A robust and sensitive microarray system microarray system for profiling of miRNAs Life Sciences - Genomics Stephanie Fulmer-Smentek RNA Applications R&D RNA Applications R&D Group Manager Agilent miRNA profiling solution August 28, 2007
A robust and sensitive microarray systemmicroarray system
for profiling of miRNAs
Life Sciences - Genomics
Stephanie Fulmer-Smentek
RNA Applications R&DRNA Applications R&DGroup Manager
Agilent miRNA profiling solution
August 28, 2007
Overview
microRNA biology
Agilent miRNA platform
S P i t t h l- SurePrint technology- miRNA protocol workflow- Probe and Array Design- Data Extraction and Quality Control
Performance Data
- Dynamic Range & sensitivity- Dynamic Range & sensitivity- Specificity- Data linearity
R d ibilit- Reproducibility- Comparison to qRT-PCR
Agilent miRNA products
Agilent miRNA profiling solution
August 28, 2007
miRNA Biology
Agilent miRNA profiling solution
August 28, 2007
microRNAs (miRNA)Long precursor (pri-miRNA)
Proteins• Key regulators of cell development
• ~22nts single-stranded RNAs
Hairpin miRNA precursorpre-miRNA (~70nts)
Proteins
• Regulate mRNA translation
• Found in animals, plants, & viruses ( total > 5000)
• >700 identified in humans (Sanger miRBase 10 0)
Mature miRNA(~22nts, single-stranded)
Proteins
• >700 identified in humans (Sanger miRBase 10.0)
• May regulate >30% of human genes
• Tissue-specific expression patterns
Combinatorial Regulation
Proteins
• Different cancers have distinct miRNA expressions
• Diagnostic and prognostic potential being explored
Translation inhibitionProteins
Agilent miRNA profiling solution
August 28, 2007
No Protein Production Figures not drawn to scale
S ll i
Challenges in miRNA Profiling• Small size
• High sequence homology
• Presence of larger RNAs with highly homologous sequences• Presence of larger RNAs with highly homologous sequences
• Expressed with large dynamic range
• Growing & changing databaseGrowing & changing database
miRNA Sequence #NThsa-let-7a ugagguaguagguuguauaguu 22hsa-let-7a ugagguaguagguuguauaguu 22hsa-let-7b ugagguaguagguugugugguu 22hsa-let-7c ugagguaguagguuguaugguu 22hsa-let-7d agagguaguagguugcauagu 21
21
hsa-let-7b ugagguaguagguugugugguu 22hsa-let-7c ugagguaguagguuguaugguu 22hsa-let-7d agagguaguagguugcauaguu 22
hsa-let-7e ugagguaggagguuguauagu 21hsa-let-7f ugagguaguagauuguauaguu 22hsa-let-7g ugagguaguaguuuguacagu 21h l t 7i 21
Sanger miRBase 10.0
hsa-let-7e ugagguaggagguuguauaguu 22hsa-let-7f ugagguaguagauuguauaguu 22hsa-let-7g ugagguaguaguuuguacaguu 22
Agilent miRNA profiling solution
August 28, 2007
hsa-let-7i ugagguaguaguuugugcugu 21 miRBase 10.0hsa-let-7i ugagguaguaguuugugcuguu 22
Agilent’s miRNA platformplatform
Agilent miRNA profiling solution
August 28, 2007
Agilent miRNA Platform HighlightsLow sample input 100 ngLow sample input - 100 ng
Total RNA used for labeling – NO small RNA isolation requiredq
Efficient, direct labeling method linked to specialized miRNA probe design methods
Simple protocol – results in < 2 days
High sequence and size specificity
Multiplex format – 8 arrays per slide
Multiple probe sequences and probe replicates per miRNA
One-color analysis
Enabled by Agilent SurePrint inkjet technology
Agilent miRNA profiling solution
August 28, 2007
y g j gy
Agilent SurePrint Inkjet Printing
Benefits:
• Features are physically isolated from each other -- NO issue of light leakingother NO issue of light leaking
