Identifying Dysregulated Genes in Autoimmune Disease Chris Cotsapas - PowerPoint PPT Presentation

Identifying Dysregulated Genes in Autoimmune Disease Chris Cotsapas PhD Yale Neurology/Genetics Broad Institute cotsapas@broadinstitute.org

Causal l Identifying Dysregulated Genes in Autoimmune Disease Chris Cotsapas PhD Yale Neurology/Genetics Broad Institute cotsapas@broadinstitute.org

Multiple sclerosis GWAS 47 new ¡ hit 15,000 Immunochip 200,000 SNPs ¡ (targeted) ¡ 10,000 25 new ¡ hits WTCCC2 8000 650,000 SNPs ¡ MS 6000 Subjects Meta-­‑Analysis 3 new ¡ hits 2.6 million ¡SNPs ¡ Meta v2.5 4000 6 new ¡ hits 2.6 million ¡SNPs ¡ IMSGC GWAS 1 new ¡ hit 345,000 SNPs ¡ 2000 ANZ ¡GWAS 2 hits 550,000 SNPs ¡ IMSGC NEJM 2007 De Jager et al. Nat Genet 2009 2007 2008 2012 2009 2010 2011 Rubio et al. Nat Genet 2009 Date of comple/on ¡ IMSGC Nature 2011 Patsopoulos et al. Ann Neurol 2011

… and that’s not all! Meta-­‑Analysis v3.0 16K MS cases / 26K controls Replica;on 18K MS cases / 18K controls 100 new ¡hits 47 new ¡ hit 15,000 Immunochip 200,000 SNPs ¡ (targeted) ¡ 10,000 25 new ¡ hits WTCCC2 8000 650,000 SNPs ¡ MS 6000 Subjects Meta-­‑Analysis 3 new ¡ hits 2.6 million ¡SNPs ¡ Meta v2.5 4000 6 new ¡ hits 2.6 million ¡SNPs ¡ IMSGC GWAS 1 new ¡ hit 345,000 SNPs ¡ 2000 ANZ ¡GWAS 2 hits 550,000 SNPs ¡ IMSGC NEJM 2007 De Jager et al. Nat Genet 2009 2007 2008 2012 2009 2010 2011 Rubio et al. Nat Genet 2009 Date of comple/on ¡ IMSGC Nature 2011 Patsopoulos et al. Ann Neurol 2011

GWAS signals are enriched in regulatory DNA Maurano et al Science 2012

MS GWAS hits enriched in transcription factor binding sites Farh et al Nature 2015

NFKB1 locus in MS GWAS Plotted SNPs r 2 10 100 rs7665090 0.8 0.6 0.4 8 80 0.2 Recombination rate (cM/Mb) ● − log 10 (p − value) 6 60 ● ● ● ● 4 40 ● ● ● ● 2 20 ● ● ● ● ● ● ● ● ● ● ● ● ● 0 0 NFKB1 MANBA UBE2D3 103.4 103.5 103.6 103.7 Position on chr4 (Mb) IMSGC, Nat Genet, 2013 Housley et al STM 2015

Supplementary File IKZF3/ORMDL3 locus in MS GWAS Supplementary Figure 42. Discovery phase rs12946510. A B IMSGC, Nat Genet, 2013 A) Regional Association and B) Forest Plot 72

Approach Total gene posterior Gene GP = Σ ( CP DHS,gene x PPA DHS ) Gene-DHS correlation posterior CP x PPA Regulatory DHS posterior RP = Σ ( PPA DHS ) Posterior probability of association PPA SNP Parisa Shooshtari

Problem 1: DHS-gene correlations Gene 1 Gene 2 CD 3 CD 3 CD 14 CD 14 CD 34 CD 34 Thymus Thymus Kidney Kidney Lung Lung Heart Heart Brain Brain Parisa Shooshtari

Aligning DHSs Over Samples Parisa Shooshtari

Identify detectable DHS clusters Scenario 1 Scenario 2 C1 1 C1 2 C2 1 C2 2 Tissues Hotspot peaks C3 1 C3 1 1 T1-Rep1 C4 0 C4 0 2 C5 0 C5 0 T1-Rep2 1 C6 1 C6 2 1 T2-Rep1 1 T2-Rep2 0 1 T3-Rep1 Unreliable Cluster Reliable Cluster 2 T3-Rep2 1 1 T4-Rep1 2 T4-Rep2 1 Align over 57 tissue replicates from REP 1 T5-Rep1 1 1,079,138/1,994,675 (54.1%) clusters pass 0 T5-Rep2 Cover 8% of genome (cf. 14% of all DHS) 1 T6-Rep1 2 T6-Rep2 1 NB singletons, low power Parisa Shooshtari

