Post on 11-Oct-2020
Quality and Use of Genome-wide Assays for methylation
and RNA Analysis
Illumina Seminar Series, Munich,
06.07.2009
Bernhard Korn, Genomcis & Proteomics Core Facility, German Cancer Research Center (DKFZ), Heidelberg, Germany
Page 2
Bernhard KornGenomics & Proteomics Core Facility Scientific Services at DKFZ
• Genomics• Sanger Sequencing• “Next Generation” Sequencing• Expression Profiling• Genotyping• Methylation Analysis• Clone Repository
• Proteomics & Structure Analysis• Peptide synthesis• Protein Interaction Screening• Protein Analysis, 2D• Protein Analysis MALDI, esiMS/MS• Mass Spectrometry: Small molecules and protein
modifications• NMR• Molecular Modeling
Page 3
Bernhard KornGenomics & Proteomics Core Facility
• Genome-wide methylation analysis• Evaluation of technology• Preliminary data on lymphomas
• Expression profiling using ‚deraded RNA‘• WG-DASL technology• Potential
Page 4
Bernhard KornGenomics & Proteomics Core Facility What is epigenetics?
• Study of heritable changes in gene function that do NOT involve changes to the nucleotide sequence of DNA
• When a cell undergoes mitosis or meiosis, the epigenetic information is stably transmitted to the subsequent generation
• Epigenetic controls add an ‘extra layer’ of transcriptional control
• Epigenetic changes include:• DNA methylation• Histone modifications• RNA interference
• DNA methylation
Page 5
Bernhard KornGenomics & Proteomics Core Facility DNA Methylation
• Methyl group introduced in the 5’ position of cytosine• In mammals, occurs in CpG dinucleotides• Catalyzed by DNA methyltransferases (DNMT)
N
N
NH2
O
N
N
NH2
O
CH3EnzSSAM AdoHcy
Page 6
Bernhard KornGenomics & Proteomics Core Facility DNA methyltransferases (DNMTs)
• DNMT1 maintenance methyltransferases• DNMT3a, DNMT3b de novo methyltransferases
◄daughter strand
◄daughter strand
Page 7
Bernhard KornGenomics & Proteomics Core Facility
Where does DNA methylation occur in the genome?
GENE EXPRESSION
E1E1 E2 E3
GenePromoter & CpG island Body of the gene
DNA repeats
•Unmethylated CpG
•Methylated CpG
•Unmethylated CpG
•Methylated CpG
CpG islands2% of the genome
Body of the genes and DNA repeats
GENE SILENCING
E1 E2 E3
x GENE SILENCING
E1E1 E2 E3
xTissue-specific DNA
methylation
Changes in gene expression
Hypermethylation of DNA repeats confers
genomic stability
Page 8
Bernhard KornGenomics & Proteomics Core Facility Coat Color defined by Avy Methylation
Geneticallly identical mice
Page 9
Bernhard KornGenomics & Proteomics Core Facility Genomic Hypomethylation and Cancer
DNA methyltransferase 1 (DNMT1)down to 10%
Reduced DNA methylation
Poor survival rate
Mice develop T cell lymphomasmany have instable T cell receptor ß locus
Gaudet (2003), Science, 300, 489-92
Page 10
Bernhard KornGenomics & Proteomics Core Facility Genomic Hypomethylation and Cancer
•Which genes are involved?•How is their promoter status?
•The methylation of which CpGs (pattern of CpGs) does have predictive information?
Page 11
Bernhard KornGenomics & Proteomics Core Facility Chromosome-wide Methylation View
Eckhardt et al. Nat Genet 2006: 1.9 million CpGs in 12 different tissuesMethylation within a promoter is homogeneous
Page 12
Bernhard KornGenomics & Proteomics Core Facility DNA Methylation Analysis
Methylation within a promoter is homogeneous
• Consequences• Test individual CpGs!
• use genotyping assay• Test only few CpGs per promoter?
