Zhang et al.: Journal of AOAC International Vol. 100, No.

1, 2017  133

FOOD COMPOSITION AND ADDITIVES

Determination of Bovine Lactoferrin in Food by HPLC with a

Heparin Affinity Column for Sample Preparation
Yin Zhang1
Denuo Product Testing Co., Ltd, Shanghai, 1550 Jiang Chang Xi Rd, Jingan District, Shanghai, 200436 China
Fei Lou1
Agri-Products Quality and Safety Testing Center of Shanghai, No. 28, Ln 1528, Xinfu Zhong Rd, Huaxin Town, Qinpu District,
Shanghai, 201708 China
Wei Wu, Xin Dong, and Jia Ren
Denuo Product Testing Co., Ltd, Shanghai, 1550 Jiang Chang Xi Rd, Jingan District, Shanghai, 200436 China
Qiuguang Shen
Agri-Products Quality and Safety Testing Center of Shanghai, No. 28, Ln 1528, Xinfu Zhong Rd, Huaxin Town, Qinpu District,
Shanghai, 201708 China

An HPLC method was developed for the quantitative bLF in dairy or other food products. Various methods have been
determination of bovine lactoferrin (bLF) in sterilized developed for bLF determination, including ELISA (11), sodium
milk, modified milk, fermented milk, infant formula, dodecyl sulfate (SDS) polyacrylamide gel electrophoresis (PAGE),
adult formula, rice cereal, vitamin function drink, and capillary electrophoresis (CE; 12, 13), HPLC (14–16), and
protein powder products. bLF was first extracted LC-MS (17). ELISA is widely used because of its good sensitivity
with a phosphate buffer (pH 8), underwent cleanup and universality, however, it has poor reproducibility and detection
in a heparin affinity column, and was detected by is expensive. SDS-PAGE requires laborious sample pretreatment,
HPLC with a C4 column and diode-array detector sophisticated and time-consuming procedures, and large amounts
at a wavelength of 280 nm. The proposed method of sample, and is not very stabile or sensitive. CE is widely used
provided a linear detection range of 10.0–1000 μg/mL in protein analysis, however, it is limited by capillary, voltage,
with an LOD of 0.6 mg/100 g in liquid samples and other conditions, which may not be sensitive enough for the
and 3 mg/100 g in solid samples and an LOQ of separation of proteins with small differences in molecular weight.
2 mg/100 g in liquid samples and 10 mg/100 g in In addition, CE has poor reproducibility for the high-throughput
solid samples. In addition, the method showed analysis of bLF. LC-MS is highly sensitive and selective, but
good recovery for various samples, ranging from cannot be widely applied for the routine detection of bLF in product
76 to 96%. The method had several remarkable quality control (QC) because the costs of detection are prohibitive.
advantages, including ease of handling, high In our previous patented method for bLF detection in dairy
sensitivity and accuracy, good reproducibility, products based on an HPLC method (18), our experimental results
and low-cost detection. Based on the distinctive demonstrated that the proposed HPLC method was not robust as a
properties presented here, we believe the proposed standard method for the daily QC testing of bLF. With authorization
HPLC assay holds great promise for the oversight from the Ministry of Health and Family Planning Commission of
and detection of bLF in testing organizations, dairy China, we further improved the HPLC method to develop a simple,
enterprises, and regulatory authorities. universal, and robust method for the accurate determination of bLF
in various products during QC, governmental quality supervision,
import–export product testing, and risk assessment. The major

