Outline

1
(No Transcript)
2
Outline

• Primer on Regulatory Variation
• Methods for Screening Regulatory Variation
• What have we learned?
• Regulatory Variation and Disease

3
Regulatory Variation rSNPs and rHAPs
rSNP
cSNP
C
A
G
G
G
C
T
T
Exons
rHAP
Regulatory SNPs Polymorphisms that affect gene
regulation
Regulatory Haplotypes Haplotypes that correlate
with expression
Location Upstream, promoter, intronic, exonic,
downstream, etc.
4
Rationale for rSNP discovery

• regulatory variation may have stronger impact on
  phenotypic diversity as compared to coding
  variation
• number of human disease associations without
  deleterious coding variants is increasing
• naturally occurring regulatory variation may
  assist in assigning function to non-coding DNA
  sequence

5
Cis-acting regulatory polymorphisms
AAAA
AAAA
AAAA
AAAA
AAAA
AAAA
AAAA
C
T
AAAA
Heritable, simple or complex (ie. VNTRs)
Affect transcription or degradation of message
6
Parental imprinting
AAAA
AAAA
AAAA
AAAA
AAAA
T
C
AAAA
Parent-of-origin dependent, allele-specific
methylation / chromatin modifications
7
Random inactivation
AAAA
AAAA
AAAA
AAAA
AAAA
C
T
AAAA
AAAA
AAAA
AAAA
AAAA
AAAA
C
T
AAAA
Sequence variation / parent-of-origin
independent, epigenetic modifications
8
Multiple approaches to study allele-specific
expression
Differential Replication Timing
Differential Post-translational Modification Of
Histones
Differential Production Of hnRNA
CHO
Differential Concentrations of Nuclear mRNA
Differential Concentrations of Expressed Protein
CHO
P II
P II
P II
H3C
H3C
Differential Recruitment Of TFs
Differential Concentrations of Cytoplasmic mRNA
Differential Methylation Of DNA
NUCLEUS
CYTOPLASM
9
Allelic Imbalance Studies

• Source of RNA
• Lymphoblastoid cell lines
• - 60 CEPH parents in HapMap panel
• - Five 3-generation CEPH-families
• 100 adipose tissue samples from
  epiploic/subcutaneous fat
• 50 Bone Marrow cultures (osteoblasts) (from Elin
  Grundberg, H. Brandstrom and A. Kindmark -
  Uppsala)

10
Allelic Imbalance Detection

• scalable / high throughput
• Easy to use
• Reliable
• Cost effective
• preservation of sequence context
• epigenetic changes detectable
• Sensitive

11
Association of allele with total expression
NOTE 1. Highly influenced by environment 2.
Marker SNP must be in LD with the rSNP
LCLs AgtG
Human Osteoblasts AgtG
12
Allelic Imbalance A deviation in the expected
relative abundancy of allelic transcripts
AAAA
AAAA
AAAA
AAAA
C
C
AAAA
AAAA
T
T
AAAA
AAAA
T
T
AAAA
AAAA
C
C
AAAA
AAAA
T
T
13
Allelic Imbalance A deviation in the expected
relative abundancy of allelic transcripts
NOTE 1. Minimizes the influence of
environment 2. Cis-effect can be detected in
absence of LD between Marker SNP and the rSNP
AAAA
AAAA
AAAA
AAAA
C
C
AAAA
AAAA
T
T
AAAA
AAAA
T
T
AAAA
AAAA
C
C
AAAA
AAAA
T
T
14
Detection of Allelic Imbalance
rSNP
cSNP
C/T
T/G
Exons