Parisa Shooshtari DHS clusters are (250-400bp vs 150bp wider than DHS peaks Caveat QC+ DHS clusters capture most MS heritability Hilary Finucane Alkes Price DHS_Peaks_Hotspot DHS_Clusters Status Penis_Foreskin_Melanocyte_Primary_Cells Penis_Foreskin_Keratinocyte_Primary_Cells Penis_Foreskin_Fibroblast_Primary_Cells Pancreas Mobilized_CD4_Primary_Cells Mobilized_CD34_Primary_Cells Mobilized_CD3_Primary_Cells IMR90 H9 H1_Derived_Mesenchymal_Stem_Cells H1_BMP4_Derived_Trophoblast_Cultured_Cells H1_BMP4_Derived_Mesendoderm_Cultured_Cells H1 Gastric Fibroblasts_Fetal_Skin_Upper_Back Fibroblasts_Fetal_Skin_Scalp Fibroblasts_Fetal_Skin_Quadriceps_Right Proportion of Heritability Fibroblasts_Fetal_Skin_Quadriceps_Left Fibroblasts_Fetal_Skin_Biceps_Right Fibroblasts_Fetal_Skin_Biceps_Left Fibroblasts_Fetal_Skin_Back Fibroblasts_Fetal_Skin_Abdomen Fetal_Thymus Fetal_Testes Fetal_Stomach Fetal_Spinal_Cord Cell Fetal_Renal_Pelvis_Right Fetal_Renal_Pelvis_Left Fetal_Renal_Pelvis Fetal_Renal_Cortex_Right Fetal_Renal_Cortex_Left Fetal_Renal_Cortex Fetal_Placenta Fetal_Muscle_Trunk Fetal_Muscle_Leg Fetal_Muscle_Back Fetal_Muscle_Arm Fetal_Lung_Right Fetal_Lung_Left Fetal_Lung Fetal_Kidney_Right Fetal_Kidney_Left Fetal_Kidney Fetal_Intestine_Small Fetal_Intestine_Large Fetal_Heart Fetal_Brain Fetal_Adrenal_Gland CD8_Primary_Cells CD56_Primary_Cells CD4_Primary_Cells CD3_Primary_Cells CD19_Primary_Cells CD14_Primary_Cells Breast_vHMEC 0.4 0.2 0.0 Proportion of h2g

Challenge 2: Gene expression correlation QQ plot for P Value of Correlation Between One DHS and 14000 Genes Correlation Structure of the Gene Expression Data After Before Correction Correction Parisa Shooshtari

Approach Total gene posterior Gene GP = Σ ( CP DHS,gene x PPA DHS ) Gene-DHS correlation posterior CP x PPA Regulatory DHS posterior RP = Σ ( PPA DHS ) Posterior probability of association PPA SNP Parisa Shooshtari

Application to MS GWAS Gene GP Chr 6 MDN1 0.555 90.5-91.5Mb BACH2 0.162 RP = 0.945 GABRR2 0.106 RRAGD 0.065 GJA10 0.029 MAP3K7 0.028 Parisa Shooshtari IMSGC NG 2013

IKZF3/ORMDL3 locus Gene GP Chr 17 ORMDL3 0.029 34.5-35.5Mb PIP4K2B 0.022 RP = 0.295 IGFBP4 0.018 IKZF3 0.015 Supplementary File GSDMB 0.014 Supplementary Figure 42. Discovery phase rs12946510. SMARCE1 0.013 A CCR7 0.013 TNS4 0.01 ZPBP2 0.009 MED1 0.009 MED24 0.009 KRT24 0.009 PNMT 0.008 CDK12 0.007 RPL23 0.007 PSMD3 0.007 PLXDC1 0.006 TOP2A 0.006 RARA 0.006 B A) Regional Association and B) Forest Plot 72

MS GWAS hits enriched in transcription factor binding sites Farh et al Nature 2015

MS GWAS risk effect: NFKB1 locus

MS patients show altered NF κ B signaling in CD4 + T cells Figure 1 . Naïve CD4 cells from patients with MS exhibit increased phospho-p65 NF κ B . Flow cytometry of PBMCs from age-matched healthy ex vivo CD4 + T cells show higher CD4 + T cells from MS patients control (HC) and relapsing-remitting MS (RRMS) p-p65 (Housley et al , STM 2015) patients stained for CD4, CD45RA, CD45RO, and proliferate more rapidly after pS529 p65 NF κ B. MFI of p65 results are shown stimulus (Kofler et al JCI 2014) gated on naïve CD4 + CD45RA + CD45RO - T-cells.

MS risk effect near NFKB1 alters signaling in CD4+ cells a b p= 0.0091 p= 0.019 p= 0.029 p= 0.027 100 2000 80 1500 60 % I κ B α pNF κ B 1000 40 500 20 0 0 rs228614 GG AG AA GG AG AA GG AG AA GG AG AA GG AG AA No Stim TNF- α PMA TNF- α PMA Nuclear localization p= 0.05 30 p= 0.037 GG 30 AA 20 p50 NFkB 20 10 10 0 0 0 15 30 GG AA Housley et al, STM 2015 Minutes

MS variant in TNFRSF1A alters TNF α -dependent NF κ B signaling Housley, unpublished

GWAS loci harbor many NF κ B genes Housley, unpublished

Model: NF κ B signaling variation p50 Activation p65 P-p50 *NF κ B Proliferation External Survival stimulus p50 Activation Proliferation p65 P-p50 *NF κ B Survival Broader phenotype? GV in NF κ B pathway New gene activation patterns by NF κ B GV in NF κ B TFBS

Systematic dissection I 25 NFKB1 risk variant homozygotes 25 NFKB1 non-risk variant homozygotes TNF α α CD3/CD28 CD4 + T cells Timing 0 15m 30m 2h 3d Event exposure signaling translocation gene activation cell phenotype CyTOF H3K27Ac-seq RNA-seq Brad Bernstein Chris Cotsapas NF κ B CHiP-seq David Hafler Will Housley Phospho-flow NF κ B

Acknowledgements • IMSGC • Yale labs – David Hafler – Parisa Shooshtari – Phil De Jager – Mitja Mitrovic – Steve Hauser – Alex Casparino – Adrian Ivinson – Will Housley – Nikos Patsopoulos – Many, many others • Partners – David Hafler – Phil De Jager – Brad Bernstein – John Stamatoyannopoulos

8.6e-09 1.6e-08 3.2e-09 1.5e-08 1.5e-08

Genes DHS1 DHS2 DHS4 DHS5 DHS3 Burdened DHSs