainly associated with differences in the major
component of the grain, starch. Association of rice quality variation with rice genome variation can be
conducted at the gene or whole-genome level. Re-sequencing of specific genes or whole genomes can be
used depending on the extent to which candidate genes for the traits of interest are known. Amplicon
sequencing of genes involved in starch metabolism can help in targeted discovery of the molecular genetic
basis of differences in starch related quality attributes. Whole-genome re-sequencing can complement this,
when the genetic basis of the trait is expected to be outside the coding region of starch metabolism genes.
These approaches have been used successfully to understand the rice genome at specific loci and over the
whole genome.

Key words Amplicon sequencing, Association analysis, Genotyping, Grain quality, Re-sequencing,
Rice, SNPs, Starch

1 Introduction

Sequencing has been utilized to identify genetic variations in genes

determining starch quality in rice. Rapid advances in sequencing
technologies have enabled the discovery of millions of polymor-
phisms in the genome such as Single-Nucleotide Polymorphisms
(SNPs) and Insertion-Deletion Polymorphisms (InDels) by com-
paring the whole-genome sequences of genotypes with high-­
quality reference genome sequences. Re-sequencing can be used to
identify differences in the genome that explain the genetic basis of
grain quality variation in rice [1]. Differences in the grain and
cooking quality of rice are strongly associated with differences in
the major component of the grain, starch [2]. Association of varia-
tion in rice quality with variation in the genome of the rice geno-
type can be conducted either at specific gene(s) or at the

Nese Sreenivasulu (ed.), Rice Grain Quality: Methods and Protocols, Methods in Molecular Biology, vol. 1892,
https://doi.org/10.1007/978-1-4939-8914-0_12, © Springer Science+Business Media, LLC, part of Springer Nature 2019

201
202 Gopala Krishnan Subbaiyan et al.

whole-genome level. Re-sequencing of specific genes [3] or whole

genomes [4] can be used depending on the extent to which candi-
date genes for the traits of interest are known.
Starch chemistry is a key contributor to rice grain quality since
starch forms the major component of the rice grain. In rice, as
many as 18 genes involved in starch biosynthesis have been identi-
fied and their sequence information is also available. Amplicon
sequencing of these genes involved in starch metabolism can help
in targeted discovery of the molecular genetic basis of differences
in starch related quality attributes [5, 6]. Whole-genome re-­
sequencing can not only be used to study variations in the genes
already known to be involved in starch biosynthesis but also in
identifying the genetic basis of the trait at the genome level thereby
enabling us to discover novel polymorphisms influencing starch
quality other than the already known genes involved in starch
metabolism genes. These approaches have been used successfully
to understand the rice genome at specific loci [7] and over the
whole genome [8] to unravel the genetic basis of starch quality in
rice. In this chapter, various sequencing approaches used in analyz-
ing the genomic variations determining starch quality in rice are
discussed with details of methods for sequencing, data analysis, and
association of genotypic data and phenotypic data.

2 Materials

2.1 Re-sequencing Chemicals—DNeasy Plant Mini Kit (QIAGEN), QIAshredder

of Starch Mini spin columns, TBE buffer, BioRadiProof TM High-Fidelity
Metabolism Genes DNA polymerase Kit, iProof polymerase, Ethidium bromide/
PicoGreendsDNA quantification kit (Invitrogen), iPLEX gold
reaction cocktail.
Instruments—Tissue lyser, Thermal cyclers, ABI Sequencer,
Illumina Sequencer, Nanodrop/Spectrometer.
Sequencher (Gene Codes Corporation, Ann Arbor, MI) and
CLUSTAL W (http://www.ebi.ac.uk/Tools/clustalw2/index.
html), Bio Edit (http://www.mbio.ncsu.edu/bioedit/bioedit.
html), Primer3 (http://bioinfo.ut.ee/primer3-0.4.0/). However,
we suggest the Clone Manager V9.1 (Sci-Ed Software, Cary, NC)
(http://www.scied.com/pr_cmbas.htm), CLC bio.

2.2 Whole-Genome DNA extraction for whole-genome re-sequencing by CTAB

Re-sequencing method
1. 1 M, Tris–HCl, pH 8.0: Dissolve 121.1 g of Tris base in
800 mL of distilled water, adjust the pH to 8.0 with concen-
trated HCl and add distilled water to a final volume of 1 L.
2. 0.5 M Ethylenediaminetetraacetic acid (EDTA). Dissolve
93.05 g of EDTA disodium salt in 400 mL, adjust the pH to
8.0 then adjust the volume to 500 mL.
 Re-sequencing Resources to Improve Starch and Grain Quality in Rice 203

3. CTAB extraction buffer (2% CTAB, 100 mM Tris–HCl,

pH 8.0; 20 mM EDTA; 1.4 M NaCl): Dissolve 20 g of CTAB
in 200 mL of 1 M Tris–HCl, 100 mL of 0.5 M EDTA and
500 mL of distilled water with a magnetic stir bar.
4. 3 M sodium acetate, pH 5.0. Dissolve 24.6 g of anhydrous
sodium acetate in 80 mL of distilled MilliQwater, adjust the
pH to 5.0 with glacial acetic acid and then the volume to
100 mL with distilled MilliQ water.
5. TE buffer (10 mM Tris–HCl, pH 8.0, 1 mM EDTA): Take
10 mL of 1 M Tris–HCl and add 2 mL of 500 mM EDTA
pH 8.0. Adjust the final volume to 1 L with distilled water.
6. Mortar and pestle.
7. Liquid nitrogen.
8. Steel spatula.
9. 50 mL Falcon tubes.
10. Fume hood.
11. 65 °C water bath.
12. Thermometer.
13. Chloroform:isoamylalcohol (24:1).
14. Chloroform.
15. RNase A 10 mg/mL, aqueous solution.
16. Ethanol (Ethyl alcohol, ≥99.5% pure).
17. 70% ethanol (v/v) ethanol/distilled water.
18. Centrifuge for 50 mL Falcon tubes and 12,000 × g.
19. DNA electrophoresis apparatus and gel documentation
system.
20. DNA stain compatible with agarose gel (e.g., ethidium bro-
mide or SYBRsafe).
21. Software—CLC bio Genomics Workbench, CLUSTAL W
(http://www.ebi.ac.uk/Tools/clustalw2/index.html), Bio
Edit (http://www.mbio.ncsu.edu/bioedit/bioedit.html).

2.3 SNP Genotyping 1. Sequenom®, MassARRAYAnalyzer Compact mass

spectrometer.

2.4 Association 1. Phenotypic data of the genotypes on starch quality.

of SNP 2. Genotypic data from amplicon sequencing/whole-genome
with Phenotypes re-sequencing.
3. Software—TASSEL (http://www.maizegenetics.net/#!tassel/
c17q9).
204 Gopala Krishnan Subbaiyan et al.

3 Methods

3.1 DNA Extraction The protocol describes how to extract DNA from rice plants for
and Quantification amplification of target genes and DNA for whole-genome re-­
sequencing. Extracting DNA through a rapid and inexpensive
DNA extraction protocol is suitable for amplification of target
genes but this method is not advisable for high-throughput
sequencing. The DNA for next-generation sequencing must be of
high quality; minimal shearing (no observable shearing on the gel)
and with an absorbance ratio between 1.8 and 2.0 [9]. Therefore,
protocols such as Cetyltrimethylammonium bromide (CTAB)
based DNA extraction protocols which extract high-quality DNA
from many tissues and most species at high yield for minimal
cost must be used for DNA extraction.

3.1.1 DNA Extraction A fast and easy DNA extraction method can guarantee accurate
from Rice for Amplification PCR amplification of the target gene. The DNeasy Plant Mini
of Target Genes Kit (or Maxi kit) (QIAGEN) is recommended for high-through-
put extraction and purification of genomic DNA from plant tis-
sues. The kit is supplied in several sizes (see Note 1 with Mini
Spin Columns which are suitable depending on the number of
samples to be extracted). Kits contain several contents which
can be found at: https://www.qiagen.com/au/resources/
resourcedetail?id=95dec8a9-ec37-4457-8884-5dedd8ba9448
&lang=en
A complete and quick disruption of leaf tissues is necessary to
achieve high-quality DNA free from degradation.
The manufacturer’s DNA extraction protocol is outlined below:
1. Sterilize 5–10 seeds with 10% hypochloride (commercial)
solution for 3 min. A drop of Tween 20 or commercial dish
washing liquid can assist sterilization.
2. Rinse the seeds with tap water and finally with sterilized water
for 5 min.
3. Wash the sterilized seeds in 70% Ethanol for 1 min.
4. Rinse again with sterilized water for 2 min.
5. Place the sterilized seeds (5–10 seeds) on wet Watson paper in
the bottom of a petri dish. Ensure the seeds are not placed
over each other and there is sufficient space to grow the seeds.
6. Incubate petri dishes in a germinator at 25–28 °C with 12 h
day/night photoperiod.
7. Check the petri dishes regularly and discard fungi-infected
seeds/peri dishes.
8. Grow the seeds for a week until the seedlings are 10–15 cm in
height.
Re-sequencing Resources to Improve Starch and Grain Quality in Rice 205

9. Normally, 100 mg, 50 mg, and 1 g of plant leaf tissues (fresh

weight) are sufficient for Mini, DNeasy 96, and Maxi kits,
respectively.
10. Plant materials can be kept for an extended period of time at
−80 °C until DNA extraction.
11. Mechanically disrupt the plant materials to powder using a
mortar and pestle and liquid nitrogen.
12. Add liquid nitrogen as required to keep the sample frozen and
grind until obtaining a fine powder.
13. For a larger number of samples, the Tissue Ruptor with 3 mm
tungsten carbide beads in a 2 mL safe-lock microcentrifuge
tube (see Note 2) containing 100 mg of fresh leaf tissue should
be used.
14. Liquid nitrogen is added to the tube (safe-lock microcentri-
fuge tube), and the sample frozen for 30 s. The sample should
be kept submerged in liquid nitrogen and disrupted for
approximately 30 s at full speed of Tissue Ruptor/TissueLyser.
15. Allow the liquid nitrogen to evaporate and immediately pro-
ceed to the next step.
16. Add 400 μL Buffer AP1 and 4 μL RNaseA stock solution
(100 mg/mL) to a maximum of 100 mg (wet weight) or
20 mg (dried) disrupted leaf tissue in tubes, then vortex vigor-
ously (see Note 3).
17. Incubate the mixture for 10 min at 65 °C and mix the tubes two
or three times during incubation by inverting to lyse the cells.
18. Add 130 μL Buffer P3 to the lysate, mix by inverting, and
incubate the tubes for 5 min on ice.
19. Centrifugation of the lysate for 5 min at 20,000 × g
(14,000 rpm) in room temperature is recommended.
20. Place the QIAshredder Mini spin column(s) (lilac colour) in
2 mL collection tubes.
21. Pipet supernatants into the QIAshredder Mini spin column
(see Note 4).
22. Centrifuge for 2 min at 20,000 × g (14,000 rpm).
23. Supply new tubes (not included in the kit) and transfer the flow-
through fraction (supernatants). Be careful not to disturb the
debris pellet. Normally, up to 450 μL of lysate is recovered.
24. Prepare buffer AW1 with addition of ethanol with the appro-
priate amount of ethanol (96–100%) as indicated on the bottle
to obtain a working solution (see Note 5).
25. Add 1.5 volumes of Buffer AW1 to the cleared lysate, and mix
by pipetting. For example, if 450 μL of lysate is recovered, add
675 μL of Buffer AW1. Quickly pipette Buffer AW1 onto the
cleared lysate and mix immediately.
206 Gopala Krishnan Subbaiyan et al.

26. Place the DNeasy Mini spin column in 2 mL collection tubes

(supplied).
27. Pipet 650 μL of the mixture from step 25, including any
formed precipitates, into DNeasy Mini spin columns.
28. Centrifuge the Mini spin columns for 1 min at ≥6000 × g
(corresponds to ≥8000 rpm for most microcentrifuges), and
discard the flow-through (see next step)*.
29. *Repeat steps 27 and 28 by reusing the collection tubes until
all lysates are processed. Finally, discard the last flow-through
along with collection tube.
30. Add Ethanol (96–100%) to Buffer AW2 (96–100%) as indi-
cated on the bottle to obtain a working solution.
31. Place the DNeasy Mini spin column into a new 2 mL collec-
tion tube (supplied in the kit).
32. Add 500 μL Buffer AW2.
33. Centrifuge the tube(s) for 1 min at ≥6000 × g (≥8000 rpm)
and discard the flow-through and reuse the collection tube(s)
for the next step.
34. Add 500 μL Buffer AW2 to the DNeasy Mini spin column
again.
35. Centrifuge for 2 min at 20,000 × g (14,000 rpm). This will
dry the DNeasy Mini spin column membrane which contains
the DNA.
36. Remove the DNeasy Mini spin column and discard flow-­
through and collection tube very carefully to ensure no resid-
ual ethanol is carried over to the next step.
37. Supply 1.5 or 2 mL microcentrifuge tube(s) (not provided in
the kit).
38. Place the dried DNeasy Mini spin column (containing DNA
from the previous step) onto the new tube(s).
39. Add 100 μL Buffer AE directly onto the DNeasy membrane.
40. For elution, incubate the tubes for 5 min at room temperature
(15–25 °C), and then centrifuge for 1 min at ≥6000 × g
(≥8000 rpm).
41. A second elution by repeating the steps 39 and 40 is recom-
mended (see Note 6).
DNA is extracted from each individual sample separately and
pooled equimolarly to create a mega pool for the whole
population.

3.1.2 DNA Extraction Next-generation sequencing requires the use of an appropriate

for Whole-Genome DNA extraction method [10]. There are many rapid and inexpen-
Re-sequencing sive DNA extraction protocols that extract poor quality DNA, but
 Re-sequencing Resources to Improve Starch and Grain Quality in Rice 207

they are not suitable for high-throughput sequencing. Commercial

kits generally extract a high yield with superior quality DNA, how-
ever, they are relatively expensive and this is an issue when survey-
ing plant breeding populations. Cetyltrimethylammonium bromide
(CTAB)-based DNA extraction protocols extract high-quality
DNA from many tissues and most species at high yield for minimal
cost. The DNA for next-generation sequencing has to be of very
good quality; minimal shearing (no observable shearing on the
gel) and with an absorbance ratio between 1.8 and 2.0.

3.1.3 CTAB Method 1. Heat a water bath to 65 °C (see Notes 7 and 8). For each
for Extraction of DNA sample, add 2 mL of CTAB extraction buffer to a 15 mL
Falcon tube and pre-heat to 650 °C by placing in the water
bath.
2. Take 2–3 g of young leaf material and place in a mortar. Pour
liquid nitrogen over the sample (see Notes 9 and 10). As the
liquid nitrogen evaporates, grind the material with a pestle.
Ensure the sample remains frozen for the entire procedure by
adding liquid nitrogen within 20 s of the liquid nitrogen evap-
orating (see Note 11).
3. Using a small spatula, scrape the ground powdered leaf mate-
rial into a 50 mL Falcon tube containing 8 mL of CTAB
extraction buffer which has been preheated to 65 °C.
4. Incubate the mixture for 1 h at 65 °C, manually mix the con-
tents gently by inverting the tubes at least every 10 min.
5. After the 1 h’s incubation, remove the tubes containing the
plant/buffer mixture from the 65 °C water bath and centri-
fuge for 5 min at 12,000 × g.
6. Carefully use a pipette to remove the supernatant from the
tube to a fresh 50 mL Falcon tube, taking care not to disturb
the plant material pellet.
7. Add RNase A to a final concentration of 1 μg/mL, mix well
gently by inverting the tubes three or four times, and then
incubate at 37 °C for 15 min (see Note 12).
8. In a fume hood, extract the supernatant with an equal volume
of chloroform:isoamylalcohol (24:1) (see Note 13). The
chloroform:isoamylalcohol is added to the supernatant and
mixed gently by inverting the tubes about 50 times. This step
is to be carried out in a fume hood (see Note 14).
9. Centrifuge for 1 min at 12,000 × g (see Note 15).
10. In a fume hood, remove the upper aqueous phase using a
suitable pipette. Take care not to remove any
chloroform:isopropanol. In the fume hood, discard the waste
chloroform:isopropanol into an appropriate waste bottle with
a lid which is resistant to organic solvents and is gas tight.
11. Repeat steps 7–9.
208 Gopala Krishnan Subbaiyan et al.

12. Precipitate the nucleic acids by the addition of a 1/10 volume

of 3 M sodium acetate (pH 5.0) and three volumes of 95%
ethanol (see Notes 16 and 17). Mix gently by inverting the
tubes several times.
13. Centrifuge at 12,000 × g. Orient the tubes so the side of the
tube which faced the outside of the centrifuge can be identi-
fied following centrifugation.
14. The DNA will be near the bottom of the tube on the side of
the tube which was oriented toward the outside of the centri-
fuge. Remove and discard the supernatant, taking care not to
disturb the DNA pellet.
15. Add 5 mL of 70% ethanol to remove residual salts from the
DNA pellet. Incubate at room temperature for 10 min.
16. Centrifuge at 12,000 × g. Orient the tubes so the side of the
tube which faced the outside of the centrifuge can be identi-
fied following centrifugation.
17. The DNA will be near the bottom of the tube on the side of
the tube which was oriented toward the outside of the centri-
fuge. Remove and discard the supernatant, taking care not to
disturb the DNA pellet.
18. Partially dry the DNA pellet at room temperature for a few
minutes. Over drying the DNA or the presence of ethanol in
the undried DNA can lead to difficulty in dissolving the DNA
in TE buffer.
19. Dissolve the DNA pellet in 200 μL of TE buffer (Tris-EDTA,
pH 8.0). Leaving the tubes overnight at room temperature
will assist in dissolving the DNA. This can be accelerated by a
65 °C incubation which has the added benefit of removing any
nuclease activity.
20. Determine the quality of the DNA by measuring the absor-
bance ratio (should be between 1.8 and 2.0) using a spectro-
photometer. Resolve the extracted DNA using electrophoresis
on a 0.7% or 1% agarose gel, to determine the quantity (by
comparing the intensity to a known amount high-molecular
weight DNA e.g., lambda DNA) and the level of shearing of
the extracted DNA.
21. Store the DNA at −20 °C.

3.1.4 Measurement There are a number of ways of measuring DNA concentration

of Genomic DNA which differ by convenience, accuracy, and precision. These meth-
Concentrations ods include simple UV (260 nm) absorbance, UV fluorescence of
dyes bound to double stranded DNA, and real-time PCR probes.
Although simple UV absorbance at 260 nm using devices such as
Nanodrop or conventional cuvette-based spectrophotometer is
convenient, accuracy is hampered by other UV absorbance of con-
Re-sequencing Resources to Improve Starch and Grain Quality in Rice 209

taminating molecules and the pH of the solution. Probe-based

real-time PCR quantification methods are extremely accurate.
However, they require significant time and effort to optimize. UV
fluorescence of dyes bound to double stranded DNA strikes an
appropriate balance between accuracy and convenience and is
described here. We have used both ethidium bromide stained DNA
in agarose gels and Pico-Green (PicoGreendsDNA Quantification
kit, Invitrogen) to measure the concentration of PCR products.
The protocol for quantification of the amplified DNA with
Ethidium bromide stained DNA in agarose gel electrophoresis is
detailed below:
1. The combs must create wells which have a volume which is
greater than the volume of the sample to be analyzed. A well
volume of around 20 μL will suffice for most analytical pur-
poses. Determine the volume of gel required to fill the gel tray
to a depth that satisfies this requirement.
2. Take a gel tray and seal it with masking tape at each end. It is
important to ensure agarose gel does not leak from the tray
(see Note 18).
3. Mix 1 g of agarose per 100 mL of Tris-Borate EDTA (TBE)
buffer in a Schott bottle. Mix by shaking (see Note 19).
4. Place the bottle in a microwave oven with the lid slightly open.
Heat the mixture on high setting, mixing occasionally, until
the agarose has melted. Check the agarose has melted by swirl-
ing the liquid, holding it up to the light and looking for small
lenticular, clear pieces of unmelted agarose.
5. When the agarose has melted, carefully pour it into the pre-
pared gel tray. Take care to avoid bubbles and make sure the
combs are covered to sufficient depth.
6. Let the agarose set by leaving it at room temperature for at
least 30 min.
7. Remove the masking tape and place the gel tray in an electro-
phoresis tank of the appropriate size, making sure the wells are
covered by TBE buffer.
8. Mix the samples with loading buffer on a piece of parafilm or
in Eppendorf or PCR tubes. Loading buffer contains dense
material such as glycerol or ficoll which ensures the sample
falls to the bottom of the well, and dyes such as bromophenol
blue and xylene cyanol so the sample can be seen. Depending
on the efficiency of the PCR, 2 μL of sample should be
sufficient.
9. Carefully load the samples into the wells of the gel. This is best
done by holding the pipette at approximately 45 °C, using
both hands, one to steady the pipette tip, and letting the sam-
ple drop into the well.
210 Gopala Krishnan Subbaiyan et al.

10. Load at least one of the wells with a DNA sample of known
size and concentration. More accurate concentration estima-
tion is achieved by using a dilution series of the sample of
known concentration.
11. Close the lid on the gel tank. Ensure the electrodes are cor-
rectly orientated (see Note 20).
12. Set the power pack to 5 V/cm. For example, if the distance is
between the electrodes is 20 cm, set the voltage to 100 V.
13. Run the gel until the bromophenol blue dye reaches the end
of the gel.
14. Remove the gel from the gel tank and place in a TBE solution
containing 2–5 μg/mL ethidium bromide (see Note 21). Use
TBE as the staining solution because this allows the gel to be
re-run if the results are not satisfactory.
15. Allow the gel to stain for 15 min.
16. Rinse the gel with tap water. This removes excess ethidium
bromide from the surface of the gel. Excess ethidium bromide
can be removed by placing the gel in TBE which does not
contain ethidium bromide. The DNA will retain the dye while
the surrounding gel destains, improving the contrast between
the DNA and background.
17. Take the gel and use a trans-illuminator to irradiate the gel
with UV light. Record the results of the experiment by captur-
ing an image of the gel digitally or on film.
18. Estimate the concentration of each amplicon by comparison
with the DNA standards.
There are a number of fluorescent dsDNA binding dyes which
can be used to measure the concentration of DNA, many of which
are proprietary. The Invitrogen Quant-iT™ PicoGreen®
(Picogreen) dye is a popular option because of its high sensitivity
and specificity for dsDNA and its linear response over several orders
of magnitude, even in the presence of contaminates such as salts,
detergents, proteins, and ssDNA (oligonucleotides) used in PCR.
In common with all other DNA binding dyes, Picogreen is a
potential mutagen and carcinogen. Picogreen is supplied in a
DMSO solution which should be handled with care as DMSO
facilitates entry of organic molecules into human tissues. Suitable
safety glasses, laboratory coat, and gloves should be worn at all
times. The Picogreen/DMSO undiluted stock solution should be
opened and diluted in a fume hood. Picogreen/DMSO must be
disposed of in accordance with the relevant regulations which gov-
ern the laboratory in which they are used.
The protocol for quantification of the amplified DNA with
PicoGreendsDNA Quantification kit, Invitrogen is explained
below:
Re-sequencing Resources to Improve Starch and Grain Quality in Rice 211

Table 1
Standard curve dilution series

Quant-iT™
(2 μg/mL DNA) Stock DNA concentration
TE volume (μL) PicoGreen® (μL) (μL) (ng/mL)
0 100 100 1000
90 10 100 100
99 1 100 10
999 1 1000 1
100 0 100 0 (Blank)

1. Warm the Picogreen to room temperature before opening the

vial.
2. If using a Invitrogen Quant-iT™ PicoGreen® kit, prepare a 1×
TE buffer (10 mM Tris–HCl, 1 mM EDTA, pH 7.5) by dilut-
ing the concentrated 20× buffer which comes with the kit
20-fold with sterile, distilled, DNase-free water. If using
Picogreen which is not sourced from a kit, take care to ensure
the TE buffer used to dilute the Picogreen is free of nucleic
acids since they bind with Picogreen, even at very low
(picogram) concentrations and interfere with the assay
­
(see Note 22).
3. Prepare the Picogreen by diluting the concentrated solution
1 in 200 in 1× TE (e.g., 100 μL Picogreen in 19.9 mL TE or
10 μL Picogreen in 1.9 mL) (see Note 23).
4. First dilute the DNA standard to a final concentration of 2 μg/
mL (or 2 ng/μL). This 2 μg/mL is the Stock DNA.
5. Make a series of duplicate dilutions in the first two rows of
each 96-well microplate, or the first row of a 384-well micro-
plate, which correspond to the amounts of TE, 2 μg/mL
Stock DNA, and Picogreen as described in Table 1.
6. If using a 100 μL/well 96-well microplate, add 4 μL of DNA
to each well followed by 96 μL of TE and 100 μL of the
Picogreen. If using a 25 μL/well 384 well microplate, add
1 μL of DNA to each well followed by 24 μL of TE and 25 μL
of the Picogreen (see Note 24).
7. Mix standards and amplicon samples by pipetting up and
down several times, ensuring no air bubbles are formed in the
wells.
8. Incubate at room temperature for 5 min protected from light
to ensure the DNA binding sites for Picogreen are fully satu-
rated with Picogreen.
212 Gopala Krishnan Subbaiyan et al.

9. Measure DNA standard and fluorescence with a microplate

reader at the fluorescein wavelengths of ~480 nm (excitation)
and ~520 nm (emission). To ensure readings remain in the
detection range of the fluorometer, ensure the instrument’s
gain is set so the DNA standard solution of 1 μg/mL has fluo-
rescence intensity just below the plate reader maximum.
10. Subtract fluorescence of the reagent blank from DNA stan-
dards. Either create a standard curve by plotting fluorescence
versus DNA concentration or, depending on the plate reader,
enter the data into an Excel spreadsheet which has a macro
that carries out the appropriate calculations.
11. Determine DNA concentration with reference to the standard
curve or by entering the data into an Excel spreadsheet which
has a macro that carries out the appropriate calculations.

3.2 Re-sequencing This protocol describes how to bulk the DNA, primer design, and
of Starch amplification of target genes by long-range PCR, amplicon pool-
Biosynthesis Genes ing, sequencing, and polymorphism (SNP) identification.
If the ultimate objective of measuring starch biosynthesis gene
allele frequencies is to be realized, it is important to ensure bias
toward any one sample, either random or systematic, is minimized
at each step of the process. The risk of bias arising during amplicon
sequencing is minimized when all amplicons are present in pools in
equimolar amounts.

3.2.1 Designing of PCR It is essential to choose the target genes which have the greatest
Primers effects on starch synthesis, composition, and quality. For this pur-
pose, a complete literature review was done and specific starch
quality genes chosen for targeted amplification and deep sequenc-
ing. The correct gene sequence (gDNA and cDNA) is needed to
design primers for amplification (see Note 25). The genes involved
in starch biosynthesis and the sequence of the primers of each of
the rice starch genes have been reported by Kharabian-Masouleh
et al. [11] and are publicly available here: http://onlinelibrary.
wiley.com/doi/10.1111/j.1467-7652.2011.00629.x/suppinfo
(Table S3).
The following major databases can be used to find the sequence
of target genes (see Note 26):
1. National Centre of Biotechnology Information (NCBI):
(http://www.ncbi.nlm.nih.gov/).
2. Rice Genome Annotation Project: (http://rice.plantbiology.
msu.edu/cgi-bin/putative_function_search.pl).
3. The sequence data can be saved in separate MS Word files or in
FASTA format in Notepads.
4. For alignment purposes, it is preferable to save the data in
FASTA format in Notepad files with recognizable naming.
 Re-sequencing Resources to Improve Starch and Grain Quality in Rice 213

3.2.2 Primer Design 1. This protocol can be used for any gene varying in length from
for Long-Range PCR 1 to 20 kb or even beyond.
(LR-PCR) Capture 2. Each gene sequence can be split into half or more if the length
and Amplification of Target is greater than >5 kb. However, this can be changed depend-
Gene Sequence ing on the performance and efficiency of polymerase used for
amplification. High-fidelity polymerases enzymes will produce
better results because they can amplify genes with greater
speed and accuracy. This will be explained later in this
chapter.
3. The following criteria should be met during the primer design:
Each sequence should include ~500–1000 bp extra from up-
and downstream of the coding regions to ensure that no vari-
ant will be missed from 5′ and 3′ UTRs. Approximately,
150 bp overlap at the end of neighboring fragments is required
to minimize the bias which normally occurs at the both end of
sequencing sides. This will enhance accuracy of contig
construction.
4. Specific primers can be designed for each gene (of gene frag-
ment) using software available for free online such as Primer3
(http://bioinfo.ut.ee/primer3-0.4.0/). However, we recom-
mend Clone Manager V9.1 (Sci-Ed Software, Cary, NC)
(http://www.scied.com/pr_cmbas.htm), which produces
high quality and efficient primers and also encompasses a com-
prehensive set of tools for sequence manipulation and analysis,
sequence alignment, and primer design.
5. To begin primer design, divide the whole sequence (in length)
into half. For example, if gene X is 10,000 bp in length, it can
be divided into two fragments of approximately 5000 bp, with
a 150 bp overlap in the middle (Fig. 1).
6. Save the halved sequence in the same or different files on your
computer.
7. Run Clone manager software, click New (New Molecule
Wizard) on top menu bar (left corner) to create a new
molecule.
8. Choose “Enter Complete Information.”
9. Insert a molecule name; choose DNA from drop down list.
Molecule description and notes are optional.
10. Click next; then paste the DNA molecule sequence and length.
11. Click next; then choose linear DNA.
12. In the next step, the molecule features are optional. Click next
and now your molecule has been created.
13. Go to “Primer” on top bar menu and then click on “Design.”
214 Gopala Krishnan Subbaiyan et al.

Fig. 1 The gene of interest (blue color) may be divided into two sections, depending on the length of gene. This
is recommended for genes with length more than 5 kb. The red segments are 3′ and 5′ UTRs. A 150 bp (or
less) overlap might be in the middle

14. Give your primer design details such as primer type (PCR
Primer Pair), choose primer length for forward and reverse
(usually ~20–25 bp) and then click next.
15. In “Source Molecule,” browse and choose your molecule that
you already have created.
16. For primer target position type number 1 and then 100.
17. Click finish.
18. The program will match the created DNA sequence and posi-
tion a primer of the length you specify at the initial position
you specify. You can then move the primer along the sequence
(by tapping on left and right arrows of keyboard) or adjust its
length until the primer evaluation meets the criteria set.
19. Order primers.

3.2.3 Long-Range PCR For amplification, a semi- to long-range PCR (LR-PCR) needs to
to Amplify Target Genes be employed in which each gene is amplified with 1–2 pairs of
primers (see Note 27).
The following procedure should be used for accurate amplifi-
cation of starch genes:
1. Concentration of extracted DNA should be quantified, using
the automated flurometric protocol of PicoGreen
(PicoGreendsDNA Quantification kit; Invitrogen, San Diego,
CA).
2. The concentration of each individual DNA sample must be
diluted to 10–20 ng/μL. DNA of 20 individuals should be
mixed for each reaction.
3. Each fragment gene (halved) can be amplified separately to
find the optimum PCR condition. However, as the number of
gene fragments and genotype individuals are often quite large,
it is better to use a unified LR-PCR approach for all genes and
fragments.
Re-sequencing Resources to Improve Starch and Grain Quality in Rice 215

4. It is preferable to apply the LR-PCR protocol to one standard-

ized DNA sample (e.g., Nipponbare) (see Note 28). This will
help determine whether the PCR conditions are optimum and
the primers are functional or not.
5. After finding the suitable protocol and efficient primers for
amplification of fragments, the protocols and primers can be
applied to different DNA samples.
6. The BioRadiProof TM High-Fidelity DNA polymerase kit
(BioRad Laboratories, Hercules, CA) is strongly recommended
for LR-PCR (see Note 29).
7. TheBioRadiProof TM High-Fidelity DNA polymerase kit con-
tains two amplification buffers (HF and GC). The HF (High
Fidelity) is used initially and if the fragment does not amplify
the GC buffer can be used. The GC buffer is normally used for
GC-rich, difficult to amplify fragments.
The details of reaction mixture preparation for LR-PCR ampli-
fication is presented below:
1. DNA of 20 individuals (15 ng/μL) or less should be mixed for
each reaction.
2. All primers (Forward and Reverse separately) should be diluted
to 2.5 μM and stored in −20 °C until use.
3. Prepare a master mix then dispense into separate 1.5 mL tube
and add different primers (Forward and Reverse for each tar-
get gene amplicon), as required.
4. Two sets of reactions can be prepared; first with HF. If some
primers did not amplify then a master mix should be prepared
with GC buffers as follows:

Volume for 20 μL
Components Concentration reaction (μL)
Buffer HF and GC 1× (exactly as provided in kit) 4
dNTPs (mix) 10 mM 0.4
Primer 1 (Forward) 2.5 μM 4
Primer 2 (Reverse) 2.5 μM 4
DNA template 15 ng/μL 2
iProof DNA 0.02 U/μL 0.2
polymerase
Sterile water 5.4
Total volume 20
216 Gopala Krishnan Subbaiyan et al.

5. HF buffer can amplify most starch synthesis genes. However,

some such as SSIIa may need GC buffer for amplification.
LR-PCR amplification thermo-cycling parameters are pre-
sented below:
1. Elevated denaturation and annealing temperatures might be
required for iProof as its reaction buffers contain more salts.
2. Each gene/primer has its specific annealing temperature.
Therefore, too many different cycling temperatures might be
“optimum” for each primer. To avoid this problem and shorten
the time to find the optimum thermo-cycling condition, we
recommend a hot touch-down protocol to amplify all genes in
one cycling condition (see Note 30).

Number
Cycle step Temp (°C) Time of cycles
Initial denaturation 98 1 min 1
Touch down 98 10 s
denaturation
Touch down annealing 72–62 20 s (in each cycle temp 10 Cycles
will reduce 1 °C)
Touch down extension 72 4 min
Denaturation 98 10 s
Annealing 62 20 s 28 Cycles
Extension 72 4 min
Final extension 72 10 min

3. Prior to pooling the DNA for Illumina sequencing, the PCR

products should be subjected to Sanger sequencing using
BigDye Terminator version 3.1 (Applied Biosystems, Foster
City, CA) to ensure that the target gene is captured.
4. The sequences generated through Sanger sequencing should
be aligned with the reference sequence. This will help ensure
the target gene has been captured and avoid the cost of high-­
throughput sequencing of non-target sequences.

3.2.4 Quantification Precise quantification of amplicons is a prerequisite for creation of

of Amplicons for Creation equimolar pool as any error in this step may lead to bias in the
of Equimolar Pool representation of the samples in the sequences. Although there are
several methods for measurement of DNA concentration, Pico-­
Green (PicoGreendsDNA Quantification kit, Invitrogen) is one of
the preferred methods.
 Re-sequencing Resources to Improve Starch and Grain Quality in Rice 217

3.2.5 Amplicon Pooling A uniform DNA pooling approach should be applied for all sam-
(Pooling of Amplified DNA) ples, before and after amplification. This is called equimolar pool-
ing, in which normalized DNA samples (15–20 or 15 ng/μL)
from each sample should be pooled in equimolar concentration to
make a megapool. For amplification (capture stage), DNA of 20
individuals (or less) should be mixed for each reaction tube (gene
segment). The following procedure should be followed after
amplification:
1. The concentration of PCR products should be measured, pref-
erably using Pico-Green (PicoGreendsDNA Quantification
kit, Invitrogen).
2. Prepare a second pool for each amplicon (amplified fragment).
This means that PCR products of all tubes (each tube contains
DNA of 20 individuals) related to one amplified fragment
should be pooled. This will facilitate the final equimolar pool-
ing of PCR products to prepare the megapool.
3. After mixing PCR products related to each fragment (second
pool), concentration of second pools must be measured again
using Pico-Green (PicoGreendsDNA Quantification kit,
Invitrogen) or Nanodrop, using the protocol here: http://
nanodrop.com/Library/CPMB-1st.pdf.
4. The concentration of each second pool should be normalized
to 25 ng/μL, using distilled-deionized water.
5. For preparation of mega pools, the length of each fragment
(bp) must be taken into consideration. Longer amplicons
require more DNA to be added than smaller fragments to
ensure the numbers of fragment copies are equal. The relative
amount added is directly proportional to length, twice as much
of a 10,000 bp a fragment is added relative to a 5000 bp
fragment.
6. As 2.5 ng of amplicons in 2500 μL (Megapool including all
amplicons of all fragments) are needed for massive parallel
sequencing, the following formula must be used for
calculations:
Amount of amplification product required (μL) = (Length
of each fragment (bp))/(Total length of all genes (bp)) × 2500
(μL).
7. Example: The length of first fragment (half of gene) of
GBSSIH1 is 2164 bp. If we consider the total length including
all genes that needs to be sequenced (e.g., 17 genes with totally
218 Gopala Krishnan Subbaiyan et al.

34 halved fragments) as 125,582 bp (~125 kb) then the

amount required will be calculated as follows:
Amount required for GBSSIH1 (μL) = (2164 (bp))/
(125,582 (bp)) × 2500 = 43.079 μL.
8. All other fragments must be calculated as above and added to
the final Megapool for Illumina sequencing.

