Jang et al.

BMC Genomics 2011, 12:144

http://www.biomedcentral.com/1471-2164/12/144

METHODOLOGY ARTICLE Open Access

Quantitative miRNA Expression Analysis Using

Fluidigm Microfluidics Dynamic Arrays
Jin Sung Jang1, Vernadette A Simon2, Rod M Feddersen2, Fariborz Rakhshan2, Debra A Schultz2,
Michael A Zschunke3, Wilma L Lingle3, Christopher P Kolbert2, Jin Jen1,2*

Abstract
Background: MicroRNAs (miRNAs) represent a growing class of small non-coding RNAs that are important
regulators of gene expression in both plants and animals. Studies have shown that miRNAs play a critical role in
human cancer and they can influence the level of cell proliferation and apoptosis by modulating gene expression.
Currently, methods for the detection and measurement of miRNA expression include small and moderate-
throughput technologies, such as standard quantitative PCR and microarray based analysis. However, these
methods have several limitations when used in large clinical studies where a high-throughput and highly
quantitative technology needed for the efficient characterization of a large number of miRNA transcripts in clinical
samples. Furthermore, archival formalin fixed, paraffin embedded (FFPE) samples are increasingly becoming the
primary resource for gene expression studies because fresh frozen (FF) samples are often difficult to obtain and
requires special storage conditions. In this study, we evaluated the miRNA expression levels in FFPE and FF samples
as well as several lung cancer cell lines employing a high throughput qPCR-based microfluidic technology. The
results were compared to standard qPCR and hybridization-based microarray platforms using the same samples.
Results: We demonstrated highly correlated Ct values between multiplex and singleplex RT reactions in standard
qPCR assays for miRNA expression using total RNA from A549 (R = 0.98; p < 0.0001) and H1299 (R = 0.95; p <
0.0001) lung cancer cell lines. The Ct values generated by the microfluidic technology (Fluidigm 48.48 dynamic
array systems) resulted in a left-shift toward lower Ct values compared to those observed by ABI 7900 HT (mean
difference, 3.79), suggesting that the microfluidic technology exhibited a greater sensitivity. In addition, we show
that as little as 10 ng total RNA can be used to reliably detect all 48 or 96 tested miRNAs using a 96-multiplexing
RT reaction in both FFPE and FF samples. Finally, we compared miRNA expression measurements in both FFPE and
FF samples by qPCR using the 96.96 dynamic array and Affymetrix microarrays. Fold change comparisons for
comparable genes between the two platforms indicated that the overall correlation was R = 0.60. The maximum
fold change detected by the Affymetrix microarray was 3.5 compared to 13 by the 96.96 dynamic array.
Conclusion: The qPCR-array based microfluidic dynamic array platform can be used in conjunction with
multiplexed RT reactions for miRNA gene expression profiling. We showed that this approach is highly
reproducible and the results correlate closely with the existing singleplex qPCR platform at a throughput that is 5
to 20 times higher and a sample and reagent usage that was approximately 50-100 times lower than conventional
assays. We established optimal conditions for using the Fluidigm microfluidic technology for rapid, cost effective,
and customizable arrays for miRNA expression profiling and validation.

* Correspondence: [email protected]
1
Department of Pulmonary and Critical Care Medicine, 200 First Street SW,
Rochester MN 55905, USA
Full list of author information is available at the end of the article

© 2011 Jang et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons
Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in
any medium, provided the original work is properly cited.
Jang et al. BMC Genomics 2011, 12:144 Page 2 of 8
http://www.biomedcentral.com/1471-2164/12/144