• Synthesis efficiency greater than 99.5% for 60-mer oligonucleotides: increased signal-to-noise because probe fidelity is critical for o se because p obe de y s c ca obinding interactions
• Our feature size and printing technology allow us to have perfect registration from layer to p g ylayer (no blurry edges)
• Our feature size allows us to get sufficient pixels per feature after scanning to perform
Back toPerformance
Back toBenchmarks
p p g ppixel level statistics that can eliminate outlier pixel populations and help estimate confidence in the measurement
Agilent miRNA profiling solution
August 28, 2007
miRNA Protocol workflowTotal RNA (100ng)Total RNA (100ng)
Dephosphorylated RNA
Phosphatase treatment, 30 min, 37ºC
* OHOH
OHP
CyAdd DMSO
Heat, iceAssemble labeling reaction, 16ºC 2hr
OH
CCy
P P
Cy
Time to results <2 days with
Labeled RNA
Desalt with spin column
Desalted Labeled RNA
* OHCP P
yyminimal
hands-on time
Assemble hybridization mixture
Heat ice
* Speed vac, ~1hr, 45ºC
* Sample can be t d f t 80oC
Heat, ice
Hybridize 20 hours, 55ºC, 20RPM
Wash, scan
Agilent miRNA profiling solution
August 28, 2007
miRNA Profile stored frozen at -80oC, if necessary
Probe Design Strategy
Utilize the Cincorporated during labeling for additional G-C base
Start design with full-length miRNA-probe sequence, additional G C base
pair on 3’end of miRNA to increase stability
p q ,attached to a stilt sequence.
Sequentially shorten target-probe base pairing from 5’ end
Incorporate hairpin structure on probes to increase sizeof miRNA during
preliminary Tm balancing by calculation.
increase size specificity and probe:target stability.
Final Step: Select Tm-balanced probes for each miRNA empirically using microarray data
Agilent miRNA profiling solution
August 28, 2007
Select Tm balanced probes for each miRNA empirically using microarray data.
Array DesignH iRNA Mi 1 0Human miRNA Microarray, v1.0:
Content from Sanger miRBase release 9.1
Probes for 470 human and 64 human viral miRNAs
Each miRNA has ≥ 2 different probe sequences, each replicated multiple times on the microarray: at least 20 features/miRNA
Eight identical, separately hybridizable, arrays per slide
Where possible*, probes have been empirically Tm balanced.
Multiple probe sequences and replicates allow for robust miRNA level data summarization
Agilent miRNA profiling solution
August 28, 2007
* When miRNA has been present in tested samples
Human miRNA microarray
Agilent miRNA profiling solution
August 28, 2007
Scanning and Data Extraction -XDRA t t d Xt d d D i R S iAutomated eXtended Dynamic Range Scanning
Automatically scans twice, with high sensitivityand low sensitivity
High Low
Feature Extraction
Automatically combines 2 scan dataand generates each array’s QC reportand text output
Extraction 9.5.3
Agilent miRNA profiling solution
August 28, 2007
8 sets of text outputs and QC Reports
Regular Data txt
Feature Extraction Data Processing – 2 text files
Background SubtractionFeature FindingCookie Cutter
Pixel Rejection Outlier flagging
Regular Data .txt File
inte
nsity
abundance
gNumPix gBGNumPix gMeanSig gBGMeanSiggMedianSig gBGMedianSiggPixSDev gBGPixSDev
gIsSaturatedgIsFeatNonUnifOL gIsBGNonUnifOL gIsFeatPopnOL gIsBGPopnOL
gBGSubSignal
l
gPixSDev gBGPixSDev
GeneView Data .txt File: Total Gene
Signal Calculation
File:simplified format5 columns
