Systems Genetics: Developing a Systems-Level Understanding of Human Disease Genetics Charles R. Farber, Ph.D. Associate Professor, PHS and BMG Associate Director, Center for Public Health Genomics @cr_farber [email protected] farberlab.com UVa School of Medicine Research Retreat February 4th, 2017 Slides on speakerdeck.com 

No content

Systems Genetics 

Systems Genetics 

Systems Genetics “understanding how genetic information is integrated, coordinated, and ultimately transmitted through molecular, cellular, and physiological networks to enable the higher-order functions and emergent properties of biological systems.” Nadeau and Dudley. 2011. Science 

Variation in Bone Strength Genetic Variation Seeman, E., & Delmas, P. D. (2006). The New England Journal of Medicine, 354(21), 2250–2261. Systems Genetics of Bone 

• Osteoporosis is a condition of reduced bone strength • ~44 million Americans either have or are at high risk • 1.5 million fractures annually - ~$18 billion in health care expenditures Osteoporosis 

Osteoporosis Bone Strength Osteoblast Activity Systems Genetics in the Farber Lab Experimental Approaches Computational Approaches 

Genome-Wide Association Studies 0 2 4 6 8 10 −log10 (p−value) 0 20 40 60 80 100 Recombination rate (cM/Mb) ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● rs11228262 0.2 0.4 0.6 0.8 r2 CHKA SUV420H1 C11orf24 LRP5 PPP6R3 GAL MTL5 CPT1A MRPL21 IGHMBP2 68 68.2 68.4 68.6 Position on chr11 (Mb) Plotted SNPs • Bone Mineral Density (BMD) • ~350 associations Estrada et al. 2012. Nature Genetics Kemp et al. submitted • Treasure trove of new genes and novel biology 

Genome-Wide Association Studies 0 2 4 6 8 10 −log10 (p−value) 0 20 40 60 80 100 Recombination rate (cM/Mb) ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● rs11228262 0.2 0.4 0.6 0.8 r2 CHKA SUV420H1 C11orf24 LRP5 PPP6R3 GAL MTL5 CPT1A MRPL21 IGHMBP2 68 68.2 68.4 68.6 Position on chr11 (Mb) Plotted SNPs • Which genes are causal and how do they impact BMD? • How do the causal genes impact systems-level function? 

Osteoblast activity Osteoclast activity Genes that influence a disease are often functionally related 

Osteoblast activity Osteoclast activity Genes that influence a disease are often functionally related Functionally related genes are often co- expressed 

Osteoblast activity Osteoclast activity Genes that influence a disease are often functionally related Functionally related genes are often co- expressed 

1. Causal gene predictions 2. Functional annotation 3. How key genetic drivers impact bone networks Mepe Kcnma1 Meox1 Hoxc6 Hdac5 Sost Dkk1 Fubp3 Btbd3 Mef2c Wnt5b Hic1 Fam3c Dnm3os Cped1 Sptbn1 Lrp5 Wnt16 Smg6 Sp7 Tnfrsf11b Mark3 Lrp4 Tmem263 Insc Samd12 Ppp6r3 Wls Wnt4 Galnt3 Dlg5 Mertk Cyld 

64 SNPs P<5.0 x 10-8 234 Genes GEFOSII BMD GWAS (N~80K) (Estrada, K., et al. 2012. Nature Genetics) 

64 SNPs P<5.0 x 10-8 234 Genes GEFOSII BMD GWAS (N~80K) (Estrada, K., et al. 2012. Nature Genetics) Bone transcriptomic profiles from 96 inbred mouse strains (Calabrese et al. 2012. PLOS Genetics) Genome-wide co-expression network 

