Integrating genetic and epigenetic variation in schizophrenia - - PowerPoint PPT Presentation

Jonathan Mill www.epigenomicslab.com

Integrating genetic and epigenetic variation in schizophrenia

1

Voineagu et al, Nature, 2011 Pidsley et al, Genome Biology, 2014

Functional annotation of human brain genome

Genomics

(GWAS)

Epigenomics

(5mC, 5hmC, histone modifications) Transcriptomics (RNA-seq)

Human brain tissue Biomarkers Clinical cohorts Animal models Cell models (iPSC) Genome editing Single-cell profiling

Systems-level analyses

Integrated –omics approach: heterogeneous etiology, convergent molecular pathology

Lunnon et al, Nature Neuroscience, 2014 Function

2

A suite of epigenetic modifications act to fine-tune genomic function

Zhou et al (2011) Addi6onal DNA modifica6ons: 5hmC, 5fC, 5cC

3

Roadmap Epigenomics Consortium, Nature, 2015 PsychENCODE Consortium, Nature Neuroscience, 2016

Regulatory elements Enhancers Generate hypotheses about function

Functional annotation of regulatory variation in the genome

4

Multi-centre clinical cohorts UCL (n = 675) ABER (n = 971) KCL (n = 801) CARDIFF (n = 950) DUBLIN (n = 713) EDIN (n = 540)

Integrated genetic-epigenetic analysis of schizophrenia

SZ- discordant monozygotic twin pairs (n = 99 pairs) Adult post- mortem brain (matched PFC, striatum, hippocampus, cerebellum) Human fetal brain (23 to 184 days post- conception) (n = 179)

• Differences in DNA methylation associated with

schizophrenia

• Differences in DNA methylation associated with

high polygenic burden for schizophrenia

• Epigenetic consequences of genetic variants

associated with schizophrenia

• Illumina 450K/EPIC DNA methylation data
• Genotype data (imputed to 1000G phase 3)

5

Control (n=23) Schizophrenia (n=20) 70 75 80 85 90 DNA methylation (%) p=1.16e−07 Control (n=23) Schizophrenia (n=20) 45 50 55 60 65 70 75 DNA methylation (%) p=1.25e−07 Control (n=23) Schizophrenia (n=20) 5 10 15 20 DNA methylation (%) p=2.4e−07

cg00903099 (HTR5A)

Control (n=23) Schizophrenia (n=20) 40 45 50 55 60 65 DNA methylation (%) p=2.85e−07

cg08171022 (PPFIA1)

Dopamine D2S Receptor-mediated MAPK signaling Serotonin receptor Axon guidance

Pidsley et al, Genome Biology, 2014 Viana et al, in review

PFC

6

Schizophrenia-associated differentially methylated regions – consistent signals across different brain regions

Viana et al, in review

RPH3AL: calcium- dependent exocytosis

7

GRMI KALRN SRRT CRHRX LOXLX MICALI DPPB SLCXEAE DNMTDA CAMKXD HLA-L KIAAXDXO ABLIMX

SHANKI

PACSINX RICHI PPPIRIC CALMD

EPHBB

BRD6

UNCU6A

KIRRELD

PVRLX CACNAXC CPLXD;LMANXL LHPP INTSX

RGSXI

PRKDC

MMPXE CUXI

CTBPX

RIMBPI ITPKA

GNAOX MEGFXX TMEMXDID

PITPNMI

LRFNI VOPPX GLTSCRX

IGHMBPI

TMEMUBB;SAPSX CPSFX MUCB TSNAREX VPS3D

X XO-E XO-X6 XO-IO

Key for node colour

Rbest p-value within 3kb of gene coding region in largest GWAS to date CSchizophrenia Working Group of the Psychiatric Genomics Consortium Nature IOX6)

Pidsley et al, Genome Biology, 2014 Viana et al, in review

PFC co-methylated modules associated with schizophrenia

8

Normal Brain Development

Mutations Polygenic variation Environmental insults Stochastic factors

Neur Neurodevelopmental origins of mental il

• developmental origins of mental illness

lness

Schizophrenia Autism

9

179 human fetal brain samples profiled for DNA methylation 100 male, 79 females 23 to 184 days post-conception DPC calculated by Carnegie staging and fetal foot length

Fetal brain tissue from 191 elective abortions

DNA methyla6on DNA hydroxymethyla6on RNA-seq (Nick Bray) Gene6c varia6on Single-cell transcriptomics

10

>28,000 DMPs (annotated to >5,000 genes) significantly associated with fetal brain development Hypermethylation Hypomethylation

Spiers et al (2015)

http:epigenetics.iop.kcl.ac.uk/fetalbrain

11

• 5

10 15 20 −log10 (p value)

• P<1E−07

P<1E−06 P<1E−05 P>1E−05 2660000 2670000 2680000 2690000 2700000

Coordinates (Genome build 37)

• −0.5

−0.3 −0.1 0.0 0.1 Regression coefficient

5’ TTYH3 3’ Clustered regions of developmentally-coordinated DNA methylation in the human fetal brain.