l lAgilent miRNA profiling solution
August 28, 2007
gene level summary
Calculation of miRNA TotalGeneSignal in Feature ExtractionExtraction•Multiple probe sequences/miRNA•Multiple replicates/probe sequence
Step 1 –Reject Outlier FeaturesStep 2 –Average non-outlier feature
Probe AProbe B
replicates/probe sequence
X OutlierStep 3 –Multiply that average by the total # pixels representing that
d l i l b “ i h ”Step 4 – Sum TotalProbeSignals for all probes for each miRNA
sequence and multiply by “weight”
+ + + + + + + + )( / 9[ ] *10*(#pixels/feature)*W=TotalProbeSignal (Probe A)
*10*(#pixels/feature)*W=TotalProbeSignal (Probe B)+ + + + + + + )( / 8[ ]
TotalProbeSignal (Probe A)+TotalProbeSignal (Probe B)=TotalGeneSignalNote: The “weight” factor scales the total signals back to a similar scale as intensity
Agilent miRNA profiling solution
August 28, 2007
Note: The weight factor scales the total signals back to a similar scale as intensity values, to better fit with downstream analysis
Metrics and tools for assessment of miRNA profiling data qualityprofiling data quality
2 methods for in-process quality control of miRNA microarray experiments:
• QC report generated with each microarray run• QC metric chart, plots key miRNA specific metrics across
all microarrays in a given Feature Extraction run
QC metrics can be customized by the user using the free QC metric toolthe free QC metric tool
Agilent miRNA profiling solution
August 28, 2007
Sample Microarray QC Report
Header
Net Signal Statistics
Grid Placement
Outlier Statistics
Histogram of BackgroundBackground Subtracted
Signal Outlier
Distribution
Agilent miRNA profiling solution
August 28, 2007
Signal Spatial Distribution
Intra-array Reproducibility
miRNA specific QC
Metrics
Agilent miRNA profiling solution
August 28, 2007
QC Run Chart
miRNA specific pmetrics presented
for an entire feature extraction run
Agilent miRNA profiling solution
August 28, 2007
extraction run
Performance Data
Agilent miRNA profiling solution
August 28, 2007
1000000
Linear Dynamic Range of miRNA measurements
100000
1000000
67 Equal-molar synthetic miRNAs were labeled and hybridized at 0 01 amol to
10000
igna
l
hybridized at 0.01 amol to 1 fmol /miRNA per microarray.
1000
Tota
lGen
eS
10
100T
11 10 100 1000 10000 100000 1000000
1 fmol = 1000 amol1 amol = 1000 zmol
Agilent miRNA profiling solution
August 28, 2007
RNA Amount (zmol)
Before Empirical Tm-balancing: (Wang, Ach, & Curry, RNA, 13, 1-9)
Specificity of hybridizationp g
40hrHyb
miRNA miRNA miRNA miRNA miRNA miRNA miRNA miRNA7a 7b 7c 7d 7e 7f 7g 7i
Probes 7a 100 10 51 3 1 5 2 0Probes 7b 0 100 1 0 0 0 7 0Probes 7c 6 70 100 1 0 0 3 0Hyb Probes 7c 6 70 100 1 0 0 3 0Probes 7d 1 2 1 100 39 0 3 0Probes 7e 4 1 1 2 100 0 1 0Probes 7f 62 3 5 1 0 100 1 0Probes 7g 1 0 0 0 0 0 100 1
After Empirical Tm-balancing: (unpublished)
Probes 7i 0 1 0 0 0 0 4 100
40hr
miRNA miRNA miRNA miRNA miRNA miRNA miRNA miRNA7a 7b 7c 7d 7e 7f 7g 7i
Probes 7a 100 2 32 0 0 0 0 0Probes 7b 0 100 3 0 0 0 0 0Probes 7c 0 21 100 0 0 0 0 0
HybProbes 7c 0 21 100 0 0 0 0 0Probes 7d 2 0 1 100 0 0 0 0Probes 7e 2 0 0 0 100 0 0 0Probes 7f 30 0 4 0 0 100 0 0Probes 7g 0 0 0 0 0 0 100 1Probes 7i 0 0 0 0 0 0 0 100
Agilent miRNA profiling solution
August 28, 2007
Probes 7i 0 0 0 0 0 0 0 100
E i i ll T b l d P b
Effect of Hybridization Time on SpecificityEmpirically Tm-balanced Probes: (unpublished)
