Towards reconstructing personalised causal regulatory networks using - PowerPoint PPT Presentation

Towards reconstructing personalised causal regulatory networks using large-scale trans -eQTL and single-cell co-expression QTL analysis Annique Claringbould Department of Genetics University Medical Center Groningen Slides Lude Franke

Twelve years of genome-wide association studies Cause: Effect:

The black box challenge Genetic risk factors Disease Black Box Genes _ unknown >10,000 known >200 diseases Pathways _ unknown Cell-types _ unknown

Far majority of genetic risk factors affect gene expression Cell-type specific cis -eQTL effect: T-Cell B-Cell Monocyte gene X expression > P = 10 -9 P = 0.8 P = 0.9 CC CT TT CC CT TT CC CT TT genetic risk factor for type 1 diabetes Dubois et al , Nature Genetics 2010 Fehrmann et al, Nature Genetics 2015 Westra et al , Nature Genetics 2013 Zhernakova et al , Nature Genetics 2017

Effects of genetic variants on single- cell gene expression

scRNA-seq analysis in 25,000 PBMCs (45 different individuals) 1 CD4+ T 2 CD8+ T 3 CD56(dim) NK 4 CD56(bright) NK 5 cMonocyte t-SNE 2 6 ncMonocyte 7 B 8 Plasma 9 mDC 10 pDC 11 Megakaryocyte 12 HPC t-SNE 1 Monique van der Wijst et al , Nature Genetics 2018

Single-cell cis- eQTL analysis rs4821670 affects LGALS2 in cis : rs9332431 affects CHTF8 in cis : Monique van der Wijst et al , Nature Genetics 2018

Downstream effects using 31,684 samples

Goal Disease Disease Disease Disease Disease Genome-wide SNP SNP SNP SNP SNP association studies cis -eQTL mapping cis -eQTL e ff ects: A B C D E trans -eQTL mapping trans -eQTL e ff ects: X Y Tissue 1 Tissue 2 Z Key driver gene Key driver gene identification Disease

Get larger sample-sizes: meta-analysis in 5,311 samples Systemic lupus erythematosis risk factor: Chr. 7 Local expression e ff ect: IKZF1 Chr. 7 Downstream Type 1 interferon response: IFI6 IFI44L IFIT1 MX1 
 d e fi i t n e trans-eQTL d i (in Monocytes) s t c s e r f e ff ects f o e t c m a f a k e s y r i n t r s a n c m w i t e o g n D n e i s g u 3 6 , 1 4 e 0 r 3 o 2 r m o f d n fi o t s m e l p i A m a 8 s 1 e 0 r 2 o m Westra et al , Nature Genetics 2013 Zhernakova et al , Nature Genetics 2017 Bonder et al , Nature Genetics 2017

Large-scale eQTL analysis: eQTLGen networks genetic variants genes eQTLGen trans-eQTLs Large-scale eQTL analysis: 37 population based cohorts Genotype data and gene genenetwork.nl/eqtlgen expression in blood available 31,684 samples www.eqtlgen.org

Large-scale eQTL analysis: eQTLGen A eQTLGen Consortium 31,684 blood samples 10,317 trait-associated SNPs 11M SNPs (MAF ≥ 1%) 3 19,960 genes studied 5 ' ' cis -eQTL analysis: trans -eQTL analysis: Polygenic risk score analysis: 11M SNPs studied 10,317 trait-associated 1,263 traits studied (Window size 1Mb, MAF ≥ 1%) SNPs studied Disease Disease Y SNP SNP Gene expression Peter A John trans -eQTL e fg ects Kate Susan cis -eQTL e fg ect X Y Z Gene A Gene B Gene C Polygenic risk for disease > cis -eQTL analysis results: trans -eQTL analysis results: Polygenic score analysis results: 16,989 (88.3%) cis-eQTL genes 6,298 (31%) trans-eQTL genes 2,658 (13%) eQTS genes 238,340 unlinked cis -eQTL SNPs 3,853 (36%) genetic risk factors 689 (54%) traits a fg ect gene expression Võsa et al , BioRxiv 2018

Expression levels of nearly every gene are influenced by SNPs cis- eQTLs Nearly every gene is showing a significant cis -eQTL effect P = 0.54 P = 0.22 P = 0.95 P = 0.09 P = 0.15 P = 0.002 P = 0.02 P = 4 x 10 −4 P = 2 x 10 -7 P = 2 x 10 -6 pLI score 1.0 0.8 0.6 0.4 0.2 0.0 100% 90% showing cis -eQTL effect 80% Proportion of genes Proportion of genes 70% 60% 50% 40% 30% 20% 10% 0% Low High Average blood gene expression −5

Loss of function intolerant genes Gene showing no eQTL effect in blood, but showing eQTL in GTEx 34% −4 Enriched pathway P-Value Carcinoma 5 x 10 -11 RNA processing 2 x 10 -9 66% RNA splicing 2 x 10 -7 Genes showing no eQTL effect in eQTLGen nor in GTEx 100% 90% showing cis -eQTL effect 80% Proportion of genes Proportion of genes 70% 60% 50% 40% 30% 20% 10% 0% Low High Average blood gene expression −5

96.2% of lead eSNPs map within 100kb of cis -gene

Limited evidence blood cis-eQTLs pinpoint disease genes Analysis of cis -eQTLs using SMR for 16 well-powered traits Prioritized SMR genes do not overlap more often than expected with genes, prioritised using pathway enrichment method DEPICT