B
ovine lactoferrin (bLF) is a natural protein that is widely challenge in developing this method was the separation, and
found in cow’s milk. It is an iron-binding glycoprotein with accurate testing, of a low-concentration target protein from other
a single polypeptide chain of 689 amino acids and molecular dominant proteins in dairy or other food products. In this study,
weight of 77 kDa. Recent studies have provided increasing evidence we used a heparin affinity column to efficiently enrich bLF from
that bLF not only has physiological functions as an antimicrobial a complex sample matrix, used a C4 column for the analytical
and/or antiviral, immunomodulatory agent, and antioxidant, but column, and optimized detection conditions for the HPLC method.
can also pharmacologically suppress tumor metastasis (1–7). The The proposed method was successfully applied for the quantitative
European Union has approved bLF as a new type of food ingredient analysis of bLF in eight products: sterilized milk, modified milk,
(8), whereas Chinese governmental authorities have specified that fermented milk, infant formula, infant rice cereal, adult formula,
the bLF concentration limit in dairy products should be 1.0 g/kg protein powder, and vitamin function drink samples.
and that bLF dosage in infant formula is 1.0 g/L (9, 10). However,
at present, there is no standard method for the determination of
Experimental
Received August 11, 2016. Accepted by SG October 3, 2016.
Corresponding author’s e-mail: [email protected] Reagents and Materials
This work was supported by the National Health and Family
Planning Commission of China.
1
These authors contributed equally to this work. bLF was donated by Tatua Co-operative Dairy Co.
DOI: 10.5740/jaoacint.16-0259 (Batch No. SP113101). Cellulose membrane (0.45 μm) was
134  Zhang et al.: Journal of AOAC International Vol. 100, No. 1, 2017

purchased from ANPEL (ANPEL Scientific Instrument, Table 1.  Optimization of gradient elution
Shanghai, China). All chemical reagents were AR grade
Time, min Line A, % Line B, % Flow rate, mL/min
and purchased from Sinopharm Chemical Reagent Co., Ltd
(Shanghai, China). HPLC grade acetonitrile was purchased 0 30 70 1.0
from Merck (Darmstadt, Germany). All mobile phase and 5 55 45 1.0
standard solutions were prepared in water treated with a 10 60 40 1.0
Nanopure water system (Thermo Fisher Scientific Co., Ltd, 12 30 70 1.0
Dubuque, IA). A HiTrap Heparin HP column (1 mL) was
16 30 70 1.0
purchased from GE Healthcare Life Sciences (Uppsala,
Sweden). Solutions were as follows: buffer A was composed
of 0.2 mol/L Na2HPO4 (pH 8) and buffer B of 0.05 mol/L Results and Discussion
Na2HPO4 and 1.0 mol/L NaCl (pH 8).
Optimization of Chromatographic Conditions
Instrumentation
A detection wavelength of 280 nm was selected based on the
maximum absorption of bLF. In our patented HPLC method
All experiments were carried out on an Agilent HPLC 1200
(18), C18 was used as the analytical column. Our experimental
series instrument (Agilent Technologies, Waldbronn, Germany)
results showed that the baseline was unstable and retention time
equipped with a diode-array detector (DAD) and ChemStation
(RT) of bLf changed after 20 injections. Because these problems
Software. A C4 analytical column (250 × 4.6 mm, 5 μm) was
may have resulted from a high salt concentration in the sample
purchased from ANPEL and a C18 guard column (4.6 × 50 mm,
solution, we tried using the C4 column as the analytical column
5  μm) was purchased from Waters (Waters Technologies,
instead of the C18 column to potentially provide more stability
Ireland). The experimental centrifuge used was an MR23i
and higher separation efficiency. As shown in Figure 1, the
purchased from Thermo Fisher (Thermo Fisher Scientific,
gradient elution was optimized to achieve the best peak shape
Nantes, France).
and minimum amount of interference from the sample matrix.