miRNA miRNA miRNA miRNA miRNA miRNA miRNA miRNA7a 7b 7c 7d 7e 7f 7g 7i
Probes 7a 100 2 32 0 0 0 0 0
40hrHyb
Probes 7b 0 100 3 0 0 0 0 0Probes 7c 0 21 100 0 0 0 0 0Probes 7d 2 0 1 100 0 0 0 0Probes 7e 2 0 0 0 100 0 0 0Probes 7f 30 0 4 0 0 100 0 0Probes 7f 30 0 4 0 0 100 0 0Probes 7g 0 0 0 0 0 0 100 1Probes 7i 0 0 0 0 0 0 0 100
miRNA miRNA miRNA miRNA miRNA miRNA miRNA miRNA7a 7b 7c 7d 7e 7f 7g 7i
20hrHyb
7a 7b 7c 7d 7e 7f 7g 7iProbes 7a 100 4 39 0 0 1 0 0Probes 7b 0 100 5 0 0 0 0 0Probes 7c 1 30 100 0 0 0 0 0Probes 7d 2 0 1 100 0 0 0 0Probes 7e 2 0 0 0 100 0 0 0y Probes 7e 2 0 0 0 100 0 0 0Probes 7f 37 1 6 0 0 100 0 0Probes 7g 0 0 0 0 0 0 100 1Probes 7i 0 0 0 0 0 0 0 100
Agilent miRNA profiling solution
August 28, 2007
Complex sample titration study
Brain and Placenta total RNA samples, pure and in two different mixtures (25:75 and 75:25)( )
Each sample was processed in 4 replicates using standard conditions, by 4 different users
TotalGeneSignals for the miRNAs were loaded into GeneSpring GX for analysis
No “per-chip” normalization was applied
Signal response as a function of %Brain sample was g p panalyzed for selected miRNAs.
Agilent miRNA profiling solution
August 28, 2007
Selection of genes for titration test
Select miRNA’s significantly different (P<0.01) between 100% Brain and 75%Brain:25% Placenta (no fold change cut-off)Placenta (no fold change cut off)
Agilent miRNA profiling solution
August 28, 2007
Linearity of Signal Response
Up-regulated in Brain
Down-regulated in Brain
Agilent miRNA profiling solution
August 28, 2007
Confirmation of small fold changes across the titration range hsa-mir-9*; FC=0 88 fortitration range hsa-mir-9 ; FC=0.88 for
75%Brain/100% Brainai
n)m
ple/
Bra
ted
sam
o (s
elec
Rat
io
Agilent miRNA profiling solution
August 28, 2007
Reproducibility- Intra-user
Intra-slide
Inter slideAgilent miRNA profiling solution
August 28, 2007
Inter-slide
Reproducibility- Inter-user
Agilent miRNA profiling solution
August 28, 2007
Comparison of miRNA Microarray Results to qRT-PCR
y = 37 109 – 0 942xne S
igna
l) miR-34a
12
qRT PCR
12
gnal
) miR-15a
y = 37.109 0.942xR = -0.987
n To
tal G
en
11
12
11
al G
ene
Sig
(log 2
(Mea
n9
10
38 152 1 02910
(Mea
n To
ta
qPCR (Mean Ct)
Arr
ay
26 27 28 29 30
9
y = 38.152 – 1.029xR = -0.988
26 27 289
Arr
ay (l
og2( qPCR (Mean Ct)
Tissues = Placenta, Brain, Breast, Liver, Heart, Testes, Ovary,
Agilent miRNA profiling solution
August 28, 2007
26 27 28qPCR (Mean Ct)A Thymus, Skeletal Muscle
Comparison of miRNA Microarray Results to qRT-PCR, cont. h i 296)qRT PCR, cont. hsa-mir-296
ene
Sign
al)
Sign
al) hsa-mir-1557
y = 18.078 – 0.388xR = -0.381
an T
otal
Ge
10
otal
Gen
e S
6
y (lo
g 2(M
ea8
34 359 0 937g 2(M
ean
To 5
30 32 34A
rray
qPCR (Mean Ct)6 y = 34.359 – 0.937x
R = -0.985
Arr
ay (l
og
26 28 30
Tissues = Placenta, Brain, Breast, Liver, Heart, Testes, Ovary, Thymus, Skeletal Muscle
Agilent miRNA profiling solution
August 28, 2007
26 28 30qPCR (Mean Ct)
Thymus, Skeletal Muscle
Microarray Correlation to qRT-PCR for 38 miRNA testedtested
16
18
12
14
8
10
*Three miRNAs had poor correlations:
4
6
*
Three miRNAs had poor correlations:• hsa-miR-494: sequence change from
Sanger 9.1 to 10.0• hsa-miR-631: Low signals for both methods
0
2
-1 99 98 97 96 95 94 93 92 91 0 9 re
• hsa-mir-296: (now hsa-miR-296-5p), next slide
Agilent miRNA profiling solution
August 28, 2007
-1-0.