Detection of these allelic imbalances
requires Quantitative genotyping
15
AI Detection using quantitative sequencing
gDNA
RNA
AAAAA
RT-PCR using random hexamers
Locus-specific PCR
Quantitative Sequencing using PeakPicker (Bing
Ge et al., 2005)
16
AI Detection using sequencing
17
Ratio of alleles for IGF1 in separate cultures
Allelic Imbalance is reproducible over time
18
Technical variability
Individual 1 Individual 2 Individual 3
RT-PCR (n1) 10.07 3.14 2.75
PCR (n3) 6.39 3.20-9.62 7.59 4.29-9.89 5.77 4.86-7.04
Sequencing (n8-9) 4.07 1.45-6.22 3.14 0.46-7.04 4.75 1.95-7.65
Overall experimental noise of 9.9
Experimental noise is very unlikely to explain
extent of allelic imbalance larger than 5545 or
1.2-fold difference
19
Allelic Imbalance Detection(Sequencing)

• scalable / high throughput
• Easy to use
• Reliable
• Cost effective
• preservation of sequence context
• epigenetic changes detectable
• Sensitive

20
SNP genotyping on Illumina platform

• 25,000 beads on 1.5 mm diameter fiberoptic
  bundle
• Interrogates 1536 SNPs in parallel
• 96 bundles tested in parallel 100,000 genotypes

21
Allele Specific Extension Assay
Allele Specific Extension
A
Address
P1 P2
G
P3
PCR with common primers
P1 P2
P3
Product capture by hybridization to array
Readout
22
RNA vs DNA allele ratios in homozygotes
log(X/Y)RNA
Median RAA
AA
Median DBB
log(X/Y)DNA
Median DAA
4MAD DBB
Median RBB
4MAD RBB
BB
23
RNA vs DNA allele ratios in heterozygotes
24
RNA vs DNA allele ratios for BCL2
No allelic imbalance
25
RNA vs DNA allele ratios for SNRPN
Either A or B alleles are over-expressed in
heterozygotes (Imprinting)
26
RNA vs DNA allele ratios for IGF1
Over-expression of A or B
27
RNA vs DNA allele ratios for CHI3L2
Overexpression of Allele A
28
Figure 2B
2 - SNP correlations for IL1A
29
Figure 1B
Technical variability between replicates for 400
SNPs
97 of allele ratios are less than 10 different
Experimental noise is very unlikely to explain
extent of allelic imbalance larger than 5545 or
1.2-fold difference
30
Illumina ASE vs comparative sequencing
31
Allelic imbalance estimates at 400 genes related
to cancer biology
32
rs Gene chr Variance Mean meanAI Note
rs2459216 OAT 10 0.9991
rs3750105 PEG10 7 0.9930 Imprinted
rs705 SNRPN 15 0.9999 0.997 Imprinted
rs7810469 PEG10 7 0.9936 0.996 Imprinted
rs13073 PEG10 7 0.9988 0.978 Imprinted
rs5956583 BIRC4 X 0.9996 0.844 X-linked
rs1056831 CHI3L2 1 0.3828 0.833
rs8371 BIRC4 X 1.0000 0.799 X-linked
rs2734647 MECP2 X 0.9833 0.792 X-linked
rs1057403 BTK X 0.787 X-linked
rs1059701 IRAK1 X 0.9824 0.786 X-linked
rs700 BTK X 0.9999 0.777 X-linked
rs9018 FHL1 X 0.765 X-linked
rs5958343 BIRC4 X 0.8100 0.733 X-linked
rs9856 BIRC4 X 0.9996 0.704 X-linked
rs11887 VBP1 X 0.9895 0.628 X-linked
rs12877 DNASE1L1 X 0.604 X-linked
rs10798 KCNQ1 11 0.9890 0.593 Imprinted
rs17561 IL1A 2 0.9994 0.549
rs2278699 ZAP70 2 0.542
rs1056825 CHI3L2 1 0.8884 0.532
rs1304037 IL1A 2 0.9999 0.527
rs3817405 PLXDC2 10 0.6824 0.505
rs10863 MEST 7 0.9838 0.496 Imprinted
rs10336 CXCL9 4 0.9988 0.472
rs5351 EDNRB 13 0.6580   0.469  
33
Allelic Imbalance Detection(Illumina ASE)

• scalable / high throughput
• Easy to use
• Reliable
• Cost effective
• preservation of sequence context
• epigenetic changes detectable
• Sensitive