• Select CpGs carefully (stability of assay, functional relevance)
• Conclude overall methylation status of each promoter• Infinium assay for methylation analysis
Page 13
Bernhard KornGenomics & Proteomics Core Facility Infinium HumanMethylation27 Beadchip
• 27.578 CpG sites (incl. 254 miRNA CpGs)
• Representing >14.000 promoters / genes• 1-3 CpGs / promoter• 12 arrays / BeadChip
• 1 µg genomic DNA• ~200 ng bisulfite treated DNA• Infinium assay
• Two-color fluorescent scanning• Methylation measure: ß-value (0-1)
Page 14
Bernhard KornGenomics & Proteomics Core Facility Do ß-Values reflect Methylation Levels?
0.0 0.2 0.4 0.6 0.8 1.0
12
10
8
6
4
2
0
Beta values
Den
sity
Cluster analysis of 26,492 CpGs
Density plotmean ß: 0.070, 0.530, 0.894
Assays are capable to discriminate methylation statusQuantitation is possible
Page 15
Bernhard KornGenomics & Proteomics Core Facility Reproducibility and Robustness
Technical replicates, Hybs: r2=0.977 +/-0.011
Technical replicates, bisulfite treatments: r2=0.955 – 0.992
Reproducibility is goodAssay outcome is robust (chemicals!)
Page 16
Bernhard KornGenomics & Proteomics Core Facility Validation of Results
Infinium data vs. MSP
....................
....................
Primer set for methylated CpG islandPrimer set for ummethylated CpG island
Methylation-specific PCR
....................
....................
Amplification
No amplification
Amplification
No amplification
U M U M U M U M U M U M U M U M U M
NL
NLu
ng
H21
66
H44
1
H12
99
NL
IVDA42
7
H2O
Strong correlation between Infinium and MSP (and bisulfite sequencing)
Page 17
Bernhard KornGenomics & Proteomics Core Facility Meaningful Data for Cancer Research
•3,916 CpGs are hypermethylated•127 CpGs are hypomethylated
in lymphoma cell lines compared to normal controls; p<0.001
Page 18
Bernhard KornGenomics & Proteomics Core Facility Are miRNAs regulated by Methylation?
•22/254 miRNA associated CpGs are hypermethylated•No miRNA assoc. CpGs are hypomethylated
in lymphoma cell lines compared to normal controls; p<0.001
Page 19
Bernhard KornGenomics & Proteomics Core Facility Methylation Analysis Pipeline
58/46 Tm 58/71 Tm
First Hybridization
2000
4000
6000
8000
10000
12000
14000
Sig
na
l
858 1742
Bisulfite Conversion
2000
4000
6000
8000
10000
12000
14000
16000
18000
20000
22000
Sig
na
l
Golden GateInfinium
probes variation
Assay Intensity
2000
4000
6000
8000
10000
Sig
na
l
Excel-like Output with Links
percent methylation
Page 20
Bernhard KornGenomics & Proteomics Core Facility Summary
• DNA methylation plays crucial role in many diseases• Genome-wide Infinium methylation assays are
• Reproducible• Robust• Can be vaildated by other techniques• Allow accurate quantitation of methylation
• Useful for hypothesis-free screening and in-depth analysis
Page 21
Bernhard KornGenomics & Proteomics Core Facility
Page 22
Bernhard KornGenomics & Proteomics Core Facility
Expresssion Profiling of Formalin-fixed Parafin Embedded Samples
• Archived patient tissue• Follow-up data• Large patient cohorts• Ease of storage (pathology)• Microdissection
• BUT• RNA quality
• Degradation• Cross-link
• cDNA synthesis• Labelling• Relative probe position
Page 23
Bernhard KornGenomics & Proteomics Core Facility DASL Technology
• Proven for GT/customized expression profiling• Address code
• Now possible on whole genome basis?
adapted from Illumina
Page 24
Bernhard KornGenomics & Proteomics Core Facility Whole Genome DASL
• Reproducibility/Robustness• Technical• RIN dependance• Influence of RNA amounts• IVT comparison
• Setup:• Artificial degradation series of HeLa total RNA• Technical replicates• Detecion level
Page 25
Bernhard KornGenomics & Proteomics Core Facility Degradation Series
• Boiling total RNA• 0 – 30 min
9.6
8.5
7.8
6.2
4.8
2.5
2.2
(650
bp)
2.2
(350
bp)
Page 26
Bernhard KornGenomics & Proteomics Core Facility Clustering of WG-DASL
Page 27
Bernhard KornGenomics & Proteomics Core Facility Clustering of WG-DASL
• Correlation drops, when comparing high and low RIN
WHY?