GWAS gene enrichment in modules 6 and 9 0 1 2 3 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 −log10(FIsher’s P-value) Module Calabrese et al. 2017. Cell Systems Mepe Kcnma1 Meox1 Hoxc6 Hdac5 Sost Dkk1 Fubp3 Btbd3 Mef2c Wnt5b Hic1 Fam3c Dnm3os Cped1 Sptbn1 Lrp5 Wnt16 Smg6 Sp7 Tnfrsf11b Mark3 Lrp4 Tmem263 Insc Samd12 Ppp6r3 Wls Wnt4 Galnt3 Dlg5 Mertk Cyld Modules 6 and 9 contained 33 genes located in 30 GWAS loci 

Modules 6 and 9 are enriched for genes known to regulate osteoblast activity and BMD 

Modules 6 and 9 are enriched for genes known to regulate osteoblast activity and BMD Module 9 - 354 genes 1. Enriched for Osteoblast Differentiation Genes (P=3.2 x 10-5) 2. Highly expressed in osteoblasts 3. Module expression was correlated with BMD (P=1.4 x 10-6) 4. Knockout of hub genes impact osteoblast differentiation and BMD 

Modules 6 and 9 are enriched for genes known to regulate osteoblast activity and BMD Module 9 - 354 genes 1. Enriched for Osteoblast Differentiation Genes (P=3.2 x 10-5) 2. Highly expressed in osteoblasts 3. Module expression was correlated with BMD (P=1.4 x 10-6) 4. Knockout of hub genes impact osteoblast differentiation and BMD −0.2 0.0 0.2 −0.3 −0.2 −0.1 0.0 0.1 0.2 Module 9 Eigengene Module 6 Eigengene r=0.63, P=6.4 x 10-12 Calabrese et al. 2012. PLOS Genetics 

GWAS SNP Chr SNP Mbp Gene Novel rs7521902 1 22.164 WNT4 NO rs6426749 1 22.385 WNT4 NO rs17482952 1 68.174 WLS NO rs12407028 1 68.182 WLS NO rs479336 1 172.23 DNM3OS NO rs1346004 2 165.745 GALNT3 NO rs6532023 4 87.853 MEPE NO rs1366594 5 89.08 MEF2C NO rs13245690 7 121.145 WNT16 NO rs3801387 7 121.335 WNT16 NO rs2062377 8 118.995 TNFRSF11B NO rs1373004 10 52.668 DKK1 NO rs7932354 11 46.701 LRP4 NO rs3736228 11 68.434 LRP5 NO rs2887571 12 1.529 WNT5B NO rs2016266 12 53.334 SP7 NO rs736825 12 54.024 HOXC6 NO rs1564981 16 50.952 CYLD NO rs1566045 16 50.988 CYLD NO rs4792909 17 43.721 MEOX1 NO rs4792909 17 43.721 SOST NO rs227584 17 44.148 HDAC5 NO GWAS SNP Chr SNP Mbp Gene Novel rs4233949 2 54.433 SPTBN1 YES rs17040773 2 111.742 MERTK YES rs13245690 7 121.145 CPED1 YES rs3801387 7 121.335 FAM3C YES rs2062377 8 118.995 SAMD12 YES rs7851693 9 130.603 FUBP3 YES rs7071206 10 77.642 KCNMA1 YES rs7071206 10 77.642 DLG5 YES rs7108738 11 15.689 INSC YES rs3736228 11 68.434 PPP6R3 YES rs1053051 12 106.973 TMEM263 YES rs11623869 14 103.417 MARK3 YES rs4790881 17 2.166 HIC1 YES rs4790881 17 2.166 SMG6 YES rs3790160 20 10.659 BTBD3 YES Osteoblast Functional Module (OFM) KNOWN NOVEL Calabrese et al. 2017. Cell Systems 