Background quantitative PCR (qPCR) method with the microfluidic

MicroRNAs (miRNAs) are short, single-stranded, non- technology. FFPE and FF samples appear to perform
coding RNAs that regulate gene expression by interacting similarly in this platform when cDNA are generated
with or inhibiting mRNA in both plants and animals from 96 multiplexed RT reactions. We compared this
[1-3]. To date, more than 800 human miRNAs have been new approach to the standard microarray-based technol-
identified and the total number is still increasing [4]. It is ogy. To our knowledge, this study is the first report
estimated that about two thirds of all protein-coding comparing miRNA expression profiling between micro-
genes are regulated by miRNAs [5]. Although some miR- arrays and qPCR using microfluidic arrays.
NAs are yet to be characterized, biochemical and genetic
studies have indicated that miRNA regulation is essential Results
for biological processes such as development, differentia- Correlation of Reverse Transcription Efficiency between
tion, cell proliferation, and apoptosis [6-9]. Recent studies Single- and Multiplexed Primer Sets
have demonstrated that miRNA genes can be aberrantly To evaluate the efficiency of reverse transcription (RT)
expressed in human cancers and they function as either reactions using different number of primer, we first
oncogenes or tumor suppressor genes via regulation of compared the use of a single miRNA specific primer
target transcripts [10,11]. and an 11 primer-mixed set using 100 ng total RNA iso-
Although formalin fixed, paraffin embedded (FFPE) tis- lated from A549 (Figure 1A) and H1299 lung cancer
sue samples typically contain fragmented nucleic acids, cells (Figure 1B). After the RT reaction, a pre-amplifica-
they are the most commonly available clinical specimens tion PCR was performed for 10 cycles and qRT-PCR
for histology and pathological analysis and are a critical was carried out using individual TaqMan probes for all
resource for developing new molecular markers in the 11 genes. Results in Figure 1 showed a strong correla-
cancer research [12,13]. Because of their small size, tion of Ct values between the two conditions with corre-
miRNA molecules appear to be less prone to degrada- lation coefficients at 0.98 and 0.95, respectively, for
tion, in contrast to mRNA expression studies, and no sig- A549 and H1299 cell lines.
nificant differences in miRNA expression between FFPE
and FF samples have been observed [14-16]. Reproducibility of Expression Levels between 48.48
Real-time quantitative PCR (qPCR) is considered a Dynamic Arrays and ABI 7900 HT
‘gold standard’ for quantification of gene expression and As an initial evaluation to determine miRNA expression
has been widely employed as a validation method for by using 48.48 dynamic array, we tested its reproducibil-
microarray studies. However, the qPCR method is a ity by comparing Ct values observed between ABI 7900
relatively low throughput, high cost, and tedious techni- HT and Fluidigm dynamic array system using FF sam-
que typically performed in a 96 or 384 well plate format. ples. Fourteen out of 16 miRNA targets exhibited lower
The Fluidigm microfluidic technology uses the inte- Ct values in the microfluidics 48.48 dynamic array sys-
grated fluidic circuits (IFC) which contain tens of thou- tem compared to those obtained by standard qPCR
sands of microfluidic controlled valves and using the ABI 7900 HT (Figure 2A &2B). The mean Ct
interconnected channels to move molecules of biological values between the platforms were 12.48 (± 0.49) for the
samples and reagents in a variety of patterns [17]. IFCs 48.48 dynamic array and 16.21 (± 0.82) for the ABI
reduce a qPCR reaction from the routine 10-20 microli- 7900 HT (coefficient of variance CV = 0.08 and 0.06,
ter volume down to a 10 nanoliter scale making it possi- respectively) reflecting a significantly increased
ble to perform routine qPCR analysis for thousands of
reactions in a single run. This technology has been used
for gene expression, genotyping, mutation detection, and
absolute quantization of nucleic-acid sequences [17,18].
Spurgeon et al. [19] showed that microfluidic dynamic
arrays can be used to simultaneously measure 48
mRNAs in several tissues. Wang et al. [20] developed a
high throughput SNP genotyping assay demonstrating
high accuracy and call rate in human samples using a
nanofluidic platform. This new real-time PCR technol- Figure 1 Correlation scatter plots of Ct values for qPCR using
ogy makes it possible to perform validation as well as multiplexed or single-plexed RT reactions. Eleven different
high throughput gene expression measurements using miRNA primers were used in single-plex (Y-axis) or 11-plex (X-axis)
very limited sample and reagent quantities [17,19]. for reverse transcriptions using A549 (A) and H1299 (B) lung cancer
Here we show that miRNA expression profiling and cell lines. The qPCR was done with respected TaqMan probes using
the ABI 7900HT instrument.
validation are possible using high throughput real-time
Jang et al. BMC Genomics 2011, 12:144 Page 3 of 8
http://www.biomedcentral.com/1471-2164/12/144