12

rela6ve enrichment = 0.26, p <3.34E-92 rela6ve enrichment = 0.30, p < 3.34E-92 rela6ve enrichment = 0.34, p <3.34E-92

The distribution and direction of fetal brain dDMPs is not equal across genomic regions…

Percentage significant probes (%) 20 40 60 80 100 Hypermethylated Hypomethylated

Downstream region CGI Distal promoter CGI Distal promoter shore Intergenic CGI Proximal promoter CGI Downstream shore Intergenic shore Gene body CGI Proximal promoter shore Intergenic non−CGI Gene body shore Gene body non−CGI Downstream region non−CGI Intergenic shelf Distal promoter non−CGI Gene body shelf Proximal promoter shelf Distal promoter shelf Downstream shelf Proximal promoter non−CGI

Overall enrichment for hypomethylated dDMPs (P = 6.74E-53) CG-rich domains: enriched for hypermethylated dDMPs

13

• −0.50

−0.46 −0.42 −0.38 0.0 0.5 1.0 1.5 2.0

BTG3 (cg14394939)

Age Expression (log2(sample/ref))

p = 2.49E−06

• −0.50

−0.46 −0.42 −0.38 −1.0 −0.5 0.0 0.5

KLHL35 (cg02313829)

Age Expression (log2(sample/ref))

p = 2.08E−05

• −0.50

−0.46 −0.42 −0.38 −0.5 0.0 0.5

FAM49A (7311) (cg06829760)

Age Expression (log2(sample/ref))

p = 8.68E−05

• −0.50

−0.46 −0.42 −0.38 −3.0 −2.5 −2.0 −1.5

WIPF1 (cg18185980)

Age Expression (log2(sample/ref))

p = 3.75E−03

Correlation with gene expression data from Brain Cloud resource (http://braincloud.jhmi.edu)

~30% of top- ranked dDMPs annotated to a gene that is dynamically- expressed during brain development

14

• ●
• 50

100 150 8 10 14 18

cg03691818 (KRT77)

Days post−conception DNA methylation (%)

P=7.59E−69

• 50

100 150 10 20 30 40

cg12691488

Days post−conception DNA methylation (%)

P=1.66E−80

There are autosomal sex differences …

15

There are autosomal sex differences and distinct sex-specific developmental trajectories in the human fetal brain methylome

16

There are distinct modules of co-methylated loci in the developing human brain…

17

2 4 6 8 10 12 14 16 18 20 central nervous system neuron central nervous system projec6on dopamine biosynthe6c process central nervous system neuron axon part regula6on of dopamine metabolic regula6on of catecholamine metabolic neuron matura6on nervous system development catecholamine biosynthe6c process dopaminergic neuron differen6a6on neurofilament cytoskeleton central nervous system neuron soma6c stem cell maintenance neurofilament single-organism developmental process dopamine metabolic process developmental process response to amphetamine posi6ve regula6on of transcrip6on cell differen6a6on single-mul6cellular organism process nega6ve regula6on of cell differen6a6on cell morphogenesis involved in mul6cellular organismal process response to amine central nervous system development mul6cellular organismal development axon

…which are highly enriched for neurodevelopmental processes.

18

Brain development DMPs are enriched in genes linked to neurodevelopmental disorders

19

Correlation with developmental age −1.0 −0.5 0.0 0.5 1.0 Schizophrenia−associated DMPs 1e−20 1e−16 1e−12 1e−08 1e−04 1e+00 FDR value

Schizophrenia-associated DNA methylation differences in PFC are significantly enriched for neurodevelopmentally-dynamic sites.

