Author response: The missing link between genetic association and regulatory function
Year: 2022
Type: peer-review
Abstract: Article Figures and data Abstract Editor's evaluation Introduction Results Discussion Materials and methods Appendix 1 Data availability References Decision letter Author response Article and author information Metrics Abstract The genetic basis of most traits is highly polygenic and dominated by non-coding alleles. It is widely assumed that such alleles exert small regulatory effects on the expression of cis-linked genes. However, despite the availability of gene expression and epigenomic datasets, few variant-to-gene links have emerged. It is unclear whether these sparse results are due to limitations in available data and methods, or to deficiencies in the underlying assumed model. To better distinguish between these possibilities, we identified 220 gene–trait pairs in which protein-coding variants influence a complex trait or its Mendelian cognate. Despite the presence of expression quantitative trait loci near most GWAS associations, by applying a gene-based approach we found limited evidence that the baseline expression of trait-related genes explains GWAS associations, whether using colocalization methods (8% of genes implicated), transcription-wide association (2% of genes implicated), or a combination of regulatory annotations and distance (4% of genes implicated). These results contradict the hypothesis that most complex trait-associated variants coincide with homeostatic expression QTLs, suggesting that better models are needed. The field must confront this deficit and pursue this 'missing regulation.' Editor's evaluation The findings reported here are important because they address the issue of how complex traits arise from their genetic underpinnings. There is an assumption that genetically mediated variation in transcript abundance, usually detected via analysis of expressed quantitative trait loci, is key to this process, but we lack robust evidence in support of that view. This article finds limited evidence that the baseline expression of trait-related genes explains the associations between complex traits and genetic variants (as identified from genome-wide association studies), leading to the view that the field needs to confront a problem of 'missing regulation.' https://doi.org/10.7554/eLife.74970.sa0 Decision letter Reviews on Sciety eLife's review process Introduction Modern complex trait genetics has uncovered surprises at every turn, including the paucity of associations between traits and coding variants of large effect, and the 'mystery of missing heritability,' in which no combination of common and rare variants can explain a large fraction of trait heritability (Manolio et al., 2009). Further work has revealed unexpectedly high polygenicity for most human traits and very small effect sizes for individual variants. Enrichment analyses have demonstrated that a large fraction of heritability resides in regions with gene regulatory potential, predominantly tissue-specific accessible chromatin and enhancer elements, suggesting that trait-associated variants influence gene regulation (Maurano et al., 2012; Trynka et al., 2013; Gusev et al., 2014). Furthermore, genes in trait-associated loci are more likely to have genetic variants that affect their expression levels (expression quantitative trait loci, or eQTLs), and the variants with the strongest trait associations are more likely also to be associated with transcript abundance of at least one proximal gene (Nicolae et al., 2010). Combined, these observations have led to the inference that most trait-associated variants are eQTLs, and their effects arise from altering transcript abundance, rather than protein sequence. Equivalent sQTL (splice QTL) analyses of exon usage data have revealed a more modest overlap with trait-associated alleles, suggesting that a fraction of trait-associated variants influence splicing, and hence the relative abundance of different transcript isoforms, rather than overall expression levels. The genetic variant causing expression changes may lie outside the locus and involve a knock-on effect on gene regulation, with the variant altering transcript abundances for genes elsewhere in the genome (a trans-eQTL), but the consensus view is that trans-eQTLs are typically mediated by the variant influencing a gene in the region (a cis-eQTL) (GTEx Consortium, 2020). Thus, a model has emerged in which most trait-associated variants influence proximal gene regulation. Here we argue that this unembellished model—in which genome-wide association study (GWAS) peaks are mediated by the effects on the homeostatic expression assayed in tissue samples—is the exception rather than the rule. We highlight the challenges of current strategies linking GWAS variants to genes and call for a reevaluation of the basic model in favor of more complex models possibly involving