3.2.6 Amplicon Sequence the pooled amplicons using a MiSeq (Illumina) or simi-
Sequencing lar sequencing platform. Data volumes need to be large enough to
obtain sufficient coverage of the total length of amplicons in the
bulk to allow distinction of SNP represented by a single individual
(in the bulk) from a sequencing error.

3.2.7 Data Analysis After massive parallel sequencing (Illumina), the output data will
to Discovery be generated, which the data is analyzed with the software package
Polymorphisms CLC Bio. The following parameters will be given for each gene
in Population separately:
1. Total length of assembled sequence for each gene, including
coordinates and chromosome number.
2. Total number of variants for each gene, including non-­
synonymous SNPs (nsSNPs), synonymous SNPs (Functional
SNPs), and Indels (insertion deletions) in coding or noncod-
ing regions. All the information about SNP/Indels comes with
frequency of SNPs/Indels, their positions, coordinates, and
type of amino acid change that they may cause. Figure 2 has
been generated by CLC Bio and shows distribution of SNP
across the gene of interest.
The following parameters can be calculated manually, using
the data output supplied by software:
1. The total rate of polymorphism is calculated as: TSI/TL × 100,
where TSI = total number of SNPs and Indels, and TL is the
total length of each candidate gene.
2. The rate of functional (nonsynonymous SNPs) can be calcu-
lated as: NS/TL × 100, where NS = number of nonsynony-
mous SNPs in each locus and TL is the total length of each
candidate gene.
3. Number of SNPs per kb length of gene, which indicates the
variability of point mutations in specific locus of the
population.
4. Number of InDels per kb length of gene, which indicates the
variability of larger mutations in specific locus of the
population.
5. Ka/Ks ratio, which is the proportion of nonsynonymous (Ka)
relative to synonymous (Ks) SNPs. The Ka/Ks ratio indicates
 Re-sequencing Resources to Improve Starch and Grain Quality in Rice 219

14
12 SSllb
10
8
6
4
2
0
0 1 2 3 4 5 6 7 8 9
4g0624500 Os04g0624600

Fig. 2 SNP/Indel distribution and short read coverage pattern across SSIIb gene generated by CLC Bio after
massively parallel sequencing (Illumina). The X-axis indicates the length of sequenced area (genes) in kb and
Y-axis shows the number of detected SNPs/InDels. The graphs show the distribution of SNPs across the gene
(The values under zero must be regarded as zero). Graphics in the middle side show the relevant gene
(Blue = introns and Yellow = exons). Graphics in the down side (pink color) show the coverage pattern of each
gene (Adapted from Kharabian-Masouleh et al. [11])

nature of selection pressure exerted on the gene during evolu-

tion. The Ka/Ks ratio = 1 means that selective condition of
evolution (perhaps by environment or human manipulation)
has been neutral. The Ka/Ks ratio < 1 shows the negative or
purifying selection pressure, where Ka/Ks ratio > 1 indicates
positive or diversifying (adaptive) effects on gene during selec-
tion [12].

3.3 Whole-Genome Whole-genome polymorphism detection by sequencing [10, 12] is

Re-sequencing a powerful tool for association genetics. High-throughput and cost
effective sequencing technologies have aided discovery of millions
of genome-wide DNA polymorphisms such as Single-Nucleotide
Polymorphisms (SNPs) and Insertion-Deletion (InDels) which are
an invaluable resource for marker-assisted breeding [13]. SNPs
and InDels are becoming the preferred markers in molecular
breeding due to multiple advantages such as high frequency, stabil-
ity, high-throughput capability, and cost effectiveness over other
DNA markers [13]. Next-generation sequencing technologies
make possible the discovery of a massive number of DNA poly-
morphisms by comparing the whole-genome sequences of indi-
viduals with high-quality reference genome sequences [13]. SNPs
have been employed in breeding programs for marker-assisted and
genomic selection, association and QTL mapping, positional clon-
ing, haplotype and pedigree analysis, seed purity analysis, and vari-
ety identification [14].
220 Gopala Krishnan Subbaiyan et al.

3.3.1 Whole-Genome There are a number of different Next-Generation Sequencing

Sequencing of Genomic (NGS) platforms that can be used for whole-genome re-­sequencing
DNA [15, 16]. Data volumes need to be large enough to obtain suffi-
cient coverage of the total length of the genome with good depth
of coverage to allow distinction of a SNP relative to the reference
genome from a sequencing error. Briefly, the protocol for sample
preparation and whole-genome sequencing the DNA using
Illumina GAIIx is presented below:

Sample Preparation The DNA is sheared, polished, and prepared following the manu-
and Sequencing facturer’s instructions, Kit FC-102-1002 (Illumina sample prepa-
ration protocol for paired-end sequencing) and sequenced in
Illumina GAIIx.
1. About 1 μg of the sample DNA is sheared using the adaptive
focused acoustics method on a Covaris S2 device with the fol-
lowing settings: duty cycle 10%; intensity 5; cycles per burst
200 for 180 s at 6 °C.
2. Ligation products are purified by electrophoresis on Invitrogen
E-gel Size Select 2%.
3. The fragments in the range of predominantly 450-bp size are
selected from the gel.
4. PCR products are further purified with a QIAquick PCR
Purification Kit and quantified using a DNA 1000 chip on an
Agilent BioAnalyzer 2100.
5. Approximately 4 pmol per individual 76 × 2 cycles on an
Illumina Genome Analyzer (GAIIx) using sequencing kit v4.
6. Base calling was performed with Illumina software Pipeline 1.4
(Illumina, San Diego, CA, US).

3.3.2 Analysis of Whole-­ The bioinformatics tools presently available help discovery of
Genome genome-wide polymorphisms by comparing the whole-genome
Re-sequencing Data sequence of individual genotypes with high-quality reference
genome sequences [13]. CLC Genomics workbench is used to
assemble the trimmed, high-quality reads from each of the geno-
types individually to the reference genome. A schematic overview
of the steps involved in whole-genome sequence data analysis and
discovery of variants (SNPs and InDels) in comparison with the
high-quality reference genome sequence with whole-genome re-­
sequencing data is presented in Fig. 3.

3.3.3 Quality Control The Paired-end sequence reads needs to be checked for their qual-
of the Whole-Genome ity and the low-quality sequence reads, as well as the adapter
Sequence Data sequences should be removed before further analysis. In order to
achieve this, the short reads are subjected to the process of trim-
ming in CLC genomics workbench 5.0 (http://www.clcbio.com),
which involves the following steps:
 Re-sequencing Resources to Improve Starch and Grain Quality in Rice 221

Fig. 3 An overview of whole-genome sequence data analysis and discovery of

variants

1. The paired-end sequences are trimmed using trim function

with a quality score limit of 0.01.
2. The adaptor sequences are also removed using the trim option
in CLC Genomics Workbench 11.0.1 (http://www.clcbio.
com).
3. Additionally, the reads of less than 30 base pairs (bp) in length
are also discarded from the short reads, after which the short
read sequences are ready to mapping.

Mapping of Reads 1. The trimmed short-read sequences are first aligned to the
organellar genomes of the rice reference genome (Chloroplast
genome, mitochondrial genome) (see Note 31).
2. The unmapped reads are then taken up for further assembly
against the nuclear genome of the reference genome. In case of
rice, the reads were assembled to the Nipponbare reference
pseudomolecule 4.0 with CLC Genomics workbench.
3. The mapping parameters generally adopted in CLC genomics
work bench involve mismatch cost—2, insertion cost—3, dele-
tion cost—3, length fraction—0.9, and similarity—0.9 (see
Note 32).
4. Then, the reads that align to unique genomic positions need to
be identified through the removal of the reads that aligned to
222 Gopala Krishnan Subbaiyan et al.

more than one position of the reference genome using filtering

option in CLC and used for determining reads mapping to
multiple positions in the reference and unmapped reads.

Calling of Variants (SNPs The assembled contigs are screened for SNPs and InDels using
and InDels) CLC Genomics workbench 11.0.1 for detecting the variants across
all genotypes mapped to the reference genome. The SNP detec-
tion tool is helpful in discovering SNPs in the assembled contigs in
comparison to the high-quality reference genome.
1. The criteria adopted for detection of variants include: mini-
mum coverage: 10×, and minimum frequency of the least fre-
quent allele: 10%, with additional parameters of a window of
21 bp at a central base quality score of 30 with the surrounding
base quality of 20 or more for SNP detection.
2. SNPs and InDels for each of the DNA samples can also be
detected based on the contigs from individual assemblies with
minimum coverage of 4×.
3. Additionally, the minimum variant frequency is very stringent
with >90% for SNP and >30% for InDel detection in individual
DNA samples.
4. The SNPs from repetitive regions can be eliminated through
selection of “No repeats” option in the overlapping annotation
tab in CLC Genome workbench.
5. The SNPs and InDels belonging to specific starch related genes
can be filtered out based on the LOC-Ids and the genomic
coordinates (Table 2) using filtering option in CLC bio.
6. In order to estimate the gene diversity and the other parame-
ters as described in the above section, the complete sequence
of the gene for each genotype can be extracted from the assem-
bled reads as consensus assembly.
7. The SNPs/InDels and their association with the starch pheno-
type can be assessed using TASSEL software as described in the
association analysis section (Subheading 3.5).

3.4 Genotyping Usually, a large number of samples and many single-nucleotide

of SNP in Rice polymorphisms (SNPs) involve in starch bio-synthesis analyses;
Genotypes therefore a rapid, inexpensive, and highly automated multiplex
method is required to genotype the sequence variants. For exam-
ple, Kharabian-Masouleh et al. [3] discovered 66 functional (cod-
ing) SNPs in 233 Australian rice samples, encompassing 18 starch
bio-synthesis genes. Using conventional Sanger sequencing
method, a total of 66 × 233 = 15,378 separate sequencing reac-
tions are required to genotype SNP-individuals. In this section, we
explain an optimized high-throughput multiplexed SNP assay
(Sequenom®) to detect SNP polymorphisms in a large number of
samples. Then, we explain how the genotyping data can be matched
with phenotype to distinguish association between SNPs and traits.
Table 2
List of genes involved in starch biosynthesis and their chromosomal location

Genome coordinates
Chromosomal
S. No. Name of the gene/enzyme Gene ID (NCBI) Gene ID (MSU) location Start End
1 AGPS2b, [ADP-glucose pyrophosphorylase Os08g0345800 LOC_Os08g25734 8 15760599 15754206
(Small Unit)]
2 SPHOL (alpha 1,4-glucan phosphorylase) Os03g0758100 LOC_Os03g55090 3 32183093 32190581
3 GPT1 (Glucose-6-phosphate/phosphate-translocator) Os08g0187800 LOC_Os08g08840 8 5138640 5142712
4 Granule Bound Starch Synthase I (Waxy gene) Os06g0133000 LOC_Os06g04200 6 1764623 1769657
5 Granule Bound Starch Synthase II Os07g0412100 LOC_Os07g22930 7 13584483 13576435
6 Starch Synthase I Os06g0160700 LOC_Os06g06560 6 3078060 3078060
7 Starch Synthase IIa Os06g0229800 LOC_Os06g12450 6 6747562 6751981
8 Starch Synthase IIb Os02g0744700 LOC_Os02g51070 2 32125071 32119749
9 Starch Synthase IIIa Os08g0191500 LOC_Os08g09230 8 5351108 5362370
10 Starch Synthase IIIb Os04g0624600 LOC_Os04g53310 4 32149493 32158120
11 Starch Synthase Iva Os01g0720500 LOC_Os01g52250 1 31786842 31797321
12 Branching enzyme I Os06g0726400 LOC_Os06g51084 6 31775431 31782688
13 Branching enzyme IIa Os04g0409200 LOC_Os04g33460 4 20260837 20265349
14 Branching enzyme IIb Os02g0528200 LOC_Os02g32660 2 20213965 20224864
Re-sequencing Resources to Improve Starch and Grain Quality in Rice

15 Debranching enzyme-isoamylase 1 Os08g0520900 LOC_Os08g40930 8 25981756 25988347

223
224 Gopala Krishnan Subbaiyan et al.

3.4.1 DNA Extraction The DNA should be extracted from leaves of seedlings and nor-
malized for PCR reactions as described in previous sections. The
Qiagen (Valencia, CA, USA) DNeasy Plant Kit might be used
according to the instructions provided earlier.

3.4.2 Primer Design/ Primer design should be done using MassARRAY® Assay design
Generation of SNP Markers v3.1 software. The user’s manual can be accessed here: http://can-
cerseqbase.uchicago.edu/documents/AssayDesign3.1Guide.pdf
In this system, there are two primers. First, amplification prim-
ers (Forward and reverse), which amplify the target gene fragment,
including SNP(s); and second, the extension primer. Generally, the
extension primer is designed to hybridize directly adjacent to a
SNP on a single (amplified) fragment of DNA and then to extend
by one of four mass-modified nucleotides. The mass-modified
nucleotides are terminator molecules specific to the DNA nucleo-
tide 3′ of the primer (Fig. 4).
The steps must be taken for capture and extension design:
1. Install Sequenom® MassARRAY® Assay design 3.1 software
on your computer. The package usually comes with the pur-
chased platform.
2. Open the software by double-click and choose your desired
MassEXTEND Assay Type.
3. Click on Browse and select the file (.txt) that you have already
prepared in .txt format (NotePad). The text should include the
sequence of all genes and involving SNPs placed in brackets
(Fig. 5).
4. There should be a heading as SNP_ID SEQUENCE in the
first line. Each gene should have a separate name (Fig. 6).
5. Usually two assay types are offered by software: MBE Mass
Extend (Multiple Base Extension) and SBE Mass Extend
(Single-­Base Extension). These terms refer to how the extend
primers are designed.
6. For starch genes, use SBE (Single-Base Extension), which is
chemically coupled with choices for a specified stop mix
(Fig. 4c). As shown in Fig. 4c, the SBE assays analyze SNPs by
a single-nucleotide extension after the extend primer bound
into the region of DNA sequence variation.
7. Choose the multiplex level (1–20). We do not recommend the
multiplex beyond 20. This means that the maximum combined
assays will not be more than 20 in one PCR (5 μL).
8. Choose your SNP Capture settings. Set the amplicon size as
80, 100, and 120 bp, for Minimum, Optimum, and Maximum,
respectively.
9. Sequence annotation should be set to Scan and Restrict.
10. Two (10-mer) tags will be added to the 5′ end of each ampli-
fication primer by default to avoid confusion in the mass spec-
trum and to improve PCR performance (5-ACGTTGGATG-3).
Fig. 4 Application of multiplexed-MALDI-TOF MS assays to detect SNP (genotyping). (a) Double strand DNA including
SNP site (G/T). A targeted length of 80–120 bp (spanning SNP site) should be amplified. (b) This will be done using
specific forward and reverse primers (including 10-mer tag). (c) Amplification with extension primer (single-strand
DNA) and termination using mass-modified terminator nucleotide. (d) The mass of extended molecule plus mass-
modified terminator nucleotide will be detected using MALDI-TOF MS. (e) The output data will pass through a com-
puter software (Typer) and peak show the SNP genotype. The results will also be available in EXCEL spreadsheet
226 Gopala Krishnan Subbaiyan et al.

Fig. 5 Input texts file showing two SNPs in starch waxy gene (GBSSI). Please note how the headings should be
named and SNP sites must be placed in brackets [G/T]. All SNP-sequences can be placed in one file to be
imported into ASSAY Design software for primer design

Fig. 6 Preparation of trait file (Phenotypic data input)

 Re-sequencing Resources to Improve Starch and Grain Quality in Rice 227

Table 3
Preparation of PCR reactions to capture starch genes for multiplex SNP assays

Volume for 20 μL Volume for 5 μL

Components Concentration reaction (μL) reaction (μL)
10× PCR buffer 10× (provided in kit) 2 0.5
(InviTrogen)
MgCl2 50 mM 1.4 0.35
dNTPs (mix) 2.5 mM 4 1
Primer 1 (Forward) 1 μM 2 0.5
Primer 2 (Reverse) 1 μM 2 0.5
DNA template 3–5 ng/μL 1 0.25
Platinum® Taq DNA 2 U/rxn 0.3 0.075
Polymerase
Sterile water – 7.3 1.825
Total volume 20 5

11. For extend primer design (extension primer), choose the fol-
lowing: Tm (°C) 45–100, by “NN”, Length 17–28 and leave
the others as default.
12. Run the program and you will get all primers designed in an
Excel file ready to be ordered.

3.4.3 Gene Capture PCR Using the aid of amplification primers, we will be able to capture
each starch gene in uniplexed or multiplexed assay. Each -plex is
referred to a single reaction including a forward and reverse primer
able to amplify a gene fragment spanning the SNP site. Before
starting the multiplexed genotyping, if there is sufficient time and
labor, it is recommended to set up a series of uniplex reactions
(assays) to test the performance of primers. Otherwise, the multi-
plexed experiments can be conducted. The detailed instruction for
iplex genotyping can be found here: http://cancerseqbase.uchi-
cago.edu/documents/iPLEXGoldApplicationGuide.pdf
The PCR reaction mixture is recommended for an 8-plex
experiment (to capture eight SNP sites):
1. The final volume can be 5 μL but volumes more than this up
to 20 μL can be used.
2. The Platinum® Taq DNA Polymerase (Invitrogen, Carlsbad,
CA, USA) is recommended for amplifications but any other
high-quality Taq DNA Polymerases might be successfully
applied.
3. Table 3 shows how to prepare a 20 μL (only for one assay);
therefore, for a 5 μL reaction the volume must be divided by 4.
228 Gopala Krishnan Subbaiyan et al.

Table 4
Thermocycling to capture starch genes

Number of
Cycle step Temp (°C) Time cycles
Initial denaturation 94 15 min 1
Denaturation 94 20 s
Annealing 56 30 s 45
Extension 72 1 min
Final extension 72 3 min 1
Storage 4 Forever –

4. Primers 1 and 2, and template DNA of each genotype should

be added separately (should not be added to the master mix).
After preparing the master mix (PCR cocktail), the following
steps need to be carried out:
1. Dispense 15 μL of PCR cocktail (without primers and DNA
template) into each well of the 384-well plate.
2. Centrifuge the microtiter plate at 110 × g (1000 rpm) for 1 min.
3. Add relevant primers (F + R) and DNA template of genotype
to be assayed.
4. Now, the total volume will be 20 μL (or 5 μL).
5. Gently mix the plate over vortex and spin down before
thermocycling.
6. The first PCR cycling to amplify the gene is as follows (Table 4).

3.4.4 Shrimp Alkaline Before applying extension primer, remaining unincorporated dNTPs
Phosphatase (SAP) from PCR products should be neutralized by SAP incubation. This
Incubation will allow the mass modified terminators to bind adjacent to exten-
sion primer to recognize SNP variations by MALDI-TOF MS.

Preparation of SAP Enzyme 1. Use a 1.5 mL tube to prepare the SAP enzyme solution
Master Mix ­(master mix) (Table 5).
2. Gently vortex the 1.5 mL tube for 5 s.
3. Centrifuge the 1.5 mL tube for 10 s at 2793 × g (5000 rpm).
4. Take a 96-well plate (preferably Vee bottom Sarstedt, Inc.
#82.1583).
5. Choose the last bottom row (H row) (see Note 33) and dispense
85 μL of SAP enzyme into each well of this row (see Note 34).
6. Use a 12-channel pipettor and draw the SAP enzyme from the
row H and distribute into each well in rows G to A (10 μL per
well, each time) (see Note 35).
 Re-sequencing Resources to Improve Starch and Grain Quality in Rice 229

Table 5
Preparation of SAP solution

Volume
Components Volume (1rxn) (μL) (384rxn)a (μL)
Water (HPLC grade) 1.53 810.9
SAP Buffer (10×) 0.17 90.1
SAP enzyme (1.7 U/μL) 0.30 159.0
Total volume 2.00 1060.0
This includes a 38% “overhang”
a

SAP Enzyme Addition The SAP enzyme dispensed in 96-well plate microtiter should be
used to add the enzyme into the PCR reactions. For this reason, a
liquid handler machine should be used (supplied along with
Sequenom platform) as follows:
1. Place the 96-well microtiter plate containing SAP solution on
position 1 (left-up corner) of the liquid-handler deck.
2. Spin down the 384-well microtiter sample plate containing
amplification products at 110 × g (1000 rpm) for 1 min.
3. Put the 384-well microtiter sample plate on a plate flattener
(provided by Sequenom manufacturer) and then place it on
liquid-handler deck position 3 (left-down).
4. Run the “SAP Addition (96 to 384)” method on the liquid-­
handler controller PC.
5. Enter number 1 (default) for the number of plates.
6. By default 2 μL of SAP solution will be added to each well of
384-well, containing PCR products (see Note 36).
7. Remove the 384-well sample microtiter plate and flattener
from position 3, and then seal the plate with plate film.
8. Remove the microtiter plate from its plate flattener and centri-
fuge for 1 min at 110 × g (1000 rpm).
9. Incubate the 384-well sample microtiter plate in thermocycler,
as follows: 37 °C for 40 min, 85 °C for 5 min, and 4 °C
forever.
10. When the SAP incubation is being done, try to prepare the
iPLEX Gold reaction cocktail.

3.4.5 Preparation 1. The iPLEX Gold reaction should be prepared in a 1.5 mL tube
of iPLEX Gold Reaction as described in Table 6 (see Note 37).
for Different DNAs 2. Gently vortex the 1.5 mL tube for 5 s.
of 96–384 Samples
(Well-Plate)
3. Centrifuge the 1.5 mL tube for 10 s at 2793 × g (5000 rpm).
4. Take a 96-well plate (preferably Vee bottom Sarstedt, Inc.
#82.1583).
230 Gopala Krishnan Subbaiyan et al.

Table 6
Preparation of multiplexed iPLEX Gold reaction cocktail

Components Concentration in 9 μL Volume (1rxn) (μL) Volume (384rxns)a (μL)

Water (HPLC grade) N/A 0.7395 391.97
iPLEX Buffer Plus (10×) 0.222× 0.200 106.00
iPLEX Termination mix 0.5× 0.100 53.00
Primer mix (14 μM) 1.25 μM 0.94 498.12
iPLEX enzyme 0.5× 0.0205 10.87
Total volume 2.000 1060.0
This includes 38% “overhang”
a

5. Choose the last bottom row (H row) and dispense 85 μL of

iPLEX Gold reaction cocktail into each well of this row.
6. Use a 12-channel pipettor and draw the iPLEX Gold reaction
cocktail from the row H and distribute into each well in rows
G to A (10 μL per well, each time) (see Notes 38 and 39).
7. Centrifuge the microtiter plate for 1 min at 280 × g (1600 rpm).

Addition of iPLEX Gold 1. Place the 96-well microtiter plate containing iPLEX Gold reac-
Reaction Cocktail into PCR tion solution on position 1 (left-up corner) of the liquid-­
Amplification Products handler deck.
from Gene Capture Stage 2. Spin down the 384-well microtiter sample plate containing
Using amplification products at 110 × g (1000 rpm) for 1 min.
the Liquid-Handler Deck
3. Put the 384-well microtiter sample plate on a plate flattener
(provided by Sequenom manufacturer) and then place it on
liquid-handler deck position 3 (left-down).
4. Run the “Cocktail Addition (96 to 384)” method on the
liquid-­handler controller PC.
5. Enter number 1 (default) for the number of plates.
6. By default 2 μL of iPLEX Gold cocktail solution will be added
to each well of 384-well, already SAP-treated PCR products
from gene capture stage (see Note 36).
7. Remove the 384-well sample microtiter plate and flattener
from position 3, and then seal the plate with plate film.
8. Remove the microtiter plate from its plate flattener and centri-
fuge for 1 min at 110 × g (1000 rpm).

Addition of iPLEX Gold The same procedure as above can be manually repeated using a
Reaction Cocktail into PCR 12-channel pipettor. 2 μL of iPLEX Gold reaction cocktail should
Amplification Products be added to each well of the sample microtiter plate.
from Gene Capture Stage
Using 12-Channel Pipettor
(Manual)
 Re-sequencing Resources to Improve Starch and Grain Quality in Rice 231

3.4.6 Thermocycling In this stage, the extension primer(s) binds adjacent to SNP site
the Mixture of iPLEX Gold and the mass-modified terminator terminates the extension reac-
Reaction and PCR tion. The mass of amplified molecule will be detected by MALDI-­
Amplification Products TOF, which will be converted by computer program to reveal the
(Gene Captured) SNP genotype of individual.
The microtiter plate should be placed in a thermocycler pro-
grammed with the parameters described in Table 7.

Cleaning Up the Reaction The next step is to clean up the iPLEX Gold reaction products
Products using Clean Resin (resin). This step optimizes the mass spectrom-
etry analysis.
1. Take a 6 mg resin 384-well dimple plate (reservoirs).
2. Transfer the resin from its container onto the 6 mg resin 384-­
well dimple plate (provided by Sequenom Company).
3. Spread the resin into the wells of the dimple plate using a scraper
and make sure that resin sits inside each well.
4. This can be done by scraping the resin back and forth.
5. Remove excessive resin of the dimple plate using the same
scraper.
6. The excessive resin should be returned to the container for fur-
ther uses.
7. Leave the resin in the dimple plate for at least 20 min.

Adding Water to the 384-­ 1. Place a reservoir of 80 mL including nanopure water on posi-
Well Sample Microtiter tion 1 of the liquid-handler deck (top-left corner).
Plate (iPLEX Gold Reaction 2. In the same time, centrifuge the 384-well sample microtiter
Products) plate for 1 min at 110 × g (1000 rpm).

Table 7
Thermocycling of iPLEX Gold reaction and PCR amplification products (see
Note 40)

Number of
Cycle step Temp (°C) Time cycles
Initial denaturation 94 30 s 1
Denaturation 94 5s 40 Cycles
Annealing 52 5s 5 Cycles
Extension (for extend 80 5s
primer)
Final extension 72 3 min 1
Storage 4 Forever –
232 Gopala Krishnan Subbaiyan et al.

3. Remove the plate sealing film and place the 384-well sample
microtiter plate on a plate flattener. Then, place sample
microtiter and flattener on position 3 of the liquid-handler
deck (Down-left side).
4. Run the “16 μL Water Addition (384)” method from the liq-
uid handler controller PC.
5. Leave the default setting unchanged, for 1 destination plate.
6. The liquid handler will add 16 μL of water into each well of
sample microtiter plate.
7. Remove the 384-well sample microtiter plate and flattener
from position 3, then seal the 384-well sample plate with seal-
ing film.
8. Remove the plate flattener.
9. Centrifuge the microtiter plate for 30 s at 280 × g (1600 rpm)
to remove air bubbles.

Addition of Resin 1. Remove the sealing film from the sample microtiter.
to the Sample 2. Carefully place the sample microtiter plate, upside-down, onto
Microtiter Plate the dimple plate (including resin) (see Note 41).
3. After flipping over, gently tap the dimple plate. This makes the
resin fall out from the dimple plate into the wells of the microti-
ter plate.

Rotate and Centrifuge 1. Rotate the sample microtiter plate (360° about its long axis)
the iPLEX Gold Reaction on a rotator at room temperature for 5 min.
Products (Resin Added) 2. Remove the plate from rotator and centrifuge the sample
microtiter plate at 3200 × g for 5 min.
3. Now, the iPLEX Gold reaction products should be transferred
to a SpectroCHIP, using the MassARRAYNanodispenser. For
instructions, see the MassARRAYNanodispenser User’s Guide
(Samsung).

3.4.7 Defining Assays This should be done using Typer 4 software after nanodispensing.
and Plates Please see the “Defining Assays” and “Defining Plates” chapters in
the MassARRAYTyper Software User’s Guide.

3.4.8 Acquiring Spectra After defining assays and plates, spectra (genotyping) should be
and Genotyping acquired using the MassARRAYAnalyzer Compact mass spectrom-
eter. For instructions, see the “Acquiring Spectra” chapter in the
MassARRAYAnalyzer Compact User’s Guide.
All the above-mentioned information can be accessed here:
http://cancer-seqbase.uchicago.edu/documents/iPLEXSoftwa-
reGuide.pdf
The results of genotyping will come out as Excel sheet tables.
 Re-sequencing Resources to Improve Starch and Grain Quality in Rice 233

3.5 Association With the advent of new genotyping technologies such as next-­
of SNP with Phenotype generation sequencing (NGS), the genetic information has become
widely available to scientists. Many markers have been used to
assess the important agronomic and quality traits. SNP markers are
one of the most important and precise molecular markers which
have been widely used. However, there is not enough information
about association mapping and linkage between these markers and
phenotypic data. Relating molecular markers and phenotypic traits
is called association analysis or mapping. For this analysis, we need
precise phenotypic and sequencing data about the plant’s popula-
tion and its structure. TASSEL is one of the most robust software
packages that can be used for association analysis. It is available as
stand-alone free software which must be downloaded and installed
to your computer.
The process is relatively simple. We have phenotypic and geno-
typic data (phenotype and SNP variants, respectively) in Excel
spread sheets and combine them in TASSEL to find the association
among SNPs and traits as follows:
1. Download and install TASSEL http://www.maizegenetics.
net/#!tassel/c17q9 (see Note 42).
2. TASSEL Version 2.1 is suitable for SNP association study and
also supports SSR analysis. The newer versions (v3 and higher)
do not support SSR marker analysis (see Note 43).
3. Prepare the trait data in a simple EXCEL sheet using the fol-
lowing format (see Notes 44–47) (Fig. 6).
4. Click “Save as” and save this phenotypic file in Text (with Tab
delimited format). Use a suitable name and save it in your
desired folder (e.g., folder A).
5. Prepare the genotypic data in the following format (see Note
48–51) (Fig. 7).
6. Click “Save as” and save this genotypic input file in Text (with
Tab delimited format). Use a suitable name and save it in your
desired folder (e.g., folder A).
7. Now, run TASSEL by double clicking on file sTASSEL.jar in
the TASSEL 2-1 directory (wherever you already downloaded
and installed).
8. Click on data icon, then a File Loader window will be opened.
9. Choose the option button “I will make my best guess and try”
the click OK.
10. When the “Open file” window prompted, choose your pheno-
type input file (Text Tab delimited).
11. Repeat the previous step to load the genotypic file (Text Tab
delimited).
234 Gopala Krishnan Subbaiyan et al.

Fig. 7 Preparation of genotypic input file using SNP variant calls

12. Once the data were loaded, they will be appeared in the “Data
Tree Panel” (see Note 52).
13. Click on “A:a Genotype” button. This will run the “Genotype
Converter” to convert the raw format into genotype state.
14. On the “Genotype Converter,” choose “Create alignment
based on genotypic state (e.g., A:a>Aa)” and click OK.
15. This will create a new dataset file, automatically named
“GenoStates” in Data Tree Panel.
16. Ctrl + click (highlight to choose) the phenotypic, genotypic,
and GenoStates files together in Data Tree Panel (Left-top side).
17. Click on the “U Join” in the “Options Panel.”
18. A new dataset named “GenoStates+Phenotypic+Genotypic”
will appear in Data Tree Panel (Left-top side).
19. Select/highlight all files already created in Data Tree Panel
holding Ctrl + click.
20. Click on “Analysis” button and choose “GLM” (General
Linear Model) and then tick “Analyze Each Data Column
Separately.”
21. Leave the other option as default and click OK.
22. On “Define Output” window change the “#permutations” to
1000 then click Run.
23. A new dataset will be appeared in Results Panel, named
GLM+GenoStates+Traits.
24. Click on “Results” button from Option panel and view the
results in Excel sheet (see Note 53).
 Re-sequencing Resources to Improve Starch and Grain Quality in Rice 235

Fig. 8 An output file including results of association study

3.5.1 Understanding The result file in Excel format will look like the following (Fig. 8).
the Results of the Output The first column shows the traits and second column name of
File SNP variant or marker. The most important columns to be inter-
preted are F-test and R2_Marker. Generally, in both columns, what-
ever the values are bigger the association is higher. For example, the
last column (R2 Marker) shows the portion of total variation
explained by marker (see Note 54).

4 Notes

1. Please carefully check the kit storage instruction upon arrival.

Most parts of kits can be kept for 1 year at room temperature.
However, some parts such as RNaseA stock solution should be
stored at 2–8 °C (for storages more than 1 year). The DNeasy
Maxi spin columns should be stored dry at 2–8 °C upon arrival.
2. Using the terms “tube” or “tubes” interchangeably, depends
on the number of samples.
3. Buffer AP1 and RNase A should not be mixed before use.
4. The QIAshredder Mini spin column will remove precipitates
and cell debris. However, some debris may pass through the
spin column and form pellet at the bottom of collection tube.
Please do not disturb this pellet.
5. Warm up Buffer AW1 to 65 °C using a water bath or heating
block to dissolve, if you observed any precipitates created upon
storage.
6. Second elution may be done using a new 1.5 or 2 mL micro-
centrifuge tube(s) to prevent dilution of the first elute and final
concentration of extracted DNA.
236 Gopala Krishnan Subbaiyan et al.

7. A laboratory coat, gloves, and safety glasses must be worn and

correctly fitted at all times.
8. Take care to ensure the water bath has sufficient water to cover
the heating element. A laboratory oven can be used in place of
a water bath.
9. Liquid nitrogen can cause severe burns. Take great care to
ensure the liquid nitrogen is not spilt or splashed and that it
does not come in contact with any exposed skin or eyes.
10. The mortar will be very cold and capable of damaging skin so
ensure skin does not come in contact with it.
11. Freezing in liquid nitrogen ruptures subcellular organelles and
exposes DNA to endogenous nuclease activity. To prevent
DNA degradation by these enzymes, the samples must remain
frozen at all times until they enter the buffer solution which
denatures the nucleases.
12. This step destroys RNA that would otherwise co-precipitate
with the DNA at step 11 (Subheading 3.1.3).
13. Chloroform is a volatile organic solvent which causes brain
damage after long-term exposure. Use of a fume hood is there-
fore a mandatory requirement.
14. This step removes proteins from the sample by precipitation.
15. Chloroform is very volatile. The tubes must be tightly closed
to ensure the chloroform does not escape when the samples
are outside of the fume hood during centrifugation.
16. The volume of DNA/buffer solution remaining at this step
will vary depending on how much of the aqueous phase was
removed at steps 5 and 8 (Subheading 3.1.3).
17. This step separates the DNA from other dissolved molecules in
the sample.
18. It is important to disperse the agarose well, otherwise it will
form clumps that are very difficult to melt.
19. Gels of 1% or 2% are suitable for most purposes. A 1% gel is
suitable for genomic analysis of long amplicons.
20. Correct orientation is when the positive electrode is at the
opposite end of the get tank to the samples and the electrodes
are plugged into the correct positions in the power pack.
21. Ethidium bromide is a mutagen. Ensure it does not come in
contact with the skin.
22. Glass may bind the Picogreen, so Picogreen should be used in
plastic labware only. Picogreen is subject to photo degradation;
therefore, it needs to be protected from light, either by storing
in the dark or by covering it with foil, when not in use. Only
prepare sufficient Picogreen which can be used on the day it is
prepared.
Re-sequencing Resources to Improve Starch and Grain Quality in Rice 237

(a) DNA quantification by Picogreen utilizes a standard

curve and when using a plate reader standard curve con-
struction takes place in parallel with PCR production
concentration data collection. A standard curve requires
a DNA sample of known concentration which can be pro-
cured from a range of suppliers such as Sigma, Roche, or
Invitrogen.
PCR products are linear and it is preferable to use lin-
ear DNA standards such as those found in DNA ladders,
which correspond or are similar to the amplicon sizes of
~5 kbp in length. Amplicon concentration will not exceed
the concentration of DNA standards supplied by commer-
cial organizations (Sigma, Roche, or Invitrogen and so on)
which are in the order of 250 μg/μL because of PCR
product inhibition which limits the final concentration of
PCR product (amplicons) (Kainz Biochimicaet Biophysica
Acta 2000).
23. Although residual salts, oligonucleotides, and gDNA from the
PCR mixture may theoretically interfere with the Picogreen
assay, all amplicons will have the same concentration of con-
taminants and will be similarly affected. Given the purpose of
quantifying amplicon concentration is to create equimolar
pools, it is more important to establish the relative amount of
amplicons, not the absolute amount, and so the (minimal)
effect of contaminants can be ignored.
24. Failure to identify the correct sequence (gDNA) will result in
amplification of the incorrect gene.
25. If there is more than one sequence available in different data-
bases, then alignment of gDNAs may be required to design the
best primer based on consensus alignment. Alignment of genes
can be by using software packages such as Sequencher (Gene
Codes Corporation, Ann Arbor, MI) and CLUSTAL W
(http://www.ebi.ac.uk/Tools/clustalw2/index.html), Bio
Edit (Free) http://www.mbio.ncsu.edu/bioedit/bioedit.
html, etc.
26. If the number of genes and genotypes are large in number, a
touch-down protocol coupled with LR-PCR can help reduce
time, labor, and costs of the study. This helps to find the opti-
mum PCR condition to capture the target genes in minimum
time.
27. It is preferable to use the same genotype, with gDNA/cDNA
sequences for standardizing the LR-PCR protocol.
28. There are benefits for using High-fidelity DNA polymerase
such as BioRadiProof TM High-Fidelity DNA polymerase. It
has a significantly low error rate of amplification (4.4 × 10−7).
The error rate is approximately 50-fold lower than normal
238 Gopala Krishnan Subbaiyan et al.

polymerases. It also has high extension efficiency of 5–30 s/kb.

