Joint disease-specificity at the regulatory base-pair level

Pushpanathan Muthuirulan,Lu Li,Liangliang Cheng,Alireza Emami,Yan Ding,Ata M Kiapour,Jiaxue Cao,Mary Sedlak,Zun Liu,Terence D Capellini,Danilo Menghini,Daniel Richard,Xinjia Ding,David Maridas,Mariel Young,Gabriela Portilla,Vicki Rosen,Shayan Hosseinzadeh,Dewei Zhao

doi:10.1038/s41467-021-24345-9

Abstract

Given the pleiotropic nature of coding sequences and that many loci exhibit multiple disease associations, it is within non-coding sequence that disease-specificity likely exists. Here, we focus on joint disorders, finding among replicated loci, that GDF5 exhibits over twenty distinct associations, and we identify causal variants for two of its strongest associations, hip dysplasia and knee osteoarthritis. By mapping regulatory regions in joint chondrocytes, we pinpoint two variants (rs4911178; rs6060369), on the same risk haplotype, which reside in anatomical site-specific enhancers. We show that both variants have clinical relevance, impacting disease by altering morphology. By modeling each variant in humanized mice, we observe joint-specific response, correlating with GDF5 expression. Thus, we uncouple separate regulatory variants on a common risk haplotype that cause joint-specific disease. By broadening our perspective, we finally find that patterns of modularity at GDF5 are also found at over three-quarters of loci with multiple GWAS disease associations.

Highlights

Given the pleiotropic nature of coding sequences and that many loci exhibit multiple disease associations, it is within non-coding sequence that disease-specificity likely exists
To date, no proteincoding mutations have been found that can explain these genomewide association studies (GWAS) associations (Supplementary Fig. 1 and Supplementary Data 1 in the Supplementary Information)[46,47]. These findings point to non-coding regulatory variants as likely causal for each disease risk at the locus, yet they remain mostly unknown, and importantly it is unclear whether GDF5-mediated joint diseases share a common pleiotropically acting causal regulatory variant, or not
One key finding from this study is that separate non-coding regulatory variants on the same risk haplotype, i.e., rs4911178 “A” in the GROW1 enhancer and rs6060369 “T” in the R4 enhancer[20], can be functionally uncoupled to independently control different pathologies (i.e., developmental dysplasia of the hip (DDH) and knee OA) in different joints (Fig. 5b)

Summary

Introduction

Given the pleiotropic nature of coding sequences and that many loci exhibit multiple disease associations, it is within non-coding sequence that disease-specificity likely exists. Common variants at this locus, which span a >100 kb risk haplotype containing much non-coding sequence[17,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,48], likely have very specific impacts on individual joints, albeit in general, less is known regarding the causal variants at GDF5 underlying its multiple GWAS associations[17,20,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45]

Methods

Results

Conclusion