34
Lessons learned

• 1. Allelic Imbalance is common
• Pastinen et al
• 27 of 241 genes showed evidence of allelic
  imbalance
• Serre et al. (Current Study)
• 20 of 345 genes showed evidence of allelic
  imbalance

35
Other Studies
Genes/SNPs Individuals Material Method AI
Yan et al. (2002) 13 genes 96 LCLs FP-single based extension 6/13
Pastinen et al. (2003) 129 genes 63 LCLs FP-single based extension 18
Lo et al. (2003) 1,063 SNPs 7 Fetal tissues Affymetrix HuSNPs chip 54
Ge et al. (2005) 40 genes 63 LCLs Quantitative sequencing of RT-PCR products na
Pant et al. (2006) 8,406 SNPs 12 White blood cells Perlegen Hybridization array 53
36
Lessons learned

1. Allelic Imbalance is common
2. Allelic imbalance can often be mapped to
   haplotypes

37
BTN3A2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
38
Systematic testing of 76 genes with allelic
imbalance
We selected 76 genes with with at least 10 of
samples showing AI and with characteristics
compatible with genetic inheritance We validated
the allelic imbalance for 64 genes in independent
LCLs (CEPH HapMap panel). 12 genes have been
dropped 5 for low informativity 2 for failed
genomic seq. assay 3 for failed cDNA seq.
assays 2 for assay design errors).
39
HapMap density at 1 kb allows efficient testing
all genes in CEPH LCLs
40
Haplotype correlations with allelic expression

• Determine Phase of Chromosomes
• 2) Correlate SNP alleles and/or haplotypes with
  high or low expression

41
Single-point data chr-association tests
42
The 76 gene test Next steps

• We generated total expression levels for the 64
  genes using real-time RT-PCR (6 - fold
  replication)
• We correlated expression values with genetic
  variant associated with AI

Can we validate that TTgtGTgtGG mRNA levels?
A T A C
C A T G
G T A G
A G C G
G A G G
A A C T
43
Total Expression vs Genotype 3 independent
cultures
44
Associated chr vs. total expression
45
Conclusions of 76-gene test
For 19 (12/64) of genes with common allele
imbalance, strong evidence has been obtained for
a rHAP, which establishes the high likelihood for
heritable cis-acting regulatory polymorphism(s)
at the locus. For 58 (37/64) of genes with
common allele imbalance, weak evidence has been
obtained for a rHAP (nominally significant
association of allelic expression and nearby
SNPs, but non-significant correlation with total
expression). For 23 (15/64) of genes with
common allele imbalance, no evidence has been
obtained for a regulatory haplotype.
46
Lessons learned

• Allelic Imbalance is common
• Allelic imbalance can often be mapped to
  haplotypes
• Allelic imbalance is more sensitive than total
  gene expression
• Cis effects less affected by environment and
  other trans-effects

47
Limitations of total expression as a test results
in underestimate of rHap evidence
Culture I Trend PNS
CAVEAT COMPARISON BETWEEN SAMPLES!
Culture II Trend PNS
Environment Trans-act. loci
Culture III Trend PNS
48
Limitations of total expression as a
test Variance of total expression for genes
without confirmation of rHap is explained by
allele-sharing in twins or sibs
49
Lessons learned

• Allelic Imbalance is common
• Allelic imbalance can often be mapped to
  haplotypes
• Allelic imbalance is more sensitive than total
  gene expression
• Allelic imbalance can be observed across tissues
  or may be tissue specific

STUDY SAMPLES METHOD/ SENSITIVITY NO. OF GENES/ HETS AI GENES/ HETS () FOLLOW-UP
Cowles et al. 3 tissues in 5 mice SBE ES/ gt1.5-fold 67 350 6 2 Validation by independent SNPs
50
Association in multiple tissues
pHOB lt 0.05
pHOB lt 0.05
pLCL lt 10-6
pLCL lt 0.01