Optimization of Sample Pretreatment

Methods

The food sample matrix was very complex, containing various

(a)  Sample extraction.—(1)  Liquid sample.—Liquid sample
chemicals and biomaterials, such as protein, carbohydrates,
(5 g) was transferred into a 50 mL centrifuge tube. Buffer A
vitamins, minerals, and additional supplementary materials. To
solution (5 mL) was added and the contents mixed on a vortex
accurately determine the amount of bLF in the food samples, an
mixer for 0.5 min and centrifuged for 10 min at 10 000 rpm at
appropriate sample pretreatment method had to be established
4°C. The supernatant was transferred to a new 50 mL centrifuge
to eliminate any interference from the sample matrix and to
tube. Buffer A solution (5  mL) was added  to the first 50  mL
preconcentrate the bLF analyte.
centrifuge tube and the above process repeated to collect
Affinity chromatography, based on the reaction between
supernatant. The two supernatants were then merged to become
the target protein molecule and the ligand on the column, is
the sample extraction solution.
an effective protein purification method. The target protein is
(2)  Solid sample.—Accurately weighed solid sample (1 g)
retained on the column as other proteins are eluted. Heparin
was transferred into a 50 mL centrifuge tube. Buffer A solution
is an acidic, negative-charged polysaccharide with a functional
(10 mL) was added and the content mixed on a vortex mixer for
group of sulfate, which has been used to purify bLF by affinity
0.5 min, followed by the same procedures described above for
chromatography (19–22). Thus, we used a HiTrap Heparin
the liquid sample.
column to clean up the sample matrix as shown in the Figure 2.
(b)  Sample cleanup.—The heparin affinity column was
rinsed with 5 mL buffer A solution at a rate of 1.0 mL/min.
The column was flushed out with sample extraction solution,
washed with 10 mL buffer A solution, and then eluted with
3.0  mL buffer B solution. The effluent was passed through a
0.45 μm cellulose membrane prior to injection into the HPLC
system. Preparation of the bLF standard solution for quantitative
analysis should follow the same procedures noted above for the
heparin affinity column.
(c)  LC conditions.—A C4 column was used as the analytical
column, coupled with a C18 guard column. The column
temperature was set to 30°C, the DAD was set at 280 nm, and the
injection volume was set at 50 μL, with a flow rate of 1.0 mL/min.
Line A in the mobile phase was pure CH3CN, whereas line B
was 0.1% (v/v) trifluoride acetic acid in water. Gradients were
set as shown in Table 1. The LOD was calculated as 3× the SD of
Figure 1.  HPLC chromatograms of bLF. (a) Standard
the noise level. All raw data were processed with ChemStation chromatography and (b) infant formula sample with cleanup. Peaks
Software. 1 is bLF. For conditions, see Experimental section.
 Zhang et al.: Journal of AOAC International Vol. 100, No. 1, 2017  135

Figure 2.  HPLC Chromatogram of bLF with and without affinity

column cleanup. (a) Infant formula sample without cleanup
and (b) infant formula sample with cleanup. (Peak 1) bLF with
interference and (peak 2) bLF. The RT for peaks 1 and 2 was 6.1 min. Figure 4.  Chromatogram of bLF in water baths at different
Chromatogram a (without cleanup) corresponds to the y-axis on the temperatures for 1 h: (a) 90°C, (b) 80°C, (c) 70°C, (d) 60°C, (e) 50°C,
right and chromatogram b (with cleanup) to the y-axis on the left. (f) 40°C, (g) 30°C, and (h) 20°C. Peak 1 is bLF at an RT of 6.1 min. For
For conditions, see Experimental section. conditions, see Experimental section.

The infant formula sample matrix could be effectively removed different temperatures—20, 30, 40, 50, 60, 70, 80, and 90°C—and
after treatment of the HiTrap Heparin affinity column. then injected directly into the HPLC system. As shown in
Solvable proteins can usually be extracted with water or the Figure 4, higher temperatures resulted in the denaturation
ingredient-specific buffer. Buffer concentration, pH, and of bLF, and the chromatographic characteristics of bLF were
environmental temperature are important factors to consider in significantly changed. In addition, the denatured bLF could not
the extraction efficiency of a solvable protein. It should be noted be retained on the HiTrap Heparin column and, thus, could not
that the biological activity of bLF should be maintained during be detected during HPLC. Therefore, for convenience, room
sample pretreatment. In this study, spiked bLF in powdered temperature was selected as the optimum temperature.
milk was extracted with water and different concentrations of Finally, ion strength and elution volume were optimized to
phosphate buffer. As shown in Figure 3, 200 mM phosphate elute the target analyte from the affinity column. As shown in
buffer almost fully extracted bLF from milk powder, with a Figure 5, a higher concentration of NaCl in the buffer would
recovery of 96.0%. The extraction efficiencies of the phosphate improve elution efficiency, therefore, 1.0 M NaCl would almost
buffer at different pH (5–9) were almost the same. In this study, fully elute bLF in milk powder. The elution curve in Figure 6
pH 8 was selected for subsequent experiments. shows that 3.0  mL effluent was enough to elute bLF from a
As a protein, bLF is denatured at high temperature. It has 1 mL affinity column.
been reported that denaturation of bLF begins at 40°C, becomes
rapidly denatured at 70°C, and then 95% bLF is denatured Analytical Parameters and Validation
at 85°C (23). Although higher temperatures would improve
extraction efficiency, at high temperatures, protein denaturation Specificity.—Specificity was investigated by comparing the
significantly changes bLF chromatography and results in chromatograms of eight different kinds of samples spiked with
deviations in the test results. In this study, 1 mL of 200 μg/mL bLF or bLF standard. As shown in Figure 7, the bLF standard
standard solution in Eppendorf PCR tubes was extracted at and spiked samples exhibited sharp and symmetric peaks at the