The kit content and specifications can be found here: http://
www.bio-rad.com/webroot/web/pdf/lsr/literature/1000
2300B.pdf.
29. Some genes may not amplify with the touch down cycling pro-
tocol; then they should be amplified separately with their own
optimum conditions.
30. Whole-genome re-sequencing also captures the organellar
(chloroplast and mitochondrial) genomes in the sequencing
process. There are some similarities in the organellar and
nuclear genomes which may lead to erroneous call of variant.
Therefore, it is very important to remove the reads mapping to
organellar genomes before subjecting the short sequence reads
to further analyses.
31. The default length fraction and similarity is 0.5 in CLC genom-
ics workbench. Since the coverage of rice genome sequence,
which is comparatively small, in an Illumina sequence run is
very good, stringent parameters are affordable. If the coverage
is relatively less (as in the case of multiplex barcode sequencing
of multiple genomes in a single run), then these parameters
have to be adjusted accordingly.
32. The H row will be used as reservoirs to distribute SAP enzyme
solution to the rest of the microtiter wells.
33. Care should be taken when pipetting the SAP to minimize loss
of solution due to viscosity.
34. After distribution of the SAP enzyme solution, each well in
rows G–A should have 10 μL of the SAP enzyme solution,
while the wells in row H (initial reservoirs) should have 15 μL.
35. The wash station must be prepared and placed on position 2 of
liquid-handler (up-right corner), according to the manufac-
turer’s instruction.
36. Add the reagents as they appeared in the table.
37. During pipetting, ensure that no droplet is placed or adhered
to the well-walls. The pipetting must be carefully placed into
the centers of microtiter plate wells.
38. After distributing iPLEX Gold reaction cocktail, each well in
rows G to A should have 10 μL of the low plexiPLEX Gold
reaction cocktail, while the wells in row H have 15 μL.
39. Please be aware that there are two cycling loops, one of five
cycles that sits inside a loop of 40 cycles. These two loops result
in a 200-cycle program.
40. There is a small-rounded post on dimple plate (including
resin). This post should be used to align the wells in microtiter
plate with the wells of dimple plate.
Re-sequencing Resources to Improve Starch and Grain Quality in Rice 239

41. TASSEL can be run only under Java environment. This means
that Java should be installed on your computer. All new
Windows operating systems install the latest versions of Java
automatically; therefore, this might already be existed on your
system. Alternatively, it is freely available online here: http://
www.oracle.com/technetwork/java/index.html
42. Preparation of phenotypic and genotypic input files is different
in version 5. Please read the instruction of each version before
preparing the input files here: https://bitbucket.org/tasselad-
min/tassel-5-source/wiki/UserManual
43. The number in the first column (233) is the number of indi-
viduals (genotypes). The second number tells TASSEL the
number of traits (13 starch quality traits). The last number (1)
indicates the number of hear rows.
44. Row number 2 includes the traits names (starch measured
characters).
45. Although the first column includes the genotype/line names
(e.g., YRR038_01-03), the column header includes the first
measured traits name (Peak1). This is due to the left-shift of
data. In fact, the column B includes the data of Peak1 mea-
surement and so on to the end of table (column N).
46. There should not be any gap or space among words in header
names. Characters, letter, and an underscore might be used, if
required.
47. The number in the first row (column A) indicates the number
of individuals/genotypes (288). The number in column B
(28:02:00) shows the number of markers and diploid nature
(2×) of species, respectively.
48. A question mark is commonly used for missing data.
49. Insert the homozygous calls (e.g., G:G) and heterozygous calls
(C:A) etc.
50. The second row includes the name of SNP markers.
51. If there are any problems in uploading input files, go back to
the Excel file and find the formatting problem (mainly in head-
ers and formatting), then save the corrected file as Text Tab
delimited and repeat the steps.
52. If the file was not opened by double click, then right click and
“save” the file. Then open by right click using “Open as” Excel
spreadsheet.
53. TASSEL Version 5 can generate statistics on association
between genes and traits such as contribution of each SNP
variant to trait, etc.
240 Gopala Krishnan Subbaiyan et al.

References
1. Anacleto R, Cuevas RP, Jimenez R, Llorente C, 9. Healey A, Furtado A, Cooper T, Henry RJ
Nissila E, Henry R, Sreenivasulu N (2015) (2014) Protocol: A simple method for extract-
Prospects of breeding high-quality rice in the ing Next-Generation Sequencing quality
post-genomic era. Theor Appl Genet 128(8): genomic DNA from recalcitrant plant species.
1449–1466 Plant Methods 10:21
2. Wang K, Henry RJ, Gilbert RG (2014) Causal 10. Henry RJ, Edwards K (2009) New tools for
relations between starch biosynthesis, structure single nucleotide polymorphism (SNP) discov-
and properties. Springer Sci Rev 2(1):15–33 ery and analysis accelerating plant biotechnol-
3. Kharabian-Masouleh A, Waters DLE, Reinke ogy. Plant Biotechnol J 7:311
RF, Ward R, Henry RJ (2012) SNP in starch 11. Kharabian-Masouleh A, Waters DLE, Reinke
biosynthesis genes associated with nutritional RF, Henry RJ (2011) Discovery of polymor-
and functional properties of rice. Sci Rep phisms in starch-related genes in rice germ-
2:557 plasm by amplification of pooled DNA and
4. Gopala Krishnan S, Waters DLE, Katiyar SK, deeply parallel sequencing. Plant Biotechnol
Sadananda AR, Satyadev V, Henry RJ (2012) J 9(9):1074–1085
Genome-wide DNA polymorphisms in elite 12. Gopala Krishnan S, Waters DLE, Henry RJ
indica rice inbreds discovered by whole- (2012) Genome-wide variations between elite
genome sequencing. Plant Biotechnol J 10: lines of indica rice discovered through whole
623–634 genome re-sequencing. In: Rangasamy SRS
5. Kharabian-Masouleh A, Waters DLE, Reinke et al (eds) 100 years of rice science and looking
RF, Henry RJ (2009) A high-throughput assay beyond. Proceedings of the international sym-
for rapid and simultaneous analysis of perfect posium held at Tamil Nadu Agricultural
markers for important quality and agronomic University, Coimbatore, Tamil Nadu, India,
traits in rice using multiplexed MALDI-­TOF 9–12 January 2012, pp 118–119
Mass Spectrometry. Plant Biotechnol J 7: 13. Krishnan SG, Daniel LE, Waters DLE, Henry
355–363 RJ (2014) A method for discovery of pair wise
6. Wang K, Hasjim J, Wu AC, Li E, Henry RJ, genome-wide SNP variations from whole
Gilbert RG (2015) Roles of GBSSI and SSIIa genome re-sequencing data. In: Furtado A,
in determining amylose fine structure. Henry RJ (eds) Cereal genomics: methods and
Carbohydr Polym 127:264–274 protocols. Springer, New York, pp 287–294
7. Wang K, Wambugu PW, Zhang B, Wu AC, 14. McCouch SR, Zhao K, Wright M, Tung CW
Henry RJ, Gilbert RG (2015) The biosynthe- (2010) Development of genome-wide SNP
sis, structure and gelatinization properties of assays for rice. Breed Sci 60:524–535
starches from wild and cultivated African rice 15. Edwards M, Henry RJ (2011) DNA sequenc-
species (Oryza barthii and Oryza glaberrima). ing methods contributing to new directions in
Carbohydr Polym 129(20):92–100 cereal research. J Cereal Sci 54:395–400
8. Krishnan SG, Waters DLE, Henry RJ (2014) 16. Henry RJ, Edwards M, Waters DLE, Gopala
Australian wild rice reveals pre-domestication Krishnan S (2012) Molecular markers for
origin of polymorphism deserts in rice genome. plants derived from large scale sequencing.
PLoS One 9:e98843 J Biosci 37(5):829–841
 Chapter 13

Quantifying Grain Digestibility of Starch Fractions

in Milled Rice
Crisline Mae Alhambra, Sushil Dhital, Nese Sreenivasulu,
and Vito M. Butardo Jr.

Abstract
Rice is one of the staple foods which serves as the major source of carbohydrate in the human diet. A typi-
cal milled rice grain is mainly composed of starch of up to 80–90%, with an average of 6–8% proteins and
some trace amounts of dietary fiber. Although cooked white rice can elicit variable glycemic response, a
portion of rice starch may evade digestion in the human small intestine. The digested portion of rice can
be estimated and characterized in vitro based on starch digestion extent and rate (kinetics). The indigest-
ible portion of starch can also be quantified. This chapter will present micro-scale methods to quantify rice
starch digestion rate and extent based on the sugar fractions released after treating the samples with diges-
tive enzymes.

Key words Rice, Grain, Digestibility, Resistant Starch, Nutrition

1 Introduction

Rice is one of the major food crops in the world, feeding almost half
of the world’s population. The major component of rice grain is
carbohydrates in the form of starch and some free sugars [1, 2]. Up
to 70% of the human energy intake comes from carbohydrates.
Starch can be classified into three fractions based on extent of
hydrolysis after treatment with digestive enzymes: (1) Rapidly
Digestible Starch (RDS), (2) Slowly Digestible Starch (SDS), and
(3) Resistant Starch (RS). RDS and SDS are the fractions of starch
which can be digested in the small intestine and subsequently
absorbed by the body within 20 min and between 20 and 120 min,
respectively. RDS represents rapidly released glucose after con-
sumption of starchy food, while SDS has been associated with
reduced sensation of hunger after ingestion due to the slowing
down of carbohydrate hydrolysis rate [3]. The undigested fraction
of starch is termed RS, which is resistant to small intestinal diges-
tion but mostly fermentable in the colon by the gut microflora

Nese Sreenivasulu (ed.), Rice Grain Quality: Methods and Protocols, Methods in Molecular Biology, vol. 1892,
https://doi.org/10.1007/978-1-4939-8914-0_13, © Springer Science+Business Media, LLC, part of Springer Nature 2019

241
242 Crisline Mae Alhambra et al.

[4, 5]. Since RS does not contribute to raise in blood glucose level,
it is known to help in the prevention of hyperglycemia, hyperinsu-
linemia, diabetes, and obesity [6–8].
Both underweight and obesity are major risk factors for the
global burden of disease [9]. In both cases, there is a need to inves-
tigate the digestibility of foods, mainly the starch portion in com-
mon foods such as rice. In this connection, the identification of
different fractions of starch in rice is important in determining its
effect on human body upon consumption considering the preva-
lence of diet-induced nutritional concerns which ranges from
caloric hunger on one end, to diabetes, overweight, and obesity on
the other end of the spectrum.
This method focuses on identifying different starch fractions in
polished rice grains to quantify and characterize rice grain digest-
ibility. Microscale methods based on Megazyme kits are presented
to determine the amount of total resistant starch (RS) as well as
total digestible carbohydrates (DC) in one assay as well as total
starch (TS) and total sugars (TSU) in a separate assay. These meth-
ods are optimum for large-scale screening of breeding and diversity
populations of rice because they utilize a micro method wherein a
tenfold decrease in the amount of sample and reagents are used to
facilitate assays using microfuge tubes. Screening for digestibility
properties in rice breeding and diversity panels are prohibitively
expensive due to cost, time, and manpower involved. The methods
outlined here greatly streamlined the process. Once superior germ-
plasm is identified, the values obtained from the screening meth-
ods can be verified using the conventional AOAC methods for
total starch (996-11) and resistant starch (2002-02). In addition,
the cooked grain amylolysis method to determine the kinetics of
starch digestibility simulating gastrointestinal tract digestion is pre-
sented in the last section as proxy measure for glycemic index (GI)
measurements. Such an in vitro method helps to characterize
diversity rice lines to estimate the values of cooked grain amyloly-
sis. The presented cooked grain amylolysis method here can be
used as a pre-screening method prior to actual clinical studies of GI
estimation.

2 Materials

2.1 Kits 1. Resistant Starch Assay Kit (Megazyme K-RSTAR 08/11),

1000 assay per kit.
2. Total Starch Assay Kit (Megazyme K-TSTA 07/11), 1000
assay per kit.
3. d-Glucose Assay Kit (Megazyme K-GLUC 07/11), 660 assays
per kit.
 Quantifying Grain Digestibility of Starch Fractions in Milled Rice 243

2.2 Reagents 1. 100 mM sodium maleate buffer pH 6.0 with 5 mM

CaCl2·2H2O (500 mL)
2. 99% Ethanol (500 mL).
3. 50% ethanol (1 L).
4. 2 M KOH (250 mL).
5. 1.2 M sodium acetate buffer pH 3.8 (1 L).
6. 100 mM sodium acetate buffer pH 4.5 (5 L).
7. 80% ethanol (2 L).
8. 50 mM maleic acid buffer pH 5.5 (500 mL).
9. 0.01 M HCl (500 mL).
10. 0.02 M NaOH (500 mL).
11. 200 mM sodium acetate buffer pH 6 with 4 mM CaCl2 and
0.49 mM MgCl2 (1 L).
12. GOPOD reagent (1 L).

2.3 Instruments 1. Analytical balance (10 mg to 220 g capacity ±0.1 mg).

2. Refrigerated Microcentrifuge (4 °C, 24-microfuge tubes
capacity).
3. Recirculating water bath (37 °C ± 0.03 °C temperature
stability).
4. UV-Vis spectrophotometer or ELISA plate reader (510 nm).
5. Magnetic stirring plate (60–134 × g).
6. Speed vacuum dryer (36-microfuge holder).
7. Dehuller (100 g per minute).
8. Miller (8–10 g capacity).
9. Ball Grinder (grinding jar with 50 g capacity).
10. Sieve (30 and 40 mesh).

3 Methods

3.1 Preparation Dehull and mill the rice paddy using standard methods to obtain
of Rice Samples milled rice. Sort the milled rice and discard the broken and discol-
ored grains. Grind polished grain samples (5 g) using a ball grinder
set in 25 rpm per second for 1 min. In IRRI, we routinely use
Retsch MM400 with 50 mL stainless steel grinder (01.462.0216)
with a 25 mm diameter grinding ball (05.368.0105). Use this sam-
ple for resistant starch and total starch analysis. For cooked milled
rice flour amylolysis, grind milled rice grains for 5 s using a ball
grinder set in 20 rpm. Pass the ground sample between 30 and 40
mesh to collect particle size of 400 to 600 μm.
244 Crisline Mae Alhambra et al.

3.2 Total Resistant In this step, the total resistant starch (RS) is measured using the RS
Starch (RS) assay procedure from Megazyme (K-RSTAR 08/11) with modifi-
cations. This micro method had been extensively validated in our
laboratory to accurately quantify RS comparable with the values
obtained using the standard method. The supernatant is also col-
lected to measure total digestible carbohydrates (DC) which is
described in Subheading 3.3.
1. Weigh 10 ± 0.1 mg of rice flour directly in a 2 mL microfuge
tube.
2. Add 400 μL of enzyme solution (10 mg/mL pancreatin with
3 U/mL amyloglucosidase (AMG) from the Megazyme kit
dissolved in 100 mM sodium maleate buffer pH 6.0 with
5 mM CaCl2·2H2O).
3. Incubate horizontally in a shaking water bath (100 strokes/
min) at 37 °C for exactly 16 h. We commonly use microfuge
rack to hold all the microfuge tubes in place using rubber
bands.
4. Remove the tubes and dry each with a paper 12 min before
the end of 16 h incubation.
5. Stop the reaction by the addition of 400 μL ethanol (99%).
6. Mix vigorously using a vortex mixer.
7. Centrifuge the tubes at 18,231 × g for 30 min.
8. After centrifugation, carefully transfer (see Note 1) the super-
natant using a pipettor and collect in a 15 mL conical tube. Set
aside for total digestible carbohydrates (DC) assay (Subheading
3.3).
9. Resuspend the pellets to 200 μL of 50% ethanol and vigor-
ously stir using a vortex mixer.
10. Add a further 600 μL of 50% ethanol and vortex mix.
11. Centrifuge at 18,231 × g for 30 min.
12. Carefully remove the supernatant using a pipettor and repeat
the suspension and centrifugation of the pellet (steps 9–11).
Collect and pool all the supernatant obtained in the previous
steps using the conical tube in step 8.
13. Prepare an ice bath over a stirring plate.
14. Place a stir bar in the tubes containing the pellet and add
200 μL of 2 M KOH while stirring in an ice bath. Stir for
approximately 20 min or until complete solubilization of the
pellet.
15. Add 800 μL of 1.2 M sodium acetate buffer pH 3.8 while
stirring.
16. Immediately add 10 μL of 3300 U/mL of AMG and mix.
 Quantifying Grain Digestibility of Starch Fractions in Milled Rice 245

17. Place the tubes in water bath at 50 °C for 30 min with inter-
mittent mixing (see Note 2).
18. After incubation, remove the stir bar and centrifuge at 18,231
× g for 10 min and get 100 μL of the supernatant (see Note 3).
19. Prepare a blank (100 μL of 100 mM sodium acetate buffer
pH 4.5) and quadruplicate standards (100 μL of 1 mg/mL
glucose, prepared and read four times).
20. Add 3 mL of GOPOD reagent supplied from the Megazyme
kit to the 100 μL supernatant, blank and standards. Incubate
at 50 °C for 20 min.
21. Read absorbance of the sample and standards against blank at
510 nm.
22. Calculate % RS (dry basis) by using the formula:
1.03 1 100 162
%RS = ∆A × F × × × ×
0.1 1000 W 180
where ΔA = absorbance of the sample against blank
1
F =
average absorbance of glucose standard
1.03
0.1 = volume correction (0.1 mL aliquot was taken from
1.03 mL final volume)
1
1000 = conversion of μg to mg

100
W = factor to present RS as % of sample weight
W = dry weigh of the sample =
 (100 moisture content ) 
 flour weight × 
 100 
162
180 = adjustment from free d-glucose to anhydrous d-glu-
cose (as occurs in starch).

3.3 Total Digestible In this step, total digestible carbohydrates (DC) refer to total digest-
Carbohydrates (DC) ible starch and glucose present in the flour sample. The supernatant
collected from Subheading 3.2 are used as starting materials.
1. Dilute the collected supernatant from RS assay (from
Subheading 3.2, step 12) to 10 mL with 100 mM sodium
acetate buffer pH 4.5.
246 Crisline Mae Alhambra et al.

2. Get 100 μL aliquot, and add 10 μL of AMG (300 U/mL in

100 mM sodium maleate buffer pH 6.0 with 5 mM
CaCl2·2H2O).
3. Incubate for 20 min at 50 °C.
4. Prepare a blank (110 μL of 100 mM sodium acetate buffer
pH 4.5) and quadruplicate standards (110 μL of 1 mg/mL
glucose, prepared and read four times).
5. Add 3 mL of GOPOD reagent the sample (step 2), blank and
standards. Incubate at 50 °C for 20 min.
6. Read absorbance of the sample and standards against blank at
510 nm.
7. Calculate % DC (dry basis) by using the formula:
10 1 100 162
% DC = ∆A × F × × × ×
0.1 1000 W 180

where
ΔA = absorbance of the sample against blank
100
F =
average absorbance of glucose standard

10
= volume correction (0.1 mL aliquot was taken from
0.1
10 mL final volume)
1
= conversion of μg to mg
1000

100
= factor to present DC as % of sample weight
W
W = dry weigh of the sample =
 (100 − moisture content ) 
 flour weight × 
 100 
162
= adjustment from free d-glucose to anhydrous d-­glucose
180
(as occurs in starch).

3.4 Total Starch (TS) In this section, TS is measured using the Total Starch assay proce-
dure (Megazyme K-TSTA 07/11) with modifications. This
method had been extensively validated in our lab to accurately
reflect the values obtained using the standard method. Total sugars
(TSu) are assayed in Subheading 3.5 using the supernatant obtained
in this section.
Quantifying Grain Digestibility of Starch Fractions in Milled Rice 247

1. Weigh 10 ± 0.1 mg of starch directly in a 2 mL microfuge

tube.
2. Add 500 μL of 80% ethanol and incubate at 85 °C for 5 min.
3. Stir using a vortex mixer after incubation and add 500 μL of
80% ethanol.
4. Mix and centrifuge at 18,231 × g for 20 min.
5. Aspirate (see Note 4) the supernatant and resuspend the pellet
by adding 1 mL of 80% ethanol.
6. Mix and centrifuge at 18,231 × g for 20 min. Collect all the
supernatant obtained for total free sugar quantification in a
2 mL microfuge tube (Subheading 3.5).
7. Prepare an ice bath over a stirring plate.
8. Place a stir bar to the obtained pellet and add 200 μL of 2 M
KOH while stirring (700 rpm) in an ice bath.
9. Stir for approximately 20 min or until the complete solubiliza-
tion of the pellet.
10. Add 800 μL of 1.2 M sodium acetate buffer pH 3.8 while
stirring.
11. Immediately add 10 μL of thermo-stable α-amylase (3000 U/
mL) and 10 μL of AMG (3300 U/mL).
12. Mix and incubate the tubes at 50 °C for 30 min (see Note 5).
13. Dilute the liquid contents with water to 10 mL (see Note 6).
14. Get 1 mL of aliquot, place in a 2 mL microfuge tube.
15. Centrifuge the aliquot at 18,231 × g for 10 min.
16. Get 100 μL of the supernatant, place in a separate tube.
17. Prepare a blank (100 μL H2O) and quadruplicate standards
(100 μL of 1 mg/mL glucose, prepared and read four times).
18. Add 3 mL GOPOD to the sample (step 16), blank and
standards).
19. Incubate at 50 °C for 20 min.
20. Read absorbance of the sample and standards against blank at
510 nm.
21. Calculate % TS (dry basis) by using the formula:
10 1 100 162
% TS = ∆A × F × × × ×
0.1 1000 W 180

where
ΔA = absorbance of the sample against blank
100
F =
average absorbance of glucose standard
1
= conversion of μg to mg
1000
248 Crisline Mae Alhambra et al.

100 = factor to present TS as % of sample weight

W
W = dry weigh of the sample =
 (100 − moisture content ) 
 flour weight × 
 100 
10
= volume correction (0.1 mL was taken from 10 mL)
0.1
162 = adjustment from free d-glucose to anhydrous d-­glucose
180
(as occurs in starch).

3.5 Total Free 1. Dry the obtained supernatant from TS assay (Subheading 3.4)
Sugar (TSu) using a speed vacuum concentrator for approximately 4 h or
until completely dry.
2. Reconstitute using 100 μL of 50 mM maleic acid buffer
pH 5.5.
3. Add 50 μL of invertase (150 U/mL in 50 mM maleic acid
buffer pH 5.5) and incubate for 30 min at 37 °C.
4. Boil in water bath for 4 min and cool immediately in ice
bath.
5. Centrifuge at 13,394 × g for 10 min, get 50 μL of the super-
natant and place in a 2 mL microfuge tube.
6. Prepare a blank (50 μL of 50 mM maleic acid buffer pH 5.5)
and quadruplicate standards (50 μL of 1 mg/mL glucose, pre-
pared and read four times).
7. Add 1.5 mL of GOPOD reagent to the sample (step 5), blank
and standards.
8. Incubate at 50 °C for 20 min.
9. Read absorbance of the sample and standards against blank at
510 nm.
10. Calculate mg of glucose by using the formula:
∆A
mg of glucose = × volume
F
where
A = absorbance of the sample against blank
F = average absorbance of glucose standard
V = 0.150 mL
Calculate % sugar (glucose and sucrose) using the formula
mg glucose
% sugar = ×100.
wt of flour
 Quantifying Grain Digestibility of Starch Fractions in Milled Rice 249

3.6 Cooked Rice In this section, the starch digestion rate of cooked rice is deter-
Amylolysis mined by the use of enzymes and conditions that mimics the bio-
chemical conditions of the human gastrointestinal digestion. Rice
samples used have particle sizes between 400 and 600 μm to mini-
mize the deviation arising from grain fragments, particle size and
to mimic the size of chewed rice during mastication. Methods and
calculations were developed based on Dhital et al. [10] and Butardo
et al. [11].
1. Weigh 500 ± 0.1 mg milled rice (400–600 μm) in a 50 mL
glass tube.
2. Add water (see Note 7) and cover with foil to minimize loss of
water.
3. Let it stand for 30 min and cook in a boiling water bath for
30 min.
4. Immediately place vertically at 37 °C water bath over a stirring
plate. If not used immediately, store at 50 °C waterbath (see
Note 8). Ensure amylolysis will be done within the day.
5. Add 2 mL of water and dislodge the clumped cooked rice
using a metal spatula. Add further 4 mL water to wash the
spatula.
6. Add a stir bar (1/2 × 5/16 in) and start stirring at 350 rpm.
7. Immediately start the digestion process (see Note 8) by adding
5 mL of pepsin (see Note 9) (1 mg/mL in 0.01 M HCl) and
incubate for exactly 30 min.
8. Add 5 mL of 2 M NaOH and incubate for exactly 1 min.
9. Add 20 mL of 200 mM sodium acetate buffer pH 6.0 with
4 mM CaCl2 and 0.49 mM MgCl2 and incubate for exactly
10 min. Get duplicate 200 μL aliquot 2 min before the end of
the incubation.
10. Add 5 mL of pancreatin/AMG mixture (2 mg/mL in pancre-
atin and 28 U/mL AMG in 200 mM sodium acetate buffer
pH 6.0 with 4 mM CaCl2 and 0.49 mM MgCl2) and get
duplicate 200 μL aliquot at desired sampling points (5, 10, 20,
30, 45, 60, 90, 120, 180 min) and place in a microfuge tube.
11. Immediately place the microfuge tube containing the aliquot
in an ice bath or liquid nitrogen to stop the reaction. Stopping
the reaction can also be done using ethanol but slight modifi-
cation in calculation has to be employed.
12. Centrifuge the aliquot at 15,720 × g at 4 °C for 10 min.
13. Aspirate 50 μL of the supernatant (see Note 10) and place in a
2 mL microfuge tube (see Note 11).
14. Prepare a blank (50 μL of 200 mM sodium acetate buffer
pH 6.0 with 4 mM CaCl 2 and 0.49 mM MgCl 2) and qua-
250 Crisline Mae Alhambra et al.

druplicate standards (55 μL of 1 mg/mL glucose, prepared

and read four times).
15. Add 5 μL of AMG (300 U/mL in 200 mM sodium acetate
buffer pH 6.0 with 4 mM CaCl2 and 0.49 mM MgCl2) to
both blank (step 14) and samples (step 13).
16. Incubate the samples, blank and standards at 50 °C for 20 min.
17. Add 3 mL GOPOD reagent and incubate at 50 °C for 20 min.
18. Read absorbance of the sample and standards against blank at
510 nm.
19. Calculate % starch hydrolysis (Sh) per sampling time by using
the formula:
  A Sx   162 
   × df  × V × 
  A std  180 
% Sh =   
 ×100
 % TS 
wt Sx ×
 100 
 

where:
ASx = absorbance of the sample at time t
Astd = average absorbance of standard
df = dilution factor at time t
V = volume at time t (in mL)
162
= adjustment from free d-glucose to anhydrous d-­glucose
180
(as occurs in starch)
wt Sx = weight of sample (mg)
% TS = % total starch (dry basis).
20. Plot the % Sh vs time to get digestion curve.
21. Get the k-value by plotting ln[(% Shx+1 − % Shx)/(tx+1 − tx)],
ln[(% Shx+2 − % Shx+1)/(tx+2 − tx+1)]…etc. vs (tx+1 − tx)/2,
(tx+2 − tx+1)/2…etc.
where:
x = 0 min
x + 1 = 5 min
x + 2 = 10 min
t = time (in the lab, we usually sample at 5, 10, 15, 20, 30, 45,
60, 90, 120, 180 min).
22. Get the equation of the line of the graph in step 19. The abso-
lute value of the slope of the line (m) is the k value or digestion
rate and is expressed as min−1.
 Quantifying Grain Digestibility of Starch Fractions in Milled Rice 251

3.7 Integrating RS and DC assay results would show the proportion of the starch
and Interpreting in rice which is resistant to digestion and digestible, respectively.
Results TS and TSu would show the total starch and soluble sugar (glucose
and sucrose) contents of rice, respectively. Starch hydrolyzed
within certain time points can be identified by using in vitro diges-
tion using the cooked grain amylolysis method. In addition, the k
value would determine the rate of digestion. Higher k value would
indicate faster rate of digestion.

4 Notes

1. Use a 1 mL pipet to aspirate.

2. Mix during the first 15 min of incubation.
3. If RS is >10%, dilute the contents to 10 mL before
centrifugation.
4. Similar to RS assay, use 1 mL pipet to aspirate.
5. Similar to RS assay, mix during the first 15 min of
incubation.
6. Quantitatively transfer the contents by washing the stir bar
with water.
7. Add water depending on amylose content (AC). The water:rice
ratio is 1.8:1, 1.6:1 and 1.4:1 for samples with AC of 25+%,
20–24%, and 0–19%, respectively. Alternatively, one can just
cook using excess water.
8. Do not allow the cooked rice to cool. Ideally, start the assay
within 30 min to avoid starch retrogradation.
9. All enzyme solution to be used should be freshly prepared.
10. Store the supernatant at −20 °C if not used within the day.
11. Dilute the supernatant using 20 mM sodium maleate buffer
pH 6.0 and dilution should be higher with increasing time.
Get only 50 μL of the diluted supernatant.

References
1. Butardo VM Jr, Sreenivasulu N (2016) substrate oxidation and feelings of hunger in
Tailoring grain storage reserves for a healthier humans. Am J Clinical Nut 72(6):1461–1468
rice diet and its comparative status with other 4. Englyst HN, Cummings JH (1987) Digestion
cereals. Int Rev Cell Mol Biol 323:31–70 of polysaccharides of potato in the small-­intestine
2. FAO (1998) Dietary carbohydrate composi- of man. Am J Clin Nutr 45(2):423–431
tion. Food and Agriculture Organization of the 5. Englyst HN, Macfarlane GT (1986)
United Nations, Rome Breakdown of resistant and readily digestible
3. Sparti A, Milon H, Di Vetta V, Schneiter P, starch by human gut bacteria. J Sci Food Agric
Tappy L, Jequier E, Schutz Y (2000) Effect of 37(7):699–706
diets high or low in unavailable and slowly 6. Ranganathan S, Champ M, Pechard C,
digestible carbohydrates on the pattern of 24-h Blanchard P, Nguyen M, Colonna P, Krempf
252 Crisline Mae Alhambra et al.

M (1994) Comparative-study of the acute 9. World Health Organization (2002) The World
effects of resistant starch and dietary-fibers on Health Report: Reducing risks, promoting
metabolic indexes in men. Am J Clin Nutr healthy life
59(4):879–883 10. Dhital S, Dabit L, Zhang B, Flanagan B,
7. Topping DL, Clifton PM (2001) Short-chain Shrestha AK (2015) In vitro digestibility and
fatty acids and human colonic function: roles of physicochemical properties of milled rice. Food
resistant starch and nonstarch polysaccharides. Chem 172:757–765
Physiol Rev 81(3):1031–1064 11. Butardo VM Jr, Anacleto R, Parween S,
8. Behall KM, Scholfield DJ, Hallfrisch JG, Samson I, de Guzman K, Alhambra CM,
Liljeberg-Elmstahl HGM (2006) Consumption Misra G, Sreenivasulu N (2017) Systems
of both resistant starch and beta-glucan improves genetics identifies a novel regulatory domain
postprandial plasma glucose and insulin in of amylose synthesis. Plant Physiol 173(1):
women. Diabetes Care 29(5):976–981 887–906
 Chapter 14

Determination of Macronutrient and Micronutrient Content

in Rice Grains Using Inductively Coupled Plasma-Optical
Emission Spectrometry (ICP-OES)
Lilia Molina, Jennine Rose Lapis, Nese Sreenivasulu,
and Rosa Paula O. Cuevas

Abstract
The rice grain endosperm is mostly composed of starch, which serves as the major source of calories for
more than half of the world’s population. Macro and micronutrients make a minor proportion of the rice
grain, which particularly gets accumulated in outer aleurone layer, which are in general eliminated upon
milling. Because rice is the major staple, it is seen as an efficient mechanism for delivering both macro- and
micronutrients, particularly for the poor who do not have ample access to diversified diets. Enriching
micronutrient and macronutrient concentrations in milled rice of endosperm and/or in brown rice, is an
important dietary intervention to create health benefits of rice consumers. Efforts are underway to increase
the nutritional content of rice through bio/fortification approaches. The plant takes up these same ele-
ments from the soil, redirect the transport of these elements into the grain. Thus besides biofortification
strategies, scientists can also use the knowledge to design proper soil nutrient management to enrich
micronutrients in the grains. Therefore, it is important to be able to determine the macro- and the micro-
nutrient composition of the vegetative parts of the rice plant and of the rice grain. In this chapter, nitric-­
perchloric acid digestion and inductively coupled plasma-optical emission spectrometry (ICP-OES)
methods routinely used in IRRI’s Grain Quality and Nutrition Services Laboratory (GQNSL) to deter-
mine the concentrations of various macro- and micronutrients found in the rice grain and the rice plant,
are described.

Key words Nitric-perchloric digestion, Inductively coupled plasma-optical emission spectrometry

(ICP-OES), Macronutrients, Micronutrients

1 Introduction

Food is a means through which nutrition is delivered. Staple crops

are of particular importance in improving consumer health because
these are consumed routinely and in large amounts. Therefore,
biofortification of these crops is a potential means to alleviate
macro- and micronutrient malnutrition, particularly for consumers

Nese Sreenivasulu (ed.), Rice Grain Quality: Methods and Protocols, Methods in Molecular Biology, vol. 1892,
https://doi.org/10.1007/978-1-4939-8914-0_14, © Springer Science+Business Media, LLC, part of Springer Nature 2019

253
254 Lilia Molina et al.

who cannot afford diversified diets [1]. Two global micronutrient

deficiency concerns majorly involve iron and zinc. The serious
health implications caused by these deficiencies are well-known
(e.g., [2–6]). Rice, for instance, is seen as a potential means to
address iron and zinc deficiencies because it is a major staple crop.
Brown rice contains high levels of proteins, essential vitamins, and
minerals [7] and it has high concentrations of iron, zinc, and man-
ganese in its aleurone layer [8]; therefore, brown rice is an option
for micronutrient supplementation. However, brown rice is not
popular with majority of consumers, who prefer milled rice for
various reasons ([7], reviewed in [9]). Milled rice is a poor source
of micronutrients, and thus needed to address the large prevalence
of micronutrient deficiency, compared to brown rice, because it
lost most of its micronutrients during the polishing process. Based
on profiling of 95 varieties, the average iron content was reported
to average at 2.28 mg/100 g (dry basis); while the average zinc
content of 57 varieties was 3.34 g/100 g (dry basis) [10].
Biofortification of iron, zinc, and pro-Vitamin A in staple crops,
including rice, are being spearheaded by HarvestPlus and IRRI
research programs [11, 12]. On the other hand, transgenic biofor-
tification is reportedly aimed to increase iron content six fold
higher and to double zinc content in the milled rice grain [13].
Quantifying the concentrations of various micro- and macro-
nutrients in the milled rice grain, therefore, is increasingly crucial
in the context of rice biofortification efforts. Various methods such
as microwave digestion, open- and closed-vessel digestion, dry ash-
ing, and wet ashing are used to digest and extract the macro and
micronutrients from plant tissues or the grain [14–17]. Profiling of
nutrients is then conducted by atomic absorption spectrophotom-
etry (AAS), laser-induced breakdown spectroscopy (LIBS), induc-
tively coupled plasma-optical emission spectrometry (ICP-OES),
and inductively coupled plasma-mass spectrometry (ICP-MS) for
quantitative nutrient analysis [15, 18–20]. Among the techniques
available, X-ray fluorescence (XRF) spectroscopy is employed for
quick scan of screening diverse germplasm and ICP-OES tech-
niques to get precise estimations of Fe and Zn concentrations in
the narrow enriched germplasm by IRRI and the HarvestPlus
group. ICP-OES is considered the gold standard because of its
high sensitivity and accuracy; however, it requires tedious sample
preparation and digestion prior to ICP-OES analysis [20]. An
alternative rapid screening, high sample throughput method was
developed by the HarvestPlus Team; it uses the bench-top Energy
Dispersive XRF for iron and zinc analysis in whole rice grain or
flour samples without the need for the sample digestion step. The
XRF method is calibrated using reference values from the ICP-­
OES method and is adopted for screening a large number of sam-
ples where high iron and zinc samples can be further verified by the
 Determination of Macronutrient and Micronutrient Content in Rice Grains Using… 255

ICP-OES technique since XRF is less accurate than ICP-OES anal-

ysis [20, 21].
In this chapter, we describe the nitric acid-perchloric acid
digestion procedure and the multi-elemental determinations by
ICP-OES using simultaneous optical systems and dual viewing of
the plasma in rice grains.