Number of sites passing FDR < 0.05 Number of permutations 10 20 30 40 50 500 1000 1500 (p=8e−04) Schizophrenia DMPs

20

60 65 70 75 80 DNA methylation (%) Control (n=23) Schizophrenia (n=20) p=1.75e−06

cg00236305 (MYT1L)

• 50

100 150 40 50 60 70 80 90

cg00236305 (MYT1L)

Time post−conception (days) DNA methylation (%)

FDR = 1.39e−13 r = −0.54

21

5hmC enriched in cerebellum 5hmC generally not in promoter CGIs

22

20 40 60 80 100 0.00 0.01 0.02 0.03 0.04 0.05 Samples passing detection threshold (%) Probe density

Fetal brain Adult ba9 Adult cerebelluum

Fetal brain Adult ctx Adult cer Proportion of samples with detectable 5hmC

Spiers et al (in prep)

High amount of sample heterogeneity in fetal brain

23

DNA hydroxymethylation (5hmC) is highly dynamic across brain development at individual loci

Spiers et al (in preparation)

• 50

100 150 5 10 15 20 25 30

cg25069807 (TK1) Days post−conception DNA hydroxymethylation (%)

P = 2.52e−09

• 50

100 150 5 10 15

cg06994572 (B3GNT4) Days post−conception DNA hydroxymethylation (%)

P = 8.66e−09

• 50

100 150 −2 2 4 6 8 10 12

cg03390187 Days post−conception DNA hydroxymethylation (%)

P = 1.12e−08

• 50

100 150 5 10 15

cg16003238 (IGDCC3) Days post−conception DNA hydroxymethylation (%)

P = 1.74e−08

• 50

100 150 −5 5 10 15 20

cg22960621 (NLGN1) Days post−conception DNA hydroxymethylation (%)

P = 1.42e−09

• 50

100 150 −5 5 10 15 20 25

cg19734190 (TBC1D14) Days post−conception DNA hydroxymethylation (%)

P = 3.69e−09

• 50

100 150 −5 5 10 15 20

cg26450149 (TNS3) Days post−conception DNA hydroxymethylation (%)

P = 7.02e−09

• 50

100 150 −5 5 10 15 20 25

cg07254421 (SLC1A3) Days post−conception DNA hydroxymethylation (%)

P = 1.13e−08

24

Hannon et al (in preparation) cerebellum cortical striatum

Temporal and spatial patterns of co-methylation in the human brain across the life-course

N = 1,426

25

– 1 body, 1 genome: all you need is a blood sample – 1 life, 1 genome: you are born with the genome you die with – Any lifestyle, 1 genome: it doesn’t matter what you’re exposed to – Any disease, 1 genome: no reverse causation – A well annotated reference genome and catalogue

• f polymorphic variants

– Methods that do as they say on the box and give results that are easy to interpret

26

– 1 body, 1 genome: all you need is a blood sample – 1 life, 1 genome: you are born with the genome you die with – Any lifestyle, 1 genome: it doesn’t matter what you’re exposed to – Any disease, 1 genome: no reverse causation – A well annotated reference genome and catalogue

• f polymorphic variants

– Methods that do as they say on the box and give results that are easy to interpret

27

To what extent can easily-accessible peripheral 6ssues/ cells (e.g. whole blood, saliva, buccal) be used as a proxy for inaccessible 6ssues/ cells (i.e. brain)?

28

Profiled five matched tissues from ~100 individuals Illumina 450K array Whole Blood (pre-mortem) Cerebellum Superior Temporal Gyrus Entorhinal Cortex Prefrontal Cortex

DramaHcally more variaHon between Hssues within an individual than between individuals within a Hssue

29

The important question for epigenetic epidemiology / biomarker research: is inter- individual variation correlated across tissues?

30

hKp://epigeneHcs.iop.kcl.ac.uk/bloodbrain/

31

Small numbers Inappropriate tissues/cells Candidate gene focused Sub-optimal study designs Hype and over-interpretation

Longitudinal prospec6ve sampling from (before) birth of disease relevant 6ssues / cells (in MZ twins who become discordant for disease)…

32

Aryl hydrocarbon receptor repressor (AHRR)

F2RL3 GFI1

Blood SZ EWAS – Phase 1 (UCL)

33

Smoking EWAS

Aryl hydrocarbon receptor repressor (AHRR) Smoking EWAS US population: 25-30% Schizophrenics: 70-90%

34

Can we infer smoking status from DNA methyla6on data?