Figure 3.  Recovery of bLF by applying different sample extraction

solutions. Error bars were calculated from six independent Figure 5.  Recovery of bLF by applying different concentrations of
experiments. NaCl. Error bars were calculated from six independent experiments.
136  Zhang et al.: Journal of AOAC International Vol. 100, No. 1, 2017

samples and 3 mg/100 g in solid samples, whereas the LOQ was

2 mg/100 g in liquid samples and 10 mg/100 g in solid samples.

Interlaboratory Validation Test

Five laboratories in China joined in the interlaboratory

validation of this study, including Shanghai Municipal
Supervisory Institute of Veterinary Drugs and Feedstuff
(Shanghai, China); Shanghai Municipal Center for Disease
Control and Prevention (Shanghai, China); Shanghai Entry-
Exit Inspection and Quarantine Bureau (Shanghai, China);
Technology Center Bright Dairy and Food Co., Ltd (Shanghai,
China); and Institute for Agri-Food Standards and Testing
Technology (Shanghai, China). Three samples were delivered
Figure 6.  Elution volume optimization The total volume is 4 mL. to each laboratory, and the same experimental procedure was
The 1st is the first 1mL, the 2nd is the second 1 mL, the 3rd is the
third 1 mL, the 4th is the forth 1mL. followed. The results are listed in Table 4.

same RT. In a sample without bLF, no peak was observed at the

specific RT. These results indicated that the bLF analyte was Table 2.  Recovery test results in the eight different kinds
of samples (n = 5)
obtained specifically from the pretreated sample.
Linearity.—Linearity was demonstrated by running standards Spiking levels,
over a range of 10.0–1000  μg/mL. Linearity was checked at Matrix mg/100 g Mean recovery, % RSD, %
regular intervals. The linear correlation coefficient (r) for all Sterilized milk 2 82 9.1
standard curves had to be ≥0.999. 5 87 4.1
Accuracy.—Accuracy of the method was determined by
10 88 2.4
overspike recoveries at approximately 100% of the fortified
levels or at levels within the analytical range of the method if 20 91 2.9
the product did not contain inherent levels of or had not been Modified milk 2 85 7.7
fortified with analytes. The products were spiked with bLF 5 86 3.0
for 3 days (Table 2). On each day, spiked solutions containing
10 90 4.3
bLF were added to ready-to-feed or reconstituted powders and
mixed well. Each spiked product was prepared and analyzed in 20 90 4.5
duplicate. Fermented milk 2 83 9.7
Precision.—Precision was determined as intermediate 5 86 5.6
precision, which is defined as the agreement between the
10 85 6.6
test results obtained using the same method on identical test
material in the same laboratory by the same operators using the 20 89 4.3

same equipment over a period of several days. It was expressed Vitamin function 2 90 2.0
in percentage RSD (Table 3). drink
5 96 1.2
LOD and LOQ.—The LOD and LOQ were determined by 10 96 2.8
calculating the concentrations corresponding to 3 and 10 times
20 95 1.7
the S/N, respectively. The LOD was 0.6 mg/100 g in liquid
Infant formula 10 88 6.2
30 92 1.7
50 90 3.5
100 94 4.1
Infant rice cereal 10 79 8.9
30 81 6.0
50 86 4.5
100 90 3.1
Adult formula 10 81 9.6
30 84 8.7
50 86 4.4
100 87 3.9
Protein powder 10 94 1.6
Figure 7.  HPLC chromatogram of bLF in the eight different 30 91 2.5
matrixes: (a) sterilized milk, (b) modified milk, (c) fermented milk,
(d) vitamin function drink, (e) infant formula, (f) infant rice cereal, 50 94 1.8
(g) adult formula, and (h) protein powder. Peak 1 is bLF at an RT of 100 95 2.6
6.2 min. For conditions, see Experimental section.
 Zhang et al.: Journal of AOAC International Vol. 100, No. 1, 2017  137