2 Materials

2.1 Glasswares 1. Glass culture tubes, 25 × 200 mm.

and Apparatus 2. Vortex mixer (Model Vortex Genie 2, Scientific Industries,
Inc., USA).
3. Polypropylene tubes, 25 × 100 mm.
4. Narrow-mouth storage bottles (low-density polypropylene)
with screw closure, 15 mL to 1000 mL capacities.
5. 2 L volumetric flask.
6. 6 L Erlenmeyer flask.
7. 2.5 L glass bottle.
8. 3.5 L container.
9. Teflon bottle.
10. Analytical balance (Model AW120, Shimadzu Corporation,
Japan).
11. Top loading balance (Model BW4200H, Shimadzu
Corporation, Japan).
12. Prototype digestor with flexi glass baffle.
13. Inductively Coupled Argon Plasma—Optical Emission
Spectrometer with WinLab32 software (Model Optima
5300DV, Perkin Elmer, USA).
14. Autosampler (Model ASX-520, CETAC Technologies, USA)
(Fig. 1).
15. Transfer pipette (Model Pipet Lite, Rainin Instruments, LLC,
USA).
16. Bottle dispenser (Model Dispensette organic easy calibration,
BrandTech Scientific, Inc., USA. Model Optifix Basic 10,
EMD Millipore, USA).

2.2 Reagent 1. Nitric acid (conc.): 65%, HNO3 (CAS No. 7697-37-2).
and Standard 2. Perchloric acid (conc.): 69–72%, HClO4 (CAS No. 7601-90-3).
Preparation
3. Hydrochloric acid (conc): 36.5–38.0% (CAS No. 7647-01-0).
4. Liquid argon gas supply: high-purity grade (99.99%).
256 Lilia Molina et al.

Fig. 1 The ASX-520 autosampler

5. Deionized water: prepared from a deionizing system with

cation and anion exchangers to produce deionized water
with 18 MΩ resistivity.
6. 1000 mg/L Sc stock standard in 2% HNO3 (100048-1,
High-­Purity Standards, USA).
7. 1000 mg/L Mn stock standard in 2% HNO3 (100032-1,
High-Purity Standards, USA).
8. 1:10 of HClO4:HNO3 mix acid. Dispense 100 mL of HClO4
(conc.) in a 2.5 L glass bottle. Add 1000 mL of HNO3 (conc.).
Mix well and transfer to the bottle dispenser labeled “1:10 of
HClO4:HNO3” (see Note 1).
9. 1% HNO3 acid. Dispense 30.76 mL of 65% HNO3 in a 2 L
volumetric flask, and dilute to the mark with deionized water.
Mix well and transfer the solution to the Teflon bottle labeled
“1% HNO3” (see Note 2).
10. 5 N HCl. Dispense 2.5 L of concentrated HCl in a 6 L con-
tainer containing 3.5 L of deionized water. Mix well (see
Note 3).
11. Calibration blank (1% HNO3 + 2.78% HClO4). In a 6 L flask,
add 92.31 mL of HNO3 (conc.) and 240 mL of HClO4
(conc.) to 400 mL deionized water. Dilute the solution to the
6 L mark with deionized water and transfer to a designated
container. This solution must be used within 1 year.
12. Mixed Calibration Standards. For a three-point calibration
curve, three stock solutions (lowest standard, middle ­standard,
and highest standard) of mixed elements (Inorganic Ventures,
USA) are used (Table 1, see Note 4).
Determination of Macronutrient and Micronutrient Content in Rice Grains Using… 257

Table 1
Three-point mixed calibration standard

Lowest standard Middle standard Highest

Element (mg/L) (mg/L) standard (mg/L)
Al 1 30 250
Ca 5 60 500
Cu 1 3 25
Fe 1 6 50
K 20 60 500
Mg 5 30 250
Mn 1 6 50
Mo 1 1.5 10
Na 2 12 100
P 10 30 250
S 5 30 250
Zn 1 3 25

13. Initial Verification Standards. Another set of three-point stan-

dards with similar concentrations as that of the three mixed
calibration standard stock solutions but from a different brand
(SM-1907-032, SM-1907-033, SM-1907-034, High Purity
Standards, USA) or source of the mixed calibration standards
(Table 1, see Note 4).
14. Continuing Calibration Verification Low standard (CCVL).
Weigh 250 g of the lowest calibration standard stock solution
in a 500 mL LDPE bottle, add prepared 1% HNO3 + 2.78%
HClO4 to make up to 500 g.
15. Continuing Calibration Verification High standard (CCVH).
Weigh 250 g of the highest calibration standard stock solution
in a 500 mL LDPE bottle. Then add the prepared calibration
blank to make up the weight to 500 g.
16. Internal Standard (1 mg/kg scandium). Weigh 1 g of
1000 mg/L Sc stock standard in a 1000 mL LDPE bottle,
add the calibration blank to make up the weight to 1000 g.
Mix well (see Note 5).
17. Instrument Performance Check Solution (10 mg/kg Mn). Weigh
10 g of 1000 mg/L Mn stock standard in a 1000 mL LDPE
bottle and add the calibration blank to make up the weight to
1000 g. Mix well. This solution must be used within a year.
258 Lilia Molina et al.

1 mg/kg Mn. Weigh 1 g of 1000 mg/L Mn stock standard in a

1000 mL LDPE bottle and add the calibration blank to make
up the weight to 1000 g. Mix well. This solution must be used
within a year.

3 Methods

3.1 Sample 1. Weigh 0.600–0.625 g of oven-dried rice grain sample in a

Preparation culture tube (25 × 200 mm) (see Note 6).
2. Add 12 mL of 1:10 of HClO4:HNO3 mixed acid solution from
the dispenser kept in the perchloric fumehood (see Notes 7
and 8). Wash as much of the sample as possible from the sides
of the tube with the acid.

3.2 Digestion 1. A sample is pre-digested from room temperature to 60 °C for

25 min in the digestor with the baffle system in place (see Note
9). Refer to steps 1–4 of the Digestion temperature scheme in
Table 2.
2. Continue digestion following step 5–21 indicated in Table 2.
Do not move the tubes and allow the digestion process to
complete until about 1 mL of colorless to slight yellow clear
digest is left (see Note 10).
3. Cool the digest and add 20 mL of 1% HNO3. Warm tubes
using the digestor for 35 s and immediately mix using a vortex
mixer. Make the final dilution to the 25 mL mark with 1%
HNO3. Mix the tubes using the vortex mixer and transfer the
sample immediately into appropriately labeled polypropylene
tubes. Analyze digests in the ICP-OES for Na, K, Ca, Mg, Fe,
Mn, Zn, Cu, Al, P, S, and Mo (see Note 11).

3.3 Analyte Detailed instructions on how to operate the ICP-OES can be

Measurement found in the instrument’s manual [22].
via ICP-OES
1. Turn on the computer and the WinLab software. Turn on the
toggle switch for the air supply. Replace tubings in peristaltic
pump and replenish water reservoir. Turn on auto sampler, and
load software.
2. Perform Hg alignment, instrument performance check, view
alignment, background equivalent concentration test (using
the instrument performance check solution, 10 mg/kg Mn)
and precision test (using calibration blank and 10 mg/kg Mn),
according to the manufacturer’s recommended procedures for
the specific instrument model.
3. Place the calibration samples, blanks, quality control samples,
standards, internal standard, and sample tubes in their desig-
nated position in the auto sampler.
Determination of Macronutrient and Micronutrient Content in Rice Grains Using… 259

Table 2
Digestion temperature scheme

Controller Set point Temp setting on Time Actual

step No. type controller (°C) (min) Temp (°C)
1 RAMP 63 25 60
2 HOLD 63 25 60
3 RAMP 89 15 80
4 HOLD 89 20 80
5 RAMP 112 10 100
6 HOLD 112 20 100
7 RAMP 135 10 120
8 HOLD 135 20 120
9 RAMP 147 7 130
10 HOLD 147 60 130
11 RAMP 158 7 140
12 HOLD 158 30 140
13 RAMP 170 7 150
14 HOLD 170 50 150
15 RAMP 193 10 170
16 HOLD 193 40 170
17 RAMP 205 7 180
18 HOLD 205 15 180
19 RAMP 257 15 225
20 HOLD 257 20 225
21 END OFF OFF

4. Perform calibration using calibration blank solution and mixed

calibration standards.
5. Subject the sample digests for Na, K, Ca, Mg, Fe, Mn, Zn, Cu,
Al, P, S, and Mo to ICP-OES. Table 3 summarizes the wave-
lengths used for each of these elements (see Note 12).
6. Once analysis is completed transfer all pump tubes into deion-
ized water to flush the system for around 5 min before shutting
down the system.
7. Data processing
The quantitative values must be reported in appropriate
units, such as milligrams per kilogram (mg/kg) for micro-­
nutrients and % for macro-nutrients. If dilutions were p
­ erformed,
260 Lilia Molina et al.

Table 3
Wavelength used per individual element

Analyte λ (nm)
Al 394.401
Ca 315.887
Cu 327.393
Fe 259.939
K 766.490
Mg 285.213
Mn 259.372
Mo 202.031
P 213.617
S 181.975
Zn 206.200
Na a
589.592, 588.873
a
There are two wavelengths used for Na. This is for the axial and radial viewing position
of the plasma, covering low and high concentration ranges of Na found from different
samples in a batch run

the appropriate corrections must be applied to the sample

values.
If appropriate, or required, calculate results for solids on a
dry-weight basis as follows:
V 
Concentration ( dry weight ) ( mg / kg ) = (C − B ) ×   ,
W 
this is the concentration in ppm.
 V × 0.0001 
Concentration ( dry weight ) ( % ) = (C − B ) ×  ,
 W 
this is how to convert ppm to %, by multiplying it with
0.0001.
where
C = Digest concentration (mg/L)
B = Average method blank (mg/L)
V = Final volume in mL after sample preparation
W = Weight in g of sample
8. Quality Control.
The following steps are performed as part of the labora-
tory’s quality control program for every ICP-OES run:
Determination of Macronutrient and Micronutrient Content in Rice Grains Using… 261

(a) 
Instrument performance check using reagent blank and
10 ppm Mn.
(b) Verify the recovery of the resulting calibration curve with
the calibration standard verification (CSV) when run as
unknown, the analyzed concentration of the CSV must be
within the acceptance criteria of ±3% of the manufacturer’s
value, otherwise recalibration must be done.
(c) Analyze the initial calibration verification (ICV) standard
after calibration. The analyte value should be ±10% of the
expected value. If the analyzed value is outside the ±10%
limit, the new calibration standards must be used and the
instrument is recalibrated prior to sample analysis.
(d) 
Dilute and rerun samples with analyte concentrations
beyond the concentration of the highest calibrant.
(e) Include at least one method blank per sample batch, where
analyte value must not exceed the method detection limit;
otherwise, the source of contamination should be identi-
fied and resolved before continuing analyses.
(f) Analyze at least one reference material with each sample
batch. The calculated recovery for each analyte should
be within ±2 s.d. of the laboratory’s long-term value. If
the analyzed value is outside this limit, the source of the
problem should be identified and resolved before continu-
ing analyses.
(g) Estimate precision by means of duplicate/replicate analy-
ses for every ten samples. Percent RPD must be ≤20%.
However, owing to the heterogeneity of some samples
submitted (e.g., whole milled or unpolished rice grain
samples), it is not always the case. For such cases, the client
is informed prior to conduct of analysis and the % RPD
obtained in replicated samples is reflected on the final
report.
D1 − D 2
RPD = × 1000
 D1 + D 2 
 
 2 

where:
D1 = First sample value
D2 = Second sample value
(h) Unusual situations occurring during sample preparation
and analysis must be documented on the work sheet, and
­non-­conformity/corrective action reports must be issued
if necessary.
262 Lilia Molina et al.

4 Notes

1. Mix and transfer acid under the perchloric hood. Solution

must be used within 6 months.
2. This is used for dilution of sample digest. Solution must be
used within 6 months.
3. This is used to acid-wash glasswares. Solution must be changed
every 6 months or as necessary.
4. These may be manually prepared from individual element
stock solutions or can be purchased as mixed standards.
5. Internal standard of 1 mg/kg Sc is used for rice grain samples.
This solution must be used within 1 year.
6. The rice grain sample can be milled, unpolished, or paddy
(whole grains or ground grains), depending on the customer’s
requirement. Rice grain samples are oven-dried for 2 h at
70 °C.
7. The mixed acid is needed for digestion since normal concen-
trated nitric acid only removes the easily oxidizable material in
the sample. The more resistant organic matter is acted upon
later by the stronger oxidizer, such as perchloric acid [23].
8. If the samples are ground material, mix immediately using a
Vortex mixer at its high-speed setting to avoid hardening of
sample into balls of dry material that may “spit” when
digesting.
9. Pre-digestion with nitric acid and perchloric acid will give pro-
tection against possible explosion during the digestion pro-
cess. An adequate amount of nitric acid must be present and
overnight pre-digestion is recommended. Pre-digestion
removes easily oxidized organic materials, including hydroxyl
compounds that give rise to unstable perchlorate esters [16].
Wear goggles, insulated gloves, and lab coats when doing acid
digestion. The improper use of HClO4 has caused numerous
explosions during digestion of organic materials. Many explo-
sions were caused by not first predigesting the sample with an
adequate amount of HNO3 [16].
10. Do not let digests boil dry. Samples high in potassium and
SiO2 may have precipitated in the final digest.
11. Most sample residue will dissolve fully if allowed to stand for
several hours. Final digest will be colorless. If there is a white
precipitate, it may be silica or KClO4. Some samples with very
high K may need to be warmed to dissolve. KClO4 may
­dissolve if it is re-warmed. After tubes are removed, rinse the
hood and exhaust stack with the wash-down system for about
30 min with the blower running.
 Determination of Macronutrient and Micronutrient Content in Rice Grains Using… 263

12. During the method development, wavelength or line selection

for each analyte is based upon avoidance of spectral interfer-
ences, detection limit requirement, and working range require-
ment. Key instrument parameters such as gas flow rate and
applied radiofrequency power must also be optimized [24].

Acknowledgment

The authors are grateful to Mariel Ong, Reah Gonzales, Marnol

Santos, Edgar Amoloza, Jose Rosales, and Karen Serdeña, for the
assistance in the SOP documentation, sample preparation, valida-
tion, and optimization of the digestion and ICP methodologies in
GQNSL. This work has been supported under the CGIAR thematic
area Global Rice Agri-Food System CRP, RICE, The Indian Council
of Agricultural Research, Taiwan support from Council of Agriculture
(COA), Ministry of Foreign Affairs (MOFA) and BBSRC Newton
grant funding Reference number BB/N013808/1.

References
1. Sharma P, Aggarwal P, Kaur A (2017) 10. Kennedy G, Burlingame B (2003) Analysis of
Biofortification: a new approach to eradicate food composition data on rice from a plant
hidden hunger. Food Rev Inter 33(1):1–21 genetic resources perspective. Food Chem
2. Walsh CT, Sandstead HH, Prasad AS, 80(4):589–596
Newberne PM, Fraker PJ (1994) Zinc: health 11. Stangoulis J (2010) Technical aspects of zinc
effects and research priorities for the 1990s. and iron analysis in biofortification of the staple
Envi Health Perspectives 102(Suppl 2):5–46 food crops, wheat and rice, in 19th World
3. Prasad AS (2001) Recognition of zinc-­ Congress of Soil Science. Soil Solutions for a
deficiency syndrome. Nutrition 17(1):61–69 Changing World, Brisbane, pp 42–44
4. Wallwork JC, Milne DB, Sims RL, Sandstead 12. Guild GE, Stangoulis J (2016) Non-matrix
HH (1983) Severe zinc deficiency: Effects on matched glass disk calibration standards
the distribution of nine elements (potassium, improve XRF micronutrient analysis of wheat
phosphorus, sodium, magnesium, calcium, grain across five laboratories in India. Front
iron, zinc, copper and manganese) in regions Plant Sci 7:784
of the rat brain. J Nutr 113(10):1895–1905 13. Trijatmiko KR, Dueñas C, Tsakirpaloglou N,
5. Yip R (1994) Iron deficiency: contemporary Torrizo L, Arines FM, Adeva C, Balindong J,
scientific issues and international programmatic Oliva N, Sapasap MV, Borrero J, Rey J,
approaches. J Nutr 124:1479S–1490S Francisco P, Nelson A, Nakanishi H, Lombi E,
6. Andrews NC (1999) Disorders of iron metab- Tako E, Glahn RP, Stangoulis J, Chadha-­
olism. New England J Med 341(26): Mohanty P, Johnson AAT, Tohme J, Barry G,
1896–1995 Slamet-Loedin IH (2016) Biofortified indica
rice attains iron and zinc nutrition dietary tar-
7. Javier EQ (2004) Let's promote brown rice to gets in the field. Sci Rep 6:19792
combat hidden hunger. Rice Today 3(1):38
14. Wu S, Feng X, Wittmeier A (1997) Microwave
8. Butardo V Jr, Sreenivasulu N (2016) Tailoring digestion of plant and grain reference materials
grain storage reserves for a healthier rice diet in nitric acid or a mixture of nitric acid and
and its comparative status with other cereals. hydrogen peroxide for the determination of
Inter Rev Cell Mol Bio 323:31–70 multi-elements by inductively coupled plasma
9. Tuaño APP, Regalado MJC, Juliano BO mass spectrometry. J Anal At Spectrom
(2016) Grain quality of rice in selected retail 12:797–806
stores and supermarkets in the Philippines. 15. Wheal MS, Fowles TO, Palmer LT (2011) A
Inter J Phi Sci Tech 9(1):15–22 cost-effective acid digestion method using
264 Lilia Molina et al.

closed polypropylene tubes for inductively 20. Stangoulis J, Guild GE (2014) Measuring trace
coupled plasma optical emission spectrometry micronutrient levels in crops. In: The 2nd
(ICP-OES) analysis of plant essential elements. global conference on biofortification: getting
Anal Methods 3:2854–2863 nutritious foods to people. Kigali, Rwanda
16. Schilt AA (1979) Perchloric Acid and 21. Paltridge NG, Palmer LJ, Milham PJ, Guild
Perchlorates. G. F. Smith Chemical Co, GE, JCR S (2012) Energy-dispersive X-ray
Columbus, OH fluorescence analysis of zinc and iron concen-
17. Gorsuch TT (1970) The Destruction of tration in rice and pearl millet grain. Plant Soil
Organic Matter, 1st edn. Pergamon Press, 361(1):251–260
Oxford 22. Perkin-Elmer Corporation (2006) Perkin Elmer
18. Hansen TH, Laursen KH, Persson DP, Pedas WinLab32 Instrument Control Software Guide.
P, Husted S, Schjoerring JK (2009) Micro-­ Perkin-Elmer Corporation, Massachusetts
scaled high-throughput digestion of plant tis- 23. Kebbekus BB (2003) Preparation of Samples
sue samples for multi-elemental analysis. Plant for Metals Analysis. In: Mitra S (ed) Sample
Methods 5:12 Preparation Techniques in Analytical Chemistry.
19. Kim G, Kwak J, Choi J, Park K (2012) John Wiley & Sons, Inc., Hoboken, NJ
Detection of nutrient elements and contamina- 24. Gaines PR (2011) ICP Operations Guide: A
tion by pesticides in spinach and rice samples Guide for using ICP-OES and
using Laser-Induced Breakdown Spectroscopy ICP-MS. Inorganic Ventures, Inc,
(LIBS). J Agric Food Chem 60:718–724 Christiansburg, VA, p 37
 Chapter 15

Determination of Cadmium Concentration in Milled

and Brown Rice Grains Using Graphite Furnace Atomic
Absorption Spectrometry
Lilia Molina, Jennine Rose Lapis, Nese Sreenivasulu,
and Rosa Paula O. Cuevas

Abstract
Heavy metal pollution is a growing public health concern since it poses a food risk to public health via
metal transfer. Cadmium is of particular concern because it is a potential carcinogen if exceed tolerable
limits in the grain. Hence, it is important to monitor the cadmium content of rice before it reaches the
market to ensure public healthy safety, especially in areas known to have high cadmium levels in soil. In this
chapter, the method used to determine the concentration of cadmium in milled and brown rice grain
samples is described. This method involves sample digestion with concentrated nitric acid and hydrogen
peroxide and analysis of cadmium by Graphite Furnace Atomic Absorption Spectrometry (GF-AAS).
Because cadmium concentrations are low in rice grains, quantification of cadmium content requires the use
of a more sensitive instrument, such as GF-AAS.

Key words Cadmium in rice, Heavy metals, GF-AAS

1 Introduction

Heavy metals (like cadmium, lead, mercury, and arsenic) in the

environment have potential deleterious effects on humans and
animals because heavy metals can be accumulated in the edible
parts of food crops, which are then ingested by humans and by
animals (reviewed in [1]). Cadmium, a heavy metal, can be taken
up by rice plants from the soil and can accumulate in different parts
of the plant, including the grain [2]. However, cadmium is mostly
accumulated in the stalk. In rice samples sourced from various
heavy metal-contaminated sites in China, the concentrations of
cadmium in the hull and in the milled grain are reported to be
mostly similar and much lower than the cadmium concentration in
the stalk [1]. On the other hand, rice samples from cadmium-con-
taminated fields in Thailand showed that the stalk had a wider

Nese Sreenivasulu (ed.), Rice Grain Quality: Methods and Protocols, Methods in Molecular Biology, vol. 1892,
https://doi.org/10.1007/978-1-4939-8914-0_15, © Springer Science+Business Media, LLC, part of Springer Nature 2019

265
266 Lilia Molina et al.

range of cadmium concentrations than the leaf and the unpolished

grain, with leaf and grain having similar concentrations [3].
Although the bioaccumulation of cadmium is lower in the rice
grain than in the stalk, cadmium is still a public health concern
because of its toxicity and because of the high risk of exposure
through continuous consumption. The critical level for cadmium in
rice grains is 0.4 mg/kg [4] and numerous studies have reported
grain cadmium concentrations higher than this (e.g., [1, 3, 5–7]),
indicating that consumption of rice cultivated from areas contami-
nated with cadmium may pose increased health risk. For instance,
elevated urinary cadmium levels are reportedly associated with renal
and urinary tract disorders, bone toxicities, and elevated blood pres-
sure (reviewed in [8]) while meta-analysis of available data suggests
that cadmium exposure is associated with risk of breast cancer,
though more studies are needed to confirm this [9]. The risk is fur-
ther exacerbated by the per capita consumption rates of rice; i.e.,
high per capita rice consumption in countries such as Sri Lanka,
India, and Bangladesh can cause cadmium bioaccumulation to
exceed the threshold [10]. Research into cadmium uptake becomes
even more important in the development of biofortified rice.
Cadmium has been reported to mimic zinc, such that cadmium can
be taken up by the plant and compete with zinc (reviewed in [11]).
Current efforts to boost zinc content in rice must ensure that cad-
mium is not co-transported to avoid the risk of increase of the heavy
metal concentration. Therefore, it is important to ensure that bio-
fortified rice does not accumulate cadmium, which can be done by
quantitatively monitoring the cadmium content of the rice grain.
Several methods for trace cadmium analysis are available such
as Flame Atomic Absorption Spectrometry (FAAS) [3], Graphite
Furnace Atomic Absorption Spectrometry (GF-AAS) [12],
Differential Pulse Anodic Stripping Voltammetry [13], Inductively
Coupled Plasma Optical Emission Spectrometry (ICP-OES) [14],
and Inductively Coupled Plasma Mass Spectrometry (ICP-MS)
[15]. However, existing methodologies have not become more
widely used because of several limitations, the methodologies may
be too complex and expensive to perform, not sensitive enough to
detect extremely low cadmium levels in rice grains and straw, or
require tedious pre-concentration steps to quantify trace cadmium
levels. Sample preparation procedures are also associated with sam-
ple spitting during digestion of agricultural sample matrices.
The method routinely used at the International Rice Research
Institute’s (IRRI) Grain Quality and Nutrition Services Laboratory
(GQNSL) to measure cadmium content is via closed vessel acid
digestion and measure it through graphite furnace-atomic absorp-
tion spectrometry (GF-AAS) [16]. The closed vessel digestion tech-
nique is preferred over the open vessel digestion since it reduces risk
of contamination and loss of volatile analyte [17], while its complex
organic matrix gets oxidized to release cadmium in solution (during
closed vessel digestion) for further quantification via GF-AAS [18].
 Determination of Cadmium Concentration in Milled and Brown Rice Grains Using… 267

2 Materials

2.1 Glassware 1. 50 mL screw-capped polypropylene tube.

and Apparatus 2. DigiPrep LS 72-tube (50 mL) block digestor (SCP Science,
USA) (Fig. 1).
3. HGA 900 Graphite Furnace with AS 800 Autosampler and
AAnalyst 400 atomic absorption spectrometer equipped with
cadmium Electrodeless Discharge Lamp (EDL, All from Perkin
Elmer, Singapore) (Fig. 2).
4. Glass volumetric flasks (1 L, 100 mL, 25 mL).
5. Horizontal shaker, Model E59100.25, Eberbach Corp., USA.
6. Calibrated dispenser, 10 mL L/I Repipet® Dispensers, Cole
Parmer, Canada (see Note 1).
7. Desiccator cabinet, Boekel, Philadelphia, USA.
8. Weighing balance.

2.2 Reagents 1. Deionized water.

and Standards 2. Concentrated Nitric Acid, HNO3 (65%, CAS No.
7697-37-2).
3. Hydrogen Peroxide, H2O2, 30% (v/v) in H2O (CAS No.
7722-84-1).
4. 4% (v/v) HNO3: Dilute concentrated nitric acid (40 mL) in
deionized water (1 L). This solution must be used within
1 year. This also serves as the calibration blank.
5. Pd stock solution (10,000 ppm) in 15% HNO3 (High Purity
Standards, USA).
6. Mg(NO3)2 (10,000 ppm, Perkin Elmer, Singapore).
7. Matrix modifier (Pd-Mg(NO3)2): Dilute an aliquot of the Pd
stock solution (3.00 g) in 15% HNO3 (prepared from 23.10 g

Fig. 1 The DigiPrep block digestor

268 Lilia Molina et al.

Fig. 2 The GF-AAS setup

of the conc. HNO3 and diluted to 100.00 g with deionized

water) then add Mg(NO3)2 (2.00 g) in a polyethylene bottle.
Make up to 10.00 g with 4% HNO3.
8. Cadmium stock solution A (1000 ppm cadmium): Single-­
element cadmium stock solution (in 2% HNO3 matrix) (High
Purity Standards, Charleston, SC, USA). This solution must
be used before the expiration date specified by the manufac-
turer (see Note 2).
9. Cadmium intermediate standard solution B (1 ppm cadmium):
Weigh 0.100 g cadmium stock solution A into a polyethylene
bottle and make up to 100.00 g with 4% HNO3. This solution
must be used within 6 months (see Note 3).
10. Cadmium working standards: three concentrations of cad-
mium solutions (0.5, 1.0, and 2.0 ppb). Weigh the required
amount of cadmium standard solution B (Table 1) into poly-
ethylene bottles and dilute to 50 g with 4% HNO3. Mix well.
The solutions must be used within 2 weeks (see Note 4).
11. National Institute for Environmental Studies (NIES) Certified
Reference Materials (CRM) #10-a and #10-b, low- and
medium-level cadmium rice flour, respectively; both of which
are unpolished. The CRM are used for method validation and
optimization.
 Determination of Cadmium Concentration in Milled and Brown Rice Grains Using… 269

Table 1
Components of cadmium working standards

Weight of Cd int. standard

Conc. (ppb Cd) solution B (g) Total weight (g)
0.5 0.025 50.00
1.0 0.050 50.00
2.0 0.100 50.00

12. In-house rice flour reference material (99G, i.e., milled rice
prepared from mixed paddy rice grain collected from ZES in
1999) used for routine cadmium analysis.
13. Continuing calibration verification (CCV) standard or QC
insert (0.5 ppb cadmium in 4% HNO3): In a polyethylene
bottle, dilute an aliquot of cadmium standard solution B
(0.025 g) to 50 g with 4% HNO3. Mix well. The solution
must be used within 2 weeks.

3 Procedure

3.1 Oven-Drying Oven-dry rice flour prepared from milled or brown rice grain sam-
the Samples ples for 2 h at 80 °C and store the dried samples packed in paper
envelope inside the desiccator cabinet prior to weighing.

3.2 Closed Tube 1. Weigh 0.30 g of oven-dried sample in a polypropylene tube

Digestion and add 2.0 mL concentrated HNO3 and 0.5 mL H2O2.
2. Repeat step 1 for the in-house rice reference material.
3. For the blank sample, dispense 2.0 mL concentrated HNO3
and 0.5 mL H2O2 in a polypropylene tube.
4. Cap the tubes and allow to stand overnight (pre-digestion) at
room temperature (22–26 °C).
5. Digest all samples in the block digestor using the settings listed
in Table 2 (see Note 5).
6. After the digestion, remove the tubes from the digestor and
allow them to cool to room temperature.
7. Add deionized water (22 mL) into the tube to allow cooling of
the diluted acid. Add the volume up to 25 mL with deionized
water (Fig. 3).
8. Re-seal the caps and agitate in a shaker (180 rpm, 5 min).
9. Allow the undissolved materials to settle (1 h). After the set-
tling time, decant into sample cups for GF-AAS run.
270 Lilia Molina et al.

Table 2
Digestor settings for the digestion of rice grain to release cadmium

Stage Temperature (°C) Time (min)

Warming 80 30
Digestion 125 120

Fig. 3 Sample of polypropylene tube with mark

3.3 Analytical 1. Start-up Perkin Elmer GF-AAS for cadmium.

Determination 2. Clean and check furnace windows and the graphite tube.
of Cadmium Content Replace the graphite tube as needed.
Using Graphite
3. Perform tip alignment, furnace alignment, and autosampler
Furnace Atomic
alignment.
Absorption
Spectrometer 4. Warm up the cadmium EDL and conduct background correc-
tion for at least 15 min.
5. Run a sensitivity check by measuring the characteristic mass,
mo of Solution B. In the instrument readings, the measured mo
should be within ±20% of the cookbook value or of the com-
parison mo (0.5 pg/0.0044A) (see Note 6).
 Determination of Cadmium Concentration in Milled and Brown Rice Grains Using… 271

6. Perform a precision check by running Solution B five times.

Relative Standard Deviation (RSD) values obtained must be <5%.
7. Prepare the sample information sheet for the sequence of the
analysis. The sequence of materials for analysis includes the
calibration standards (Table 1), the matrix modifier, the dilu-
ents, the samples, and the reference materials. Place sample
cups containing these materials in the autosampler based on
the encoded sequence.
8. Quantify the cadmium contents of the materials in the sequence
using the optimized furnace program (Table 3). Check from
time-to-time if the sample is being dispensed properly to the
graphite tube.
9. Shut down the GF-AAS after analysis.

4 Quality Control

1. Method validation for cadmium in rice grains is done using

NIES CRM #10 prepared via closed vessel digestion
(Subheading 3.2) and analyzed using the GF-AAS optimized
setting (Table 3).
2. The calibration curve must be linear thru zero with a correla-
tion coefficient ≥0.995; re-calibrate if this correlation coeffi-
cient is not achieved.
3. Each sequence run or batch of samples must have at least one
method blank, one reference material, and one duplicated
sample run for every ten samples. The result for the in-house
rice reference material, 99G, must be within ±2 s.d. of the
mean in the control chart (see Note 10).
4. The relative percent difference (% RPD) for sample repeats
must be ≤20%. However, owing to the heterogeneity of some
samples (e.g., milled grain samples), it is not always the case.
The cadmium concentrations of standards included in sequence
runs as unknowns must be within ±10% of the certified or the-
oretical value. % RSD for ICV and CCV insert must be ≤10%
RSD (see Note 11).
5. If the results are not acceptable, take corrective action or qual-
ify the data accordingly (see Note 12).
6. Unusual situations such as presence of undigested sample after
sample preparation, high method blank readings, poor sensi-
tivity signal, and poor precision between replicated samples
occurring during analysis must be documented on the work
sheet, and non-conformity/corrective action reports must be
issued if necessary (see Note 12).
272 Lilia Molina et al.

Table 3
Optimized furnace programa used for the analysis of cadmium in rice
grains

Step no. Temp. (°C) Ramp time (s) Hold time (s) Internal Ffow
1 110 5 20 250
2 120 10 20 250
3 600 10 30 250
4 20 1 10 250
5 1600 0 5 0
6 2450 1 3 250
a
Refer to Perkin Elmer AAnalyst 400 atomic absorption spectrometer equipped with
HGA 900 Graphite Furnace, with AS 800 Autosampler Protocol or the Perkin Elmer
AA WinLab Software Guide Manual [14] for the complete guide to the operation of
GF-AAS (see Note 7). The matrix modifier is Pd-Mg(NO3)2 (see Notes 8 and 9)

5 Calculations and Reporting Results

1. Results of cadmium content must be reported in μg/kg. If

dilution was used, appropriate dilution factor must be incor-
porated in the result.
2. Calculate results for rice grain as follows:
µg  vol. of digest ( mL )
cd   = ( Rs − Rb ) × .
 kg  wt. of oven − dried sample ( g )
where
Rs = μg/L Cd in sample digest obtained from standard curve
Rb = μg/L Cd in blank obtained from standard curve.

6 Notes

1. To calibrate the dispenser bottle, dispense 10 mL of the digest-

ing solution from the dispenser into a graduated cylinder.
Obtain the weight of the dispensed solution. Repeat the pro-
cedure five times. Calculate mean and RSD. The mean must
be 10 ± 0.1 g and the RSD must fall within ±2%. Otherwise,
adjust the dispenser and repeat the calibration.
2. The stock solution may also be prepared from ultra-high-­
purity-grade chemical or metal (99.99% pure or greater).
Determination of Cadmium Concentration in Milled and Brown Rice Grains Using… 273

3. The weight of the cadmium stock solution may not be exactly

0.1 g, depending on the density of the solution. Take into
account the density and recalculate the weight of stock solu-
tion needed.
4. The solution containing 1.0 ppb in 4% (v/v) HNO3 both can
be used as a calibration working standard and as a standards
sensitivity check.
5. Pressure may build up during the warming stage. Loosen the
caps to equalize the pressure and immediately re-tighten firmly
and place back to the digestor.
6. The characteristic mass (mo) is defined as the mass of analyte
that gives 1% absorption or 0.0044 absorbance [18]. If calcu-
lated mo < 20% of the comparison mo, troubleshoot the
GF-AAS for low absorbance due to possible contamination
and inaccuracy in the preparation of the calibration standards.
If calculated mo > 20% of the comparison mo, check for the
integrity of the EDL; the integration time may be too short,
the internal gas flow during atomization was not set to zero,
or the wavelength/slit width were not corrected.
7. For cadmium analysis, use 228.8 nm wavelength and argon
gas for GF-AAS.
8. For standard and sample injections, a working volume of
20 μL is used, with an additional 5 μL aliquot of matrix modi-
fier added in a combined injection. The mixing of the reagents
and the samples is performed by the autosampler. A continu-
ous light source (deuterium lamp) is used for background
correction.
9. The rice grain contains elements such as sodium, potassium,
calcium, and magnesium [19] and will require a matrix modi-
fier to suppress matrix interferences, reduce cadmium volatil-
ity, and stabilize the cadmium signal [20, 21].
10. The control chart is used to monitor how an analytical process
changes over time. Data from the reference materials used in a
batch of analyzed samples are plotted in time order.
Measurement process is within statistical control if the results
for reference materials are within ±2 s.d. in the control chart.
If the data shows out of control signal (i.e., above ±3 s.d. in
the control chart or observed QC trending), mark it on the
chart and investigate the cause. Proper documentation on the
cause and how it was corrected is required.
11. The initial calibration verification (ICV) and the continuing
calibration verification (CCV) are quality control standards
inserted after the calibration standards (ICV) to check the
validity of the calibration standards’ concentrations. The
274 Lilia Molina et al.

changes in concentrations of the calibration standards may be

attributed to changing conditions of the graphite tube such as
corrosion or formation of cracks. These calibration verification
standards are from independent standard sources and are
placed at regular intervals in the sequence run (CCV).
12. These are some corrective actions to do if the results are not
acceptable. Verify the quality and the homogeneity of the sam-
ple and the reference materials used by visual inspection, check
the quality of the calibration standards used (its concentration
and its expiry date), check for possible contamination of lab-
ware, and inspect the graphite tube if it is already old and cor-
roded and requires replacement.

Acknowledgments

The authors thank the following laboratory staff involved in the

sample preparation, in the method development and optimization,
and in the preparation of the initial documentation: Edgar
Amoloza, Marnol Santos, Karen Serdeña, and Eric Jhon Cruz.
This work has been supported under the CGIAR thematic area
Global Rice Agri-Food System CRP, RICE, The Indian Council of
Agricultural Research, Taiwan support from Council of Agriculture
(COA), Ministry of Foreign Affairs (MOFA) and BBSRC Newton
grant funding Reference number BB/N013808/1.