99-0.
98-0.
97-0.
96-0.
95-0.
94-0.
93-0.
92-0.
91 -0.9
More
Titration of hsa-miRNA-296 suggests consistent probe performance
100000
1000000probe performance
67 Equal-molar synthetic miRNAs were labeled and hybridized at 0 01 amol to
10000
100000
gnal
hybridized at 0.01 amol to 1 fmol /miRNA per microarray.
1000
talG
eneS
ig
10
100Tot
1
10
1 10 100 1000 10000 100000 1000000
1 fmol = 1000 amol1 amol = 1000 zmol
Agilent miRNA profiling solution
August 28, 2007
1 10 100 1000 10000 100000 1000000RNA Amount (zmol)
Differential Expression Comparison between miRNA microarrays and qRT-PCR
Placenta/Testes Skeletal Muscle/Breast
tes)
)
ast))
miRNA microarrays and qRT PCR
0
4
0
2
acen
ta/T
es
2(Sk
M/B
rea
-4 -2
s (L
og2(
Pla
rray
s (L
og2
Y = 0.100 – 0.934xY = 0.380 – 0.845x
qPCR (Mean deltaCt, Placenta-Testes) qPCR (Mean deltaCt, SkM-Breast)-4 0 4 8
-82 40-2
-4Arr
ays
Ar Y 0.100 0.934x
R = -0.944R = -0.979
• Log2 expression ratios of the 38 miRNAs in two different tissue pairs were determined for both qPCR and array measurements.
• Results are shown here are for placenta/testes and skeletal muscle/breast ratios
Agilent miRNA profiling solution
August 28, 2007
• Results are shown here are for placenta/testes and skeletal muscle/breast ratios.
Summary of performance data
Five orders of magnitude of linear dynamic range
Detection of miRNAs in amounts as low as 10 zmol ( 6000Detection of miRNAs in amounts as low as 10 zmol (~6000 molecules)
Highly specific hybridization with low levels of cross-Highly specific hybridization with low levels of cross-hybridization with as few as one mismatch
Accurate and consistent detection of small fold changesAccurate and consistent detection of small fold changes
Reproducible data across multiple users
Good correlation with RT PCRGood correlation with RT-PCR
Agilent miRNA profiling solution
August 28, 2007
miRNA products from Agilent TechnologiesMicroarray Platform:Microarray Platform:
-Human miRNA microarray kit (version 1.0)
-miRNA labeling reagent and hybridization kitg g y
Coming soon: Mouse, Rat and updated Human microarray kits2100 Bioanalyzer:
Total RNA Assays (for RNA integrity)
-RNA 6000 Nano Kit
RNA 6000 Pico Kit-RNA 6000 Pico Kit
Small RNA Kit (for analysis of small RNAs)
Stratagene’s qPCR:St atage e s q C-High-Specificity miRNA QRT-PCR Detection Kit
-miRNA Specific Forward Primers
Agilent miRNA profiling solution
August 28, 2007
For more information…
Technical Background:
Wang Ach & Curry RNA 13 1 9 (2007)Wang, Ach, & Curry, RNA, 13, 1-9 (2007).
Product Information:Product Information:
http://www.opengenomics.com
http://www.opengenomics.com/miRNAOverview.aspx
Agilent miRNA profiling solution
August 28, 2007