Figure 3 Correlation between matched FFPE and FF samples in

qPCR by ABI vs 48.48 dynamic array platforms. cDNAs were
synthesized using 96-plex primer sets and RNA from both normal
lung (A and C) and lung tumor samples (B and D). qPCR reactions
were carried out individually for 16 miRNA targets (A and B) by ABI
Figure 2 Ct value comparisons using the 48.48 dynamic array 7900 HT, and for 48 miRNA targets by the 48.48 dynamic array (C
and ABI 7900 HT. cDNAs were synthesized using 96-plexed primer and D). Each plot displays mean values calculated from triplicate
set and 100 ng total RNA from FF normal lung (A) and FF tumor samples.
lung (B) samples. Bars represent the means of Ct values from
replicates for the indicated miRNA targets. Open bars: ABI 7900 HT,
closed bars: 48.48 dynamic array systems. The reverse transcription and pre-amplification were
carried out in 96-plex format and the qPCR were run
using 48.48 dynamic arrays. The scatter plots shown in
sensitivity by the microfluidics array when qPCR reac- Figure 4 demonstrates that the correlation coefficients
tions are being carried out in nanoliter volumes. were essentially the same over the tested range of input
RNA for both FF (Figure 4A &4B) and FFPE samples
Comparison of miRNA Expression between FF and FFPE
Samples
We next used both ABI 7900 HT and Fluidigm 48.48
dynamic array to directly compare the qRT-PCR perfor-
mance using RNA from FF and FFPE samples over a
wide range of miRNA gene expression levels. The
cDNA was synthesized using a 96-plex primer set and
100 ng of total RNA from both sample types. The Ct
values of FFPE samples ranged from 13 to 25 while
those of FF samples ranged between 11 to 22 on the
standard ABI platform with correlations at R = 0.95 and
0.87, respectively, p < 0.0001) (Figure 3A &3B). When
the same reaction were carried out using the 48.48
dynamic arrays, FFPE samples had similarly higher aver-
age Ct values compared to FF samples reflecting the
generally lower quality of the RNA in FFPE samples
(6.84-22.6 in FFPE vs. 6.49-20.97 in FF). Again, the Ct
values from both sample types were highly correlated
(Figures 3C and 3D). Figure 4 Effect of in put RNA concentrations on Ct values. A
total of 48 miRNAs were tested using different input amounts of
total RNA from FFPE and FF samples for RT and qPCR by the 48.48
Effect of RNA Template Concentration Using 48.48 dynamic array systems. The Ct values were plotted using the
Dynamic Arrays average of the duplicated measurements and the error bar for
To evaluate the dynamic range of the multiplex RT-PCR values on Y-axis. Correlation scatter plots represent the correlation
in the 48.48 dynamic array systems, we evaluated the Ct of Ct values for 100 ng RNA (X-axis) and the Ct values for the same
values for each miRNA target using total input RNA FF (A and B) or FFPE (C and D) sample at 10 ng to100 ng
concentrations (Y-axis).
quantities at 10 ng, 25 ng, 50 ng or 100 ng per reaction.
Jang et al. BMC Genomics 2011, 12:144 Page 4 of 8
http://www.biomedcentral.com/1471-2164/12/144

(Figure 4C &4D) with the lowest correlation at R = 0.96

(p < 0.0001).

Comparison of miRNA Expression between the 96.96

Dynamic Array and Affymetrix GeneChip microRNA
Microarray
We compared miRNA expression levels measured by
the 96.96 dynamic array with those obtained from the
Affymetrix miRNA GeneChip™. The same samples, FF1
and FFPE9 analyzed by the 96.96 dynamic array were
subjected to gene expression profiling using the Affyme-
trix miRNA arrays. In the Affymetrix miRNA array ana-
lysis, 33-35% of miRNA targets were detectable in each
sample with a correlation of R = 0.99. Similarly, strong
correlations were also obtained using the 96.96 dynamic Figure 5 miRNA expression measurements by Fluidigm
array when replicate RNA samples were analyzed (R = dynamic array and Affymetrix microarray. The raw intensity
0.95). This high correlation value was consistent across values of microarray data were transformed to log2 values for
comparison to PCR Ct values generated by the qPCR platform. Gene
both FF and FFPE sample types (Table 1). Overall, the
expression differences between FF and FFPE were compared
miRNA microarray profiles generated for FFPE and FF against 59 shared genes. Fold differences by the dynamic array
samples showed high correlations across all 847 human (log2) were calculated by ΔΔCt method; ΔCt = (target miRNA log2
miRNAs (R = 0.94, p < 0.0001, data not shown). Simi- values-hsa-mir-16 log2 value), ΔΔCt = (fresh frozen ΔCt-FFPE ΔCt).
larly, the Ct values above detection threshold were
obtained for 86 (FF) and 80 (FFPE) of 94 miRNA targets
by the qPCR-based analysis using microfluidics. Fifty- characterization of miRNAs during clinical investiga-
nine probes that were called present on Affymetrix tions sometimes fall short because of low throughput,
miRNA array matched those tested by the 96.96 insufficient sensitivity, and relatively high cost. The
dynamic array. Comparison of the fold changes between recently developed Microfluidic technology enables a
the two samples obtained by Fluidigm and Affymetrix significantly higher throughput qPCR analysis for a large
arrays showed an overall correlation of R = 0.60. Signifi- number of samples; and assays in a much shorter time
cantly, the fold changes detected by Affymetrix microar- and at a lower cost compared to the conventional meth-
ray ranged between 0 to 3.5, while those by Fluidigm ods [17]. This new real-time PCR technology can be
ranged between 0 to 13 reflecting a much higher used to perform experimental validation as well as high
dynamic range by the microfluidics platform (Figure 5). throughput gene expression measurement with nearly
100-fold less input of sample and reagent [17,19]. In
Discussion this study, we examined miRNA expression measure-
MicroRNAs are potential biomarkers and novel targets ments using qPCR-based microfluidic technology using
for cancer diagnosis, prognosis and recurrence FFPE and matched FF samples. We evaluated this rela-
[10,11,21]. Conventional methods for detection and tively high throughput miRNA profiling method using