Page 28
Bernhard KornGenomics & Proteomics Core Facility Number of Genes expressed
Reduction of RIN failure to detect transcripts
Page 29
Bernhard KornGenomics & Proteomics Core Facility Data Correlation
Page 30
Bernhard KornGenomics & Proteomics Core Facility Scatter Plot
Decreasing RIN, reduced detection
10 2 10 3 10 4 10 5
RIN_85 AVG_Signal
vs RIN_85 AVG_Signal
10 2
10 3
10 4
10 5
RIN
_48
AV
G_S
igna
l
Page 31
Bernhard KornGenomics & Proteomics Core Facility Scatter Plot
Don‘t compare samples of highly different RINs
10 2 10 3 10 4 10 5
RIN_85 AVG_Signal
vs RIN_85 AVG_Signal
10 2
10 3
10 4
10 5
RIN
_25_
1200
AV
G_S
igna
l
Page 32
Bernhard KornGenomics & Proteomics Core Facility Correlation at low RIN
Excellent correlation with ‚fully‘ degraded RNA
10 2 10 3 10 4 10 5
RIN_22_400 AVG_Signal
vs RIN_22_400 AVG_Signal
10 2
10 3
10 4
10 5
RIN
_22_
650
AV
G_S
igna
l
Page 33
Bernhard KornGenomics & Proteomics Core Facility Gains and Losses
• High reproducibility• Low ‚false positives‘
Tran
scrip
ts
dete
cted
Tran
scrip
ts
over
lap
Per
cent
age
over
lap
Tran
scrip
ts
new
Per
cent
age
new
Page 34
Bernhard KornGenomics & Proteomics Core Facility Genes ‚regulated‘: RIN 9.6 – 2.2
Page 35
Bernhard KornGenomics & Proteomics Core Facility Sensitivity to RNA Amounts
• Assay is highly robust to varying amounts of RNA• No comparison to standard IVT assay possible
Page 36
Bernhard KornGenomics & Proteomics Core Facility Fold-changes: WG-DASL vs. IVT
• Reproducibility on the level of expression changes!
data from John Quackenbush, Dana Faber
r2 = 0.8931
Page 37
Bernhard KornGenomics & Proteomics Core Facility Summary WG-DASL
• Highly robust assay: technical replicatesr2 = 0.98 - 0.99 (RIN dependent)
• Only RNAs of similar RIN should be compared• Artifically adjust RNA quality?• Use more biological replicates!
• High sensitivity: significant and reproducible result down to 25 ng total RNA
• Good concordance with IVT results, based on fold-changes• Reproducible loss of number of transcripts detected, with
decreasing RIN – can we correct for it?• Requirements
• Highly standardized procedure• Calibrated equipment
• Biased RNA deradation in FFPE samples? - unsolved
Page 38
Bernhard KornGenomics & Proteomics Core Facility Outlook
• WG-DASL Core service for• FFPE samples• Microdisected samples• (partially) FACS samples
• Needs and Wishes• WG-DASL for Sentrix-6• Reproducibility down to smaller amounts of RNA
(<1000 cells)• Approach for direct labelling (avoiding RNA isolation)
Page 39
Bernhard KornGenomics & Proteomics Core Facility Acknowledgements
• University Kiel• Ole Ammerpohl• José Martin-Subero• Julia Richter• Reiner Siebert
• German Cancer Res. Center• Tamara Fries• Roger Fischer• Sabine Henze• Bernhard Korn