References

1. Zhuang P, McBride MB, Xia H, Li N, Li Z 5. Fang Y, Sun X, Yang W, Ma N, Xin Z, Fu J, Liu

(2009) Health risk from heavy metals via con- X, Liu M, Mariga AM, Zhu X, Hu Q (2014)
sumption of food crops in the vicinity of Concentrations and health risks of lead, cad-
Dabaoshan mine, South China. Sci Total mium, arsenic, and mercury in rice and edible
Environ 407:1551–1561 mushrooms in China. Food Chem
2. Chaney RL, Ryan JA, Li YM, Welch RM, 147:147–151
Reeves PG (1996) Phyto-availability and bio- 6. Simmons RW, Pongsakul P, Chaney RL,
availability in risk assessment for Cd in agricul- Saiyasitpanich D, Klinphoklap S, Nobuntou
tural environments. In: Sources of cadmium in W (2003) The relative exclusion of zinc and
the environment. Organization for Economic iron from rice grain in relation to rice grain
Cooperation and Development, Paris, cadmium as compared to soybean:
pp 49–78 Implications for human health. Plant Soil
3. Simmons RW, Noble AD, Pongsakul P, 257(1):163–170
Sukreeyapongse O, Chinabut N (2008) 7. Sun L, Xu X, Jiang Y, Zhu Q, Yang F, Zhou J,
Analysis of field-moist Cd contaminated paddy Yang Y, Huang Z, Li A, Chen L, Tang W,
soils during rice grain fill allows reliable predic- Zhang G, Wang J, Xiao G, Huang D, Chen C
tion of grain Cd levels. Plant Soil (2016) Genetic diversity, rather than cultivar
302(1):125–137 type, determines relative grain Cd accumula-
4. Codex Alimentarius Commission (2005) tion in hybrid rice. Front Plant Sci 7:1407
Report of the 37th session of the codex com- 8. Swaddiwudhipong W, Limpatanachote P,
mittee on food additives and contaminants, Mahasakpan P, Krintratun S, Punta B,
Codex Alimentarius Commission, The Hague. Funkhiew T (2012) Progress in cadmium-
p 201 related health effects in persons with high envi-
 Determination of Cadmium Concentration in Milled and Brown Rice Grains Using… 275

ronmental exposure in northwestern Thailand: of cadmium in rice grain downstream of a zinc

a five-year follow-up. Environ Res 112: mineralized area in Thailand: implications for
194–198 public health. Environ Geochem Health 27:501
9. Larsson SC, Orsini N, Wolk A (2015) Urinary 15. Shindoh K, Yasui A (2003) Changes in cad-
cadmium concentration and risk of breast mium concentration in rice during cooking.
cancer: a systematic review and dose-response Food Sci Tech Res 9(2):193–196
meta-analysis. Epidemiol Rev 16. Gunduz S, Akman S (2013) Determination of
182(5):375–380 lead in rice grains by solid sampling HR-CS
10. Meharg AA, Norton G, Deacon C, Williams P, GFAAS. Food Chem 141:2634–2638
Adomako EE, Price A, Zhu Y, Li G, Zhao F-J, 17. Hansen TH, Laursen KH, Persson DP, Pedas
McGrath S, Villada A, Sommelia A, De Silva PM, P, Husted S, Schjoerring JK (2009) Micro-­
Brammer H, Dasgupta T, Islam MR (2013) scaled high-throughput digestion of plant tis-
Variation in rice cadmium related to human expo- sue samples for multi-elemental analysis. Plant
sure. Environ Sci Technol 47(11):5613–5618 Methods 5:12
11. Palmgren MG, Clemens S, Williams LE, 18. Perkin-Elmer Corporation (1996) Analytical
Krämer U, Borg S, Schjørring JK, Sanders D methods for atomic absorption spectroscopy.
(2008) Zinc biofortification of cereals: Perkin-Elmer Corporation, Massachusetts
Problems and solutions. Trends Plant Sci 19. Juliano BO (1993) Rice in human nutrition.
13(9):464–473 In: FAO food and nutrition series. Food and
12. Shi J, Li L, Pan G (2009) Variation of grain Cd Agriculture Organization, Rome, pp 4–59
and Zn concentrations of 110 hybrid rice culti- 20. Bettinelli M, Baroni U, Pastorelli N (1988)
vars grown in a low-Cd paddy soil. J Environ Determination of arsenic, cadmium, lead, anti-
Sci 21(2):168–172 mony, selenium and thallium in coal fly ash
13. Kjuus BE (1985) Determination of cadmium, using the stabilised temperature platform fur-
copper, lead, and zinc in inorganic fertilizers by nace and Zeeman-effect background correc-
anodic stripping voltammetry. Fresenius J Anal tion. J Anal Atom Spectro 3:1005–1011
Chem 322:577–580 21. Schlemmer G, Radziuk B (1999) Analytical
14. Simmons RW, Pongsakul P, Saiyasitpanich D, graphite furnace atomic absorption spectrome-
Klinphoklap S (2005) Elevated levels of cad- try: a laboratory guide. Birkhäuser, Germany,
mium and zinc in paddy soils and elevated lelvels p 286
 Chapter 16

Analysis of Developing Rice Grain Transcriptome

Using the Agilent Microarray Platform
Mandy Püffeld, Christiane Seiler, Markus Kuhlmann,
Nese Sreenivasulu, and Vito M. Butardo Jr.

Abstract
Transcriptome analysis reflects the status quo of transcribed genetic code present in the form of mRNA,
which helps to infer biological processes and unravel metabolic status. Despite the increasing adoption of
RNA-Seq technique in recent years, transcriptome analysis using the microarray platform remains the gold
standard technique, which offers a simpler, more cost-effective, and efficient method for high-throughput
gene expression profiling. In this chapter, we described a streamlined transcriptomic analyses pipeline
employed to study developing rice grains that can also be applied to other tissue samples and species. We
described a novel RNA extraction method that obviates the problem introduced by high-starch content
during rice grain development that usually leads to reduction in RNA yield and quality. The detailed pro-
cedure of microarray analysis involved in cDNA synthesis, cRNA labeling, microarray hybridization, slide
scanning, feature extraction to QC validation has been described. The description of a newly developed
Indica- and Japonica-specific microarray slides developed from the genome information of subpopulation
to study gene expression of 60,000 genes has been highlighted. The downstream bioinformatics analyses
including expression QTL mapping and gene regulatory network analyses were mentioned.

Key words cDNA synthesis, cRNA labeling, Microarray, Rice grain, RNA extraction, Hybridization,
Slide scanning

1 Introduction

Transcriptome is the sum of all ribonucleic acid (RNA) molecules

present in a cell, which are generated from the genome through
transcription. The composition of the transcriptome is dependent
on the active transcription of genes, as well as processing and sta-
bility of the messenger RNA. The composition of a transcriptome
reflects the response of a cell or tissue and thus the status of each
individual cell can be monitored by analysis of its transcriptome.
The presence, absence, or abundance of transcripts derived from
specific marker genes can be used to estimate the developmental
status or response of an organism to environmental changes.

Nese Sreenivasulu (ed.), Rice Grain Quality: Methods and Protocols, Methods in Molecular Biology, vol. 1892,
https://doi.org/10.1007/978-1-4939-8914-0_16, © Springer Science+Business Media, LLC, part of Springer Nature 2019

277
278 Mandy Püffeld et al.

Transcripts can be detected by various methods after the extrac-

tion of the total RNA from a selected tissue at a specific time point
under a defined growth condition. Two broadly applied methods
currently exist that allow the quantification of transcripts from their
respective transcriptomes, which include RNA sequencing (RNA-
Seq) and whole-genome microarray. The RNA-Seq technology is
more recent and offers a number of advantages compared to the
older microarray hybridization technology. RNA-Seq allows unbi-
ased detection of novel transcripts because it is not limited to spe-
cies- or transcript-specific probes, novel transcripts, gene fusions,
single-nucleotide variants such as edited nucleotides, indels (small
insertions and deletions), and other previously unknown changes
which otherwise cannot be detected by microarrays. RNA-Seq also
provides a broader dynamic range of detection, which depends on
the depth of sequencing coverage. In addition, RNA-Seq technol-
ogy is able to quantify discrete, unique sequencing read counts rep-
resenting the quantity of transcripts. However, the data analysis of
RNA-Seq data previously required extensive bioinformatics exper-
tise, presenting a major hurdle to the adoption of RNA-Seq tech-
nology by biologists. Currently, several user-­ friendly tools are
available to analyze microarray-based transcriptome analysis that
can significantly simplify the analysis process, providing accessible
solutions for any researcher without requiring significant learning
curve for data analyses [1–6]. In contrast, array hybridization tech-
nology measurement is limited by oligos represented on the micro-
array which is customized and in addition signal saturation issues
need to be dealt.
The comparability of results derived from RNA-Seq and array
hybridization are still under discussion. Several parallel compari-
sons between RNA-Seq and microarray hybridization technology
have been performed which revealed a 68% overlap of commonly
detectable genes [7–13]. The comparable reliability of detectable
differential expressed genes is approximately 90% in both technol-
ogies validated by quantitative PCR. However, comparisons like
this can only be performed if sufficient genomic information of the
species under investigation is available. Because sufficient sequence
information is the prerequisite for reliable microarray hybridiza-
tion experiments, most of the well-characterized model species
such as rice have predesigned microarrays available.
In terms of cost-efficiency, microarray technology is signifi-
cantly cheaper than RNA-Seq technology based on the current
price lists from various providers. Also, the amount of data stored
from the microarray technology is much smaller and therefore eas-
ier to handle than the RNA-Seq datasets. While the data derived
from the microarray hybridization is approximately 30 MB, the
RNA-Seq data set of one sample ranges from 30 to 65 GB. This will
not only affect time of uploading and downloading of data, but it
 Rice Grain Microarray-Based Transcriptomics 279

will also significantly slow down the time needed for analysis and
increase the cost of data storage. Taken together, although RNA-
Seq is now widely used, gene expression analysis by whole-­genome
microarray is still an applicable technology especially when several
samples and multiple replications should be performed such as in
the screening of a breeding population or diversity panel.
Several studies have already utilized the power of genome-­
wide microarray to characterize gene expression profile in develop-
ing rice grains. Microarray technology was successfully utilized in
determining genes controlling nutrient partitioning [14], analyz-
ing the effect of elevated temperature on grain filling [15], identi-
fying multiple regulatory pathways involved in the formation of
chalkiness trait [16], and associating genes for grain number [17].
Because of the economic advantage of microarray technology, it
finds its application in transcriptome mapping which involves
genome-wide genetic analysis of gene expression data to identify
expression QTLs (eQTLs) to further unravel metabolic, regula-
tory, and developmental pathways [18–21]. In this method, the
microarray method newly developed in our lab are thoroughly
described to aide researchers in generating 60 K transcriptomic
data from indica or japonica subpopulation of rice.

2 Materials

2.1 Kits, Chemicals, 1. RNA Extraction Buffer (100 mM Tris–HCl pH 8.0, 150 mM
and Reagents LiCl, 50 mM EDTA, 1.5% SDS).
2. β-Mercaptoethanol (freshly add 300 μL to 20 mL RNA
Extraction Buffer immediately before use).
3. Phenol:chloroform:isoamyl (25:24:1).
4. Isopropanol.
5. Absolute ethanol.
6. 1.2 M NaCl.
7. Nuclease-Free Water.
8. Qiagen RNeasy Plant Mini Kit (50).
9. Qiagen RNase-Free DNase Set (50).
10. Qiagen RNeasy Mini Kit (50).
11. Agilent Low Input Quick Amp Labeling Kit, one-color.
12. Agilent One-Color RNA Spike-In Kit.
13. Agilent Hi-RPM Gene Expression Hybridization Kit, Large
Volume.
14. Agilent Gene Expression Wash Buffer Kit.
280 Mandy Püffeld et al.

2.2 Disposable and 1. RNA-free pipette filter tips (maximum capacity 2, 10, 20, 100,
Consumable Items 200, and 1000 μL volumes).
2. RNAse-free microfuge tubes (1.5 mL and 2.0 mL capacity).
3. Crushed ice.
4. Agilent gasket slides.
5. Agilent ozone barrier slides.
6. Agilent glass trays.
7. Agilent slide holder.

2.3 Instruments 1. Pipettor set (for 2, 10, 20, 100, 200, and 1000 μL tips).
2. Dewar flask for liquid nitrogen.
3. Liquid nitrogen storage tank.
4. Vortex mixer.
5. Crushed ice maker.
6. Heat blocks for 1.5 mL microfuge tubes (set at 50 °C).
7. Heat block for 2.0 mL microfuge tubes.
8. Ultralow temperature freezer (set at −80 °C).
9. Refrigerated microcentrifuge with rotor for 24 microcentri-
fuges (set at 4 °C).
10. Agilent DNA Microarray Scanner.
11. Hybridization Oven.
12. Mini-incubator.

3 Methods

This method describes in detail the optimized steps for One-Color

Microarray-Based Gene Expression Analysis using the Low Input
Quick Amp Labeling Protocol from Agilent. The protocol is con-
veniently subdivided into four sections: (1) RNA extraction, (2)
cDNA synthesis and cRNA labeling, (3) microarray hybridization,
and (4) microarray reading and data quality validation. Each sec-
tion can be conveniently done within 1 day but it is suggested that
all samples should all go through the same process before moving
on to the next section.

3.1 RNA Extraction This newly developed RNA extraction procedure takes into account
that developing rice grains have significantly elevated starch con-
tent that can reduce RNA quality and quantity. Interference due to
high amounts of starch is addressed in the first part of the protocol
by selective fractionation of starch pellets while ensuring optimal
solubility of RNA in the aqueous phase. Cryogenic grinding is the
most time-consuming part of this phase. It is suggested that all
 Rice Grain Microarray-Based Transcriptomics 281

samples be ground before moving on to RNA extraction proper. A

total of 24 ground samples can be conveniently subjected to RNA
extraction per day. As this step involves the use of harmful organic
solvents, all steps should be conducted inside the fume hood.
1. Grind developing rice grains to fine powder under strictly
cryogenic conditions (see Note 1). Collect the powdered sam-
ples using stainless steel spatula scoops precooled by dipping
in liquid nitrogen. Place samples in precooled 2.0 mL
microfuge tubes. Place tubes with samples in liquid nitrogen
and store immediately in ultralow temperature freezer at
−80 °C until use.
2. On the day of RNA extraction, place 750 μL RNA Extraction
Buffer and 500 μL phenol:chloroform:isoamyl (25:24:1) in
2 mL RNAse-free microfuge tubes. Vortex mix briefly.
3. Using a small metal spatula scoops previously dipped in liquid
nitrogen, add approximately 200 mg of powdered rice grain
sample in each tube. Immediately mix by vortexing. Keep on
ice until all samples in 2.0 mL microfuge tubes are thoroughly
resuspended in the extraction buffer.
4. Mix all samples simultaneously in a multi-tube vortex mixer at
room temperature for 5 min at 13,394 × g. Keep all tubes on
ice for 2 min.
5. Separate the supernatant by centrifugation at 14,000 × g for
10 min. Do all centrifugation steps for this section at 4 °C.
6. While spinning the samples by centrifugation, put 700 μL iso-
propanol and 400 μL 1.2 M NaCl in a fresh 2 mL microfuge
tube. The number of tubes to be prepared will depend on the
number of samples being centrifuged.
7. After centrifugation, collect and transfer supernatant (approxi-
mately 800 μL) into new 2.0 mL microfuge tube containing
isopropanol and NaCl solution.
8. Mix by gentle inversion 5 times. Incubate in ultralow freezer
(−80 °C) for 1 h or in regular freezer (−20 °C) overnight (see
Notes 2 and 3).
9. Pellet the RNA by centrifugation at 18,800 × g for 15 min.
10. While spinning, prepare RLT buffer (100 μL β-mercaptoethanol
in 10 mL RLT buffer) and DNAse 1 solution (50 μL frozen
DNAse stock +350 μL RDD using the Qiagen DNAse set).
11. Discard the supernatant and dissolve the pellet in freshly pre-
pared 450 μL RLT buffer. Dissolve pellet in freshly prepared
450 μL RLT buffer by simultaneous mixing in the multi-tube
vortexer at room temperature (see Note 4).
12. Transfer the solution in QIAshredder mini spin column (pur-
ple) with 2 mL collection tube. Clean the lysate by passing
282 Mandy Püffeld et al.

through the purple spin column by centrifugation for 1 min at

9000 × g.
13. Discard the purple spin column and add 0.5 volume of abso-
lute ethanol (225 μL) to the cleared lysate in the 2 mL collec-
tion tube and mix by pipetting up and down several times.
14. Transfer the solution immediately to a new RNeasy mini spin
column (pink) with 2 mL collection tube (see Note 5). Collect
the RNA in the pink column with RNAeasy silica-gel mem-
brane by centrifugation at 9000 × g for 15 s. Discard
flow-through.
15. Wash the RNA sample in the membrane with 350 μL buffer
RW1 by centrifugation at 9000 × g for 15 s.
16. Directly add 80 μL diluted DNAse 1 onto the membrane and
incubate on the benchtop for at least 15 min.
17. Remove the DNAse 1 and clean the RNA extract further by a
series of washing steps. Add 350 μL buffer RW1 and centri-
fuge at 9000 × g for 15 s. Discard flow-through. Repeat the
procedure but this time, use 500 μL buffer RPE. Wash again
with the same amount of buffer RPE and centrifuge for 2 min.
18. Dry the membrane by transferring to a new 2 mL collection
tube and spinning at 9000 × g.
19. Transfer the pink spin column to a new 1.5 mL collection
tube. Add 50 μL RNAse-free milliQ water pre-warmed at
50 °C. Incubate for 3 min at 50 °C heat block.
20. Elute the RNA samples by centrifugation at 9000 × g for
1 min. Immediately place the samples on ice.
21. Determine the concentration by measuring the absorbance of
each sample using NanoDrop. Dilute the sample accordingly
(usually 1:10) and determine the quality using BioAnalyzer.
RNA extracts with RIN value of 8.0 or higher are amenable
for microarray experiment.

3.2 cDNA Synthesis One staff can conveniently process a maximum of 24 samples per
Followed by cRNA day for this section of the experiment. The steps below are designed
Amplification, to be conducted continuously within one whole day but optional
Labeling stop points are identified whenever possible. Three heat blocks (or
and Purification small water baths) are needed for this part of the protocol. Before
commencing the protocol below, turn on each heat block at 80 °C,
65 °C and 37 °C. The 80 °C heat block should be dedicated to
2.0 mL microfuges while the other two heat blocks should
­accommodate 1.5 mL microfuges. The 37 °C heat block is set to
40 °C instead if the Spike Mix will not be prepared in step 2. The
initial temperature settings will be changed later based on the steps
below.
 Rice Grain Microarray-Based Transcriptomics 283

Table 1
Sample spreadsheet for dilution of samples for cDNA synthesis

Stock Concentration of Volume of DEPC-water Total

RNA (ng/ RIN First first dilution(ng/ RNA needed needed for volume
Sample ID μL) value dilution μL) for 50 ng (μL) 50 ng (μL) (μL)
RNA_001 1596.4 8.3 1:15 56.8 0.9 0.6 1.5
RNA_002 1709.8 8.2 1:17 57.0 0.9 0.6 1.5
RNA_003 1134.7 8.2 1:11 58.9 0.8 0.7 1.5

Prepare one-color spike mix, T7 promoter primer mix, and

cDNA master mix based on Agilent protocol. The prepared mix
will depend on the amount of starting material, which can range
from 10 to 200 ng total RNA or 5 ng PolyA RNA. In our labora-
tory, we routinely use 50 ng total RNA as starting material for
microarray hybridization. This method is described below. To pre-
pare master mixes for 24 samples, add 2 extra reactions to com-
pensate for pipetting variations.
1. Dilute each total RNA sample to 50 ng in a 1.5 mL microfuge
tube to make a final volume of 1.5 μL. A preliminary 100-fold
dilution of the original total RNA extract is usually needed to
fit within the 50–100 ng concentration range. Measure the
actual concentration of the First Dilution using Nanodrop.
Perform a second dilution of each sample using the First
Dilution to make 1.5 μL 50 ng RNA sample. To facilitate
tracking of sample and ease of dilution, the sample spreadsheet
in Table 1 can be used. Keep all diluted samples on ice.
2. Pre-warm the Spike Mix stock at 37 °C in a 2 mL heat block
for 5 min prior to serial dilution. Change the temperature to
80 °C after use. The actual serial dilution is done in 1.5 mL
microfuge tubes using the Dilution Buffer from the Agilent
Spike-In kit as described in Table 2. After each dilution, the
diluted spike mix is resuspended thoroughly by vortex mixing.
Store the First and Second Dilution of the Agilent One-Color
Spike mix positive controls at −80 °C, taking note of date
when they were prepared. They can only go through freeze/
thaw cycles up to eight times. On the other hand, the Third
and Fourth Dilution should be prepared fresh daily and stored
on ice prior to use.
3. Prepare the T7 promoter mix in a 1.5 mL microfuge tube.
The volume needed depends on the number of samples for
labeling. The volume required for 24 samples is summarized
in Table 3. Store the mix on ice prior to use.
284 Mandy Püffeld et al.

Table 2
Serial dilution of Agilent One-Color Spike Mix for Cyanine-3 labeling using
50 ng total RNA as starting material

Dilution
Dilution buffer (μL) Amount (μL) Stability
First dilution (1:20) 38 2 μL spike mix stock 2 months
Second dilution 48 2 μL first dilution 1 week
(1:25)
Third dilution (1:20) 38 2 μL second dilution Prepare fresh
Fourth dilution (1:2) 30 30 μL third dilution Prepare fresh

Note that dilutions vary when different RNA amount is used as starting material, please
refer to the Low Input Quick Amp Labeling Kit manual for details (Agilent)

Table 3
T7 Promoter Primer mix for 24 labeling reactions

Volume Per Volume Per 24 + 2

Component Reaction (μL) Reactions (μL)
T7 promoter primer (green cap) 0.8 20.8
Nuclease-free water (white cap) 1.0 26.0
Total volume 1.8 46.8

4. Prepare the primer and the template by adding each of the

components in Table 1 to Table 3, using the microfuge tube
in Table 1 containing 1.5 μL of 50 ng RNA sample. Mix the
components as specified in Table 4 by pipetting several times.
Denature the mixture by incubating at 65 °C heat block for
10 min.
5. While denaturing, pre-warm the 5× First Strand Buffer at the
first heat block at 80 °C for at least 4 min, in preparation for
step 7.
6. Immediately place the denatured primer-template mix from
step 4 on ice for 5 min.
7. While incubating on ice in step 6, quickly prepare the cDNA
master mix as detailed in Table 5. Mix each component by
gentle pipetting and keep the master mix at room temperature
prior to use. However, the AffinityScript RNAse Block Mix
should be kept in the freezer and immediately placed back
after use.
 Rice Grain Microarray-Based Transcriptomics 285

Table 4
Preparation of cDNA synthesis reaction mix

Volume Per
Components Reaction (μL)
50 ng total RNA (from Table 1) 1.5
Fourth dilution of spike mix (from Table 2) 2.0
T7 primer mix (from Table 3) 1.8
Total mix per tube 5.3

Table 5
Components of cDNA master mix

Volume Per Volume Per 24 + 2

Components Reaction (μL) Reactions (μL)
Pre-warmed 5× First Strand Buffer 2.0 52.0
(green cap)
0.1 M DTT (white cap) 1.0 26.0
10 mM dNTP mix (green cap) 0.5 13.0
AffinityScript RNAse Block Mix 1.2 31.2
(violet cap)
Total volume 4.7 122.2

8. Remove the samples from ice (step 6) and centrifuge the tubes
briefly at room temperature to spin down the contents to the
bottom in preparation for the next step.
9. Add 4.7 μL of the cDNA Master Mix from Table 5 to each
tube and mix gently by pipetting.
10. Synthesize cDNA by incubating the samples for 2 h at 40 °C
heat block. During the incubation, the 65 °C heat block can
be changed to 70 °C in preparation for the heat inactivation
(step 11).
11. Inactivate the AffinityScript enzyme by incubating at 70 °C for
15 min (Optional stopping point: The samples can be stored
at −80 °C overnight. Continue the next day after heat
inactivation).
12. Immediately transfer the tubes on ice and incubate for 5 min.
13. While incubating on ice, prepare Transcription Master Mix as
detailed in Table 6. Just keep the mix at room temperature
prior to dispensing. However, the T7 RNA polymerase blend
should be kept in the freezer and immediately placed back
after use. As the last component is light sensitive, mixing and
286 Mandy Püffeld et al.

Table 6
Components of Transcription Master Mix

Volume
Volume Per 24 + 2
Component Per Reaction (μL) Reactions (μL)
Nuclease-free water (white cap) 0.75 19.5
5× Transcription Buffer (blue cap) 3.2 83.2
0.1 M DTT (white cap) 0.6 15.6
NTP mix (blue cap) 0.1 26.0
T7 RNA Polymerase Blend (red cap) 0.21 5.5
Cyanine 3-CTP (Cy3) 0.24 6.2
Total volume 6.0 156.0

dispensing of the mix (step 15) should be done in low light

conditions. The light directly above the work bench should be
turned off. Instead, indirect lights from nearby lab benches or
windows can be used.
14. Centrifuge the samples from step 12 briefly in a microcentri-
fuge to drive down tube contents to the bottom.
15. Add 6 μL Transcription Master Mix to each sample tube, gen-
tly mixing by pipetting. The total volume of the tube is now
16 μL.
16. Transcribe and label the cDNA to produce Cy3-labeled cRNA
by incubating samples at 40 °C for 2 h. (Optional stopping
point: The samples can be stored at −80 °C after
transcription).
17. In preparation for the next step, convert the 70 °C heat block
to 55 °C and add nuclease-free water in preparation for the
elution of purified cRNA (step 27).
18. After transcription and labeling (step 16), purify the labeled
cRNA using a modified method for Qiagen RNeasy Mini Kit
as detailed below (step 19–28).
19. Adjust the volume to 100 μL of each sample by adding 84 μL
nuclease-free water.
20. Add 350 μL Buffer RLT.
21. Add 250 μL absolute ethanol.
22. Mix thoroughly by pipetting. Transfer 700 μL of each cRNA
sample to an RNAeasy mini column with 2 mL collection
tube.
 Rice Grain Microarray-Based Transcriptomics 287

23. Centrifuge samples at 4 °C for 30 s at 7,534 × g. Discard

flow-through.
24. Wash the samples by adding 500 μL Buffer RPE to each col-
umn (do not forget to add ethanol to this buffer prior to first
use). Centrifuge samples at 4 °C for 30 s at 7,534 × g. Discard
flow-through. Repeat this step once.
25. Transfer each RNAeasy column to a new collection tube.
Remove traces of RPE buffer and dry the samples by centrifu-
gation at 4 °C for 30 s at 7,534 × g.
26. Discard the collection tube and transfer the mini spin column
into a new 1.5 mL microfuge tube provided in the kit.
27. Elute the cRNA sample by adding 30 μL RNAse-free water
pre-warmed at 55 °C (step 17) directly onto the RNAeasy
filter membrane. Incubate for 60 s at the 55 °C heat block.
Centrifuge at room temperature for 30 s at 7,534 × g.
28. Remove the RNAeasy column and close each microfuge tube.
Immediately place all tubes on ice (Optional stopping point:
The samples can be stored at −80 °C after elution).
29. Quantify the cRNA using the microarray measurement feature
of Nanodrop based on the manufacturer’s method. The sample
type should be set to RNA-40. The following values are needed:
Cyanine 3 dye concentration (pmol/μL), RNA absorbance
ratio (260 nm/280 nm), cRNA concentration (ng/μL).
30. Immediately store the cRNA samples at −80 °C after Nanodrop
readings.
31. Compute for the cRNA yield using the following formula:

( cRNA concentration in ng / µ L ) × (30 µ L elution volume )

cRNA yield ( µ g ) =
1000

32. Determine the specific activity using the following formula:

concentration of Cy3 ( pmol / µ L )

Specific activity ( pmol Cy3 / µ g cRNA ) = × 1000
concentration of cRNA ( ng / µ L )

33. The computations in steps 31 and 32 can easily be done using

a spreadsheet. The recommended yield for 8-pack microarray
format is at least 0.825 μg, a 260/280 ratio of greater than 2.0
and a specific activity of 6 pmol Cy3/μg cRNA. Consult the
Agilent manual for the recommended yield and specific activ-
ity values for the other formats. A summary spreadsheet should
be prepared similar to Table 7.
288 Mandy Püffeld et al.

Table 7
Summary spreadsheet for cRNA quantification

Specific activity
Sample ID Cy3 (pmols/μL) 260/280 ratio cRNA (ng/μL) cRNA yield (μg) (pmols Cy3/μg cRNA)
RNA_001 0.95 2.18 73.49 2.20 12.9
RNA_002 1.20 2.22 82.94 2.49 14.5
RNA_003 0.81 2.19 68.79 2.06 11.8

3.3 Microarray A new custom 8x60K microarray slide for japonica (Order Code:
Hybridization 054269) and indica (Order Code: 054270) rice transcriptomes
and Scanning were designed. These are available for purchase from Agilent. The
microarray hybridization protocol can be started in the afternoon.
It will only take around 3 to 4 h to complete the hybridization of
24 samples. Up to 8 slides (64 samples) can be processed in the
washing step but one staff can optimally process only up to 4 slides
(32 samples) per day. The following morning is devoted to wash-
ing of and scanning of the microarray slides. Additional microarray
hybridizations in the afternoon of the second day can be started.
The process is repeated until all samples are hybridized and
scanned. Use color-free latex gloves to avoid contaminating the
hybridization mix and the microarray slides with colored pigments
that can interfere with the signals during the microarray scanning.
1. Prior to the actual hybridization, prepare the 10× Blocking
Agent based on the manufacturer’s instructions. The large
volume 10× Blocking Agent is resuspended in 1250 μL
nuclease-­free water. Heating the mixture for up to 5 min at
37 °C usually helps in pellet dissolution. The solution is gently
vortexed until the pellet is dissolved. Centrifuge the solution
for 5 s to drive down everything to the bottom. A 200 μL
aliquot of the 10× Blocking Agent are dispensed in 1.5 mL
microfuge tubes and stored at −20 °C freezer. Each aliquot is
stable for up to 2 months and is good for up to 40
hybridizations.
2. During the actual day for hybridization, get one tube aliquot
of 10× Blocking Agent and thaw on ice. Turn on the hybrid-
ization oven and set the temperature at 65 °C. Also equilibrate
a heat block (or a small water bath) to 60 °C for cRNA frag-
mentation. If the lab is equipped with an air purification sys-
tem like IQAir, one can also turn this on prior to starting the
hybridization protocol.
3. Prepare the fragmentation mix based on Agilent protocol. The
volume of the components will depend on the microarray for-
mat being used. For example, one can routinely hybridize
600 ng of linearly amplified Cy3-labeled cRNA to an 8-pack
 Rice Grain Microarray-Based Transcriptomics 289

Table 8
Fragmentation mix for 8-pack microarray format using 600 ng of cRNA

Volume/mass 8-pack
Components microarrays
Cyanine 3-labeled, linearly amplified cRNA 600 ng
10× Blocking Agent 5 μL
Nuclease-free water Bring volume to 24 μL
25× Fragmentation Buffer 1 μL
Total volume 25 μL

microarray format. The components for this format are enu-

merated in Table 8. Each fragmentation mix is prepared indi-
vidually in 1.5 mL microfuge tubes on ice. When all of the
components for each sample are dispensed by pipetting, each
fragmentation reaction is then mixed gently and thoroughly
using a vortex mixer. Spin the samples briefly in the
microcentrifuge.
4. Incubate the samples for exactly 30 min in the pre-­equilibrated
60 °C heatblock.
5. Immediately stop the fragmentation reaction by cooling on ice
for 1 min and proceeding to the next step.
6. Add 2× GEx Hybridization Buffer HI-RPM to each of the
tube based on the manufacturer’s instructions. The volume
depends on the microarray format. For 8-pack microarray for-
mat, add 25 μL of Hybridization Buffer to stop the fragmen-
tation reaction. Mix well by gently pipetting up and down. Be
careful not to introduce bubbles during mixing.
7. Centrifuge the tubes for 1 min at room temperature at
15,720 × g to drive all solution to the bottom and to remove
any bubbles formed during pipetting.
8. Immediately place all samples on ice and load onto the micro-
array slide as soon as possible.
9. Prepare the hybridization assembly. Details of the procedure
can be found on Agilent Microarray Chamber User Guide
enclosed in the Agilent Microarray Chamber Kit and available
for download at the Agilent website. Open the packet and
load a clean gasket slide into the Agilent SureHyb chamber
base (Fig. 1a). Ensure that the label is facing up and aligned
with the rectangular section of the chamber base. Also see to
it that the gasket slide is flushed with the chamber base and is
not misaligned.
Fig. 1 Setting up overnight hybridization: (a) setup for the Agilent SureHyb chamber, (b) dispensing of samples
in the middle of each gasket well, (c) simplified technique of placing the microarray slide on top of the gasket
slide, (d) securing the SureHyb chamber assembly, (e) ensuring sample distribution and mobility along each
hybridization chamber, (f) securing the SureHyb assembly in the hybridization oven, and (g) slide wash setup
before leaving the lab for overnight hybridization
 Rice Grain Microarray-Based Transcriptomics 291

10. Pipette out hybridization mix and dispense slowly in the mid-
dle of each gasket well (Fig. 1b). The volume needed to
hybridize will depend on the microarray format. Usually,
44 μL for the 8-pack format is required. Do not allow the
pipette tip or the hybridization mix to touch the gasket walls
to avoid leakage. Dispense until most of the mix is placed onto
the slide but do not press the pipette until the second stop to
avoid introducing bubbles. Add 44 μL of 1× Hybridization
Buffer as blank for unused wells.
11. Place a microarray “active side” down onto the SureHyb gas-
ket slide so that the “Agilent”-labeled barcode is facing down
and the numeric barcode is facing up. Make sure that the
microarray slide is parallel to the gasket slide and both are
properly aligned. Be sure not to haphazardly drop the array
slide onto the gasket to prevent the samples from mixing
between gasket wells. To ensure that this crucial step is done
seamlessly, two extra top covers of a SureHyb chamber can be
used by placing bottom up on the left and on the right slots of
the SureHyb chamber base in use. The microarray slide is
placed on top, with the two extra top covers preventing the
two slides from touching each other. Slowly but firmly pull
out the two extra top covers on opposite direction parallel to
the lab bench to place and properly align the microarray slide
onto the gasket slide (Fig. 1c). Practice this step with used
microarray slide to gain confidence before actually performing
this critical step.
12. Immediately place the SureHyb chamber cover onto the sand-
wiched glass slides. Slide the clamp assembly. Hand-tighten
the clamp onto the chamber by turning the thumbscrew
clockwise to secure the assembly (Fig. 1d).
13. Rotate the hybridization chamber assembly to distribute the
samples evenly and assess the mobility of the big bubble
formed (Fig. 1e). Stationary bubbles can be mobilized by tap-
ping the assembly onto the lab bench surface.
14. Place the hybridization chamber assembly in the rotisserie in a
hybridization oven pre-equilibrated at 65 °C (Fig. 1f). When
using the 2× GEx Hybridization Buffer, set the rotator speed
at 10 rpm. Hybridize at 65 °C for 17 h. Ensure proper balance
of weight across the rotisserie by adding empty hybridization
chamber assembly whenever necessary.
15. Before leaving the lab for the overnight 17-h hybridization,
prepare Gene Expression Wash Buffers 1 and 2 based on the
manufacturer’s protocols. Also prepare three wash chamber
assemblies as detailed in steps 16 and 17. The addition of
0.005% Triton X-102 to both buffers is purely optional but
highly recommended to reduce the incidence of microarray
292 Mandy Püffeld et al.

wash artifacts. Open the cubitainer cardboard box and remove

the outer and inner caps. Directly add 2 mL 10% Triton X-102
to the buffer and replace the original inner and outer caps.
Thoroughly mix the buffer by inverting and shaking the con-
tainer 5 to 6 times. Discard the outer and inner caps and install
the spigot provided with the wash buffer. Do this for both
Wash Buffers 1 and 2. Label each cubitainer as “Wash Buffer
1” and “Wash Buffer 2” accordingly and indicate the date
when the Triton X-102 was added.
16. Prepare two wash chamber assemblies. Each assembly should
be washed thoroughly with copious amounts of Milli-Q water
at least five times and then dried inverted in paper towel before
use. Dish 1 should be empty, without any microarray slide
rack. This will be filled with Wash Buffer 1 which will be used
to disassemble the sandwiched microarray slides (step 22).
Dish 2 should have microarray slide rack and a small magnetic
stirrer inside. It should be labeled with “Wash Buffer 1” as it
will be used to wash the microarray slides with the first buffer.
Also keep a rack holder nearby as it will be used to transfer the
rack from Wash Buffer 1 to Wash Buffer 2. Keep these two
wash chamber assemblies in laboratory bench.
17. Pre-warm a mini incubator to 37 °C and place a third wash
chamber assembly. Dish 3 should be empty like the first wash
chamber but should contain a small magnetic stirrer like Dish
2. Label this as “Wash Buffer 2” as it will be used to wash the
slides using the second buffer.
18. Obtain 500 mL of Wash Buffer 1 and Wash Buffer 2 directly
from the spigot and place individually in two sterile 1 L reagent
bottles. Leave Wash Buffer 1 at room temperature but place
Wash Buffer 2 in a 37 °C water bath overnight. Prepare an
extra bottle of Wash Buffer 1 and leave at room temperature
similar to the other Wash Buffer 1. Label it as “Wash Buffer 1
Reuse” as this will be used later as a reagent bottle to save the
Wash Buffer 1 that was used in the first wash chamber (steps
22–24). The final setup from steps 15–18 is illustrated in
Fig. 1g. This washing setup should be accomplished before
leaving the lab in preparation for the next day to ensure that
no bubbles are introduced in the wash buffers.
19. The following day, prepare the wash buffers. The summary of
wash conditions is listed in Table 9. Fill Dish 1 to three-fourths
volume with Gene Expression Buffer 1. Similarly fill Dish 2
with Gene Expression Buffer 1 enough to submerge the metal
slide rack. Dish 1 will be used to soak the microarray slide
sandwich while disassembling (step 22). Wash Buffer 1 for
hybridization chamber disassembly should be used for maxi-
mum 10–15 slides. Each microarray slide will then be trans-
 Rice Grain Microarray-Based Transcriptomics 293