Table 1 Comparison of gene expression measurements between Affymetrix microarray and Fluidigm 96.96 dynamic
array.
Platform Sample† Background Signal to Noise Positive Signal Intensity or Ct Detection Rate (%) Replicate Correlation
AVE MAX MIN MAX AVE
Microarray FF#1a 2.66 15.79 189 3.49 10530 503 33 0.997
(847 miRNA)
FF#1b 2.66 11.68 185 1.73 10313 493 33
FFPE#9a 2.26 35.38 208 2.34 9331 470 35 0.996
FFPE#9b 2.22 17.85 209 2.59 8724 464 35
96.96 dynamic array FF#1a N/A N/A - 7.9 22.4 16.7 91 0.954
(94 miRNA)
FF#1b N/A N/A - 5.0 22.7 16.2 91
FFPE#9a N/A N/A - 5.9 35.9 19.5 85 0.956
FFPE#9b N/A N/A - 4.2 34.9 19.0 85
†
Samples are as described in the method; N/A, not applicable.
Jang et al. BMC Genomics 2011, 12:144 Page 5 of 8
http://www.biomedcentral.com/1471-2164/12/144

the standard TaqMan miRNA assays on ABI 7900 HT RNA could be used to detect all 48 miRNAs with 96-
and the Fluidigm microfluidic dynamic arrays. We also multiplexing RT reaction in both FFPE and FF samples
developed and validated miRNA expression assays using (Figure 4). The minimal correlation coefficient observed
cDNA made from either singleplex or multiplex RT was 0.96 for 10 ng vs.100 ng input RNA for both FF
reactions and assessed their application for high- and FFPE samples (P < 0.0001). Lao et al. [24] suggested
throughput miRNA profiling using the microfluidic that 10 ng of a human lung sample can be assessed with
dynamic arrays. a substantial degree of accuracy without statistical varia-
TaqMan® assays are traditionally used for validation tions from stochastic effects when multiplex RT reac-
of microarray-based expression analysis [13,22-25]. The tions are employed. Therefore, our data are consistent
RT primer in each TaqMan® miRNA assay is a single- with those previously reported and indicates that 96-
stranded stem-loop RT primer which was developed to multiplexed miRNA RT reactions can provide reliable
allow cDNA detection to be more specific and sensitive miRNA profiles when using low input amounts in the
as compared to conventional linear primer reverse tran- dynamic array systems. Although just a few samples
scription [22]. Based on this result, Chen, et al. sug- were used in this study, the robustness of the assay was
gested that stem-loop RT primers can be used for demonstrated consistently for all tested assays varying at
multiplex RT reaction and small RNA cloning for better a wide range of Ct values (Figures 3 and 4).
efficiency and specificity [22,24]. In our study, the abso- To determine the concordance between qPCR-arrays
lute Ct for miRNA expression levels by singleplex were and microarrays, we compared miRNA expression using
slightly lower than those obtained in multiplex condi- 96.96 dynamic arrays and the hybridization-based miRNA
tions, but both methods showed high correlation com- array offered by Affymetrix. We used the same samples
pared to results of the qRT-PCR (A549 cells, R = 0.98 for both analyses and the overall correlation for the 59
and H1299 cells, R = 0.95; p < 0.0001) (Figure 1). shared genes was R = 0.60 (p < 0.0001) for miRNA expres-
Several studies have demonstrated that FFPE samples sion between the two platforms. This moderate correlation
can be used for miRNA profiling analysis [13,16]. How- likely reflects the use of fundamentally different principles
ever, most of the studies are based on microarray data for gene expression measurements in these two platforms.
and validated using low throughput Applied Biosystems Fluidigm uses the quantitative PCR assays which are
qPCR platforms [13,15,16]. Here, we compared the cor- highly sensitive with a dynamic range of at least 6-7 logs
relation of miRNA expression profiles between FFPE [19,22]. In contrast, Affymetrix GeneChip is primarily
and matched FF samples using both qPCR and array based on hybridization of the labeled probes to the match-
bases platforms. We observed a high correlation of ing oligonucleotides that are affixed to a matrix. The
miRNA expression levels (R = 0.95 and R = 0.98; p < dynamic range of the microarray is usually 3 to 4 logs
0.0001) measured for both sample types using the ABI [25,26]. In our hands, the maximum fold change observed
7900 HT and the 48.48 dynamic array (Figure 3). FF was around 3 for Affymetrix arrays and 13 for Fluidigm
samples appeared to contain higher levels of miRNAs dynamic array (Figure 5).
than FFPE samples in our study, consistent with results Several studies have previously reported a similar rate
observed by Leite, et al [13] using standard qPCR assays. of inter-platform concordance among different miRNA
This could be due to the loss of miRNA during paraffin microarrays and the different expression values from
embedding process or RNA extraction. Considering the each miRNA microarray platform when compared to
minimal influence on miRNA measurements, the small qPCR values [27-29]. Although highly sensitive, the
difference of Ct values between FFPE and FF samples is quality of the gene expression assessment by the Taq-
not expected to affect the result of the study, particu- Man-based method is dependent largely on the specifi-
larly for Fluidigm based studies since the reference city of the probe to discriminate among highly
miRNA targets are measured on the same array for the conserved miRNA target sequences as well as the sensi-
exact same sample. tivity of the assay probes to quantitatively measure the
To compare between qPCR platforms, we investigated target miRNA over a wide range of expression levels.
16 different miRNA targets using both ABI 7900 HT The small 18-25 nucleotide length of the miRNA targets
and Microfluidic technology with FF RNA samples. The creates a challenge to meet these requirements. Prader-
Ct value of the 48.48 dynamic array system shifted vand, et al. [27] observed that the different GC content
toward lower Ct values compared to those observed by of mature miRNA sequences contributes to higher dif-
ABI 7900 HT in a 5 μl reaction. In our hands, the mean ference values between both Affymetrix and qPCR. This
Ct value difference was 3.79 between the two systems same study also showed a lower correlation between
and the coefficient of variation across the 128 reactions qPCR and Affymetrix while reporting a higher correla-
in the 48.48 dynamic array system was 8.9% (Figure 2). tion between qPCR based assay and other platforms (R
In addition, we showed that as little as 10 ng of total = 0.8-0.9). In the study by Chen, et al. [29], a correlation
Jang et al. BMC Genomics 2011, 12:144 Page 6 of 8
http://www.biomedcentral.com/1471-2164/12/144