−4 37% of genetic risk factors for disease affect expression in trans trans- eQTLs 37% of 10,000 risk factors affect gene expression levels in trans P = 0.54 P = 0.78 P = 0.13 P = 3 x 10 -6 P = 0.005 P = 9 x 10 -4 P = 0.95 P = 0.10 P = 10 −5 P = 6 x 10 -7 pLI score 1.0 0.8 0.6 0.4 0.2 0.0 100% 90% showing trans -eQTL effect 80% 70% 60% 50% 40% 30% 20% 10% 0% Low High Average blood gene expression

Biological mechanism of trans -eQTLs Transcription Co-expression factor binding Disease Disease SNP SNP X A TF X A B trans- eQTL gene trans- eQTL gene Fold enrichment = 2.2x, P = 2 x 10 -61 Fold enrichment = 1.98x, P = 4 x 10 -83 Susceptibility locus Susceptibility locus (RegulatoryCircuits, Marbach et al , (Co-expression based on Nature Biotechnology 2016) 31,684 eQTLGen samples) Biological mechanism known for trans -eQTLs Indirect transcription 1 Expression levels of local gene 0 % 4% 3% factor binding mediate trans -eQTL e fg ect Disease Disease SNP SNP Y X A TF A B trans- eQTL gene co-expression Susceptibility Fold enrichment = 5.3x, P = 10 -67 Susceptibility X locus (Tested in 3,831 BIOS samples) locus trans- eQTL gene Fold enrichment = 3.2x, P < 10 -300 (RegulatoryCircuits, Marbach et al , Biological mechanism Nature Biotechnology 2016, co-expression based on 31,684 eQTLGen samples) unknown for trans -eQTLs 83% Disease Protein-protein interaction SNP Disease Close physical SNP proximity X X A B trans- eQTL gene trans- eQTL gene Susceptibility locus Fold enrichment = 1.19x, P = 0.05 Fold enrichment = 0.99x, P = 0.30 (Protein interactions based on InWeb) (Hi-C interactions, Rao et al, Cell 2014)

Using blood trans -eQTLs to gain insight into brain genes CAD SNP affects REST transcription factor: Trans -genes specific for brain: rs17087335 trans cis REST

Trans-eQTL effects in cancer rs116766442 Multiple myeloma Testicular rs2900333 risk factor germ cell tumor risk cis -eQTL factor trans -eQTL e fg ect e fg ects O-glycan biosynthesis GOLM1 Chromatin organization ATF7IP trans -eQTL e fg ects trans -eQTL e fg ect Multiple myeloma DDX43 GTSF1 FAM50B rs4487645 risk factor Gametocyte Male meiosis Male meiosis / speci fj c factor 1 Highly Highly expres- expressed sed in testis in testis rs7745098 Hodgkin’s lymphoma Melanoma rs1801516 risk factor risk factor missense trans -eQTL e fg ect variant Cell cycle RTKN2 DNA repair ATM trans -eQTL e fg ects trans -eQTL e fg ect HIST1H2BD HIST1H2AC HIST1H2BE Hodgkin’s lymphoma rs114865495 risk factor HIST1H1C H1F0 HIST2H2BE HIST1H4H HIST2H2BF HIST1H1PS1 HIST1H3D HIST1H2BC HIST1H4E DNA Damage / Telomere Stress Induced Senescence

Converging effects in systemic lupus erythematosis 1q25.3 rs17849501 7p12.2 IFI6 12q24.12 rs4917014 OASL rs10774625 2q24.2 IFIT1 EPSTI1 rs1990760 rs597808 HERC5 rs2111485 MX1 IFI44 OAS3 ISG15 rs35472514 RSAD2 IFI44L Interferon genes rs34572943 OAS2 rs11574637 rs2663052 rs877819 rs7097397 rs1143679 rs9888739 rs1913517 > Expression of interferon genes 16p11.2 10q11.23 EPSTI1 PARP14 PARP9 IFI6 IFIT3 OAS2 OAS3 RSAD2 XAF1 > Polygenic SLE Risk > OAS1 MX1 IFI44 IFI44L HERC5 IFIT2 HELZ2 IFIT1 EIF2AK2 DDX58 CMPK2 OASL

What is a polygenic risk score? 403 variants associated to type 2 diabetes Polygenic scores calculated for 1,263 diseases and traits Bob Carl Alice 11 risk alleles 189 risk alleles 362 risk alleles

Genetic risk scores on several metabolites PGS correlations: SLC7A1 ASNS ALKBH7 CHRM3-AS2 ANKRD36BP2 RP11-439E19.8 FBXO9 Glycine N-acteylgycine L-serine Creatine PHGDH PSAT1 AARS ANKHD1 PHDGH and PSAT1: Glycose 3-Phosphoglycerate Pyruvate enzymes in formation of 3-PGDH serine, acetylglycine, 3-Phosphohydroxypyruvate glycine and creatine PSAT 3-Phosphoserine PSPH Phospholipids L-serine SHMT N-acteylglycine Glycine Creatine Glycine is upstream on Derivative of glycine biosynthetic pathway

HDL Cholesterol: genetic risk score correlation Foam cell Liver Apo A-1 Cholesterol Nascent HDL ABCA1 SR-BI Mature ABCG1 HDL CETP LDLR LDL / VLDL SREBP2 ( SREBF2 )