Table 9
Summary of wash conditions for microarray slides

Steps Dish Wash buffer Temperature Time

Disassembly 1 GE wash buffer 1 Room temperature As fast as possible
First wash 2 GE wash buffer 1 Room temperature 1 min
Second wash 3 GE wash buffer 2 37 °C 1 min

ferred to Dish 2 and placed on the microarray slide rack as

detailed in the next step. Always use fresh Wash Buffer 1 and
Wash Buffer 2 for each batch of washing, for up to 4–5 slides.
20. After the 17 h hybridization at 65 °C (step 14), get one
hybridization chamber from the incubator. Optionally, the
start time can be recorded but as experience is gained, one can
just conduct the succeeding steps very quickly. Small bubbles
formed during hybridization can also be noted and whether
they are all rotating freely or not.
21. Disassemble the hybridization assembly on a flat surface
(Fig. 2a). This is usually done on a lab bench with lint-free
paper underneath. Loosen the thumbscrew by turning coun-
ter clockwise. Slide off the clamp and remove the chamber
cover. Remove the array-gasket sandwich from the chamber
base, ensuring that the two glass slides are still firmly adhering
to each other.
22. Transfer the sandwich to Dish 1. Keep the microarray numeric
barcode facing up as the slide sandwich is submerged onto
Wash Buffer 1 in a slanted position, ensuring that the barcode
label is not submerged in the buffer to avoid contaminating
the buffer with ink from the barcode label. Handle the slides
from their ends only and avoid touching especially the active
side of the microarray slide.
23. While maintaining the sandwich in completely submerged
position in Wash Buffer 1, separate the two glass slides from
the barcode end by inserting one of the blunt ends of a plastic
forceps between the two slides (Fig. 2b). Gently twist the for-
ceps upward or downward to separate the slides. Allow the
gasket slide to drop to the bottom of Dish 1 while ensuring
that the microarray slide is held firmly for the next step.
24. Gently lift the microarray slide side sideways and immediately
transfer the glass slide from Dish 1 to into the rack in Dish 2
containing Wash Buffer 1 (Fig. 2c). As one does this crucial
step, minimize exposure of the slide to air. Avoid touching the
active side by ensuring that only the barcode portion of the
microarray slide and its edges are touched. This step can be
Fig. 2 Washing of microarray slides: (a) disassembling hybridization chamber on a flat surface, (b) dislodging
of the microarray slide from the gasket slide while immersed in Wash Buffer 1, (c) transferring of the microar-
ray slide into the metal rack containing Wash Buffer 1, (d) distributing the microarray slides evenly along the
metal rack, (e) washing of the microarray slides in Wash Buffer 1, (f) side-by-side washing setup for Wash
Buffer 1 and Wash Buffer 2, and (g) transferring of rack from Wash Buffer 1 to (h) Wash Buffer 2
 Rice Grain Microarray-Based Transcriptomics 295

practiced using the used gasket and microarray slides to gain

confidence before actually performing the procedure.
25. Repeat steps 21 through to 24 for up to seven additional
slides to ensure uniformity of washing. Distribute the microar-
ray slides evenly along the rack (Fig. 2d). Each staff can opti-
mally do these steps unassisted for up to 4 slides.
26. Once all the slides are placed in the rack, attach the rack holder
and transfer the entire setup to the first magnetic stirrer. Stir
gently using setting 4 for exactly 1 min (Fig. 2e).
27. While stirring for 1 min, remove Dish 3 from the mini 37 °C
incubator and place on top of the second magnetic stirrer.
Also remove Wash Buffer 2 from the 37 °C water bath and
gently pour the buffer in Dish 3 placed on the top of the sec-
ond magnetic stirrer. Ensure that no bubbles are formed while
pouring. The two magnetic stirrers should be positioned side
by side, in close proximity with each other to ensure faster
transfer of slide rack from Dish 2 to Dish 3 (Fig. 2f).
28. Once the 1 min stirring is finished, gently and very slowly lift
the slide rack from Dish 2 and transfer it gently and slowly to
Dish 3 (Fig. 2g, h). Remove the rack holder and stir using set-
ting 4 for exactly 1 min.
29. Slowly remove the slide rack from Dish 3 ensuring that the
formation of buffer droplets is minimized (Fig. 3a). This step
should be done slowly and should take 5–10 s. Dry the sides
of each slide by touching gently in lint-free paper (Fig. 3b).
Place the slide in a used slide box and do the same for the
remaining slides. Dry for approximately 15 min in the dark
(Stopping point: short break).
30. Put each microarray slide in a slide holder with the Agilent
barcode facing up (Fig. 3c). The ozone levels inside the labo-
ratory should be 50 ppb (100 μg/m3) or less. In our labora-
tory, an Agilent Ozone Barrier Slide Cover is routinely used to
prevent ozone-induced degradation of cyanine dyes (Fig. 3d).
Place the Ozone Barrier Slide Cover on the top of the array.
31. Once the microarray slide is securely in place inside a slide
holder with ozone barrier, scan the slides immediately to mini-
mize the possibility of reduction of signal intensities due to
environmental oxidation. A SureScan or Agilent C microarray
scanner is needed for SurePrint G3 format microarray slides. A
feature extraction compatibility matrix for non-Agilent scan-
ners is available online.
32. Place the assembled slide holders into the scanner carousel.
Load the samples in sequence, starting from slot 1 onward
(Fig. 3e).
296 Mandy Püffeld et al.

Fig. 3 Scanning microarray slides: (a) preparing microarray slide for microarray scanning, (b) drying of micro-
array slide, (c) placing microarray slide in slide holder, (d) placing of ozone barrier slide on top of the microarray
slide, and (e) loading of the microarray slide assembly in the microarray scanner

33. Open the Scan Control software. In the main window, choose
the slot number of the first slide for Start Slot and the slot
number for the last slide as the End Slot.
34. Read the slides with the prescribed scan settings based on the
manufacturer’s specifications. The scan profile depends on the
manufacturer’s specifications. In our laboratory, we use the
 Rice Grain Microarray-Based Transcriptomics 297

Table 10
Recommended scan settings

Settings for 8x60K G3 microarray

Parameters format
Profile AgilentG3_GX_1Color
Dye channel Green
Scan region Scan area (61 × 21.6 mm)
Scan resolution (μm) 3 μm double pass
Tiff size 20 bit

profile AgilentG3_GX_1Color which can be downloaded in

the Agilent website. The scan settings are summarized in
Table 10.
35. Scan the samples when the scanner status in the main window
says Scanner Ready. Click Start Scan to commence the scan-
ning process. This procedure usually takes 30 min per slide.
36. After scanning, disassemble the glass slides and store them in
the orange slide boxes in a nitrogen purge box. Alternatively,
they can be stored sealed in aluminum foil under vacuum and
protected from light.

3.4 Microarray Once all the samples are scanned, the process of feature extraction
Reading, Feature can commence. During this step, the information from the probe
Extraction, and QC features is extracted from microarray scan data to facilitate the
Validation measurement of gene expression in the experiment. The software
can read and processed up to 100 raw microarray image files in
TIFF format. It can automatically find and place microarray grids,
reject outlier pixels, accurately determine feature intensities and
ratios, flag outlier pixels, and calculate statistical confidences. This
software is available from the Agilent website.
1. Open the Agilent Feature Extraction (FE) Program from the
PC by double-clicking on the icon. All Feature Extraction
Software from version 10.7 and above can automatically
­
download Grid Templates, protocols and QC metrics. This
can be set up under Advanced Options and clicking Use eAr-
ray server during extraction.
2. Add the .tif images to be extracted to the FE Project. This is
done either by clicking Add New Extraction Set(s) icon
found on the toolbar or by right-clicking the Project Explorer
and selecting Add Extraction. The default grid template and
the protocol are usually automatically assigned by the FE pro-
gram for each extraction set.
298 Mandy Püffeld et al.

3. Set the FE Project properties. The Operator can be set in the

General section. Verify that the default settings in the Input
section are specified. MAGE and JPEG can be selected if the
data will be imported into Rosetta Resolver. Project Properties
for MAGE and JPEG can be set as local file only. The
Overwrite Previous Results feature can be set to False to pre-
vent accidental loss of files. The customization of other optional
settings can be accomplished by consulting the user manual.
4. Check the Extraction Set Configuration tab. Verify that the
program assigned the correct grid template from the Grid
Name column. A different grid template can be assigned to an
extraction set by selecting from the pull down menu. To
update a grid template to the latest version, right-click Grid
Template Browser and select online update. Grid templates
will not be automatically available for each new microarray
slide design but the information can be downloaded from the
Agilent website. Go to Grid Template Browser and select
Add and specify the grid template code. The design file pro-
vided in .xml format can also be browsed. Click Open to load
grid template into the FE database. Every time a new grid
template is added, specify the default protocol to ensure that
the FE software will be able to automatically assign an FE pro-
tocol to an extraction set.
5. In addition, verify that the correct protocol is assigned to each
extraction set under the Protocol Name column. A different
protocol can be specified from the pull down menu. The cor-
rect protocol always begins with “GE1” for one-color analysis.
FE Protocol Browser can also import a protocol if it is not
available. This is done by browsing for an appropriate FE pro-
tocol in .xml format and loading the protocol into the FE
database by clicking Open.
6. Save the FE Project in .fep format by clicking File > Save As
and specifying the correct file directory position. Ensure that
the icons for the image files in the FE Project Window has no
more red “X” which indicates that an extraction protocol was
not selected. Click Project > Start Extracting to commence
the feature extraction procedure. This takes around 30 min
per microarray slide to finish.
7. After feature extraction, QC report for each microarray slide
extraction is available for viewing in the Summary Report
tab. The proper placement of the grid can be inspected using
Spot Finding at the four corners of the array. If a QC Metric
Set has been assigned to the FE Project, the QC results can be
viewed via (1) Project Run Summary, (2) QC Report, and (3)
QC Chart Tool. An application note for specific for this step
can be obtained from the Agilent website.
 Rice Grain Microarray-Based Transcriptomics 299

8. For downstream analysis of Agilent microarray data, use

GeneSpring GX 9.0 or later. To compare data across a set of
one-­ color microarray data, the 75th percentile scaling is
employed by the GeneSpring software for signal intensity nor-
malization. The detailed bioinformatic approaches involved in
normalization of microarray data, application of dimension
reduction techniques to detect outliers, implementation of
clustering methods to identify clusters to define coexpressed
genes, implementing statistical methods to identify differen-
tially expressed genes (DEGs), and annotating functional
pathway of DEGs were described earlier [22]. Hence, this will
not be covered as this is beyond the scope of this protocol.

4 Notes

1. In our laboratory, we use Retsch mill MM400 with cryo kit

because it can grind two samples at a time. We soak the 50 mL
stainless steel grinding jar (Retsch P/N 01.462.0216) and
25 MM grinding ball (05.368.0105) under liquid nitrogen
before placing the rice grain samples. Grinding is done for
1 min at speed of 25/s.
2. The experiment can be stopped at this point. This can be con-
tinued the following day.
3. The samples may freeze at this stage but one can proceed to
centrifugation in the next step without thawing. The samples
will thaw during the next step.
4. Some pellets might not dissolve completely within the 5 min
allocated time. Just extend the vortex mixing until the pellet is
dissolved.
5. Up to 750 μL clarified lysate can be transferred. Discard flow-­
through and repeat until all lysates are transferred.

Acknowledgments

This work has been supported under the CGIAR thematic area
Global Rice Agri-Food System CRP, RICE, Stress-Tolerant Rice
for Africa and South Asia (STRASA) Phase III funding.

References

1. Trapnell C, Roberts A, Goff L, Pertea G, Kim 2. Trapnell C, Pachter L, Salzberg SL (2009)

D, Kelley DR, Pimentel H, Salzberg SL, Rinn TopHat: discovering splice junctions with
JL, Pachter L (2012) Differential gene and RNA-Seq. Bioinformatics 25(9):1105–1111
transcript expression analysis of RNA-seq 3. Law CW, Chen YS, Shi W, Smyth GK (2014)
experiments with TopHat and cufflinks. Nat Voom: precision weights unlock linear model
Protoc 7(3):562–578
300 Mandy Püffeld et al.

analysis tools for RNA-seq read counts. 12. Zhang W, Ferguson J, Ng SM, Hui K, Goh G,
Genome Biol 15(2):R29 Lin AP, Esplugues E, Flavell RA, Abraham C,
4. Robertson G, Schein J, Chiu R, Corbett R, Zhao HY, Cho JH (2012) Effector CD4+T
Field M, Jackman SD, Mungall K, Lee S, cell expression signatures and immune-­
Okada HM, Qian JQ, Griffith M, Raymond A, mediated disease associated genes. PLoS One
Thiessen N, Cezard T, Butterfield YS, 7(6):e38510
Newsome R, Chan SK, She R, Varhol R, 13. Zhao SR, Fung-Leung WP, Bittner A, Ngo K,
Kamoh B, Prabhu AL, Tam A, Zhao YJ, Moore Liu XJ (2014) Comparison of RNA-Seq and
RA, Hirst M, Marra MA, Jones SJM, Hoodless microarray in Transcriptome profiling of acti-
PA, Birol I (2010) De novo assembly and anal- vated T cells. PLoS One 9(1):e78644
ysis of RNA-seq data. Nat Methods 14. Zhu T, Budworth P, Chen W, Provart N,
7(11):U909–U962 Chang H-S, Guimil S, Su W, Estes B, Zou G,
5. Trapnell C, Roberts A, Goff L, Pertea G, Kim Wang X (2003) Transcriptional control of
D, Kelley DR, Pimentel H, Salzberg SL, Rinn nutrient partitioning during rice grain filling.
JL, Pachter L (2014) Differential gene and Plant Biotechnol J 1(1):59–70
transcript expression analysis of RNA-seq 15. Yamakawa H, Hakata M (2010) Atlas of rice
experiments with TopHat and Cufflinks (vol 7, grain filling-related metabolism under high tem-
p 562, 2012). Nat Protoc 9(10):2513 perature: joint analysis of metabolome and tran-
6. Grabherr MG, Haas BJ, Yassour M, Levin JZ, scriptome demonstrated inhibition of starch
Thompson DA, Amit I, Adiconis X, Fan L, accumulation and induction of amino acid accu-
Raychowdhury R, Zeng QD, Chen ZH, mulation. Plant Cell Physiol 51(5):795–809.
Mauceli E, Hacohen N, Gnirke A, Rhind N, di https://doi.org/10.1093/pcp/pcq034
Palma F, Birren BW, Nusbaum C, Lindblad-­ 16. Liu X, Guo T, Wan X, Wang H, Zhu M, Li A,
Toh K, Friedman N, Regev A (2011) Full-­ Su N, Shen Y, Mao B, Zhai H, Mao L, Wan
length transcriptome assembly from RNA-Seq J (2010) Transcriptome analysis of grain-­filling
data without a reference genome. Nat caryopses reveals involvement of multiple regu-
Biotechnol 29(7):644–U130 latory pathways in chalky grain formation in
7. Bottomly D, Walter NAR, Hunter JE, Darakjian rice. BMC Genomics 11:730. https://doi.
P, Kawane S, Buck KJ, Searles RP, Mooney M, org/10.1186/1471-2164-11-730
McWeeney SK, Hitzemann R (2011) Evaluating 17. Deshmukh R, Singh A, Jain N, Anand S,
gene expression in C57BL/6J and DBA/2J Gacche R, Singh A, Gaikwad K, Sharma T,
mouse striatum using RNA-Seq and microar- Mohapatra T, Singh N (2010) Identification of
rays. PLoS One 6(3):e17820 candidate genes for grain number in rice
8. Fu X, Fu N, Guo S, Yan Z, Xu Y, Hu H, (Oryza sativa L.). Funct Integr Genomics
Menzel C, Chen W, Li YX, Zeng R, Khaitovich 10(3):339–347
P (2009) Estimating accuracy of RNA-Seq and 18. Li H, Deng HW (2010) Systems genetics, bio-
microarrays with proteomics. BMC Genomics informatics and eQTL mapping. Genetica
10:161 138(9–10):915–924
9. Marioni JC, Mason CE, Mane SM, Stephens 19. Jansen RC, Nap JP (2001) Genetical genom-
M, Gilad Y (2008) RNA-seq: an assessment of ics: the added value from segregation. Trends
technical reproducibility and comparison with Genet 17(7):388–391
gene expression arrays. Genome Res 18(9): 20. Jansen RC (2003) Studying complex biological
1509–1517 systems using multifactorial perturbation. Nat
10. Sirbu A, Kerr G, Crane M, Ruskin HJ (2012) Rev Genet 4(2):145–151
RNA-Seq vs dual- and single-channel microar- 21. Butardo VM Jr, Anacleto R, Parween S,
ray data: sensitivity analysis for differential Samson I, de Guzman K, Alhambra CM, Misra
expression and clustering. PLoS One G, Sreenivasulu N (2017) Systems genetics
7(12):e50986 identifies a novel regulatory domain of amylose
11. Xu X, Zhang YH, Williams J, Antoniou E, synthesis. Plant Physiol 173(1):887–906.
McCombie WR, Wu S, Zhu W, Davidson NO, https://doi.org/10.1104/pp.16.01248
Denoya P, Li E (2013) Parallel comparison of 22. Sreenivasulu N, Sunkar R, Wobus U, Strickert
Illumina RNA-Seq and Affymetrix microarray M (2010) Array platforms and bioinformatics
platforms on transcriptomic profiles generated tools for the analysis of plant transcriptome in
from 5-aza-deoxy-cytidine treated HT-29 response to abiotic stress. Methods Mol Biol
colon cancer cells and simulated datasets. Bmc 639:71–93
Bioinformatics 14:S1
 Chapter 17

Quantification of DNA Methylation as Biomarker

for Grain Quality
Christiane Seiler and Markus Kuhlmann

Abstract
DNA methylation is an important biomarker for gene activity. It contributes to gene silencing and is
involved in regulating various seed developmental processes in plants. Many of these processes are involved
in important traits associated with aspects of grain quality. A reliable, fast, and cheap method is the estima-
tion of DNA methylation utilizing methylation sensitive restriction enzymes (MSRE) and quantitative
real-time PCR (qPCR) for selected candidate regions. The presented method can be used to confirm an
effect of RNAi constructs on their target genes or trans-activity. Analysis of promoter regions can contrib-
ute to estimation of gene activity and related traits.

Key words DNA methylation, MSRE, qPCR, Grain quality, Epigenetic

1 Introduction

High amounts of methylated 5-methylcytosine (5-mC) can be

found in plant genomic DNA. The presence of a methyl-group at
position 5 of cytosine is associated with inactivation of genes, for-
mation of heterochromatin and displays a hallmark of epigenetic
modification per se [1, 2]. In contrast to other eukaryotes, DNA
cytosine methylation in plants appears in three sequence contexts:
CpG, CpHpG, and CpHpH, where H stands for A, C, or T. DNA
methylation in plants is species-, tissue-, organelle- and age-­specific.
It is controlled by phytohormones and changes on seed germina-
tion, flowering and under the influence of various pathogens (viral,
bacterial, fungal). DNA methylation of respective regions influ-
ences plant growth and development, with particular involvement
in regulation of gene expression and DNA replication.

Electronic supplementary material: The online version of this chapter (https://doi.org/10.1007/978-1-4939-8914-

0_17) contains supplementary material, which is available to authorized users.

Nese Sreenivasulu (ed.), Rice Grain Quality: Methods and Protocols, Methods in Molecular Biology, vol. 1892,
https://doi.org/10.1007/978-1-4939-8914-0_17, © Springer Science+Business Media, LLC, part of Springer Nature 2019

301
302 Christiane Seiler and Markus Kuhlmann

DNA methylation is controlled by three main pathways: de

novo methylation, maintenance of methylation, and active removal
of methylation marks on the genomic DNA [3]. Sequence specific-
ity of modification is conferred by short interfering RNAs (siRNA)
in a mechanism termed RNA-directed DNA methylation (RdDM)
that was first observed in tobacco plants infected with the RNA
pathogen potato spindle tuber viroid [4].
The RdDM mechanism is the main pathway contributing to de
novo DNA methylation [5, 6]. This mechanism is plant-specific
and involved in various regulatory processes such as basal heat tol-
erance [7]. DNA methylation acquired by RdDM can be associ-
ated with transcriptional gene silencing—TGS [8], genome
stability by transposon inactivation [9], hybrid vigor [10], and
DNA virus defence [11]. When the region subjected to RdDM
contains promoter elements, the transcription of the correspond-
ing gene can be supressed correlating with the amount of DNA
methylation present [12]. The term grain quality can be separated
in various aspects from different perspectives.
The quality depends on the intended use of the grain and is
manifested in the biological features and chemical composition.
Main aspects are high germination percentage (viability of seed)
and chemical composition of the grain such as starch content, fla-
vor, appearance, and healthy compounds. During fertilization and
grain development massive transcriptional changes occur, which
are controlled by epigenetic regulators and chromatin remodeling
factors [13–15]. These events of remodeling are accompanied by
histone and DNA modification and/or Polycomb-complexes as
mechanisms affecting the gene expression.
As several physiological parameters are changed during the
early phase of grain development one of the most important com-
pounds with respect to grain quality is the starch content. During
embryo development a decline of starch level was detected in
Arabidopsis thaliana seeds [16]. A similar regulation was proposed
for the monocot species Brachypodium dystachion [17] followed by
a strong accumulation of starch in the germinating embryo. Within
the same study it is concluded that epigenetic modifications may
play an important role in the organs transmitting stimuli to the
embryo during seed maturation, desiccation, and germination.
Nevertheless up to now no target gene regulated via DNA meth-
ylation of its promoter was identified in a crop plant. The reason
for this is most probably the lack of knowledge on the genome
sequence for several species and methods estimating DNA meth-
ylation. Using MSAP (methylation-sensitive amplified polymor-
phism) markers an endosperm-specific decrease of CG and CHG
methylation could be identified in Sorghum [18]. This indicates
that processes involving DNA methylation and demethylation are
contributing to gene regulation at this phase of seed development
 Quantification of DNA Methylation as Biomarker for Grain Quality 303

in the endosperm but also contributing to the proper germination

of the embryo.
The described method can be used to validate a methylation
pattern of a putative target gene and correlate the level of methyla-
tion with an evaluated quality trait. The presence of DNA methyla-
tion in a promoter region [19] is generally related to transcriptional
suppression of a target gene.
Another valuable approach is the detection of DNA methyla-
tion as footprint of a functional RNAi-construct at the target gene
level. The identification of DNA methylation within the target
region is much faster and simpler than detection of small RNAs as
hallmark of functional RNAi: After cleavage (dicing) of the tran-
scribed hairpin RNA the produced small RNAs will result in DNA
methylation as a consequence of the RdDM mechanism. This
DNA methylation can be detected in all genomic regions homolo-
gous to the 24mers siRNAs derived from the RNAi-construct. If
the selected RNAi target gene is related to grain quality, DNA
methylation is an easy detectable and reliable marker. But also for
several crop genes related to grain development and germination
an epigenetic regulation was proposed [20, 21].

1.1 Experimental For the analysis of DNA methylation within a defined region vari-
Approaches ous methods based on different principles are available:
to Quantify DNA Methylated DNA ImmunoPrecipitation (MeDIP) is based on
Cytosine Methylation the immunological detection of the 5-mC-Epitope by a specific
antibody. The bound DNA can be precipitated and based on the
assay quantified by PCR, displayed by slot blotting procedure or
used for microarray hybridization and next-generation
sequencing.
The second method: bisulfite sequencing is based on the
chemical property of the methyl-group attached to the cytosine’s
to protect them against chemical modification with sodium bisul-
fite. Based on this technique sodium bisulfite is used to reduce
unmodified cytosines to uracil. After PCR amplification all uracils
are converted into thymines. The products can be sequenced or
analyzed by other methods (e.g., COmbined Bisulfite Restriction
Analysis, COBRA [22]). With technical methods available this
technique can be applied genome wide. The classical method of
methylation analysis is based on the property of restriction enzymes
to be unable to cleave methylated DNA. This property is based on
the evolutionary use of restriction enzymes developed in bacteria
and archea. In these prokaryotes restriction enzymes protect the
organism from foreign DNA and virus particles by recognizing
DNA sequences and cleaving them into nonfunctional pieces.
Endogenous sequences are masked by a modification enzyme (a
methyltransferase) that modifies the host DNA and blocks cleavage
of the corresponding cleaving enzyme. Such two component sys-
tem is referred to as restriction modification system [23].
304 Christiane Seiler and Markus Kuhlmann

The quantification of the proportion of cleaved to not cleaved

DNA can be performed in several ways. Cleavage can be displayed
directly by Southern blotting and hybridization of a probe related
to the target region [12] or quantified by qPCR. The advantage of
using methylation sensitive restriction enzymes in combination
with quantitative real-time PCR is relatively simple handling and
low analysis costs. This will allow testing a high number of indi-
vidual plants. With this, the methylation level of a particular target
region in a small population, cultivar or breeding line can be esti-
mated and used as biomarker.
One drawback of the method is that only the selected restric-
tion site can be monitored.

2 Materials

2.1 Extraction 1. Any DNA extraction kit.

of Genomic DNA 2. Water bath or heating block for heating at 65 °C and 80 °C.
3. Vortexer.
4. Ethanol (96–100%).
5. Microcentrifuge with rotor for 2 ml tubes.

2.2 Methylation 1. Water bath or heating block for heating at 37 °C and 85 °C.
Sensitive Restriction 2. Methylation sensitive restriction enzymes (e.g., HpaII) with
Digestion reaction buffer.
(a) 
A list of methylation sensitive restriction enzymes and
specificities of their methylation sensitivity are described at
REBASE.NEB.COM [24] and can be found in
Supplementary Table 1.

2.3 Real-Time PCR 1. Real-time PCR cycler.

2. 96- or 384-Well plates.
3. Optical seal.
4. SyBr-green-based PCR mix.
5. Specific forward and reverse primer for control and target
region.

3 Method

DNA methylation assay based on cytosine methylation-sensitive

restriction cleavage and quantitative PCR
An exemplary quantification of DNA cytosine methylation
present at a particular sequence by PCR-based quantification of
 Quantification of DNA Methylation as Biomarker for Grain Quality 305

that respective DNA part that is resistant to cleavage by cytosine

methylation-sensitive restriction enzymes will be shown as an
example. The analyzed target region that will be analyzed is a
genic, coding region, which is not methylated under any environ-
mental condition in a wild-type barley plant (Fig. 1). After trans-
formation of a hairpin-RNAi-construct the post transcriptional
gene silencing of the corresponding target gene was monitored. As
by-product of the PTGS related small RNAs also 24mer small
RNAs are generated. This class of small RNAs is targeting the
RdDM machinery to homologous sequences. As a consequence,
the genomic DNA encoding the hairpin sequence as well as the
homologous target gene region in the nucleus is subjected to DNA
methylation. For the exemplary analysis two cytosine methylation
sensitive enzymes, HpaII (CCGG) and NheI (CGTACG), with
single-recognition sites in this region (Fig. 1) are selected. It is
generally suggested to include an enzyme that is insensitive to
cytosine methylation to confirm completeness of the cleavage reac-
tion if a suited restriction site is included in the region analyzed.
The level of methylation will be displayed as amplification relative
to the uncut control, respective control cleavage.

3.1 Cleavage Genomic DNA preparations from individual plants should be iso-
of Genomic DNA lated according to the manufacturer’s protocols. Uniform sam-
pling of selected tissue is recommended. Plants should be of similar
age, developmental stage, and grown under similar controlled
environmental conditions without pathogens (all these factors
have influences on DNA methylation) unless specific effects should
be tested. To get statistical meaningful data it is recommended to
test 5 biological replicates.
1. For each DNA preparation:
0.1 μg genomic DNA is diluted in 500 μl H2O
separate aliquots of 87 μl reaction tubes (~17.5 ng), e.g.,

Tube I Tube II Tube III

No enzyme HpaII NheI

2. Add 10 μl buffer (10x).

3. Add 3 μl enzyme (30 U, 10 U/μl).
4. Incubate over night at 37 °C.
5. Inactivate restriction enzymes for 5 min at 85 °C (do also for
“no enzyme” control).
6. Add 900 μl bidistilled water to achieve a final volume of
1000 μl.
Standards that can be used for quantitative PCR calibration
306 Christiane Seiler and Markus Kuhlmann

Fig. 1 Principle of methylation analysis of target (Gene of interest) and Control region by MSRE-qPCR. (a)
Genomic DNA is cleaved by methylation sensitive restriction enzymes and selected regions are amplified by
qPCR (gray: gene of interest/green: not methylated control). The region homologous to the RNAi construct is
methylated by RNA directed DNA methylation. The arrows indicate the forward and reverse primer defining the
amplicon used for quantitative PCR. (b) The relative level of amplification indicates the level of protection of the
restriction sites by methylation. DNA was extracted from barley leaves. The level of amplification is given rela-
tive to the uncleaved control (1 equals 100%) as mean with standard deviation. Green bars show the negative
cleavage of the control region (HpaII: dark green, NheI: light green). Gray bars indicate the relative amplification
after analytical cleavage of the target region (HpaII: dark gray, NheI: light gray). N = 5 biological replicates. The
level of estimated DNA methylation in the region homologous to the RNAi construct is ∼50%

gDNA serial dilution: 0.01 μg/10 μl

0.001 μg/10 μl
0.0001 μg/10 μl
0.00001 μg/10 μl
Water-control

3.2 Setup 1. Use 10 μl (∼0.000225 μg) of cleaved or control DNA prepara-

of Quantitative tion per 25 μl PCR setup:
“Real-Time” PCR 10 μl template.
12.5 μl SyBr green Supermix.
1.25 μl 100 pmol/μl forward primer.
1.25 μl 100 pmol/μl reverse primer.
25 μl total volume per reaction.
2. Prepare a master-mix for all samples (SyBr green Supermix, for.
& rev. primers).
3. Set up samples for PCR in 3 technical replicates of a 96-well
plate.
4. Transfer into real-time PCR device.
 Quantification of DNA Methylation as Biomarker for Grain Quality 307

3.3 PCR Program 1. 5 min 95 °C.

2. 15 s 95 °C.
3. 30 s 62 °C.
4. 30 s 72 °C Detection.
5. Repeat steps 2–4 40×.
Melting curve
6. 65 °C Detection.
7. +0.5 °C.
8. Repeat step 7 60×.

3.4 Evaluation Evaluation of real-time PCR data should always be performed

of Data according to the MIQE (Minimum Information for publication of
Quantitative real-time PCR Experiments) guidelines as a minimal
standard [25]. The calculation of the relative amount of amplified
target is done according to [26] with the 2−ΔΔCT formula. After the
PCR run, the melting curve of the PCR products provides an
important control for the homogeneity of the amplification prod-
ucts (Fig. 2). A sharp, narrow peak is desired. Shoulders or signals
beside the peak at higher temperatures are hints for unspecific
amplification “background.” As these unspecific products also
contribute to the fluorescence signal on which quantification is
based, PCR conditions should be carefully optimized for each
primer combination. The values for the correlation coefficient of
the PCR should be around 0.9 and the PCR efficiency around
100%. In routine operation, values between 80% and 110% are
acceptable and might vary according to pipetting errors or sample
quality. The CT (cycle threshold) is the value at which the fluores-
cence signal curve of a sample crosses the preset threshold of

8000
7000
6000
5000
- d(RFU) / dT

4000
3000
2000
1000
0
–1000
62 64 66 68 70 72 74 76 78 80 82 84 86 88 90 92 94 96
Temperature[°C]

Fig. 2 Melting-cure of the PCR product obtained for a fragment of 460 bp length and a melting point at
88.5–89 °C. The melting curve analysis should result in one uniform peak. Additional peaks indicate additional
products and the reaction should not be used for further evaluation. Peaks in the lower temperature range
(60–70 °C) are usually results of primer dimer formation
308 Christiane Seiler and Markus Kuhlmann

60000 60000
55000 55000
50000 50000
PCR Base Line Subtracted CF RFU

45000 45000
40000 40000
35000 35000
30000 30000
25000 25000
20000 20000
15000 15000
10000 10000
5000 5000
0 0
–5000 –5000
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42
Cycle

Fig. 3 “Real-time” fluorescence signals during PCR amplification. The graph indicates two technical replicates
of cleaved (blue/brown) to uncleaved reactions (red/green). The line indicates the Cycle threshold level (CT)
that should be positioned in the exponential amplification phase of the PCR reaction

amplification (orange line ∼15,000) in a particular analysis (Fig. 3).

The CT should be set in the range of the exponential amplification
phase of the PCR for all samples.
The “CT value” of a PCR reaction is defined as the number of
amplification cycles until the fluorescence signal passes the CT line.
Measurements are usually done in triplicate and the mean value of
CT values for each sample will be used for further calculation. The
CT values generated for the different samples can be compared
with the help of the “delta CT method” [26].
Relative amount of PCR target in sample1 compared to
sample2:
−( CT sample1−CT sample 2 )
sample1 / sample2 = 2

[Accordingly, percentage methylation = 100% × 2−(CT sample cleaved

].
− CT sample uncleaved)

4 Notes

1. The cleavage of the genomic DNA should be tested. The entire

genome will never be cleaved completely. The presence of
residual uncut DNA can be tested by adding non-methylated
PCR products or plasmid DNA as tester. Alternatively, a non-­
methylated region containing a restriction site of the selected
enzyme can be used as reference.
 Quantification of DNA Methylation as Biomarker for Grain Quality 309

2. It is recommended to analyze reference regions with described

DNA methylation levels. Meanwhile, several regions of crop
plants are analyzed by bisulfite sequencing and available in
public databases [27].
3. To make general statements about the level of methylation it
has to be considered that CG methylation is in general higher
abundant than asymmetric DNA methylation. This is caused
by the maintenance mechanism acting in the methylated
region based on symmetric sequences. Using a restriction
enzyme such as HpaII (CCGG) methylation on one of the
four cytosines present in the cleavage recognition motif will
be detected.
4. Using enzymes with a longer recognition motif will increase
specificity and completeness of the genome cleavage.
5. If DNA methylation levels of high repetitive sequence should
be analyzed a higher dilution of genomic DNA might be
required.