of 0.44 (p < 0.0001, N = 84) was reported between Taq- 100 nM dNTP, 0.2 μl of RNase inhibitor 20 U/μl, 1.5 μl
Man qPCR-array and microarray. The variation of reverse transcriptase (50 U/μl), 8 μl of 11-or 96-plex
observed was thought to result from the low abundance reverse primer (mixed to allow a final concentration of
miRNAs reflecting the different sensitivities of the two 0.05X of each) and 1.6 μl of H2O. 2 μl of 5 ng of total
platforms. Therefore, the assessment of miRNA expres- RNA and 3 μl of 5X reverse primer were used in the
sion in a given system should be interpreted with cau- single-plex RT reaction. All reagents were purchased
tion and it requires validation using at least two from Applied Biosystems, Inc. (Foster City, CA). The
different platforms when the correlation is poor or the reaction mixture was mixed with RNA and incubated as
expression level of the target is low. follows; 16°C for 30 min, 42°C for 30 min and then 85°
C for 5 min. A list of all 96 tested assays is available
Conclusion upon request. The 11 primer set that was used for the
We demonstrated that multiplexing RT reactions with RT reaction included: RNU66, RNU6B, mir-135a, mir-
stem-loop primers can be adapted with relative ease to a 564, mir-29b, mir-339, mir-138, mir-425, mir-191, let-7
new qPCR-array based microfluidic platform to profile g and mir-566 (Applied Biosytems, Foster City, CA).
miRNA expression profiling. We showed that this
approach is highly reproducible and correlates closely Pre-PCR amplification
with standard ABI7900 systems yet offers higher For pre-amplification of cDNA, we pooled 11- or 96
throughput, with much lower sample input, and reagent TaqMan Assays at a final concentration of 0.2X for each
usage. We believe that the microfluidic dynamic array assay. Pre-PCR amplification reaction was done at 10 μl
technology could be used to develop cost effective and containing 5 μl TaqMan PreAmp Master Mix (2X), 2.5
customized assays with rapid turn-around for profiling μl of 11- or 96-pooled TaqMan assay mix (0.2X) and
and validating of miRNA expression. 2.5 μl of cDNA. The pre-amplification PCR performed
at one cycle 95°C for 10 min, 10 cycles at 95°C for 15
Methods sec and then 60°C for 4 min. After pre-amplification
RNA extraction from FFPE, FF samples and lung cancer PCR, the product was diluted 1:5 with dH2O and stored
cell lines at -80°C until needed.
FF and corresponding FFPE samples were obtained from
lung cancer (FF1 and FFPE9) or non-diseased lung tissues Real-time qPCR
(FF4 and FFPE7) that had been preserved between 2007 qRT-PCR was carried out using ABI 7900 HT Real-time
and 2008 following the approved Mayo Clinic Institutional PCR system in a 384 well plate format. PCR reaction of
Review Board protocol. FFPE samples were cut to 10 μm 5 μl contained 2.5 μl of TaqMan PCR Master Mix-UNG
thickness and several tissue slices were put into a 1.5 ml (2X), 0.25 μl of each TaqMan assay probe (20X), 1.25 μl
tube. One milliliter of xylene was added for deparaffiniza- of diluted cDNA and 1 μl of H2 O. The PCR was per-
tion followed by mixing twice with a high speed vortex for formed at 95°C for 10 min, followed by 40 cycles at 95°
3 min at room temperature. Total RNA was then C for 15 sec and 60°C for 1 min. The data was analyzed
extracted with the Qiagen miRNeasy FFPE kit (Valencia, with ABI RQ Manager software (Foster City, CA) after
CA) and/or RecoverAll (Ambion Inc. Austin, TX) follow- exportation as a SDS file.
ing manufacturers’ protocols. Fresh-frozen tissues were
extracted using Qiagen miRNeasy kit (Valencia, CA) fol- miRNA expression analysis using 48.48 and 96.96
lowing manufacturer’s protocols. The isolation procedure dynamic array
for FF and FFPE samples were performed in duplicate to Reverse transcription was carried out as described above
derive samples FF1a and FF1b, and FFPE9a and FFPE9b. using pooled miRNA primers with 10 ng, 25 ng, 50 ng
Human lung cancer cell lines H1299 and A549 were cul- and 100 ng of total input RNA. Pre-amplification was
tured in RPMI 1640 growth media with 10% fetal bovine performed with a 96 pooled (final 0.2X of each) Taq-
serum and 1% penicillin (50 IU/mL) and streptomycin (50 Man assay. After pre-amplification PCR, the product
μg/ml). Total RNA was isolated from TRIzol (Invitrogen, was diluted 1:5 with dH 2 O and stored in -80°C until
Carlsbad, CA). RNA quantity was determined using Nano- needed. qPCR was carried out using the 48.48 or 96.96
drop (Thermo Scientific, Waltham, MA) and the quality dynamic array (Fluidigm Corporation, CA, USA) follow-
was assessed by Agilent 2100 Bioanalyzer (Agilent Tech- ing the manufacturer’s protocol [19]. Specifically, a 5 μl
nologies, Santa Clara, CA). sample mixture was prepared for each sample contain-
ing 1 × TaqMan Universal Master Mix (No UNG), 1 ×
MicroRNA reverse transcription GE Sample Loading Reagent (Fluidigm PN 85000746)
The 15 μl reverse transcription reaction contained 2 μl and each of diluted pre-amplified cDNA. 5 μl of Assay
of either 5, 12.5, 25 or 50 ng/μl of total RNA, 0.2 μl of mix was prepared with 1 × each of TaqMan miRNA
Jang et al. BMC Genomics 2011, 12:144 Page 7 of 8
http://www.biomedcentral.com/1471-2164/12/144