context-specificity with respect to cell types, developmental stages, cell states, or the constancy of expression effects. Our argument begins with several observations that challenge the unembellished model. One challenge is the difference between spatial distributions of eQTLs, which are dramatically enriched in close proximity to genes, and GWAS peaks, which are usually farther away (Stranger et al., 2007; Farh et al., 2015; Mostafavi et al., 2022). Another is that expression levels mediate a minority of complex trait heritability (Yao et al., 2020). Finally, many studies have designed tools for colocalization analysis: a test of whether GWAS and eQTL associations are due to the same set of variants, not merely distinct variants in linkage disequilibrium. If the model is correct, most trait associations should also be eQTLs, but across studies, only 5–40% of trait associations colocalize with eQTLs (Giambartolomei et al., 2014; Chun et al., 2017; Giambartolomei et al., 2018; Hormozdiari et al., 2016). Despite the doubts raised, the fact that most GWAS peaks do not colocalize with eQTLs cannot disprove the predominant, unembellished model. In a sense, negative colocalization results are confusing because their hypothesis is too broad. If we predict merely that GWAS peaks will colocalize with some genes' expression, it is not clear what is meant by a peak's failure to colocalize with any individual gene's expression. Thus, a narrower, more testable hypothesis requires identifying genes we believe a priori are biologically relevant to the GWAS trait. If these trait-linked genes have nearby GWAS peaks and eQTLs, failure to colocalize would be a meaningful negative result. Earlier studies tested all GWAS peaks; when a peak has no colocalization, the model is inconclusive. But trait-linked genes that fail to colocalize reveal that our method for detecting non-coding variation is, with current data, incompatible with our model for understanding it. With this distinction in mind, we created a set of trait-associated genes capable of supporting or contradicting the model of non-coding GWAS associations acting as eQTLs. For this purpose, the selection of genes becomes extremely important. Because the model attempts to explain the genetic relationship between traits and gene expression, true positives cannot be selected based on measurements of genetic association to traits (GWAS) or expression (eQTL mapping). With this restriction, one source of true positives is to identify genes that are both in loci associated with a complex trait and are also known to harbor coding mutations tied to a related Mendelian trait or the same complex trait. Using a model not based on expression, Mendelian genes are enriched in common-variant heritability for cognate complex traits (Weiner et al., 2022). The genes and their coding variants may be detected in familial studies of cognate Mendelian disorders or by aggregation in a burden test on the same complex phenotypes as GWAS (Backman et al., 2021). For genes whose coding variants can cause detectable phenotypic change, the strong expectation is that a variant of small effect influences the gene identified by its rare coding variants. As an example, APOE and LDLR are both low-density lipoprotein receptor genes (Schneider et al., 1981; Goldstein and Brown, 1973). Coding variants in APOE and LDLR can lead to the Mendelian disorder familial hypercholesterolemia (Goldstein and Brown, 1973; Cenarro et al., 2016). Even in the absence of a Mendelian coding variant, experiments in animal models have found that the overexpression of these genes reduces cholesterol levels (Shimano et al., 1992a; Shimano et al., 1992b; Kawashiri et al., 2001). GWAS on human subjects have found significant associations near APOE and LDLR, so it seems reasonable to suspect that any non-coding effects in these loci may be mediated by these genes. This general relationship between Mendelian and complex traits is supported by several lines of evidence summarized in Appendix 1. Results To test the model that trait-associated variants influence baseline gene expression, we assembled a list of putatively causative genes for seven polygenic common traits with available large-scale GWAS data, each of which also has an extreme form in which coding variants of large effect alter one or more genes with well-characterized biology (Table 1). Our selection included four common diseases: type II diabetes (T2D) (Mahajan et al., 2018), where early-onset familial forms are caused by rare coding mutations (insulin-independent MODY; neonatal diabetes; maternally inherited diabetes and deafness; familial partial lipodystrophy); ulcerative colitis (UC) and Crohn disease (CD) (Liu et al., 2015; Goyette et al., 2015), which have Mendelian pediatric forms characterized by severity of presentation; and breast cancer (BC) (Zhang et al., 2020), where germline coding mutations (e.g., BRCA1) or somatic tissue (e.g., PIK3CA) are sufficient for disease. We also chose three quantitative traits: low- and high-density lipoprotein levels (LDL and HDL); and height. Between known Mendelian genes and those from Backman et al., 2021, our analysis included 220 unique gene-trait pairs (Figure 1). Table 1 Putatively causative Mendelian genes. Each gene includes reference(s) to the known biological role of its coding variants, as established in familial studies, in vitro experiments, and/or animal models. Genes from Backman et al., 2021 are not included here, but can be found in Figure 2. PhenotypeGenesLow-density lipoproteinAPOB Soria et al., 1989; Pullinger et al., 1995 APOC2 Hegele et al., 1991 APOE de Knijff et al., 1994 LDLR Brown and Goldstein, 1976 LPL Heizmann et al., 1991; Clee et al., 2001 PCSK9 Abifadel et al., 2003High-density lipoproteinABCA1 Brooks-Wilson et al., 1999; Bodzioch et al., 1999; Rust et al., 1999; Ordovas et al., 1986 APOA1 Ordovas et al., 1986 CETP Glueck et al., 1975 LIPC Isaacs et al., 2004; Grarup et al., 2008; Iijima et al., 2008 LIPG Yamakawa-Kobayashi et al., 2003 PLTP Jiang et al., 1999 SCARB1 Tai et al., 2003; McCarthy et al., 2003Height ANTXR1 Stránecký et al., 2013; Bayram et al., 2014 ATR O'Driscoll et al., 2003; Ogi et al., 2012 BLM Ellis et al., 1995; Foucault et al., 1997 CDC6 Bicknell et al., 2011a CDT1 Bicknell et al., 2011a; Guernsey et al., 2011 CENPJ AlDosari et al., 2010 COL1A1 Wallis et al., 1990 COL1A2 Spotila et al., 1992; De Paepe et al., 1997 COMP Briggs et al., 1995; Mabuchi et al., 2003 CREBBP Menke et al., 2016; Menke et al., 2018; Angius et al., 2019 DNA2 Shaheen et al., 2014 EP300 Woods et al., 2014; Tsai et al., 2011 EVC Polymeropoulos et al., 1996; Ruiz-Perez et al., 2003 EVC2 Ruiz-Perez et al., 2003; Galdzicka et al., 2002 BN1 Faivre et al., 2003; Le Goff et al., 2011; Horn and Robinson, 2011; Takenouchi et al., 2013 FGFR3 Hyland et al., 2003; Toydemir et al., 2006; Makrythanasis et al., 2014 FKBP10 Alanay et al., 2010; Kelley et al., 2011; Barnes et al., 2013 HR Berg et al., 1993; Woods et al., 1996; Goddard et al., 1995; Ayling et al., 1997 KRAS Aoki et al., 2005; Schubbert et al., 2006; Carta et al., 2006 NBN Varon et al., 1998; Tanzanella et al., 2003 NIPBL Tonkin et al., 2004; Krantz et al., 2004 ORC1 Bicknell et al., 2011a; Guernsey et al., 2011; Bicknell et al., 2011b RC4 Guernsey et al., 2011; Bicknell et al., 2011b ORC6L Bicknell et al., 2011a; de Munnik et al., 2012 PCNT Rauch et al., 2008; Griffith et al., 2008; Piane et al., 2009 PLOD2 van der Slot et al., 2003; HaVinh et al., 2004; Puig-Hervás et al., 2012 PTPN11 Tartaglia et al., 2001; Maheshwari et al., 2002; Kosaki et al., 2002 RAD21 Deardorff et al., 2012; Kruszka et al., 2019; Goel and Parasivam, 2020 RAF1 Pandit et al., 2007; Razzaque et al., 2007 RECQL4 Lindor et al., 2000; Beghini et al., 2003; Wang et al., 2003 RIT1 Aoki et al., 2013; Bertola et al., 2014; Gos et al., 2014 ROR2 Afzal et al., 2000; van Bokhoven, 2000; Tufan et al., 2005 SLC26A2 Hästbacka et al., 1993; Rossi and Superti-Furga, 2001; Barreda-Bonis et al., 2018 SMAD4 Le Goff et al., 2012; Caputo et al., 2012; Lindor et al., 2012 SRCAP Hood et al., 2012; Le Goff et al., 2013 WRN Yu et al., 1996; Goto et al., 1997; Yu, 1997Crohn diseaseATG16L1 Hampe et al., 2007 CARD9 Rivas et al., 2011 IL10 Fowler, 2005 IL10RA Gasche et al., 2003; Mao et al., 2012 IL10RB Glocker et al., 2009; Begue et al., 2011 IL23R Duerr et al., 2006; Libioulle et al., 2007; Glas et al., 2007 IRGM McCarroll et al., 2008; Craddock et al., 2010; Prescott et al., 2010 NOD2 Ogura et al., 2001; Hugot et al., 2001 PRDM1 Ellinghaus et al., 2013 PTPN22 Diaz-Gallo et al., 2011 Ulcerative colitisATG16L1 Fowler et al., 2008 CARD9 Rivas et al., 2011 IL23R Glas et al., 2007; Fisher et al., 2008 IRGM McCarroll et al., 2008 PRDM1 Ellinghaus et al., 2013 PTPN22 Diaz-Gallo et al., 2011 RNF186 Rivas et al., 2016; Beaudoin et al., 2013 Type II diabetesABCC8 Reis et al., 2000 BLK Borowiec et al., 2009 CEL Bengtsson-Ellmark et al., 2004; Raeder et al., 2006 EIF2AK3 Harding et al., 2001; Brickwood et al., 2003; Durocher et al., 2006 GATA4 Shaw-Smith, 2014 GATA6 Yorifuji et al., 2012; De Franco et al., 2013 GCK Froguel et al., 1993 GLIS3 Senée et al., 2006 HNF1A Yamagata et al., 1996b; Vaxillaire et al., 1997 HNF1B Horikawa et al., 1997; Lindner et al., 1999 HNF4A Yamagata et al., 1996a; Stoffel and Duncan, 1997 IER3IP1 Poulton et al., 2011; Abdel-Salam et al., 2012; Shalev et al., 2014 INS Støy et al., 2007 KCNJ11 Hani et al., 1998; Gloyn et al., 2004 jKLF11 Neve et al., 2005 LMNA Cao and Hegele, 2000 NEUROD1 Malecki et al., 1999 NEUROG3 Gradwohl et al., 2000; Rubio-Cabezas et al., 2011; Pinney et al., 2011 PAX4 Shimajiri et al., 2001; Mauvais-Jarvis et al., 2004; Plengvidhya et al., 2007 PDX1 Stoffers et al., 1997; Macfarlane et al., 1999; Hani et al., 1999 PPARG Deeb et al., 1998; Savage et al., 2002 PTF1A Sellick et al., 2004 RFX6 Smith, 2010; Sansbury