GWAS SNP Chr SNP Mbp Gene Novel rs7521902 1 22.164 WNT4 NO rs6426749 1 22.385 WNT4 NO rs17482952 1 68.174 WLS NO rs12407028 1 68.182 WLS NO rs479336 1 172.23 DNM3OS NO rs1346004 2 165.745 GALNT3 NO rs6532023 4 87.853 MEPE NO rs1366594 5 89.08 MEF2C NO rs13245690 7 121.145 WNT16 NO rs3801387 7 121.335 WNT16 NO rs2062377 8 118.995 TNFRSF11B NO rs1373004 10 52.668 DKK1 NO rs7932354 11 46.701 LRP4 NO rs3736228 11 68.434 LRP5 NO rs2887571 12 1.529 WNT5B NO rs2016266 12 53.334 SP7 NO rs736825 12 54.024 HOXC6 NO rs1564981 16 50.952 CYLD NO rs1566045 16 50.988 CYLD NO rs4792909 17 43.721 MEOX1 NO rs4792909 17 43.721 SOST NO rs227584 17 44.148 HDAC5 NO GWAS SNP Chr SNP Mbp Gene Novel rs4233949 2 54.433 SPTBN1 YES rs17040773 2 111.742 MERTK YES rs13245690 7 121.145 CPED1 YES rs3801387 7 121.335 FAM3C YES rs2062377 8 118.995 SAMD12 YES rs7851693 9 130.603 FUBP3 YES rs7071206 10 77.642 KCNMA1 YES rs7071206 10 77.642 DLG5 YES rs7108738 11 15.689 INSC YES rs3736228 11 68.434 PPP6R3 YES rs1053051 12 106.973 TMEM263 YES rs11623869 14 103.417 MARK3 YES rs4790881 17 2.166 HIC1 YES rs4790881 17 2.166 SMG6 YES rs3790160 20 10.659 BTBD3 YES Osteoblast Functional Module (OFM) KNOWN NOVEL Prediction: Many OFM genes are likely to be responsible for their respective GWAS associations Calabrese et al. 2017. Cell Systems 

SNP Group OFM Lead SNPs (N=30) NOFM Lead SNPs (N=34) OFM matched random SNP set 0 10 20 30 40 50 H3K27ac H3K4me1 H3K4me2 H3K4me3 H3K79me2 H3K36me3 H3K9me3 H3K27me3 Overlap (%) P=0.02 P=0.02 P<0.001 P<0.001 P<0.001 Comparing OFM and non-OFM loci Histone Modifications in Primary Osteoblasts ENCODE Project Consortium. (2012). Nature. 

Comparing OFM vs non-OFM loci OSTEOBLASTS CHONDROCYTES MESENCHYMAL STEM CELLS CD184+ ENDODERM CELLS PREFRONTAL CORTEX FORESKIN FIBROBLAST 10 20 30 40 −0.1 0.0 0.1 0.2 0.3 Difference (% OFM overlap - % NOFM overlap) Overlap (%) H3K27ac OFM Overlap (%) Roadmap Epigenomics Consortium (2015). Nature, 518(7539), 317–330 • These data link OFM loci to gene regulation in osteoblasts supporting an involvement for OFM genes 

GWAS SNP Chr SNP Mbp Gene Novel rs7521902 1 22.164 WNT4 NO rs6426749 1 22.385 WNT4 NO rs17482952 1 68.174 WLS NO rs12407028 1 68.182 WLS NO rs479336 1 172.23 DNM3OS NO rs1346004 2 165.745 GALNT3 NO rs6532023 4 87.853 MEPE NO rs1366594 5 89.08 MEF2C NO rs13245690 7 121.145 WNT16 NO rs3801387 7 121.335 WNT16 NO rs2062377 8 118.995 TNFRSF11B NO rs1373004 10 52.668 DKK1 NO rs7932354 11 46.701 LRP4 NO rs3736228 11 68.434 LRP5 NO rs2887571 12 1.529 WNT5B NO rs2016266 12 53.334 SP7 NO rs736825 12 54.024 HOXC6 NO rs1564981 16 50.952 CYLD NO rs1566045 16 50.988 CYLD NO rs4792909 17 43.721 MEOX1 NO rs4792909 17 43.721 SOST NO rs227584 17 44.148 HDAC5 NO GWAS SNP Chr SNP Mbp Gene Novel rs4233949 2 54.433 SPTBN1 YES rs17040773 2 111.742 MERTK YES rs13245690 7 121.145 CPED1 YES rs3801387 7 121.335 FAM3C YES rs2062377 8 118.995 SAMD12 YES rs7851693 9 130.603 FUBP3 YES rs7071206 10 77.642 KCNMA1 YES rs7071206 10 77.642 DLG5 YES rs7108738 11 15.689 INSC YES rs3736228 11 68.434 PPP6R3 YES rs1053051 12 106.973 TMEM263 YES rs11623869 14 103.417 MARK3 YES rs4790881 17 2.166 HIC1 YES rs4790881 17 2.166 SMG6 YES rs3790160 20 10.659 BTBD3 YES Osteoblast Functional Module (OFM) KNOWN NOVEL Prediction: Many OFM genes are likely to be responsible for their respective GWAS associations 