• a subset of targets identified in LCLs also
  measurable in HOBs

51
Use of EST databases to find allelic imbalance
Allele frequencies for 2678 SNPs genotyped in all
HapMap populations (x-axis) were compared to EST
allele-counting derived allele frequencies
(y-axis). The Caucasian allele (CEU top left
graph) frequencies show relatively good
concordance with those observed in the ESTs.
Ge et a, Genome Research, 2005
52
Link gene expression level with disease
Association studies rSNPs/rhaplotypes that
correlate with high or low expression
Leukemia Colon Cancer Asthma Diabetes Coronary
Artery Disease Obesity
53
Genetic Studies of the Vitamin D Receptor
(in Asthma)
54
Vitamin D Receptor and Bone Density
Morrison NA et al., 1994 Australia - bone
density Houston et al., 1996 Scotland - bone
density Uitterlinden AG et al., 1997 radiographic
osteoarthritis at the knee Sainz et al.,
1997 US skeletal development, variation in size,
volume and density of bone Lorentzon et al.,
2000 body constitution at birth, BMD, bone
area at age 17 and 19. Yamagata Z et al.,
1994 Japan - BMD Tokita A et al., 1996 Japan -
BMD, turnover in premenopausal women Kiel DP
et al., 1997 bone mass density Uitterlinden AG
et al., 2002 bone biology
55
Vitamin D Receptor, Immune Disease and Infection
Hitman G.A et al., 1997 Indian Asians -
IDDM Saruta T et al., 2003 Japanese - Onset
pattern of type 1 diabetes Chuang L et al.,
2000 Chinese (Taiwanese)- IDDM Jewell D.P. et
al., 2000 Europeans - IBD (CD, UC) Liu W. et al.,
2003 Chinese - Tuberculosis Davidson R.N. et al.,
2000 Gujarati Asians - Tuberculosis Bellamy R. et
al., 1999 Africans - Tuberculosis, chronic
hepatitis B Selvaraj P. et al., 2000 Indians -
Tuberculosis
VDR - Candidate Gene for Asthma?
56
987 6 5 4 3 2 1c/1d
1b/1a/1e 1f

VDR
A
B
Asthma
C
Atopy
Poon, Hudson et al, AJRCC, 2004
57
VDR replication in Nurses Health Study
Raby, Weiss et al, AJRCC, 2004
58
VDR LD, Haplotype Structure, and Asthma
59
Associated chr vs. total expression
60
VDR LD, Haplotype Structure, and Asthma
61
VDR Knock-out - Protective for asthma
62
Science, Feb 23, 2006
63
Other Next Steps

• Expand study to
• More genes
• More tissues
• Stimulated Tissues
• Streamline rSNP
• identification

64
Future analyses (1)

• Better understand the biology of cis-acting
  variation affecting gene expression in the human
  genome
• extent of allelic imbalance in an unbiased
  sampling of genes
• relative importance of genetic and epigenetic
  mechanisms
• (identify regulatory polymorphisms)
• (evolutionary analyses)
• ? Screen ENCODE/Chr21 genes for allelic imbalance
• (extensive genotyping, extensive data on
  methylation/splicing/, comparative genomics)

65
Future analyses (2)

• Gene Regulation in Diseases (GRID)
• - regulatory haplotypes linked to allelic
  imbalance (using LCLs)
• - test association with particular phenotype on
  appropriate cohorts
• - investigate different tissues
• Test 80 genes with evidence from linkage or
  association studies but without segregating
  cSNPs

66
Acknowledgements
Genome Quebec Regulatory Genetics Consortium

• Lab Heads Allelic Imbalance Pipeline
• Thomas Hudson (U. McGill) Tomi Pastinen
• Daniel Sinnett (U. Montreal) David Serre
• Alan Peterson (U. McGill) Robert Sladek
• Celia Greenwood (U. McGill) Pierre Lepage
• Damian Labuda (U. Montreal) Bing Ge
• Kenneth Morgan (U. McGill) Scott Gurd
• Ken Dewar (U. McGill) Laura Oksanen
• Marie-Claude Vohl (U. Laval) Donna Sinnett
• Anna Naumova Andrei Verner
• Tiffany Gaudet
• Amelie Villeneuve
• Eef Harmsen