et al., 2015 SLC19A2 Labay et al., 1999, Oishi et al., 2002; Shaw-Smith et al., 2012 SLC2A2 Laukkanen et al., 2005; Sansbury et al., 2012 WFS1 Strom et al., 1998; Hardy et al., 1999; Khanim et al., 2001 ZFP57 Mackay et al., 2008; Boonen et al., 2013 Breast cancer (selected using MutPanning; Dietlein et al., 2020)AKT1ARID1AATMBRCA1BRCA2CBFBCDH1CDKN1BCHEK2CTCFERBB2ESR1FGFR2FOXA1GATA3GPS2HS6ST1KMT2CKRASLRRC37A3MAP2K4MAP3K1NCOR1NF1NUP93PALB2PIK3CAPTENRB1RUNX1SF3B1STK11TBX3TP53ZFP36L1 Figure 1 with 2 supplements see all Download asset Open asset Putatively causative genes identified by each method category. The leftmost column in each half of the plot displays the entire group of putatively causative genes for our Mendelian set of genes and our (Backman et al., 2021) set of genes, respectively, as well as noting how many are unique to each set or shared between the two sets. The second column in each half indicates how many genes from each set have a nearby GWAS peak or have both a nearby GWAS peak and an expression QTL (eQTL). The remaining columns indicate how many genes were identified through colocalization, transcriptome-wide association studies (TWAS), or chromatin methods, while noting how many of these genes are unique vs. shared between the Mendelian and Backman sets. In well-powered GWAS, even relatively rare large-effect coding alleles (mutations in BRCA1 that cause breast cancer, for instance) may be detectable as an association to common variants, which could make the effect of a coding variant appear to be regulatory instead. To account for this possibility, we computed association statistics in each GWAS locus conditional on coding variants. We applied a direct conditional test to datasets with available individual-level genotype data (height, LDL, HDL); for those studies without available genotype data, we computed conditional associations from summary statistics using COJO (Yang et al., 2011; Yang et al., 2012; 'Materials and methods'). With both methods, the resulting GWAS associations should reflect only non-coding variants. After controlling for coding variation, we examined whether these genes are more likely than chance to be in close proximity to variants associated with the polygenic form of each trait. In agreement with existing literature (Freund et al., 2018), we observe a significant enrichment for all traits in our combined Mendelian and Backman et al., 2021 gene sets (Figure 1—figure supplement 1). Of our 220 genes, 147 (67%) fell within 1 Mb of a GWAS locus for the cognate complex trait, over three times as many as the 43 predicted by a random null model (95% confidence interval: 31.5–54.5). Our window of 1 Mb represents roughly the upper bound for distances identified between enhancer–promoter pairs, but most pairs are closer (Nasser et al., 2021), so we would expect enrichment to increase as the window around genes decreases; this proves to be the case. At a distance of 100 kb, we find 104 putatively causative genes (47%), though the null model predicts only 11 (95% CI 4.5–17.0), an order-of-magnitude enrichment (Figure 1—figure supplement 1). Given their known causal roles in the severe forms of each phenotype, these results suggest that the 147 genes near GWAS signals are likely to be the targets of trait-associated non-coding variants. For example, we see a significant GWAS association between breast cancer risk and variants in the estrogen receptor (ESR1) locus even after controlling for coding variation; the baseline expression model would thus predict that non-coding risk alleles alter ESR1 expression to drive breast cancer risk. We next looked for evidence that the trait-associated variants were also altering the expression of our 147 genes in relevant tissues. Controlling for the number of tests we conducted, 134 (91.1%) of these genes had an eQTL in at least one relevant tissue at a false discovery rate (FDR) of Q < 0.05 ('Materials and methods'). If these variants act through changes in gene expression, phenotypic associations should be driven by some of the same variants as eQTLs in relevant tissue types. We therefore looked for co-localization between our GWAS signals and eQTLs in relevant tissues (Table 2) drawn from the using three (Giambartolomei et al., et al., and et al., 2016). We found support for the colocalization of trait and eQTL association for only genes of 147 for for and for for this represents only putatively causative genes or putatively causative genes near GWAS peaks, even without ('Materials and which is not better than random We that of the fraction of GWAS associations with eQTLs Giambartolomei et al., 2014; Chun et al., 2017; Hormozdiari et al., 2016; et al., do not the to find causative genes. our of the number of putatively causative genes that colocalize with eQTLs tests the of our and Table 2 were selected for each trait based on a priori of disease Mendelian II of our approach is the of our to tissues. We