MARK3 is predicted to be causal gene at BMD GWAS locus on Chr14q32.32 0 1 2 3 4 5 6 7 rs11623869 ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● 0.2 0.4 0.6 0.8 r2 GEFOSII Femoral Neck BMD EIF5 SNORA28 MARK3 CKB TRMT61A BAG5 APOPT1 103.75 103.8 103.85 103.9 103.95 104 Position on chr14 (Mb) -log10(P-value) 

MARK3 is predicted to be causal gene at BMD GWAS locus on Chr14q32.32 0 1 2 3 4 5 6 7 rs11623869 ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● 0.2 0.4 0.6 0.8 r2 GEFOSII Femoral Neck BMD EIF5 SNORA28 MARK3 CKB TRMT61A BAG5 APOPT1 103.75 103.8 103.85 103.9 103.95 104 Position on chr14 (Mb) -log10(P-value) 9.00 9.25 9.50 9.75 10.00 0.06 0.07 0.08 0.09 0.10 Femoral BMD (g/cm2) Bone Mark3 Expression (log 2 (INT)) r=-0.25, P=0.036 

MARK3 GTEX Thyroid 0 10 20 30 40 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0.2 0.4 0.6 0.8 r2 EIF5 SNORA28 CKB TRMT61A BAG5 APOPT1 103.75 103.8 103.85 103.9 103.95 104 PP4=97.4% PP3=2.5% 0 1 2 3 4 5 6 7 ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● 0.2 0.4 0.6 0.8 r2 EIF5 SNORA28 MARK3 CKB TRMT61A BAG5 APOPT1 103.75 103.8 103.85 103.9 103.95 104 Position on chr14 (Mb) GEFOSII Femoral Neck BMD -log10(p−value) Position on chr14 (Mb) MARK3 MARK3 local eQTL rs11623869 “G” allele associated with increased BMD and decreased MARK3 GTEx Consortium. 2015. Science 

Primary calvarial osteoblasts Mark3 siRNAs Differentiate Quantify Mineralization SC Mark3_1 Mark3_2 0.09 0.12 0.15 0.18 0.21 Alizarin Red (mM) SC Mark3 1 Mark3 2 P<0.05 0.00 0.25 0.50 0.75 1.00 SC Mark3 1 Mark3 2 Mark3 Relative Expression P<0.01 

Mark3 knockdown in vivo increases cortical BMD 0.00 0.25 0.50 0.75 1.00 1.25 +/+ +/- -/- Genotype Mark3 Relative Expression P<0.05 Mark3 KOMP gene trap mouse 

0.050 0.052 0.054 0.056 12 16 Age Spine BMD (g/cm2) 0.065 0.070 0.075 0.080 12 16 Age Femur BMD (g/cm2) * * 0.00 0.25 0.50 0.75 1.00 1.25 +/+ +/- -/- Genotype Mark3 Relative Expression Genotype +/+ +/- Genotype +/+ +/- P<0.05 Mark3 knockdown in vivo increases cortical BMD 0.00 0.25 0.50 0.75 1.00 1.25 +/+ +/- -/- Genotype Mark3 Relative Expression P<0.05 Mark3 KOMP gene trap mouse 