Table 3.  Results of the spike test in the eight different Table 5.  Comparison of the proposed HPLC method and
kinds of samples (n = 12) the LC-MS/MS method

Spiking levels, LC-MS/MS Amount

Matrix mg/100g Mean recovery, % RSD, % Mix process HPLC method method added
Matrix No. technology mg/100 g mg/100 g mg/100 g
Sterilized milk 10 89 3.6
1 Dry-mix 37.3 61.7 30–50
Modified milk 10 92 5.4
2 Dry-mix 36.8 56.1 30–50
Infant formula 10 88 4.8
3 Dry-mix 420 507 400–500
Vitamin function
drink 10 96 2.1 4 Dry-mix 490 491 400–500
Protein powder 10 93 3.3 5 Wet-mix 6.82 184 200
Adult formula 10 82 7.7 6 Wet-mix 7.57 200 200
Infant rice cereal 10 80 6.7 7 Wet-mix 10.0 230 200
Fermented milk 10 84 6.1 8 Wet-mix 7.23 246 200

Comparison of LC and LC-Tandem MS (MS/MS) extraction and eluent conditions. bLF was effectively extracted
Methods from samples with 200 mM phosphate buffer at pH 8.0, retained
by the heparin affinity column, and eluted with a phosphate
buffer and 1.0 M NaCl. All analytical parameters of this method
As shown in Table 5, there was a substantial deviation
complied with regulatory requirements (25). It should be pointed
between the proposed HPLC and LC-MS/MS methods (17).
To test for bLF in infant formula products that were produced out that once bLF was thermally denatured, its chromatographic
in a dry-mix process (24), the testing results of the proposed characteristics were significantly changed and, thus, bLF could
HPLC method fell into the added-bLF range, whereas higher not be retained by the heparin affinity column. This method can
results were achieved with the LC-MS/MS method. However, accurately detect natural bLF because thermally denatured bLF
the proposed method had very low results when infant formula is washed out as sample matrix.
products were produced with a wet-mix process. During a wet-
mix process, bLF is exposed to high temperatures and, thus, Acknowledgments
the majority of bLF would be denatured in the final product,
subsequently failing to be retained on the affinity column and We thank the Shanghai Municipal Supervisory Institute of
thereby not detectable. Because the LC-MS/MS method was Veterinary Drugs and Feedstuff; Shanghai Municipal Center for
based on a signature peptide and multiple reaction monitoring Disease Control and Prevention; Shanghai Entry-Exit Inspection
mode, the method could not distinguish natural bLF from and Quarantine Bureau; Technology Center Bright Dairy and
denatured bLF, therefore, denatured bLF was included in the Food Co., Ltd; and Institute for Agri-Food Standards and
final results. The LC method provided accurate natural bLF Testing Technology for their assistance with the interlaboratory
results, whereas the LC-MS/MS results included both natural validation experiment.
and denatured bLF.
References
Conclusions
(1) Levay, P.F., & Viljoen, M. (1995) Haemtologica 80, 252–267
In this study, an accurate and robust HPLC method was (2) Jenssen, H., & Hancock, R.E.W. (2009) Biochimie 91, 19–29.
developed for the determination of bLF in eight different doi:10.1016/j.biochi.2008.05.015
(3) van der Kraan, M.I.A., Jasper, G., Kamran, N., Veerman, E.C.I.,
kinds of products. The key points in this method were sample
Bolscher, J.G.M., & Amerongen, A.V.N. (2004) Peptides 25,
cleanup with a heparin affinity column and the optimization of
177–183. doi:10.1016/j.peptides.2003.12.006
(4) Wakabayashi, H., Yamauchi, K., & Takase, M. (2006) Int. Dairy
J. 16, 1241–1251. doi:10.1016/j.idairyj.2006.06.013
Table 4.  Comparison of real samples in the five laboratories (5) González-Chávez, S.A., Sigifredo, A.G., & Quintín, R.C. (2009)
(n = 3) Int. J. Antimicrob. Agents 33, 301.e1–301.e8. doi:10.1016/j.
Matrix ijantimicag.2008.07.020
(6) Redwan, E.M., Uversky, V.N., El-Fakharany, E.M., &
Infant formula 1 Infant formula 2 Modified milk Al-Mehdar, H. (2014) C. R. Biol. 337, 581–595. doi:10.1016/j.
Detected Detected Detected crvi.2014.08.003
content, content, content, (7) Tomita, M., Wakabayashi, H., Shin, K., Yamauchi, K.,
Lab No. mg/100 g RSD, % mg/100 g RSD, % mg/100 g RSD, % Yaeshima, T., & Iwatsuki, K. (2009) Biochimie 91, 52–57.
1 30.9 2.5 31.8 1.7 5.3 2.9
doi:10.1016/j.biochi.2008.05.021
(8) Commission Implementing Decision 2015/568 (2015) Off. J.
2 30.9 2.6 32.1 2.1 5.4 3.7
Eur. Union L93, 71–71.
3 30.6 3.7 32.5 2.2 5.2 3.8 (9) Chinese National Standards (2012) Standard for the Use of
4 31.7 2.7 32.5 2.2 5.2 3.8 Nutritional Fortification Substances in Foods, GB 14880-2012
(10) National Health and Family Planning Commission of the PRC
5 31.5 1.8 32.3 2.1 5.4 3.2
(2013) Announcement on approval expanding the dosage scope
138  Zhang et al.: Journal of AOAC International Vol. 100, No. 1, 2017