Phase 1 Phase 2

35

Non-smokers vs Smokers DNAm age Cell counts (blood and brain)

36

Blood SZ EWAS – Phase 1 (UCL) – model controlling for sex, age, cell counts, smoking 25 DMPs with P < 1x10-7 1,223 DMPs with P < 5 x 10-5

37

Using principal components to test for additional confounding in data

38

166 human fetal brain samples with genetic and DNA methylation data 56-166 days post-conception Prefrontal cortex, striatum and cerebellum samples from adult brain 16,809 mQTLs at a conservative Bonferroni-corrected significance threshold of P < 3.69x10-13 Imputation to 1,000 genomes project identified an additional 256,040 mQTLs Median DNA methylation change per allele across all identified mQTLs = 6.69% (interquartile range = 3.17%-8.96%)

39

Position of DNA methylation site (Chr) Position of SNP (Chr) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

The majority of fetal brain mQTLs (96.3%) involve SNPs and DNA methylation sites on the same chromosome

epigenetics.iop.kcl.ac.uk/mQTL

40

Trans-mQTLs in the developing human brain

~700 inter-chromosomal (trans) mQTLs (P < 3.69x10-13)

41

1 H3K4me1 H3K4me3 H3K9me3 H3K27me3 H3K36me3 1 Cerebellum DHS Cerebrum Frontal DHS Frontal Cortex DHS 1

6.55E-08 1.07E-03 1.44E-04 1.39E-15 ns 5.92E-03 6.35E-05 1.96E-04

2.96E-11 1.66E-09 1.88E-09 1.10E-06 1.12E-06 2.99E-06 6.07E-06 1.33E-04 2.01E-04 2.98E-04

Fetal Brain ChIP-seq Brain DHS sites ENCODE TFBSs Repressive marks Ac6ve transcrip6on

42

There is a high-correlation of mQTL effects between fetal brain and adult brain regions…

43

• CC

CA AA CC CA AA CC CA AA CC CA AA 10 20 30 40 50 % DNA Methylation

rs10447470 − cg07900658

het P = 7.23e−40 Fetal PFC STR CER

Examples of fetal-specific genetic effects on regulatory variation

Fetal brain Adult cortex Adult striatum Adult cerebellum

44

• CC

CA AA CC CA AA CC CA AA CC CA AA 40 50 60 70 80 90 het P = 8.39e−36 % DNA Methylation

rs2108854 − cg21577356

het P = 8.39e−36 Fetal PFC STR CER

…and opposite-effect effects across tissues and developmental stage.

Fetal brain Adult cortex Adult striatum Adult cerebellum

45

10 20 30 40 50 0.0 0.1 0.2 0.3 0.4 0.5 0.6 −log10P Density

SNPs associated with DNA methylation are more significantly associated with gene expression than non-mQTL variants.

Fetal brain mQTL Not mQTL

Wilcoxon rank sum test P < 2.2x10-16

eQTL data from BRAINIAC (Mike Weale)

46

Highly-significant enrichment of genome-wide significant schizophrenia risk variants amongst fetal brain mQTLs