assay and 1 × Assay Loading Reagent (Fluidigm PN References

1. Ambros V: The functions of animal microRNAs. Nature 2004,
85000736). An IFC controller was used to prime the 431(7006):350-355.
fluidics array (chip) with control line fluid and then with 2. Bartel DP, Chen CZ: Micromanagers of gene expression: the potentially
samples and assay mixes in the appropriate inlets. After widespread influence of metazoan microRNAs. Nature reviews 2004,
5(5):396-400.
loading, the chip was placed in the BioMark Instrument 3. Lim LP, Lau NC, Garrett-Engele P, Grimson A, Schelter JM, Castle J,
for PCR at 95°C for 10 min, followed by 40 cycles at 95° Bartel DP, Linsley PS, Johnson JM: Microarray analysis shows that some
C for 15 sec and 60°C for 1 min. The data was analyzed microRNAs downregulate large numbers of target mRNAs. Nature 2005,
433(7027):769-773.
with Real-Time PCR Analysis Software in the BIO- 4. Bentwich I, Avniel A, Karov Y, Aharonov R, Gilad S, Barad O, Barzilai A,
MARK instrument (Fluidigm Corporation, CA, USA). Einat P, Einav U, Meiri E, et al: Identification of hundreds of conserved
and nonconserved human microRNAs. Nature genetics 2005,
37(7):766-770.
miRNA microarray 5. Friedman RC, Farh KK, Burge CB, Bartel DP: Most mammalian mRNAs are
The miRNA microarray profiling was performed using conserved targets of microRNAs. Genome research 2009, 19(1):92-105.
Affymetrix GeneChip miRNA arrays (Santa Clara, CA, 6. Cheng AM, Byrom MW, Shelton J, Ford LP: Antisense inhibition of human
miRNAs and indications for an involvement of miRNA in cell growth and
USA) according to manufacturer’s recommended proto- apoptosis. Nucleic acids research 2005, 33(4):1290-1297.
col. Briefly, 1 μg of total RNA was labeled by polyA 7. Karp X, Ambros V: Developmental biology. Encountering microRNAs in
polymerase addition using the Genisphere FlashTag cell fate signaling. Science (New York, NY) 2005, 310(5752):1288-1289.
8. Sempere LF, Sokol NS, Dubrovsky EB, Berger EM, Ambros V: Temporal
HSR kit following the manufacturer’s recommendations regulation of microRNA expression in Drosophila melanogaster
(Genisphere, Hatfield, PA). RNA was hybridized to the mediated by hormonal signals and broad-Complex gene activity.
Affymetrix miRNA array as recommended by the ven- Developmental biology 2003, 259(1):9-18.
9. Xu P, Guo M, Hay BA: MicroRNAs and the regulation of cell death. Trends
dor. Standard Affymetrix array cassette staining, washing Genet 2004, 20(12):617-624.
and scanning was performed using the post-hybridiza- 10. Calin GA, Croce CM: MicroRNA signatures in human cancers. Nat Rev
tion kit (#900720; Affymetrix) and GeneChip Scanner Cancer 2006, 6(11):857-866.
11. Esquela-Kerscher A, Slack FJ: Oncomirs-microRNAs with a role in cancer.
3000. Feature extraction was performed using Affyme- Nat Rev Cancer 2006, 6(4):259-269.
trix Command Console software. The raw data were 12. Lehmann U, Kreipe H: Real-time PCR analysis of DNA and RNA extracted
treated by the following workflow: background detec- from formalin-fixed and paraffin-embedded biopsies. Methods (San Diego,
Calif) 2001, 25(4):409-418.
tion, RMA global background correlation, quantile nor- 13. Leite KR, Canavez JM, Reis ST, Tomiyama AH, Piantino CB, Sanudo A,
malization, median polish and log2-transformed with Camara-Lopes LH, Srougi M: miRNA analysis of prostate cancer by
miRNA QC tool software (Affymetrix). quantitative real time PCR: Comparison between formalin-fixed paraffin