References
1. Cokus SJ et al (2008) Shotgun bisulphite 10. Shen H et al (2012) Genome-wide analysis of
sequencing of the Arabidopsis genome reveals DNA methylation and gene expression changes
DNA methylation patterning. Nature in two Arabidopsis ecotypes and their recipro-
452(7184):215–219 cal hybrids. Plant Cell 24(3):875–892
2. Lister R et al (2008) Highly integrated single-­ 11. Wang MB et al (2012) RNA silencing and
base resolution maps of the epigenome in plant viral diseases. Mol Plant-Microbe Interact
Arabidopsis. Cell 133(3):523–536 25(10):1275–1285
3. Matzke M et al (2007) Targets of RNA-­ 12. Fischer U et al (2008) Local DNA features
directed DNA methylation. Curr Opin Plant affect RNA-directed transcriptional gene
Biol 10(5):512–519 silencing and DNA methylation. Plant
4. Wassenegger M et al (1994) RNA-directed de J 53(1):1–10
novo methylation of genomic sequences in 13. Shi J, Dong A, Shen WH (2014) Epigenetic
plants. Cell 76(3):567–576 regulation of rice flowering and reproduction.
5. Chan SW, Henderson IR, Jacobsen SE (2005) Front Plant Sci 5:803
Gardening the genome: DNA methylation in 14. Zhang H, Ogas J (2009) An epigenetic per-
Arabidopsis thaliana. Nat Rev Genet 6(5): spective on developmental regulation of seed
351–360 genes. Mol Plant 2(4):610–627
6. Matzke MA, Birchler JA (2005) RNAi-­ 15. Zhang L et al (2011) ABA treatment of germi-
mediated pathways in the nucleus. Nat Rev nating maize seeds induces VP1 gene expres-
Genet 6(1):24–35 sion and selective promoter-associated histone
7. Popova OV et al (2013) The RdDM pathway is acetylation. Physiol Plant 143(3):287–296
required for basal heat tolerance in Arabidopsis. 16. Andriotis VM et al (2010) Starch turnover in
Mol Plant 6(2):396–410 developing oilseed embryos. New Phytol
8. Castel SE, Martienssen RA (2013) RNA inter- 187(3):791–804
ference in the nucleus: roles for small RNAs in 17. Wolny E, Braszewska-Zalewska A, Hasterok R
transcription, epigenetics and beyond. Nat Rev (2014) Spatial distribution of epigenetic modi-
Genet 14(2):100–112 fications in Brachypodium distachyon embryos
9. Bucher E, Reinders J, Mirouze M (2012) during seed maturation and germination. PLoS
Epigenetic control of transposon transcription One 9(7):e101246
and mobility in Arabidopsis. Curr Opin Plant 18. Zhang M et al (2011) Tissue-specific differ-
Biol 15(5):503–510 ences in cytosine methylation and their associa-
310 Christiane Seiler and Markus Kuhlmann

tion with differential gene expression in and their impact on genome evolution. Nucleic
sorghum. Plant Physiol 156(4):1955–1966 Acids Res 29(18):3742–3756
19. Kuhlmann M et al (2014) DNA methylation 24. Roberts RJ et al (2015) REBASE--a database
maintenance consolidates RNA-directed DNA for DNA restriction and modification: enzymes,
methylation and transcriptional gene silencing genes and genomes. Nucleic Acids Res
over generations in Arabidopsis thaliana. Plant 43(Database issue):D298–D299
J 80(2):269–281 25. Bustin SA et al (2009) The MIQE guidelines:
20. Bai F, Settles AM (2014) Imprinting in plants minimum information for publication of quan-
as a mechanism to generate seed phenotypic titative real-time PCR experiments. Clin Chem
diversity. Front Plant Sci 5:780 55(4):611–622
21. Li J et al (2013) Integration of epigenetic and 26. Livak KJ, Schmittgen TD (2001) Analysis of
genetic controls of seed size by cytokinin in relative gene expression data using real-time
Arabidopsis. Proc Natl Acad Sci U S A 110(38): quantitative PCR and the 2(−Delta Delta
15479–15484 C(T)) method. Methods 25(4):402–408
22. Xiong Z, Laird PW (1997) COBRA: a sensitive 27. Li X et al (2012) Single-base resolution
and quantitative DNA methylation assay. maps of cultivated and wild rice methylomes
Nucleic Acids Res 25(12):2532–2534 and regulatory roles of DNA methylation in
23. Kobayashi I (2001) Behavior of restriction-­ plant gene expression. BMC Genomics
modification systems as selfish mobile elements 13:300
 Chapter 18

CRISPR-Cas9-Mediated Genome Editing of Rice Towards

Better Grain Quality
Anindya Bandyopadhyay, Xiaojia Yin, Akshaya Biswal, Robert Coe,
and William Paul Quick

Abstract
With continued economic development in Asia the demand for high yielding varieties with premium grain
quality traits is set to increase. This presents a significant challenge to plant breeders because varieties must
be tailored to meet regional preferences. It is already apparent that traditional breeding techniques cannot
meet this challenge and so emerging genomics technologies will have to be utilized. Genome editing tools
afford the ability to efficiently and precisely manipulate the genome. Among these, the bacterial clustered,
regularly interspaced, short palindromic repeat (CRISPR) associated protein 9 (Cas9) or CRISPR-Cas9
has emerged as the easiest, most economic, and efficient technology to undertake genome editing in rice.
This technique allows precise site-specific gene modification or integration. In this chapter we present a
method for utilizing CRISPR-Cas9 for improving grain quality traits in rice; this should enable molecular
breeders to quickly and efficiently produce high yielding rice varieties tailored to meet specific cultural and
regional requirements for grain quality.

Key words Genome, Rice, Grain quality

1 Introduction

The market value of rice and the adoption of new varieties by

farmers is strongly driven by grain quality traits [1, 2]. Underlying
these are consumer preferences for particular physical, nutritional,
cooking, and eating characteristics [3–5]. Desirable physical prop-
erties include uniform grain length, width, weight, color (white-
ness and translucence), and chalk content. Cooking and eating
(organoleptic) characteristics, such as cooking time (gelatinization
temperature and viscosity [6]), the ability of rice to remain soft
after cooking (gel consistency [7]), textural properties of the
cooked rice (amylose content [8–10] and aroma [11]) are also very
important. Aroma is considered to be the premium quality trait
that brings the highest economic gain for the farmer [11–15].

Nese Sreenivasulu (ed.), Rice Grain Quality: Methods and Protocols, Methods in Molecular Biology, vol. 1892,
https://doi.org/10.1007/978-1-4939-8914-0_18, © Springer Science+Business Media, LLC, part of Springer Nature 2019

311
312 Anindya Bandyopadhyay et al.

The current benchmark varieties with premium grain quality

traits were developed during the 1960–1980s [16]. These variet-
ies, although low yielding and susceptible to stress, continue to be
grown today because of their superior grain quality and high accep-
tance among consumers. The release of new varieties has been lim-
ited by a focus on increasing yield and the difficulty associated with
combining this with superior grain quality [17]. This presents a
challenge for breeders as continued economic growth in Asia [18,
19] will only increase the demand for premium quality rice [1, 20].
Meeting this demand is complicated by the fact that there are dis-
tinctive cultural differences in the traits that are preferred [21],
meaning that breeders must tailor rice varieties to meet regional
preferences. The genomics era brings many opportunities to help
breeders to address this challenge.
While not complete, our understanding of the genetic basis
underlying many grain quality traits is growing [3, 22]. Quantitative
trait loci (QTL) controlling for chalkiness (reviewer in [22]),
appearance (reviewed [3]), protein content [23], weight, and
length [24–26] have been reported. Regions on chromosome 3, 6,
and 8 have been identified as hotspots for QTLs controlling a vari-
ety of grain quality traits [22, 27, 28], indicating that there remain
numerous novel genes to be identified. The genes associated with
these QTLs are starting to be identified and characterized, such as
BETADINE ALDEHYDE DEHYDROGENASE 2 (BADH2)
controlling the flavor and aroma of Jasmine and Basmati rice
(reviewed in [3, 29]), the Waxy locus associated with amylose con-
tent and grain texture of japonica and indica rice [30] and GRAIN
INCOMPLETE FILLING 1 (GIF1) controlling grain size [31].
As a deeper understanding of the genotype underlying varieties
with diverse grain quality traits becomes available, the potential to
use biotechnology to engineer these traits into high-yielding vari-
eties will become a reality [32–35]. The discovery and fine-­mapping
of QTLs can be used for genomic selection of varieties with desir-
able traits. Genes can be cloned to unravel the allelic variation in
the polymorphisms or transformed into high-yielding lines to sat-
isfy regional preferences in grain quality. However, as it is now
possible to sequence the entire genomes of a large number of rice
accessions [36], breeders will soon be able to identify the genes
underlying grain quality traits and to mine for variation in alleles.
This affords an unprecedented opportunity to understand how
allelic variation relates to phenotypic variation. The challenges for
molecular breeders then become how best to use this information.
The major drawbacks with conventional transformation technolo-
gies are consumer acceptance and the fact they cannot eliminate
the endogenous target gene, thus there is an emerging role for
modern tools such as targeted genome editing technologies.
Genome editing allows precise, targeted changes to be made in
the genome by inserting, replacing, or removing DNA to induce a
 CRISPR-Cas9-Mediated Genome Editing of Rice Toward Better Grain Quality 313

Fig. 1 Illustration of CRISPR-Cas9 and its mode of action. CRISPR-Cas9 is composed of Cas9 nuclease and a
sgRNA. The 20 nt guide sequence binds to the target sequence upstream to the PAM (in yellow) and the Cas9
nuclease creates a DSB. The DSB is then repaired by either NHEJ that leads to indels (in red), or by HDR that
could confer precise genome editing (in blue)

desired trait. Several artificially engineered nucleases have emerged

in the last decade, including customized homing endonucleases
(mega-nulceases) [37], zinc finger nucleases [ZFNs; [38–40], and
transcription activator like effector nucleases [TALENs; [40–47].
These techniques generate double strand breaks (DSBs) at specific
loci in the genome, which are then repaired by either nonhomolo-
gous end joining (NHEJ) or homology-directed repair (HDR; see
Fig. 1). NHEJ is a DSB repair pathway that ligates two broken
ends together leading to insertions or deletions (indels) at the site
of the break. These often induce mutations that alter, or lead to
frame-shifts that knock out gene function. Although uncommon,
HDR is a template-dependent pathway for DSB repair which can
be exploited to insert gene(s) of interest (GOIs) or single-­
nucleotide substitutions at a target locus. A site-specific nuclease is
used to cleave the DNA, a homology-containing donor template is
then provided that is faithfully inserted by HDR.
While these technologies offer an unprecedented ability to effi-
ciently and precisely modify the genome they are technically chal-
lenging, time consuming, and costly. The emergence of the
bacterial clustered, regularly interspaced, short palindromic repeat
(CRISPR) associated protein 9 (Cas9) or CRISPR-Cas9 [48–57]
from Streptococcus pyogenes offers a much easier, more economic,
and efficient technology for genome editing. CRISPR-Cas9 is a
microbial adaptive immune mechanism that utilizes RNA-guided
nucleases to cleave foreign genetic elements [48–51]. The system
314 Anindya Bandyopadhyay et al.

is composed of the nuclease Cas9, the mature CRISPR RNA

(crRNA), and a partially complementary trans activating crRNA
(tracrRNA). The crRNA contains a ∼20 nucleotides (nt) sequence
that guides Cas9 to a complementary DNA sequence within the
genome. This DNA sequence must be adjacent to a short sequence
known as protospacer-adjacent motif (PAM). Cas9 then creates
DSB at the target locus that is then repaired in a manner similar to
ZFNs and TALENs. The crRNA and tracrRNA can be artificially
fused together resulting in a chimeric RNA molecule known as
single-guide RNA (sgRNA). The sgRNA can be directed toward
any locus of interest in the genome by editing the 20-nt guide
sequence to match a preselected PAM sequence. This makes the
system much simpler than other engineered nuclease systems.
The CRISPR-Cas9 system has been used for genome editing
of both model and crop plants [58–69]. Multiple rice genes such
as the phytoene desaturase gene (OsPDS), betaine aldehyde dehy-
drogenase (OsBADH2), mitogen-activated protein kinase
(OsMPK2), bacterial blight susceptible genes/sugar transporters
(OsSWEETs), Myb family transcription factor (OsMYB1), rice out-
ermost cell-specific gene 5 (ROC5), stromal processing peptidase
(SPP), young seedling albino gene (YSA), stress-responsive
ricemitogen-­ activated protein kinase (OsMPK5), chlorophyll A
oxygenase-1 (CAO1), LA1/rice tiller angle controlling gene
(LAZY01), and rice stomatal developmental gene Epidermal
Patterning Factor Like-9 (EPFL9) have been modified [60, 62–65,
69–71]. There remains an opportunity to utilize this technology to
enhance grain quality traits. An example of the potential is to mod-
ify grain texture by manipulating the amylose content of the endo-
sperm [72]. The lower the amylose content the stickier the rice.
Then enzyme granule bound starch synthase (GBSS), encoded by
the waxy locus on chromosome 6, is required for amylose synthe-
sis. Two wild-type alleles are found in rice, Wxa is associated with
high amylose content of indica rice, and Wxb associated with lower
amylose content of japonica rice [30]. In Wxb a single-­nucleotide
polymorphism (SNP) change of G to T at +1 position of the intron
1 reduces the ability of the enzyme to excise the intron 1 from the
Waxy pre-mRNA, reducing the abundance of the mature Waxy
transcript, GBSS protein, and amylose content. The CRISPR-Cas9
system could offer breeders the potential to engineer high yielding
varieties to meet local preferences for stickiness by artificially mim-
icking natural variation and inducing a SNP. The system could also
be used to introduce non-native genes into the rice genome.
Homology directed insertion allows these to be introduced into a
safe loci, such as intergenic regions, where there is minimal effect on
the adjacent genes. This should reduce the pleiotropic effects associ-
ated with random gene insertion associated with Agrobacterium
or biolistic mediated transformation techniques that are currently
widely used.
 CRISPR-Cas9-Mediated Genome Editing of Rice Toward Better Grain Quality 315

2 Materials

2.1 Cloning 1. PCR thermocycler (Q-CyclerII, Quanta Biotech, UK).

2. Water bath (SW22, Julabo, Germany).
3. Liquid nitrogen flask.
4. NanoDrop spectrophotometer (Thermo Fisher Scientific,
USA).
5. Digital Gel Documentation Systems (Uvitec, Cambridge,
UK).
6. Laminar flow hood (ELITE LH-48, Elite Scientific &
Diagnostic International Supplies Corp, Philippines).
7. Horizontal gel electrophoresis systems (Thermo Fisher
Scientific, USA).
8. Plating spreader (glass or plastic).
9. Microcentrifuge tubes (2 mL, 1.5 mL, and 0.6 mL, Axygen,
Corning, USA).
10. PCR tubes (0.2 mL, Axygen, Corning, USA).
11. 96-Well PCR plates (Axygen, Corning, USA).
12. Pipette tips (1000, 200, and 10 μL, Axygen, Corning, USA).
13. 94 mm × 16 mm petri dishes (Biologix Plastics Co., Ltd.,
China).
14. Liquid nitrogen.
15. SOB Medium (Pronadisa, 1541).
16. Bacto agar (BD, 214010).
17. SOC Medium (Pronadisa, 2019).
18. Miller’s LB Broth (Pronadisa, 1551).
19. Miller’s LB Agar (Pronadisa, 1552).
20. YEB Agrobacterium Growth Medium (bioT, 30627031-1).
21. Bacto agar (BD, 214010).
22. Kanamycin sulfate (Sigma Aldrich, K1377).
23. Rifampicin (Sigma Aldrich, R3501).
24. Agarose, low EEO (ASIAGEL, 8010).
25. Goodview DNA dye (SBS Bio, China).
26. 1 kb Plus DNA ladder (Invitrogen, USA).
27. AarI (Thermo Fisher Scientific, ER1581 or ER1582).
28. HindIII-HF® (NEB, R3104S).
29. EcoRI-HF® (NEB, R3101S).
30. KpnI-HF® (NEB, R3142S).
31. SpeI-HF® (NEB, R3133S).
316 Anindya Bandyopadhyay et al.

32. AarI (Thermo Fisher Scientific, USA).

33. T4 DNA ligase (NEB, M0202S).
34. DNA Polymerase I, Large (Klenow) Fragment (NEB,
M0210S).
35. PCR reagent (U-taq Kit, SBS Bio, China).
36. dNTP set, 100 mM × 0.4 mL each (SBS Bio, China).
37. GenUP™ Plasmid Kit (Biotechrabbit, BR0700203).
38. GenUP™ Gel Extraction Kit (Biotechrabbit, BR0700403).
39. Plasmid: IRS154a (modified pCAMBIA1300).
40. Plasmid: pOsU3-sgRNA.
41. Plasmid: pJIT163-2NLSCas9.
42. E. coli DH5α chemically competent cells.
43. Agrobacterium tumefaciens LBA4404 competent cells.

2.2 Rice 1. Environmentally controlled growth chamber (Conviron,

Transformation Winnipeg, Canada).
2. Laminar flow hood (NUAIRE, Plymouth, USA).
3. Stereo microscope (Olympus, Tokyo, Japan).
4. 28 °C and 37 °C incubator (LM-570RD, Yihder Co., Ltd).
5. Cell density meter (Ultrospec10, Biochrom Ltd., UK).
6. Micropipettors (EPPEDORF, Hamburg, Germany).
7. pH meter(BP3001, Trans Instrument, Singapore).
8. Rotator AG (FINE PCR, South Korea).
9. Water bath (JULABO TW20, Julabo GmbH, Germany).
10. Centrifuge (Thermo Fisher Scientific, USA).
11. Weighing balance (Ohaus, USA)
12. Steri 350 (Inotech, USA).
13. Graduated cylinder (glass, Corning, USA).
14. Beaker (glass, Corning, USA).
15. Forceps and scalpels (KFI K-14 stainless steel).
16. Nicrome Loop.
17. 50 mL Falcon tubes (Thermo Fisher Scientific, USA).
18. Tween20 (M147, AMRESCO).
19. Ethanol (E193, AMRESCO).
20. 94 mm × 16 mm petri dishes (Biologix Plastics Co., Ltd.,
China).
21. Filter paper (GE Healthcare Life Sciences, 3030–917).
22. 1.5 mL microfuge tubes (Axygen, Corning, USA).
23. 1 mL pipette tips (Axygen, Corning, USA).
 CRISPR-Cas9-Mediated Genome Editing of Rice Toward Better Grain Quality 317

24. 3 M Micropore tape (3 M, USA) (Tables 1 and 2).

25. Reagents for transformation are listed in Tables 1 and 2.

2.3 Plant Verification 1. PCR thermocycler (Q-CyclerII, Quanta Biotech, UK).

2. Water bath (SW22, Julabo, Germany).
3. NanoDrop spectrophotometer (Thermo Fisher Scientific,
USA).
4. Digital Gel Documentation Systems (Uvitec, Cambridge,
UK).
5. Horizontal gel electrophoresis systems (Thermo Fisher
Scientific, USA).
6. Standard microcentrifuge tubes (2 mL, 1.5 mL, and 0.6 mL,
Axygen, Corning, USA).
7. PCR tubes (0.2 mL, Axygen, Corning, USA).
8. Pipette tips (1000, 200, 10 μL).
9. Agarose, low EEO (ASIAGEL, 8010).
10. Goodview (SBS Bio, China).
11. 1 kb plus DNA ladder (Invitrogen, USA).
12. Surveyor® Mutation Detection Kit for Standard Gel
Electrophoresis (IDT, 706020).
13. Phusion® High-Fidelity DNA Polymerase (NEB, M0530S).
14. Zero Blunt® pCR Cloning Kit (Invitrogen, K2700).
15. GenUP™ Plasmid Kit (Biotechrabbit, BR0700203).
16. GenUP™ Gel Extraction Kit (Biotechrabbit, BR0700403).
17. 100 mM dNTP set (Invitrogen, 10,297,018).
18. E. coli DH5α chemically competent cells.
19. SOB Medium (Pronadisa, 1541).
20. Bacto agar (BD, 214010).

3 Methods

The work flow for the CRISPR-Cas9 system is illustrated in Fig. 2.

1. Identify the target sequence in rice genome. Perform in silico
evaluation of the target sequence to eliminate possible off target
effects. Synthesize the guide sequence.
2. Clone the guide sequence into a binary vector harboring
sgRNA scaffold, Cas9, and a hygromycin resistant gene that
acts as a selectable marker.
3. Transform the sgRNA-Cas9 binary construct into immature
rice embryos using Agrobacterium-mediated transformation.
318 Anindya Bandyopadhyay et al.

Table 1
Stock solutions 1 (store at 4 °C)

Stock solution Ingredients Amount for 1 L solution

AA SALT MACRO CaCl2·2H2O 1.47 g (10 mM)
MgSO4·7H2O 2.46 g (10 mM)
NaH2PO4·2H2O 1.7 g (10.87 mM)
KCl 29.82 g (400 mM)
AA SALTS MICRO CoCl2·6H2O 23.79 mg (0.1 mM)
CuSO4·5H2O 24.97 mg (0.1 mM)
H3BO3 3g (48.5 mM)
KI 750 mg (4.5 mM)
MnSO4·H2O 1g (5.9 mM)
Na2MoO4·2H2O 250 mg (1 mM)
ZnSO4·7H2O 2g (7 mM)
B5 MINOR-1 FeSO4·7H2O 2.78 g (10 mM)
Na2EDTA·2H2O 3.72 g (10 mM)
B5 MINOR-2 MnSO4·4H2O 1g (4.5 mM)
ZnSO4·7H2O 200 mg (0.7 mM)
H3BO3 200 mg (4.85 mM)
B5 MINOR-3 KI 75 mg (0.45 mM)
B5 MINOR-4 CuSO4·5H2O 2.5 mg (0.01 mM)
Na2MoO4·2H2O 25 mg (0.1 mM)
CoCl2·6H2O 2.5 mg (0.01 mM)
B5 VITAMINS Thiamine HCl 20 mg (0.06 mM)
Pyridoxine HCl 2 mg (0.01 mM)
Nicotinic acid 2 mg (16 μM)
i-Inositol 200 mg (1.1 mM)
GLYCINE Glycine 7.5 g (100 mM)
MS-1 KNO3 95 g (940 mM)
NH4NO3 82.5 g (1.03 M)
MS-2 MgSO4·7H2O 37 g (150 mM)
MnSO4·4H2O 2.23 g (10 mM)
ZnSO4·7H2O 860 mg (3 mM)
(continued)
 CRISPR-Cas9-Mediated Genome Editing of Rice Toward Better Grain Quality 319

Table 1
(continued)

Stock solution Ingredients Amount for 1 L solution

CuSO4·5H2O 2.5 mg (0.01 mM)
MS-3 CaCl2·2H2O 44.1 g (300 mM)
KI 83 mg (0.5 mM)
CoCl2·6H2O 2.5 mg (0.01 mM)
MS-4 KH2PO4 17 g (125 mM)
H3BO3 620 mg (10 mM)
Na2MoO4·2H2O 25 mg (0.1 mM)
MS VITAMINS Nicotinic acid 1 mg (8.1 μM)
Pyridoxine HCl 1 mg (5 μM)
Thiamine HCl 200 μg (0.6 μM)
Glycine 4 mg (53 μM)
Myoinositol 200 mg (1.07 mM)
N6 MAJOR-1 KNO3 141.5 g (1.4 M)
N6 MAJOR-2 MgSO4·7H2O 18.5 g (75 mM)
(NH4)2SO4 46.3 g (350 mM)
N6 MAJOR-3 KH2PO4 40 g (294 mM)
N6 MAJOR-4 CaCl2·2H2O 16.6 g (113 mM)
YCS MAJOR-1 NH4NO3 914 g (11.4 M)
YCS MAJOR-2 NaH2PO4·2H2O 403 g (2.58 M)
YCS MAJOR-3 K2SO4 714 g (4.1 M)
YCS MAJOR-4 CaCl2 886 g (8 M)
YCS MAJOR-5 MgSO4·7H2O 3240 g (13.1 M)
YCS MINOR MnC2·4H2O 15 g (75.8 mM)
(NH4)6Mo7O24·4H2O 740 mg (0.6 mM)
H2BO3 9.34 g (153.5 mM)
ZnSO4·7H2O 350 mg (1.22 mM)
CuSO4·5H2O 310 mg (1.24 mM)
FeCl3·6H2O 77 g (285 mM)
Citric acid (monohydrate) 119 g (566 mM)
320 Anindya Bandyopadhyay et al.

Table 2
Stock solutions 2

Stock solution Amount for 50 mL

name solution Preparation notes
2,4-d 50 mg Dissolve in 2 mL of 1 M KOH, and
dilute with Nanopure H2O to a
final volume of 50 mL, filter,
sterilize and store at 4 °C
NAA 50 mg Dissolve in 2 mL of 1 M NaOH,
and dilute with Nanopure H2O to
a final volume of 50 mL, filter-­
sterilize and store at 4 °C
6BA 50 mg Dissolve in 1 mL of 1 M NaOH,
and dilute with Nanopure H2O to
final 50 mL, filter-sterilize and
store at 4 °C
Kinetin 50 mg Dissolve in 1 mL of 1 M NaOH,
and dilute with Nanopure H2O to
final 50 mL, filter-sterilize and
store at −20 °C
Cafotaxime 10 g Dissolve in sterilized H2O, filter-­
sterilize and store at −20 °C
Cabernicillin 5g Dissolve in autoclaved Nanopure
H2O, filter-sterilize and store at
−20 °C
Hygromycin B Supplied as Supplied as sterilized 1 mg/mL
solution solution, store at 4 °C

4. Analyze the regenerated plants to verify that the targeted

genome sequence has been edited.
If homologous recombination HDR was used in the
introduction-­mediated gene insertion to be performed, an extra
donor vector containing the gene of interest, flanked by a long
homology arm, is required. Homology arms of the donor vector are
homologues to the sequences flanking both sides of the target site.

3.1 Generation In order to establish a robust CRISPR-Cas9 rice genome editing

of the CRISPR-Cas9 system that is optimized for rice transformation, the CRISPR-Cas9
Binary Vector binary vector is inserted into the pCAMBIA background contain-
ing a hygromycin selectable marker (pCAMBIA1300, http://
www.cambia.org/daisy/cambia/585.html). The pCAMBIA back-
bone is well tested and provides for efficient transformation of rice
[75, 76]. Scaffold-sgRNA can be acquired from any of the avail-
able scaffold vectors. Similarly, the Cas9 cassette can be acquired
 CRISPR-Cas9-Mediated Genome Editing of Rice Toward Better Grain Quality 321

Identify the target

(a) sequence in rice genome.
Perform in-silico
evaluation of the target
sequence to eliminate
possible off target effects.
Synthesize the guide
sequence.

Guide Sequence Clone the guide sequence

(b) into a binary vector
harboring sgRNA scaffold,
Cas9 and a hygromycin
resistant gene that acts as
a selectable marker.

Working vector

Transform the sgRNA-Cas9

(c) binary construct into
immature rice embryos
using Agrobacterium
mediated transformation.

(d) Analyze the regenerated

plants are analyzed to
verify that the targeted
genome sequence that
has been edited.

Fig. 2 Work flow for plant genome editing using CRISPR-Cas9

322 Anindya Bandyopadhyay et al.

from any of the available Cas9 vectors, for example, pOsU3-sgRNA

and pJIT163-2NLSCas9 [73, 74], available from Gao lab and
Addgene]. The following describes a method of generating the
binary vector:
Multiple sgRNA scaffold vectors are available from non-profit
plasmid depositories such as “Addgene” (https://www.addgene.
org/CRISPR/) or directly from the laboratories that developed
them.
1. Digest the pCAMBIA1300 and pOsU3-sgRNA plasmid DNA
with the restriction enzymes HindIII-HF® and EcoRI-HF®
in the reaction mixture shown below. Incubate the mixture at
37 °C for 1 h, then add 3 μL of 6x loading dye (containing 1%
SDS) and incubate at 65 °C for 10 min. Centrifuge at maxi-
mum speed for 5 min.

Component Amount
snpH2O X μL (12.5 − X) μL
10× CutSmart Buffer
®
1.5 μL
HindIII-HF® 0.5 μL
Eco RI-HF® 0.5 μL
pDNA X μL (1000 ng)
Total 15 μL

Run the digested mixture on 0.8% agarose gel and excise

the 8858 bp band of pCAMBIA 1300 (pCAMBIA-HindIII-
EcoRI) and the 558 bp band of pOsU3-sgRNA
­(pOsU3-sgRNA-HindIII-­EcoRI). Purify the digested prod-
uct using GenUP™ Gel Extraction Kit and elute into a maxi-
mum total volume of 20 μL.
2. Ligate pCAMBIA-HindIII-EcoRI and pOsU3-sgRNA-­
HindIII-­EcoRI (1:1 molar ratio) at 16 °C for 30 min in the
reaction mixture shown below.

Component Amount
IRS154a-HindIII-EcoRI X μL
pOsU3-sgRNA-HindIII-EcoRI (7.5 − X) μL
10 mM ATP 1 μL
10× T4 DNA ligase reaction buffer 1 μL
T4 DNA ligase 0.5 μL
Total 10 μL
CRISPR-Cas9-Mediated Genome Editing of Rice Toward Better Grain Quality 323

3. Transform the ligated product into E. coli using the heat shock
method (5 μL per transformation). Also transform 3 μL of
linearized pCAMBIA-HindIII-EcoRI plasmid as a negative
control.
4. Screen colonies using OsU3-F + sgRNA-R primers, the PCR
product should be 455 bp. Select 2 positive colonies and culture
in SOB medium containing 50 mg/L kanamycin overnight.
5. Isolate plasmid DNA (pCAMBIA_sgRNA) from the over-
night culture and digest the plasmid DNA with HindIII-HF®
and EcoRI-HF® at 37 °C for 45 min for verification.
Successfully ligated pCAMBIA_sgRNA should be digested,
producing one DNA fragment of 8858 bp and 558 bp.

Component Amount
snpH2O (12.5 − X) μL
10× CutSmart Buffer
®
1.5 μL
HindIII-HF® 0.5 μL
EcoRI-HF® 0.5 μL
pDNA X μL (200 ng)
Total 15 μL

6. Linearize pCAMBIA_sgRNA by digesting with HindIII-HF®

at 37 °C for 1 h.

Component Amount
snpH2O (13.0 − X) μL
10× CutSmart® Buffer 1.5 μL
HindIII-HF ®
0.5 μL
pDNA X μL (1000 ng)
Total 15 μL

7. Digest pJIT163-2NLSCas9 with KpnI-HF® and SpeI-HF® at

37 °C for 1 h.

Component Amount
snpH2O (12.5 − X) μL
10× CutSmart® Buffer 1.5 μL
KpnI-HF® 0.5 μL
SpeI-HF ®
0.5 μL
pDNA X μL (1000 ng)
Total 15 μL
324 Anindya Bandyopadhyay et al.

8. Run the digested mixture on 0.8% agarose gel and excise the
9418 bp band of pCAMBIA_sgRNA (pCAMBIA_sgRNA-­
HindIII) and the 5720 bp band of pJIT163-2NLSCas9
(pJIT163-2NLSCas9-KpnI-SpeI). Purify the two digested
products using GenUP™ Gel Extraction Kit and elute each
into a maximum total volume of 20 μL.
9. Blunt both fragments in the reaction mixture shown below.
Incubate the mixture at 25 °C for 20 min, then add 1 μL of
150 mM EDTA and incubate at 70 °C for 20 min to inactivate
the enzyme.

Component Amount
DNA fragment 12 μL
10× T4 DNA ligase reaction buffer 1.4 μL
10 mM dNTP mix 0.1 μL
DNA Polymerase I, Large (Klenow) Fragment 0.5 μL
Total 14 μL

10. Ligate the two blunted fragments (1:1 molar ratio) in the
reaction mixture below at 16 °C for 3 h.

Component Amount
snpH2O 1.5 μL
Blunt-pC_sgRNA-HindIII X μL
Blunt-pJIT163-2NLSCas9-KpnI-SpeI (10 − X) μL
10 mM ATP 1.5 μL
10× T4 DNA ligase reaction buffer 0.5 μL
T4 DNA ligase 1.5 μL
Total 15 μL

11. Transform 5 μL of the ligated product into E. coli using the

heat shock method.
12. Screen colonies using Cas9-F + Cas9-R primers, the PCR prod-
uct should be 390 bp. Select 2 positive colonies and culture in
SOB medium containing 50 mg/L kanamycin overnight.
13. Isolate the plasmid DNA. Successful cloning should be
­verified by sequencing. The Cas9 cassette could have been
inserted in either orientation as it was cloned with a
blunt end.
 CRISPR-Cas9-Mediated Genome Editing of Rice Toward Better Grain Quality 325

14. Store the verified CRISPR-Cas9 backbone vector (pCRISPR_

UniDir) as the E. coli glycerol stock (500 μL of E. coli culture
and 250 μL of 50% glycerol, deep in liquid nitrogen before
keeping at −80 °C).

3.2 Cloning Cas9 is directed to the targeted loci in plant cells by the help of the
the Working Vector 20 nt guide sequence of the sgRNA. Target sequence is identified
with Desired sgRNA according to experimental purpose. In silico evaluation needs to be
performed to analyze the on-target score, as well as the possible
off-target effects.
1. The 20 nt guide sequence should be the same sequence as the
20 nt at the 5′ upstream of the NGG PAM.
2. Digest the plasmid DNA (pCRISPR_UniDir) with restriction
enzyme AarI in the following reaction mixture, incubated at
37 °C for 5 h.

Component Amount
snpH2O (17.1 − X) μL
10× Buffer AarI 2 μL
pCRISPR_UniDir X μL (1000 ng)
50× oligonucleotide 0.4 μL
AarI 0.5 μL
Total 20 μL

3. Run the digested mixture on a 0.7% agarose gel and excise the
15,089 bp band. Purify the digested product using GenUP™
Gel Extraction Kit. Elute purified product (pCRISPR_UniDir-­
AarI) in a maximum total volume of 20 μL.
4. The guide sequence can be synthesized as single-strand oligos.
Forward oligo (FP) sequence is GGCA(N)20, of which the
(N)20 sequence is the 20 nt identified target sequence at the 5′
upstream of the PAM. Reverse oligo (RP) sequence is
AAAC(N)20, of which the (N)20 sequence is complementary to
the 20 nt identified target sequence. The two oligos are diluted
to 10 μM. Mix 10 μL of each oligos, and heat to 95 °C for
3 min on a heating block and then turn it off to cool down to
room temperature naturally. The annealed oligos have sticky
ends that are compatible to the AarI digested pCRISPR_
UniDir, as shown in Fig. 3.
5. Ligate the linearized (AarI digested) backbone plasmid
(pCRISPR_UniDir-AarI) and the annealed oligos in the reac-
tion mixture shown below, for 30 min at 16 °C.
326 Anindya Bandyopadhyay et al.

FP 5’-GGCANNNNNNNNNNNNNNNNNNNN-3’
||||||||||||||||||||
RP 3’-NNNNNNNNNNNNNNNNNNNNCAAA-3’

Fig. 3 Annealed oligos that have compatible sticky ends to AarI digested
pCRISPR_UniDir

Component Amount
pCRISPR_UniDir-AarI 3 μL
annealed oligos (FP + RP) 4.5 μL
10 mM ATP 1 μL
10× T4 DNA ligase reaction buffer 1 μL
T4 DNA ligase 0.5 μL
Total 10 μL

6. Transform 5 μL of the ligated product into E. coli using the

heat shock method. Also transform 3 μL of linearized
pCRISPR_UniDir-­AarI plasmid as a negative control.
7. To identify the successful insertion of the guide sequence, use
OsU3-F and RP primers for colony PCR. The band size
should be 315 bp. Select 2 positive colonies and culture in
SOB medium containing 50 mg/L kanamycin overnight.
8. Isolate plasmid DNA from the overnight culture and validate
the vector by sequencing using the OsU3-F primer.
9. Transform the sequence-verified plasmid into Agraobactrium
using the freeze thaw method. Two days after transformation,
perform colony PCR with the OsU3-F and RP primer.
10. Inoculate 1 positive colony and extract plasmid DNA from the
Agrobacterium and then transform it back into E. coli. Verify
the back-transformed plasmid by PCR (OsU3-F and RP
primer) and sequencing.
11. Store the verified working CRISPR-Cas9-sgRNA vector as the
E. coli and Agrobacterium glycerol stock (500 μL of bacterium
culture and 250 μL of 50% glycerol, deep in liquid nitrogen
before keeping at −80 °C) respectively.

3.3 Preparation 1. Infection medium: For 1 L total volume, add 100 mL of AA

of Medium salt Macro, 1 mL of AA SALTS MICRO, 10 mL of B5
for Transformation MINOR-1, 1 mL of B5 VITAMIN, 1 mL of 100 mM glycine,
of Rice 876 mg of l-glutamine, 266 mg of aspartic acid, 174 mg of
Arginine, 500 mg of Casamino Acid, 20 g of Sucrose, and
10 g of d-­glucose. Adjust pH to 5.2 and filter-sterilize the
solution. Store the solution at 4 °C for a maximum of 2 weeks.
CRISPR-Cas9-Mediated Genome Editing of Rice Toward Better Grain Quality 327

For transformation of 300 rice immature embryos (IEs),

25 mL Infection medium should be prepared with 25 μL of
100 mM of fresh Acetosyringone (490.5 μg dissolved in 25 μL
of DMSO).
2. Co-cultivation medium: For 1 L total volume, add 20 mL of
N6 MAJOR-1, 10 mL of N6 MAJOR-2, 10 mL of N6
MAJOR-3, 10 mL of N6 MAJOR-4, 10 mL of B5 MINOR-
1, 10 mL of B5 MINOR-2, 10 mL of B5 MINOR-3, 10 mL
of B5 MINOR-­4, 5 mL of B5 VITAMINS, 500 mg of casa-
mino acid, 500 mg of l-proline, 20 g of sucrose, and 10 g of
d-glucose. Adjust to a pH of 5.2 and add 5.5 g of ttype I
agarose. Autoclave the mixture for 15 min, cool to 50 °C, and
add 2 mL of 2,4-d, 1 mL of NAA, 1 mL of 6BA and 1 mL of
fresh 100 mM acetosyringone prior to dispensing in to the
petri dishes (25 mL per 94 mm × 16 mm petri dish). Seal petri
dishes with micropore tape and store at 4 °C for a maximum
of 1 week.
3. Resting medium: For 1 L total volume, add 20 mL of N6
MAJOR-1, 10 mL of N6 MAJOR-2, 10 mL of N6 MAJOR-
3, 10 mL of N6 MAJOR-4, 10 mL of B5 MINOR-1, 10 mL
of B5 MINOR-2, 10 mL of B5 MINOR-3, 10 mL of B5
MINOR-­4, 5 mL of B5 VITAMINS, 500 mg of casamino
acid, 500 mg of l-proline, 300 mg of l-glutamine, 36 g of
mannitol, 20 g of maltose. Adjust pH to 5.8 then add 5 g of
gelrite. Autoclave the mixture for 15 min, cool to 50 °C and
add 1 mL of 2,4-d, 1 mL of NAA, 200 μL of 6BA, 1 mL of
cafotaxime, and 1 mL of c­ abernicillin right prior to dispensing
in to the petri dishes (25 mL per 94 mm × 16 mm petri dish).
Seal petri dishes with micropore tape and store at 4 °C for a
maximum of 2 weeks.
4. Selection medium: For 1 L total volume, add 20 mL of N6
MAJOR-1, 10 mL of N6 MAJOR-2, 10 mL of N6 MAJOR-
3, 10 mL of N6 MAJOR-4, 10 mL of B5 MINOR-1, 10 mL
of B5 MINOR-2, 10 mL of B5 MINOR-3, 10 mL of B5
MINOR-­4, 5 mL of B5 VITAMINS, 500 mg of casamino
acid, 500 mg of l-proline, 300 mg of l-glutamine, 36 g of
mannitol, and 20 g of maltose. Adjust pH to 5.8 then add 5 g
of gelrite. Autoclave the mixture for 15 min, cool to 50 °C and
add 1 mL of 2,4-d, 1 mL of NAA, 200 μL of 6BA, 1 mL of
cafotaxime, 1 mL of cabernicillin, and 0.6 mL of hygromycin
prior to dispensing into the petri dishes (25 mL per
94 mm × 16 mm petri dish). Seal petri dishes with micropore
tape and store at 4 °C for a maximum of 2 weeks.
5. Pre-regeneration medium: For 1 L total volume, add 20 mL of
MS-1, 10 mL of MS-2, 10 mL of MS-3, 10 mL of MS-4,
5 mL of MS VITAMINS, 10 mL of B5 MINOR-1, 30 g of
maltose, and 20 g of sorbitol. Adjust pH to 5.8 and then add
10 g of ttype I agarose. Autoclave the mixture for 15 min, cool
328 Anindya Bandyopadhyay et al.

to 50 °C, add 2 mL of kinetin, 500 μL of NAA, 1 mL of cafo-

taxime, and 1 mL of hygromycin prior to dispensing into the
petri dishes (25 mL per 94 mm × 16 mm petri dish). Seal petri
dishes with micropore tape and store at 4 °C for a maximum
of 2 weeks.
6. Regeneration medium: For 1 L medium, add 20 mL of MS-1,
10 mL of MS-2, 10 mL of MS-3, 10 mL of MS-4, 5 mL of MS
VITAMINS, 10 mL of B5 MINOR-1, and 30 g of sucrose.
Adjust pH to 5.8 and then add 3 g of Gelrite. Autoclave the
mixture for 15 min, cool to 50 °C, and add 2 mL of kinetin,
1 mL of NAA, 1 mL of cafotaxime and 1 mL of hygromycin
prior to dispensing into the petri dishes (25 mL per
94 mm × 16 mm petri dish). Seal petri dishes with micropore
tape and store at 4 °C for a maximum of 2 weeks.
7. YCS solution: For 10 L solution, add 12.5 mL of YCS
MAJOR-­1, 12.5 mL of YCS MAJOR-2, 12.5 mL of YCS
MAJOR-3, 12.5 mL of YCS MAJOR-4, 12.5 mL of YCS
MAJOR-5, and 12.5 mL of YCS MONOR. Adjust pH to 5.0.
Store at room temperature.