of eight kinds of food additives such as ascorbic acid sodium, (18) Meng, J., Zheng, X., Han, H., Wang, X., Ma, Y. & He, Y. (2012)
No. 11, National Health and Family Planning Commission of Chinese patent No. CN201210030215.9
the PRC Bulletin, Beijing, China (19) van Veen, H.A., Geerts, M.E.J., van Berkel, P.H.C., &
(11) Dupont, D., Arnould, C., Rolet-Repecaud, O., Duboz, G., Nuijens, J.H. (2002) Anal. Biochem. 309, 60–66.
Faurie, F., Martin, B., & Beuvier, E. (2006) Int. Dairy J. 16, doi:10.1016/S0003-2697(02)00273-7
1081–1087. doi:10.1016/j.idairyj.2005.09.012 (20) Noppe, W., Plieva, F.M., Galaev, I.Y., Vanhoorelbeke, K.,
(12) Li, J., Ding, X., Chen, Y., Song, B., Shan, Z., & Zhi, W. Bo, M., & Deckmyn, H. (2006) J. Chromatogr. A 1101, 79–85.
(2012) J. Chromatogr. A 1244, 178–183. doi:10.1016/j. doi:10.1016/j.chroma.2005.09.064
chroma.2012.05.004 (21) Caldwell, E.E.O., Nadkarni, V.D., Fromm, J.R., Linhardt, R.J., &
(13) Riechel, P., Weiss, T., Weiss, M., Ulber, R., Buchholz, H., & Weiler, J.M. (1996) Int. J. Biochem. Cell Biol. 28, 203–216.
Scheper, T. (1998) J. Chromatogr. A 817, 187–193. doi:10.1016/ doi:10.1016/1357-2725(95)00123-9
S0021-9673(98)00445-2 (22) Chen, L., Guo, C., Guan, Y., & Liu, H. (2007) Separ. Purif.
(14) Palmano, K.P., & Elgar, D.F. (2002) J. Chromatogr. A 947, Tech. 56, 168–174. doi:10.1016/j.seppur.2007.01.019
307–311. doi:10.1016/S0021-9673(01)01563-1 (23) Parris, N., & Baginski, M.A. (1991) J. Dairy Sci. 74, 58–64.
(15) Xia, M., & Ying, T. (2010) J. Food Sci. 31, 165–167 doi:10.3168/jds.S0022-0302(91)78143-5
(16) Elgar, D.F., Norris, C.S., Ayers, J.S., Pritchard, M., Otter, D.E., (24) Chinese National Standards (2010) National Food Safety
& Palmano, K.P. (2000) J. Chromatogr. A 878, 183–196. Standard - Good Manufacturing Practice for Powdered
doi:10.1016/S0021-9673(00)00288-0 Formulae for Infants and Young Children, GB 23790-2010
(17) Zhang, J., Lai, S., Cai, Z., Qi, C., Huang, B., & Ren, Y. (2014) (25) Chinese National Standards (2008) Criterion on Quality Control
Anal. Chim. Acta 829, 33–39. doi:10.1016/j.aca.2014.04.025 of Laboratories - Chemical Testing of Food, GB/T 27404-2008