embedded and fresh-frozen tissue. Urologic oncology 2009.
14. Mortarino M, Gioia G, Gelain ME, Albonico F, Roccabianca P, Ferri E,
Statistical Analysis Comazzi S: Identification of suitable endogenous controls and
Statistical Analysis was performed using GraphPad differentially expressed microRNAs in canine fresh-frozen and FFPE
lymphoma samples. Leukemia research 2009, 34(8):1070-7.
Prism 4 (GraphPad Software, Inc.). The Pearson correla- 15. Xi Y, Nakajima G, Gavin E, Morris CG, Kudo K, Hayashi K, Ju J: Systematic
tion coefficient(R) was employed to determine the corre- analysis of microRNA expression of RNA extracted from fresh frozen and
lation of efficiency of RT reaction and expression of formalin-fixed paraffin-embedded samples. RNA (New York, NY) 2007,
13(10):1668-1674.
FFPE and FF samples. 16. Zhang X, Chen J, Radcliffe T, Lebrun DP, Tron VA, Feilotter H: An array-
based analysis of microRNA expression comparing matched frozen and
formalin-fixed paraffin-embedded human tissue samples. J Mol Diagn
Acknowledgements 2008, 10(6):513-519.
We thank Stacy Johnson and Cindy Dilworth for editorial assistance and Drs. 17. Melin J, Quake SR: Microfluidic large-scale integration: the evolution of
Julie Cunningham and Eric Wieben for scientific input. This work is design rules for biological automation. Annual review of biophysics and
supported by funds from Mayo Cancer Center and the Center for biomolecular structure 2007, 36:213-231.
Individualized Medicine. 18. Thorsen T, Maerkl SJ, Quake SR: Microfluidic large-scale integration.
Science (New York, NY) 2002, 298(5593):580-584.
Author details 19. Spurgeon SL, Jones RC, Ramakrishnan R: High throughput gene
1
Department of Pulmonary and Critical Care Medicine, 200 First Street SW, expression measurement with real time PCR in a microfluidic dynamic
Rochester MN 55905, USA. 2Gene Expression Shared Resource, Advanced array. PloS one 2008, 3(2):e1662.
Genomics Technology Center and the Mayo Clinic Cancer Center, 200 First 20. Wang J, Lin M, Crenshaw A, Hutchinson A, Hicks B, Yeager M, Berndt S,
Street SW, Rochester MN 55905, USA. 3Department of Experimental Huang WY, Hayes RB, Chanock SJ, et al: High-throughput single
Pathology, 200 First Street SW, Rochester MN 55905, USA. nucleotide polymorphism genotyping using nanofluidic Dynamic Arrays.
BMC genomics 2009, 10:561.
Authors’ contributions 21. Johnson SM, Grosshans H, Shingara J, Byrom M, Jarvis R, Cheng A,
JSJ performed the research, analyzed the data and wrote the paper. VS, RMF, Labourier E, Reinert KL, Brown D, Slack FJ: RAS is regulated by the let-7
FR and DS participated in the Fluidigm and/or microarray experiments. MZ microRNA family. Cell 2005, 120(5):635-647.
performed RNA preparation and WL participated in study discussions. JJ 22. Chen C, Ridzon DA, Broomer AJ, Zhou Z, Lee DH, Nguyen JT, Barbisin M,
directed research and analyzed the data. RMF, CPK and JJ revised the Xu NL, Mahuvakar VR, Andersen MR, et al: Real-time quantification of
manuscript. All authors read, corrected and approved the final manuscript. microRNAs by stem-loop RT-PCR. Nucleic acids research 2005, 33(20):
e179.
Received: 13 May 2010 Accepted: 9 March 2011 23. Lao K, Xu NL, Sun YA, Livak KJ, Straus NA: Real time PCR profiling of 330
Published: 9 March 2011 human micro-RNAs. Biotechnology journal 2007, 2(1):33-35.
Jang et al. BMC Genomics 2011, 12:144 Page 8 of 8
http://www.biomedcentral.com/1471-2164/12/144

