Functional Testing of Thousands of Osteoarthritis-Associated Variants for Regulatory Activity

Abstract

17To date, genome-wide association studies have implicated at least 35 loci in osteoarthritis, but 18 due to linkage disequilibrium, we have yet to pinpoint the specific variants that underlie these 19 associations, nor the mechanisms by which they contribute to disease risk. Here we functionally 20 tested 1,605 single nucleotide variants associated with osteoarthritis for regulatory activity using 21 a massively parallel reporter assay. We identified six single nucleotide polymorphisms (SNPs) 22 with differential regulatory activity between the major and minor alleles. We show that our most 23 significant hit, rs4730222, drives increased expression of an alternative isoform of HBP1 in a 24 heterozygote chondrosarcoma cell line, a CRISPR-edited osteosarcoma cell line, and in 25 chondrocytes derived from osteoarthritis patients. 27Main text 28 29 Genome-wide association studies (GWAS) have successfully implicated thousands of genetic 30 loci in common human diseases. Most of the underlying signal is believed to derive from 31 variation in non-coding regulatory sequences. However, because of linkage disequilibrium (LD), 32 it has been extraordinarily challenging for the field to identify the variants that causally underlie 33 each association. 35Over the past decade, we and others developed massively parallel reporter assays (MPRAs) to 36 increase the throughput at which regulatory sequences can be tested for functional potential [1][2][3][4] . 37An MPRA involves cloning thousands of candidate regulatory sequences to a single reporter was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.The copyright holder for this preprint (which . http://dx.doi.org/10.1101/379727 doi: bioRxiv preprint first posted online Jul. 30, 2018; 3 scoring quintile includes 240 genomic regions, 67 of which overlap putative enhancers from at 81 least one dataset. In contrast, the least active quintile includes 239 genomic regions, only 37 of 82 which overlap putative enhancers from at least one dataset (1.8-fold difference, chi-square 83 p=9.7e-4). Altogether, these enrichments demonstrate that at least a subset of the 1,605 genomic 84 regions tested here correspond to enhancers in OA-relevant tissues. All activity scores are 85 included in Table S2. 87We next sought to ask whether any alleles are differentially active, focusing on the 752 SNPs for 88 which we successfully measured activity scores for both alleles (Table S3). Overall, activity 89 scores for two alleles of a given SNP were highly correlated, with an overall Spearman 90 correlation of 0.96 (Fig. 1G). This was reassuring, given that each pair of alleles was separately 91 synthesized and cloned, and therefore at non-identical abundances in the STARR-seq library. 92After correcting for multiple testing with Benjamini-Hochberg (BH) at a 5% FDR, we identified 936 SNPs whose alleles ...

Multi-tissue epigenetic analysis of the osteoarthritis susceptibility locus mapping to the plectin gene PLEC

“…The mapping of risk loci for polygenic traits is now a relatively straightforward procedure, with the next major step in complex trait analysis being the transition from association signal to functional characterization 32 . For OA, considerable strides have been made in this regard in recent years 33 , 34 , 35 . Our interest is in the epigenetic dimension of OA genetic risk and especially the role of DNA methylation.…”

Systematic identification of cis-regulatory variants that cause gene expression differences in a yeast cross

“…MPRAs have been used to probe DNA sequences for their ability to drive transcription ( Arnold et al, 2013 ; Kheradpour et al, 2013 ; Wang et al, 2018 ), dissect the importance of individual bases in regulatory elements ( Patwardhan et al, 2009 ), and examine the combined effects of multiple elements in regulatory ‘grammars’ ( Davis et al, 2020 ; Kosuri et al, 2013 ; Mogno et al, 2013 ; Sharon et al, 2012 ; Smith et al, 2013 ) in promoters ( Kotopka and Smolke, 2020 ; Lubliner et al, 2015 ; Sharon et al, 2012 ; Weingarten-Gabbay et al, 2019 ), UTRs ( Cuperus et al, 2017 ; Dvir et al, 2013 ; Rabani et al, 2017 ; Shalem et al, 2015 ) and enhancers ( Arnold et al, 2013 ; Klein et al, 2019 ; Melnikov et al, 2012 ; Patwardhan et al, 2009 ). Other applications have assayed sequences that promote splicing ( Cheung et al, 2019 ; Rosenberg et al, 2015 ), translation ( Goodman et al, 2013 ; Weingarten-Gabbay et al, 2016 ), DNA methylation ( Krebs et al, 2014 ) and RNA editing ( Safra et al, 2017 ).…”

Pathogenesis of Osteoarthritis: Risk Factors, Regulatory Pathways in Chondrocytes, and Experimental Models

“…By the end of 2019, 90 genome-wide significant OA risk loci have been identified with GWAS, most of which are enriched near genes involved in skeletal development and morphogenesis [ 81 , 82 ]. These include the Wnt pathway genes HBP1 (HMG-Box transcription factor 1) and BCL9 (B cell CLL/lymphoma 9 protein), and TGFβ pathway genes TGFB1 , latent transforming growth factor beta binding protein 1 (LTBP1), LTBP3 , SMAD3 , and the recently identified ROCR long non-coding RNA (lncRNA) that acts upstream of SOX9 during chondrogenic differentiation [ 83 , 84 , 85 ]. Genetic overlaps with height, hip shape, bone area, and developmental dysplasia of the hip are observed in these sensitive loci, which might alter joint biomechanics and predispose the individual to OA later in life [ 86 , 87 ].…”