3.4 Rice Transform the CRISPR-Cas9-sgRNA vector into immature

Transformation embryos (IEs) of Oryza sativa IR64 using Agrobacterium
tumefaciens-­mediated transformation [75, 76].
1. Agrobacterium preparation: Streak Agrobacterium glycerol
stock onto a YEB plate containing 50 mg/L kanamycin and
incubate at 28 °C for 2 days. One hour before infecting the
IEs, take the Agrobacterium from the YEB plate using a loop
and mix well with 5 mL of infection medium. Adjust the
Agrobacterium suspension to an OD of 0.3 using a cell density
meter. Incubate the Agrobacterium suspension at 25 °C for
1 h in the dark without shaking.
2. Preparation of immature embryos: Collect immature seeds
(milk to soft dough stage) from panicles harvested 12 days
after anthesis. Soak de-hulled immature seeds in 70% ethanol
for 1 min and then rinse with sterilized distilled water. Sterilize
immature seeds with 1% sodium hypochlorite solution (con-
taining one drop of Tween20) in a 50 mL falcon tube for
10 min. Rinse the seeds with sterilized distilled water until all
the sodium hypochlorite has been removed. Isolate IEs and
place 50 on a co-cultivation medium plate and air dry for 3 h
in laminar flow hood to prevent the over spreading of
bacterium.
3. Infection: Rearrange IEs on a co-cultivation medium plate
with the scutellum facing up. Drop 5 μL of Agrobacterium
suspension on top of each IE, seal the plates with micropore
tape, and incubate at 25 °C for 7 days in the dark.
 CRISPR-Cas9-Mediated Genome Editing of Rice Toward Better Grain Quality 329

4. Blotting: After the co-cultivation, remove elongated shoots

from IEs using a sterilized scalpel. Blot the IEs gently on ster-
ile filter paper to prevent the overgrowth of bacterium.
Transfer the blotted IEs to resting medium with the scutellum
facing up. Seal the plates with micropore tape and incubate at
30 °C for 5 days under continuous illumination.
5. Selection 1: Cut each IE into 4 pieces and place them on selec-
tion medium containing 30 mg/L hygromycin with scutellum
facing up (10 IEs/plate, i.e., 40 pieces/plate). Four pieces
from a single IE should be grouped together on a single plate.
Seal the plates with micropore tape and incubate at 30 °C for
10 days under continuous illumination.
6. Selection 2: Transfer all the cut IEs onto fresh selection medium
containing 30 mg/L hygromycin. Seal the plates with micro-
pore tape and incubate at 30 °C for 10 days under continuous
illumination. .
7. Selection 3: Select hygromycin-resistant calli that are well pro-
liferated and yellowish to white in color and transfer onto fresh
selection medium containing 30 mg/L hygromycin. Label
each calli corresponding to the cut IE number. Seal the plates
with micropore tape and incubate at 30 °C for 10 days under
continuous illumination.
8. Regeneration: Transfer (8 calli/plate) hygromycin-resistant
calli onto pre-regeneration medium containing 50 mg/L
hygromycin and incubate at 30 °C for 10 days. Transfer the
proliferating calli with visible green tissue onto regeneration
medium ­containing 50 mg/L hygromycin. Allow the calli to
grow on regeneration medium for 10 to 15 days or until roots
are 2 mm long.
9. Hydroponics: Regenerated plantlets should be transferred to
Yoshida Culture Solution (YCS) until large enough to plant
into soil. The YCS solution should be changed weekly.

3.5 Plant Verification 1. Extract and normalize the concentration of genomic DNA,
Using Surveyor Assay from both transformed plants and tissue culture controls.
2. PCR should be performed using Phusion® High-Fidelity
DNA Polymerase with primers that flank the target site. One
primer should be designed 200–300 bp away from the target
site, and the second primer 400–500 bp away from the oppo-
site side. This is to ensure that the digested bands can easily be
distinguished from each other (Fig. 4). The primers should
have a high melting temperature (Tm) to avoid the formation
of primer dimmers [77].
3. Take 6 μL of PCR product from each sample and mix with
6 μL of WT PCR product to produce heteroduplex DNA
using the following protocol:
330 Anindya Bandyopadhyay et al.

Temperature Time Temperature ramp

95 °C 10 min
95 °C to 85 °C (−2.0 °C/s)
85 °C 1 min
85 °C to 75 °C (−0.3 °C/s)
75 °C 1 min
75 °C to 65 °C (−0.3 °C/s)
65 °C 1 min
65 °C to 55 °C (−0.3 °C/s)
55 °C 1 min
55 °C to 45 °C (−0.3 °C/s)
45 °C 1 min
45 °C to 35 °C (−0.3 °C/s)
35 °C 1 min
35 °C to 25 °C (−0.3 °C/s)
25 °C 1 min
4 °C Hold ∞

4. Prepare Surveyor Nuclease reaction mixture as shown below.

Incubate the mixture at 42 °C for 60 min.

Component Amount
Hybridized heteroduplex DNA 12 μL
Surveyor Enhancer S 1 μL
Surveyor Nuclease S 2 μL
Total 15 μL

5. Add 1.5 μL of Stop Solution and run the digested product on

a 2% agarose gel. Herteroduplex DNA with mismatched base
pairs will be digested and should render a banding pattern
similar to that shown in Fig. 4, this indicates a Cas9-induced
mutation.
6. Samples that showed digested bands should be amplified by
Phusion® High-Fidelity DNA and PCR products should be
sequenced directly. The sequencing results can be analyzed
using TIDE [78] and Poly Peak Parser [79].
 CRISPR-Cas9-Mediated Genome Editing of Rice Toward Better Grain Quality 331

Fig. 4 The illustration of Surveyor Assay for detection of Cas9-induced mutations. PCR is performed on trans-
formed and tissue culture controls plants (a) and the products are hybridized to form hetroduplex DNA (b). The
surveyor nuclease cuts the mismatched base pairs from hetroduples DNA (c) yielding a distinctive banding
pattern, allowing plants carrying the Cas9-induced mutation to be identified (d)

4 Notes

1. When designing the target sequence, introns and the region

near the 3′ end of the coding sequence should be avoided.
2. Once the 20 nt target sequence has been identified, off targets
should be identified using either Blast search (http://blast.
ncbi.nlm.nih.gov/) or web tools such as Cas Offinder (http://
www.rgenome.net/cas-offinder/, [80]) or CRISPR-P
(http://cbi.hzau.edu.cn/crispr/, [81]).
3. If HDR-based gene insertion is to be performed, a double
stranded DNA donor molecule containing the long homology
arm is required. Single-stranded DNA oligos with a shorter
homology arm can also be used [73, 82].
4. Prior to transformation, a transient expression assay of the
protoplast should be performed in order to validate the activ-
ity of the guide sequence that has been designed. The binary
vector containing the guide sequence and Cas9 can be trans-
332 Anindya Bandyopadhyay et al.

formed into protoplasts, followed by transient expression anal-

ysis and genomic DNA can be extracted from the transformed
protoplast to find the desired mutation. If HDR-mediated
gene insertion is targeted, then a donor vector needs to be co-­
transformed along with the binary vector.
5. Rice transformation and subsequent tissue culture should be
performed in an aseptic environment at a stable temperature.
Proper timing (days after anthesis) of collecting immature
seeds and immature embryos with the correct size is crucial for
the success of the transformation experiments.
6. PCR and restriction enzyme combined assay may also be under-
taken to test for the presence of a mutation. If this is done then the
target locus should include a restriction enzyme site that is destroyed
by the induced mutation. Once the targeted region is amplified
from the transformed plants the mutated amplicon will be resistant
to the restriction digestion but the unmodified amplicon will get
digested. This can be used as an alternative to the Surveyor assay.
7. Instead of using Ubiquitous promoters, germ-line-specific
promoters [83] can be used to drive the expression of Cas9
and the sgRNA. The use of germ-line-specific promoters has
been reported to improve the heritability of the mutations in
the T1 and T2 generations resulting in fewer chimeric plants.

Acknowledgments

The authors thank Dr. Tobias Kretzschmar for insightful com-

ments and discussions. This work has been supported by a grant
from International Rice Research Institute and the Bill and Melinda
Gates foundation.

References

1. Wang H, Pandey S, Velarde O, Hardy B (2012) 6. Juliano B (2003) Rice chemistry and qual-
Patterns of varietal adoption and economics ity. International Rice Research Institute, Los
of rice production in Asia. International Rice Baños, Laguna
Research Institute, Los Baños 7. Philpot K, Martin M, Butardo V, Willoughby
2. Dalton TJ (2004) A household hedonic model D, Fitzgerald M (2006) Environmental factors
of rice traits: economic values from farmers in that affect the ability of amylose to contribute
West Africa. Agric Econ 31:149–159 to retrogradation in gels made from rice flour.
3. Fitzgerald MA, McCouch SR, Hall RD (2009) J Agric Food Chem 54:5182–5190
Not just a grain of rice: the quest for quality. 8. Champagne ET, Bett KL, Vinyard BT,
Trends Plant Sci 14:133–139 McClung AM, Barton FE, Moldenhauer K,
4. Juliano BO (1979) The chemical basis of rice Linscombe S, McKenzie K (1999) Correlation
grain quality. Chemical aspects of rice grain between cooked rice texture and rapid
quality. International Rice Research Institute, visco analyser measurements. Cereal Chem
Los Baños 76:764–771
5. Juliano BO, Villareal CP (1993) Grain qual- 9. Cagampang GB, Perez CM, Juliano BO
ity evaluation of world rices. International Rice (1973) A gel consistency test for eating quality
Research Institute, Los Baños, Laguna of rice. J Sci Food Agric 24:1589–1594
 CRISPR-Cas9-Mediated Genome Editing of Rice Toward Better Grain Quality 333

10. Pandey MK, Rani NS, Madhav MS, Sundaram 23. Peng B, Kong H, Li Y, Wang L, Zhong M, Sun
RM, Varaprasad GS, Sivaranjani AK, Bohra L, Gao G, Zhang Q, Luo L, Wang G, Xie W,
A, Kumar GR, Kumar A (2012) Different Chen J, Yao W, Peng Y, Lei L, Lian X, Xiao J,
isoforms of starch-synthesizing enzymes con- Xu C, Li X, He Y (2014) OsAAP6 functions as
trolling amylose and amylopectin content an important regulator of grain protein content
in rice (Oryza sativa L.). Biotechnol Adv and nutritional quality in rice. Nat Commun
30:1697–1706 5:4847
11. Fitzgerald MA, Sackville Hamilton NR, 24. Fan C, Xing Y, Mao H, Lu T, Han B, Xu C,
Calingacion MN, Verhoeven HA, Butardo VM Li X, Zhang Q (2006) GS3, a major QTL for
(2008) Is there a second fragrance gene in rice? grain length and weight and minor QTL for
Plant Biotechnol J 6:416–423 grain width and thickness in rice, encodes a
12. Champagne ET (2008) Rice aroma and flavor: putative transmembrane protein. Theor Appl
a literature review. Cereal Chem 85:447–456 Genet 112:1164–1171
13. Chen S, Yang Y, Shi W, Ji Q, He F, Zhang Z, 25. Song XJ, Huang W, Shi M, Zhu MZ, Lin HX
Cheng Z, Liu X, Xu M (2008) Badh2, encod- (2007) A QTL for rice grain width and weight
ing betaine aldehyde dehydrogenase, inhib- encodes a previously unknown RING-type E3
its the biosynthesis of 2-acetyl-1-pyrroline, a ubiquitin ligase. Nat Genet 39:623–630
major component in rice fragrance. Plant Cell 26. Weng J, Gu S, Wan X, Gao H, Guo T, Su N,
20:1850–1861 Lei C, Zhang X, Cheng Z, Guo X, Wang J,
14. Huang TC, Teng CS, Chang JL, Chuang HS, Jiang L, Zhai H, Wan J (2008) Isolation and
Ho CT, Wu ML (2008) Biosynthetic mecha- initial characterization of GW5, a major QTL
nism of 2-acetyl-1-pyrroline and its relation- associated with rice grain width and weight.
ship with Δ1-pyrroline-5-carboxylic acid and Cell Res 18:1199–1209
methylglyoxal in aromatic rice (Oryza sativa 27. Kwon SW, Cho YC, Lee JH, Suh JP, Kim
L.) callus. J Agric Food Chem 56:7399–7404 JJ, Kim MK, Choi IS, Hwang HG, Koh HJ,
15. Kovach MJ, Calingacion MN, Fitzgerald MA, Kim YG (2011) Identification of quantitative
McCouch SR (2009) The origin and evolution trait loci associated with rice eating qual-
of fragrance in rice (Oryza sativa L.). Proc Natl ity traits using a population of recombinant
Acad Sci 106:14444–14449 inbred lines derived from a cross between
16. Leung H, Hettel GP, Cantrell RP (2002) two temperate japonica cultivars. Mol Cells
International Rice Research Institute: roles 31:437–445
and challenges as we enter the genomics era. 28. Liu X, Wan X, Ma X, Wan J (2011) Dissecting
Trends Plant Sci 7:139–142 the genetic basis for the effect of rice chalki-
17. Peng S, Khush GS, Virk P, Tang Q, Zou ness, amylose content, protein content, and
Y (2008) Progress in ideotype breeding to rapid viscosity analyzer profile characteristics
increase rice yield potential. Field Crops Res on the eating quality of cooked rice using the
108:32–38 chromosome segment substitution line popu-
lation across eight environments. Genome
18. Lee JW, Hong K (2012) Economic growth in 54:64–80
Asia: determinants and prospects. Japan World
Econ 24:101–113 29. Bradbury LMT, Fitzgerald TL, Henry RJ, Jin
Q, Waters DLE (2005) The gene for fragrance
19. FAO, WFP, IFAD (2012) The State of Food in rice. Plant Biotechnol J 3:363–370
Insecurity in the World 2012. Economic
growth is necessary but not sufficient to accel- 30. Isshiki M, Morino K, Nakajima M, Okagaki
erate reduction of hunger and malnutrition. RJ, Wessler SR, Izawa T, Shimamoto K (1998)
FAO, Rome A naturally occurring functional allele of the
rice waxy locus has a GT to TT mutation at
20. ADB (2012) Food Security and Poverty in Asia the 5′ splice site of the first intron. Plant
Pacific: Key Challenges and Policy Issues. Asian J 15:133–138
Development Bank, Los Baños
31. Wang E, Wang J, Zhu X, Hao W, Wang L,
21. Son JS, Do VB, Kim KO, Cho MS, Suwonsichon Li Q, Zhang L, He W, Lu B, Lin H, Ma
T, Valentin D (2014) Understanding the effect H, Zhang G, He Z (2008) Control of rice
of culture on food representations using word grain-filling and yield by a gene with a poten-
associations: The case of ‘rice’ and ‘good rice. tial signature of domestication. Nat Genet
Food Qual Prefer 31:38–48 40:1370–1374
22. Sreenivasulu N, Butardo VM, Misra G, Cuevas 32. Lucca P, Hurrell R, Potrykus I (2001) Genetic
RP, Anacleto R, Kavi Kishor PB (2015) engineering approaches to improve the bio-
Designing climate-resilient rice with ideal grain availability and the level of iron in rice grains.
quality suited for high-temperature stress. Theor Appl Genet 102:392–397
J Exp Bot 66:1737–1748
334 Anindya Bandyopadhyay et al.

33. Paine JA, Shipton CA, Chaggar S, Howells human pluripotent cells using TALE nucleases.
RM, Kennedy MJ, Vernon G, Wright SY, Nat Biotechnol 29:731–734
Hinchliffe E, Adams JL, Silverstone AL, Drake 44. Reyon D, Tsai SQ, Khayter C, Foden JA,
R (2005) Improving the nutritional value of Sander JD, Joung JK (2012) FLASH assembly
Golden Rice through increased pro-vitamin A of TALENs for high-throughput genome edit-
content. Nat Biotechnol 23:482–487 ing. Nat Biotechnol 30:460–465
34. Lucca P, Poletti S, Sautter C (2006) Genetic 45. Boch J, Scholze H, Schornack S, Landgraf A,
engineering approaches to enrich rice with iron Hahn S, Kay S, Lahaye T, Nickstadt A, Bonas
and vitamin A. Physiol Plant 126:291–303 U (2009) Breaking the code of DNA binding
35. Bhullar NK, Gruissem W (2013) Nutritional specificity of TAL-type III effectors. Science
enhancement of rice for human health: The 326:1509–1512
contribution of biotechnology. Biotechnol Adv 46. Moscou MJ, Bogdanove AJ (2009) A simple
31:50–57 cipher governs DNA recognition by TAL effec-
36. The 3,000 rice genomes project (2014) The tors. Science 326:1501
3,000 rice genomes project. Gigascience 3:7 47. Sanjana NE, Cong L, Zhou Y, Cunniff MM,
37. Gao H, Smith J, Yang M, Jones S, Djukanovic Feng G, Zhang F (2012) A transcription
V, Nicholson MG, West A, Bidney D, Falco activator-­like effector toolbox for genome
SC, Jantz D, Lyznik LA (2010) Heritable tar- engineering. Nat Protoc 7:171–192
geted mutagenesis in maize using a designed 48. Deveau H, Garneau JE, Moineau S (2010)
endonuclease. Plant J 61:176–187 CRISPR/Cas system and its role in phage-­bacteria
38. Miller JC, Holmes MC, Wang J, Guschin interactions. Annu Rev Microbiol 64:475–493
DY, Lee YL, Rupniewski I, Beausejour CM, 49. Horvath P, Barrangou R (2010) CRISPR/Cas,
Waite AJ, Wang NS, Kim KA, Gregory PD, the immune system of bacteria and archaea.
Pabo CO, Rebar EJ (2007) An improved Science 327:167–170
zinc-­finger nuclease architecture for highly 50. Makarova KS, Haft DH, Barrangou R, Brouns
specific genome editing. Nat Biotechnol SJJ, Charpentier E, Horvath P, Moineau S,
25:778–785 Mojica FJ, Wolf YI, Yakunin AF, van der Oost
39. Sander JD, Dahlborg EJ, Goodwin MJ, Cade J, Koonin EV (2011) Evolution and classifi-
L, Zhang F, Cifuentes D, Curtin SJ, Blackburn cation of the CRISPR-Cas systems. Nat Rev
JS, Thibodeau-Beganny S, Qi Y, Pierick CJ, Microbiol 9:467–477
Hoffman E, Maeder ML, Khayter C, Reyon D, 51. Bhaya D, Davison M, Barrangou R (2011)
Dobbs D, Langenau DM, Stupar RM, Giraldez CRISPR-Cas systems in bacteria and archaea:
AJ, Voytas DF, Peterson RT, Yeh JR, Joung versatile small RNAs for adaptive defense and
JK (2011) Selection-free zinc-finger-nuclease regulation. Annu Rev Genet 45:273–297
engineering by context-dependent assembly
(CoDA). Nat Methods 8:67–69 52. Cong L, Ran FA, Cox D, Lin S, Barretto R,
Habib N, Hsu PD, Wu X, Jiang W, Marraffini
40. Wood AJ, Lo T-W, Zeitler B, Pickle CS, Ralston LA, Zhang F (2013) Multiplex genome engi-
EJ, Lee AH, Amora R, Miller JC, Leung neering using CRISPR/Cas systems. Science
E, Meng X, Zhang L, Rebar EJ, Gregory 339:819–823
PD, Urnov FD, Meyer BJ (2011) Targeted
genome editing across species using ZFNs and 53. Mali P, Yang L, Esvelt KM, Aach J, Guell M,
TALENs. Science 333:307 DiCarlo JE, Norville JE, Church GM (2013)
RNA-guided human genome engineering via
41. Christian M, Cermak T, Doyle EL, Schmidt C, Cas9. Science 339:823–826
Zhang F, Hummel A, Bogdanove AJ, Voytas
DF (2010) Targeting DNA double-strand 54. Jinek M, East A, Cheng A, Lin S, Ma E,
breaks with TAL effector nucleases. Genetics Doudna J (2013) RNA-programmed genome
186:757–761 editing in human cells. Elife 2:1–9
42. Zhang F, Cong L, Lodato S, Kosuri S, Church 55. Cho SW, Kim S, Kim JM, Kim JS (2013)
GM, Arlotta P (2011) Efficient construction of Targeted genome engineering in human cells
sequence-specific TAL effectors for modulat- with the Cas9 RNA-guided endonuclease. Nat
ing mammalian transcription. Nat Biotechnol Biotechnol 31:230–232
29:149–153 56. Garneau JE, Dupuis MÈ, Villion M, Romero
43. Hockemeyer D, Wang H, Kiani S, Lai CS, Gao DA, Barrangou R, Boyaval P, Fremaux C,
Q, Cassady JP, Cost GJ, Zhang L, Santiago Horvath P, Magadán AH, Moineau S (2010)
Y, Miller JC, Zeitler B, Cherone JM, Meng The CRISPR/Cas bacterial immune sys-
X, Hinkley SJ, Rebar EJ, Gregory PD, Urnov tem cleaves bacteriophage and plasmid
FD, Jaenisch R (2011) Genetic engineering of DNA. Nature 468:67–71
 CRISPR-Cas9-Mediated Genome Editing of Rice Toward Better Grain Quality 335

57. Mazumdar S, Quick WP, Bandyopadhyay A 69. Yin X, Biswal AK, Dionora J, Perdigon KM,
(2016) CRISPR-Cas9 mediated genome edit- Balahadia CP, Mazumdar S, Chater C, Lin
ing in rice, advancements and future possibili- HC, Coe RA, Kretzschmar T, Gray JE, Quick
ties. Indian J Plant Physiol 21:437–445 WP, Bandyopadhyay A (2017) CRISPR-Cas9
58. Li J, Norville JE, Aach J, McCormack M, and CRISPR-Cpf1 mediated targeting of a sto-
Zhang D, Bush J, Church GM, Sheen J (2013) matal developmental gene EPFL9 in rice. Plant
Multiplex and homologous recombination-­ Cell Rep 36:1–13
mediated genome editing in Arabidopsis and 70. Xie K, Yang Y (2013) RNA-Guided genome
Nicotiana benthamiana using guide RNA and editing in plants using a CRISPR-Cas system.
Cas9. Nat Biotechnol 31:688–691 Mol Plant 6:1975–1983
59. Nekrasov V, Staskawicz B, Weigel D, Jones JD, 71. Kumar V, Jain M (2015) The CRISPR-Cas
Kamoun S (2013) Targeted mutagenesis in system for plant genome editing: advances and
the model plant Nicotiana benthamiana using opportunities. J Exp Bot 66:47–57
Cas9-guided endonuclease. Nat Biotechnol 72. Fitzgerald MA (2004) Starch. In: Champagne
31:691–693 ET (ed) Rice Chemistry and Technology, 3rd
60. Shan Q, Wang Y, Li J, Zhang Y, Chen K, Liang edn. American Association of Cereal Chemists,
Z, Zhang K, Liu J, Xi JJ, Qiu JL, Gao C (2013) Eagan, pp 109–133
Targeted genome modification of crop plants 73. Shan Q, Wang Y, Li J, Gao C (2014) Genome
using a CRISPR–Cas system. Nat Biotechnol editing in rice and wheat using the CRISPR/
31:686–688 Cas system. Nat Protoc 9:2395–2410
61. Belhaj K, Chaparro-Garcia A, Kamoun S, 74. Chen K, Gao C (2015) Developing CRISPR
Nekrasov V (2013) Plant genome editing technology in major crop plants. Adv New
made easy: targeted mutagenesis in model and Technol Target Modif Plant Genomes
crop plants using the CRISPR/Cas system. 145–159
Plant Methods 9:39 75. Hiei Y, Komari T (2008) Agrobacterium-­
62. Feng Z, Zhang B, Ding W, Liu X, Yang mediated transformation of rice using imma-
DL, Wei P, Cao F, Zhu S, Zhang F, Mao Y, ture embryos or calli induced from mature
Zhu JK (2013) Efficient genome editing in seed. Nat Protoc 3:824–834
plants using a CRISPR/Cas system. Cell Res 76. Slamet-Loedin IH, Chadha-Mohanty P,
23:1229–1232 Torrizo L (2014) Agrobacterium-mediated
63. Jiang W, Zhou H, Bi H, Fromm M, Yang B, transformation: rice transformation. Methods
Weeks DP (2013) Demonstration of CRISPR/ Mol Biol 1099:261–271
Cas9/sgRNA-mediated targeted gene modifi- 77. Ran FA, Hsu PDP, Wright J, Agarwala V, Scott
cation in Arabidopsis, tobacco, sorghum and DA, Zhang F (2013) Genome engineering
rice. Nucleic Acids Res 41:e188 using the CRISPR-Cas9 system. Nat Protoc
64. Mao Y, Zhang H, Xu N, Zhang B, Gou F, Zhu 8:2281–2308
JK (2013) Application of the CRISPR-Cas sys- 78. Brinkman EK, Chen T, Amendola M, van
tem for efficient genome engineering in plants. Steensel B (2014) Easy quantitative assessment
Mol Plant 6:2008–2011 of genome editing by sequence trace decompo-
65. Miao J, Guo D, Zhang J, Huang Q, Qin G, sition. Nucleic Acids Res 42(22):e168
Zhang X, Wan J, Gu H, Qu LJ (2013) Targeted 79. Hill JT, Demarest BL, Bisgrove BW, Su YC,
mutagenesis in rice using CRISPR–Cas system. Smith M, Yost HJ (2014) Poly peak parser:
Cell Res 23:1233–1236 method and software for identification of
66. Sugano SS, Shirakawa M, Takagi J, Matsuda unknown indels using sanger sequencing of
Y, Shimada T, Hara-Nishimura I, Kohchi T polymerase chain reaction products. Dev Dyn
(2014) CRISPR/Cas9-mediated targeted 243(12):1632–1636
mutagenesis in the liverwort Marchantia 80. Bae S, Park J, Kim J (2014) Cas-OFFinder: a fast
polymorpha L. Plant Cell Physiol and versatile algorithm that searches for poten-
55:475–481 tial off-target sites of Cas9 RNA-guided endo-
67. Jiang H, Wang N (2014) Targeted genome nucleases. Bioinformatics 30(10):1473–1475
editing of sweet orange using Cas9/ 81. Lei Y, Lu L, Liu H, Li S, Xing F, Chen L
sgRNA. PLoS One 9:e93806 (2014) CRISPR-P: a web tool for synthetic
68. Wang Y, Cheng X, Shan Q, Zhang Y, Liu J, single-guide RNA design of CRISPR-system in
Gao C, Qiu JL (2014) Simultaneous editing of plants. Mol Plant 7(9):1494–1496
three homoeoalleles in hexaploid bread wheat 82. Strouse B, Bialk P, Niamat RA, Rivera-Torres
confers heritable resistance to powdery mildew. N, Kmiec EB (2014) Combinatorial gene edit-
Nat Biotechnol 32:947–951
336 Anindya Bandyopadhyay et al.

ing in mammalian cells using ssODNs and line-specific CRISPR-Cas9 systems to improve
TALENs. Sci Rep 4:3791 the production of heritable gene modifications
83. Mao Y, Zhang Z, Feng Z, Wei P, Zhang H, in Arabidopsis. Plant Biotechnol J 14:519–532.
Botella JR, Zhu JK (2016) Development of germ- https://doi.org/10.1111/pbi.12468
 Index

A G
Amplicon sequencing������������������������������� 202, 203, 212, 218 Gelatinization temperature (GT)���������������������vii, 21, 23–24,
Amylose content (AC)�������������������������20–23, 25, 26, 31, 32, 26, 27, 32, 35, 38, 109, 111, 126, 137, 147, 148, 162
35, 37, 44, 76, 109, 111, 126, 137, 139, 140, 143, Genome��������������������������� 12, 45, 77, 138, 201, 277, 302, 312
147–149, 151, 159, 162, 163, 170, 251, 311, 312, 314 Genomics�������������� vii, 7, 44–46, 202–204, 208–212, 219–222,
Appearance�������������������������������� 6, 8, 10, 12, 13, 28, 302, 312 236, 238, 278, 301, 303–306, 308, 312, 329, 332
Aroma��������������������������������� vii, 25, 28, 32–35, 188, 311, 312 Genotyping���������������������������������45, 110, 138, 203, 222–232
Association analysis�������������������������������������������������� 222, 233 Grain�����������������������1, 19, 59, 75, 89, 99, 109, 137, 151, 170,
188, 201, 241, 265, 279, 302, 311
B Grain elongation����������������������������������������������������������26, 99
Breeding��������������������������vii, 2, 11–14, 19, 22–24, 32–35, 39, Grain quality������������������������������ vii, 7–10, 12, 19–46, 63, 75,
44–46, 75–77, 90, 94, 95, 109, 111, 138, 151, 152, 90, 99, 100, 109, 114, 137, 138, 148, 159, 162,
188, 207, 219, 242, 279, 304 187–197, 201–239, 266, 306, 318
Grain yield���������������������������������������������������7, 63, 76, 77, 111
C Graphite Furnace Atomic Absorption
Spectrometry (GF-AAS)�������266, 268, 269, 271, 273
Cadmium in rice�������������������������������������������������������265–274
Growth staging system�������������������������������������������������65, 67
cDNA synthesis������������������������������������������������ 280, 282–287
Chain-length distribution (CLD)������ 30, 170–172, 179–182 H
Chalk������������������������������������������������������������ 5–7, 10–14, 311
Chalkiness�������������������������2, 5–7, 12, 13, 76, 86, 90, 99–107, Head rice recovery (HRR)����������������������� 12, 13, 89–97, 100
126, 279, 312 Head rice yield (HRY )��������������������������������������� vii, 1–14, 63
Cooking�����������������������������������vii, 1, 6, 20–24, 26, 27, 30, 31, Heavy metals������������������������������������������������������ vii, 265, 266
33, 34, 42, 44–46, 101, 104, 109, 111, 126, 137, Hybridization����������������������������278–280, 283, 288–293, 304
152–155, 159, 161, 163, 165, 201, 311
I
cRNA labeling�������������������������������������������������� 280, 282–287
Image analyses�������������������������������������������������6, 8, 10, 77, 99
D Inductively coupled plasma-optical emission spectrometry
Dehulling���������������������������������������� 2, 3, 5, 10, 90, 93, 94, 96 (ICP-OES)��������������������������������������������� 41, 253–263
Digestibility������������������������������������ vii, 22, 29, 35, 39, 43, 44, Inferior grains��������������������������������������������������������� 63, 72, 76
169, 170, 183
M
DNA methylation������������������������������������������������������������308
Drying�������������������������2–5, 63, 89–91, 93–96, 100, 138, 158, Macronutrients��������������������������������������������� 35, 39, 253–263
197, 208, 269, 296 Metabolite profile analyses����������������������� 187, 189, 190, 195
Metabolites����������������������������������������������vii, 39, 43, 187–197
E Methylation sensitive restriction enzymes
Eating��������������������������������vii, 20, 21, 24, 26, 31–34, 45, 147, (MSRE)��������������������������������������������������������304–309
148, 151, 152, 161, 163, 311 Microarray�������������������������������������������������������� 277–299, 303
Epigenetics����������������������������������������������������� vii, 7, 301–303 Micronutrients��������������������� vii, 29, 35, 40–41, 188, 253–263
Milling�������������������������� vii, 1, 8, 9, 20, 21, 25, 89, 90, 94, 97,
F 100, 149, 164, 175
Milling quality����������������������������������� vii, 1, 8, 20, 89, 90, 100
Fissures��������������������������������������������������������� 3–5, 7, 9–13, 63
Multivariate analyses������������������������������������������ 42, 115, 196

Nese Sreenivasulu (ed.), Rice Grain Quality: Methods and Protocols, Methods in Molecular Biology, vol. 1892,
https://doi.org/10.1007/978-1-4939-8914-0, © Springer Science+Business Media, LLC, part of Springer Nature 2019

337
 Rice Grain Quality: Methods and Protocols
338 Index
  
N Rice grain����������������������������� 1, 20, 63, 89, 99, 112, 137, 152,
172, 188, 202, 241, 266, 280
Near-infrared spectroscopy (NIRS)��������������9, 25, 26, 42, 43, Rice quality������������������������������ 2, 7, 25, 32, 34, 121, 188, 201
109–133 RNA extraction��������������������������������������������������������279–281
Nitric-perchloric digestion��������������������������������������� 255, 262
Nutrition��������������������22, 45, 46, 90, 100, 114, 188, 253, 266 S
O Sampling������������������������� 67, 72, 96, 112, 194, 249, 250, 305
Seed����������������������� vii, 2, 4, 11, 13, 57–73, 76, 100, 103, 187,
Oryza sativa L.�����������������������������������������������������������������328 188, 204, 219, 301, 302, 328, 332
Seed development������������������������������������������������� vii, 57, 302
P
Sensory��������������������������������� 6, 20, 21, 24, 27, 28, 31–34, 39,
Panicle architecture������������������������������������������ vii, 12, 75–87 42–45, 152
Physical�����������������������������������vii, 7, 8, 10, 11, 13, 20, 45, 99, Single-nucleotide polymorphisms (SNPs)�����37, 38, 45, 201,
101–103, 111, 126, 311 203, 212, 218–220, 222–235, 239, 314
Physical traits����������������������������������������������� vii, 99, 101–103 Size and shape�����������������������������������������������7, 12, 40, 83, 94
Protein content (PC)�������������������������� 25, 28, 32, 39, 76, 111, Size-exclusion chromatography (SEC)����������������������� 25, 26,
119, 125–129, 162, 312 30, 41, 170, 172–174, 176–182
Slide scanning��������������������������������������������������� 288, 295–297
Q Starch������������������������4, 21, 57, 111, 142, 165, 169, 187, 201,
Quality evaluation������������������������������������������������ 7, 121, 137 241, 280, 302, 314
Quantitative real-time PCR (qPCR)��������������� 304, 306, 307 Starch structure�������������vii, 25, 41, 42, 44, 45, 170, 171, 174,
177–178
R Superior grains�������������������20, 34, 45, 46, 63, 67, 72, 76, 312

Rapid Visco-Analyser (RVA)�������������������23, 27, 30–32, 137, T

141, 146, 147, 149
Reproductive development������������������������������� 59, 64, 68, 70 Texture������������������������vii, 24–27, 31, 32, 42, 43, 45, 46, 111,
Re-sequencing����������������������������������������������������������201–239 126, 151, 312, 314
Resistant starch (RS)��������� vii, 29, 39, 44, 169, 241–245, 251 Texture profile analyses (TPA)27, 32, 42, 151, 152, 154–157,
Rheometry����������������������������� 25, 44, 151, 152, 154, 156–159 159, 161–163, 165
Rice�������������������������1, 19, 57, 75, 89, 99, 137, 151, 169, 188, Timing����������������������������������������������������������������� 64, 72, 332
201, 241, 265, 278, 311
V
Rice breeding�������������������vii, 2, 14, 19, 22–24, 33, 39, 44–46,
75, 109, 242 Viscoelastic properties������������������������������������ 5, 31, 152, 159