P = 3.0x10-6

47

Fetal PFC STR CER

rs10422819 ; cg03709012 rs10422819 ; cg20644253 rs10282 ; cg03709012 rs10282 ; cg20644253 rs1054284 ; cg03709012 rs1054284 ; cg20644253 rs4808200 ; cg03709012 rs4808200 ; cg20644253 rs2965198 ; cg26470696 rs2965198 ; cg03709012 rs2965198 ; cg20644253 rs15622 ; cg26470696 rs15622 ; cg03709012 rs15622 ; cg20644253 rs2285627 ; cg03709012 rs2285627 ; cg20644253 rs2905421 ; cg20644253 rs769267 ; cg26470696 rs769267 ; cg03709012 rs769267 ; cg20644253 rs1009136 ; cg03709012 rs1009136 ; cg20644253 rs3934667 ; cg26470696 rs3934667 ; cg03709012 rs3934667 ; cg20644253 rs2074303 ; cg26470696 rs2074303 ; cg03709012 rs2074303 ; cg20644253 rs7212447 ; cg02206767 rs7212447 ; cg04398451 rs2955372 ; cg02206767 rs2955372 ; cg04398451 rs2955355 ; cg02206767 rs2955355 ; cg04398451 rs2955383 ; cg02206767 rs2955383 ; cg04398451 rs2955378 ; cg02206767 rs2955378 ; cg04398451 rs9913277 ; cg02206767 rs9913277 ; cg04398451 rs4584886 ; cg02206767 rs4584886 ; cg04398451 rs4368210 ; cg02206767 rs4368210 ; cg04398451 rs4459604 ; cg02206767 rs4459604 ; cg04398451 rs4072739 ; cg02206767 rs4072739 ; cg04398451 rs7219320 ; cg02206767 rs11078408 ; cg02206767 rs11078408 ; cg04398451 rs8079321 ; cg09796270 rs11078400 ; cg09796270 rs302321 ; cg14258853 rs7085104 ; cg24592962 rs7085104 ; cg11784071 rs7085104 ; cg08772003 rs7096169 ; cg24592962 rs7096169 ; cg11784071 rs7096169 ; cg08772003 rs743572 ; cg24592962 rs743572 ; cg11784071 rs743572 ; cg08772003 rs6163 ; cg11784071 rs6163 ; cg24592962 rs6163 ; cg08772003 rs6461049 ; cg19624444 rs10038174 ; cg00585072 rs10050455 ; cg00585072 rs10037757 ; cg00585072 rs10223116 ; cg00585072 rs3822346 ; cg00585072 rs1030166 ; cg00585072 rs3733707 ; cg00585072 rs3756338 ; cg00585072 rs12055222 ; cg00585072 rs2563263 ; cg00585072 rs2531352 ; cg00585072 rs2563302 ; cg26395211 rs2535627 ; cg11645453 rs10910078 ; cg13556452 rs10910078 ; cg02275930 rs10910078 ; cg07700843 rs10910078 ; cg02850689 rs4648845 ; cg02275930 rs867810 ; cg02275930

−0.1 0.05 Value

Color Key

Fetal mQTLs in schizophrenia-associated regions have larger effects on DNA methylation during neurodevelopment than the in adult brain

PFC P = 0.0420, STR P = 0.00226, CER P = 0.00998

heterogeneity

48

Bayesian co-localization analysis across 105 autosomal regions associated with schizophrenia

• posterior probabilities for 65 regions (involving 296 DNA methylation sites in

306 pairs) were supportive of a co-localized association signal for both schizophrenia and DNA methylation in that region

• 26 of these pairs (covering 15 regions associated with schizophrenia) strongly

supportive for both schizophrenia and DNA methylation being associated with the same causal variant

RPEL1 NT5C2 CNNM2 AS3MT BORCS7−ASMT BORCS7 CYP17A1 WBP1L

rs11191419

chr10:104535135 105006335 SNPs 450K Probes SCZ GWAS signal

49

Bayesian co-localization approaches to identify variants associated with both disease and genomic regulation

50

51

SMR analysis for SZ / DNA methylation

Brain Blood Green = probes that pass both steps of the SMR analysis (pleiotropy not LD)

52

53

Increased polygenic risk burden in schizophrenia brain samples

54

Increased polygenic risk burden in schizophrenia blood samples

P = 3.34x10-27 P = 2.09x10-31 Phase 1 - UCL Phase 2 - Aberdeen

55

Examples of PRS-associated DNA methylation in brain

Viana et al (in review)

Striatum Cross-brain region model

56

DMPs associated with schizophrenia diagnosis are disHnct from those associated with polygenic burden Schizophrenia EWAS Polygenic risk score EWAS

PRS-associated varia6on does not appear to be mediated by mQTL effects

57

Schizophrenia project acknowledgments

Complex Disease Epigenetics Group Eilis Hannon Helen Spiers Joana Viana Emma Dempster Therese Murphy Joe Burrage Adam Smith Ruby MacDonald Kings College London Gerome Breen Robin Murray Claire Troakes James MacCabe Garven Institute Ruth Pidsley McGill University Gustavo Turecki UCL Nick Bass Hugh Gurling Andrew McQuillin Aberdeen University David St Clair University of the Highlands and Islands Colette Mustard National Institute for Health and Welfare, Helsinki Sebastian Therman Jaakko Kaprio University of Essex Leonard Schalkwyk Cardiff University Nick Bray Mick O’Donovan Eli Lilly and Company David Collier University of Hong Kong Timothea Toulopoulou University Medical Center Utrecht Hilleke E Hulshoff Pol Marc M. Bohlken Rene S. Kahn Jena University Hospital Igor Nenadic Karolinska Institutet Christina M Hultman Trinity College Dublin Derek Morris Aiden Corvin

58

www.epigenomicslab.com Twitter: @PsyEpigenetics

59