0.050 0.052 0.054 0.056 12 16 Age Spine BMD (g/cm2) 0.065 0.070 0.075 0.080 12 16 Age Femur BMD (g/cm2) * * 0.00 0.25 0.50 0.75 1.00 1.25 +/+ +/- -/- Genotype Mark3 Relative Expression Genotype +/+ +/- Genotype +/+ +/- P<0.05 20 30 40 50 +/+ -/- Genotype BA/TA (%) 0.05 0.10 0.15 0.20 Genotype Ct.Th (mm) 1000 1050 1100 1150 1200 Genotype TMD (mg HA/cm3) * * * Femur (cortical bone) +/+ -/- +/+ -/- Mark3 knockdown in vivo increases cortical BMD 0.00 0.25 0.50 0.75 1.00 1.25 +/+ +/- -/- Genotype Mark3 Relative Expression P<0.05 Mark3 KOMP gene trap mouse 

0.050 0.052 0.054 0.056 12 16 Age Spine BMD (g/cm2) 0.065 0.070 0.075 0.080 12 16 Age Femur BMD (g/cm2) * * 0.00 0.25 0.50 0.75 1.00 1.25 +/+ +/- -/- Genotype Mark3 Relative Expression Genotype +/+ +/- Genotype +/+ +/- P<0.05 20 30 40 50 +/+ -/- Genotype BA/TA (%) 0.05 0.10 0.15 0.20 Genotype Ct.Th (mm) 1000 1050 1100 1150 1200 Genotype TMD (mg HA/cm3) * * * Femur (cortical bone) +/+ -/- +/+ -/- 0 3 6 9 Genotype BV/TV (%) 0 1 2 3 4 Genotype Tb.N (1/mm) 0.00 0.01 0.02 0.03 0.04 0.05 Genotype Tb-Th (mm) 0.0 0.1 0.2 Genotype Tb.Sp (mm) * Femur (trabecular bone) +/+ -/- +/+ -/- +/+ -/- +/+ -/- Mark3 knockdown in vivo increases cortical BMD 0.00 0.25 0.50 0.75 1.00 1.25 +/+ +/- -/- Genotype Mark3 Relative Expression P<0.05 Mark3 KOMP gene trap mouse 

• Serine-Theronine Kinase • Member of the AMPK-related kinase family (Bright et al. 2009. Acta Physiologica) • Mammalian homologue of Par1 (C. elegans) in mammals (along with Mark1, -2 and -4 (Kemphues et al. 1998. Cell; Guo & Kemphues. 1995. Cell) • Implicated in metabolism MARK3 (microtubule affinity regulating kinase 3) 

• Serine-Theronine Kinase • Member of the AMPK-related kinase family (Bright et al. 2009. Acta Physiologica) • Mammalian homologue of Par1 (C. elegans) in mammals (along with Mark1, -2 and -4 (Kemphues et al. 1998. Cell; Guo & Kemphues. 1995. Cell) • Implicated in metabolism MARK3 (microtubule affinity regulating kinase 3) Lennerz et al. 2010. Molecular and Cellular Biology 

0 2 4 6 8 −log10 (p−value) 0 20 40 60 80 100 Recombination rate (cM/Mb) ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● rs2010281 0.2 0.4 0.6 0.8 r2 1 gene omitted MARK3 EIF5 SNORA28 CKB TRMT61A BAG5 APOPT1 103.75 103.8 103.85 103.9 103.95 104 104.05 Position on chr14 (Mb) Plotted SNPs Locke et al. 2015. Nature BMI MARK3 is associated with Body Mass Index (BMI) in humans rs11623869 “G” allele MARK3 transcript levels BMD BMI 

Osteoblast-derived effects of Mark3 on BMD and Fat Mass? Mark3 KOMP gene trap mouse Osteoblast-specific Knockout Osteocalcin-cre 

Osteoblast-derived effects of Mark3 on BMD and Fat Mass? Mark3 KOMP gene trap mouse Osteoblast-specific Knockout Osteocalcin-cre 0.10 0.12 0.14 0.16 0.18 0.20 Crosectional Cortical Thickness (mm) ( ) + -/ , 2%2 031 2%2 031 % 