24. Lao K, Xu NL, Yeung V, Chen C, Livak KJ, Straus NA: Multiplexing RT-PCR
for the detection of multiple miRNA species in small samples.
Biochemical and biophysical research communications 2006, 343(1):85-89.
25. Tang F, Hajkova P, Barton SC, O’Carroll D, Lee C, Lao K, Surani MA: 220-plex
microRNA expression profile of a single cell. Nature protocols 2006,
1(3):1154-1159.
26. Shippy R, Sendera TJ, Lockner R, Palaniappan C, Kaysser-Kranich T, Watts G,
Alsobrook J: Performance evaluation of commercial short-oligonucleotide
microarrays and the impact of noise in making cross-platform
correlations. BMC genomics 2004, 5(1):61.
27. Pradervand S, Weber J, Lemoine F, Consales F, Paillusson A, Dupasquier M,
Thomas J, Richter H, Kaessmann H, Beaudoing E, et al: Concordance
among digital gene expression, microarrays, and qPCR when measuring
differential expression of microRNAs. BioTechniques 2010, 48(3):219-222.
28. Sato F, Tsuchiya S, Terasawa K, Tsujimoto G: Intra-platform repeatability
and inter-platform comparability of microRNA microarray technology.
PloS one 2009, 4(5):e5540.
29. Chen Y, Gelfond JA, McManus LM, Shireman PK: Reproducibility of
quantitative RT-PCR array in miRNA expression profiling and comparison
with microarray analysis. BMC genomics 2009, 10:407.

doi:10.1186/1471-2164-12-144
Cite this article as: Jang et al.: Quantitative miRNA Expression Analysis
Using Fluidigm Microfluidics Dynamic Arrays. BMC Genomics 2011 12:144.

Submit your next manuscript to BioMed Central

and take full advantage of:

• Convenient online submission

• Thorough peer review
• No space constraints or color figure charges
• Immediate publication on acceptance
• Inclusion in PubMed, CAS, Scopus and Google Scholar
• Research which is freely available for redistribution

Submit your manuscript at

www.biomedcentral.com/submit