Osteoblast-derived effects of Mark3 on BMD and Fat Mass? Mark3 KOMP gene trap mouse Osteoblast-specific Knockout Osteocalcin-cre 0.10 0.12 0.14 0.16 0.18 0.20 Crosectional Cortical Thickness (mm) ( ) + -/ , 2%2 031 2%2 031 % ( ) + - + - 2,6 - /04 / ( ) + -/ , 2%2 031 2%2 031 % * * Body Weight (g) Adiposity (%) 

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0.07 0.09 9.2 9.4 9.6 9.8 10.0 Femoral BMD (g/cm2) Mark3 Expression (log2) ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −0.2 0.0 0.2 9.0 9.2 9.4 9.6 9.8 10.0 Mark3 Expression (log2) REF rs29145351 ([email protected] Mbp) Genotype Mark3 Expression 0 200 400 600 800 Adj. P=0.002 M9 Eigengene Expression Non-REF rs29145351 Mark3 M9 M9 Mark3 M9 Causal Model, score = 0.95 Reactive Model, score = -3.10 Independent Model, score = -4.20 rs29145351 Mark3 rs29145351 Cd1d1 Fgfr2 LOC385603 Rerg Mfap2 Prss35 Trpv4 Col22a1 Dcbld1 Sp7 Praf2 Ift20 Adam12 Tmtc2 Tmem45a Mbtps1 Ifitm5 Tmem16a Marveld1 Loxl2 A930017M01Rik Adamts2 Nupr1 Mark3 rs29145351 Flrt3 Rgs3 Col1a1 Gng4 Syt9 Gsg1l Anxa5 Rcn3 Shox2 Pvrl3 Pthr1 Unc5b C530001D20Rik Ggcx Zbtb7c Lox 9430011C21Rik Slc35a2 Loxl4 9130024F11Rik Mxra8 Entpd3 Slc36a2 Maged2 4932409I22Rik Lman1 Sgms2 Flvcr2 9130227L01Rik Olfr1036 EG385454 Tsc22d1 2010003D24Rik Cgref1 Robo2 Srpx Col6a2 Gstm2 Kdelr3 C1qtnf6 Adamts9 Slc22a17 Lamp1 Fkbp9 Lrp4 Fstl1 Col11a1 Nhsl1 1110048P06Rik Syn1 Nav1 Lifr Apbb1 Serpinh1 Ppic Smpd3 2310069G16Rik Dcn Cpz Col5a2 Crtap Pltp Sdc3 Fkbp14 Dpysl3 Akp2 Lepre1 Fkbp10 Calu Slc41a2 Lhfp Gpx8 Colec12 BC065085 Vkorc1 Nkd2 Wfdc6a Lpp 0610007N19Rik Slc30a3 Tpm1 Scara3 Islr LOC380980 Sec31a Bmp1 Col11a2 Maged1 Tmem119 Wnt4 D430041D05Rik Ppm1e Mmp23 Mmp16 AI505012 Mrc2 LOC237831 Creb3l1 Col2a1 Gja1 Pard6g Cntn2 Col5a1 4930544G21Rik Fat3 6720467C03Rik Fkbp11 Gpx7 1110001M20Rik 

Summary • MARK3 is a network hub and is at least partially responsible for a BMD GWAS locus on Chr. 14. • Integrating GWAS and network data can inform GWAS and provide systems-level context Genetics Systems Genetics Systems-Based Precision Therapeutics 

Farber Lab • Gina Calabrese • Larry Mesner, Ph.D. • Olivia Sabik • Basel Al-Barghouthi • Ingrid Braenne, Ph.D. • Robert Rotzin Maine Medical Center Research Institute • Cliff Rosen, M.D. Yale • Mark C. Horowitz, Ph.D. • Steve Tommasini, Ph.D. University of Maryland • Joe Stains, Ph.D. John Hopkins • Tom Clemens, Ph.D.