believe this is in to the of each gene–trait in each a large number of false positives or a severe that may lead to false However, to the set of tissues with a known biological role and available expression data tissues with in of the tissues we do have their to eQTLs with effects. To address from the available of tissue we the et al., is a method that genetic association summary statistics across a of by of across the summary statistics of each individual In our an eQTL is to find in a tissue of information from tissues may it. this method summary statistics that to a We on every locus in our analysis using data from all tissues ('Materials and methods'). with these statistics the number of across all genes by to However, the 100 identified only four putatively causative genes (Figure 1—figure supplement These results indicate that selection not the in our association studies et al., 2015; Gusev et al., 2016; et al., 2017; et al., are of methods applied to identify causative genes GWAS peaks using gene expression. genetic between traits and is not designed to caused by which it in the of or causative variants et al., However, while analyses typically sets that many false positives and are to the number of tissue et al., Results from the of et al., across all tissues identified our putatively causative genes as likely tied to the GWAS peak in loci However, only genes were identified by when we the analysis to relevant tissues. Given the paucity of GWAS peaks, we whether GWAS variants in our study loci in likely regulatory the genes in the Mendelian of putatively causative genes, we each nearby GWAS association using the et al., For of these genes, we identified at least one variant as a variant with or than within 100 of the We tested whether these variants within regulatory by et al., a or et al., or as an by and 2012; and 2017; 'Materials and methods'). As many as genes identified this have a variants within a regulatory region across all the tissue or when to tissues (Figure 1 and Despite this strong evidence that these GWAS associations arise from effects on regulatory only loci to expression effects in our eQTL Figure 2 Download asset Open asset Genes identified as associated with a complex trait by each and indicate whether a gene is from the Mendelian set of putatively causative genes, the Backman et or columns indicate whether a gene identified as a using each of our transcriptome-wide association studies (TWAS), and chromatin In to more our regulatory analysis to our eQTL we computed of the method that a a regulatory region to a gene (Nasser et al., et al., 2019; 'Materials and Figure each with the we of our Mendelian of genes. As even when a GWAS association and gene are in the same they are to whether using eQTLs or current to chromatin data with genes. Figure Download asset Open asset causative gene the of GWAS variants, three methods were The identified variants within regions as by The second identified variants within and their using an that combined the of the the of the or with its distance to the gene of ('Materials and methods'). The both of for variants within a region and the Figure data 1 results for expression and Download Discussion our results are with the that complex traits are by non-coding genetic variants whose effects on are mediated by their to the regulation of nearby genes. However, these same results are with a model that such effects on gene regulation arise from the genetic influences on baseline gene expression that are by eQTLs. The enrichment of putatively causative based on existing biological GWAS peaks their role in complex the enrichment of GWAS variants in likely regulatory regions by and chromatin support to the general model of GWAS associations from effects on gene regulation and expression. However, the of methods to GWAS associations and expression the that the causative GWAS variants exert their effects via the homeostatic or effects on gene expression that are by eQTLs of the by have for the limited of methods to explain the of GWAS variants. most linking designed to be of a priori biological and This is for discovery because it the number of but it is for for negative do not what were it is to explain not find it. only loci for which is a we were better to distinguish which from identifying links using expression As a we that a number of when variant-to-gene methods are not in the we These lack of for GWAS, the absence of eQTLs for relevant genes, and methods for linking expression to GWAS (Table Table for negative results the unembellished model. have for GWAS associations that are not by This the with our which are in the column and on the involving more models of gene regulation can be found in Table of the here involve of the of our and complex trait If coding and non-coding variants affect different biological or trait associations
Language: English
Subfield: Genetics
Domain: Life Sciences
Open Access status: gold
OpenAlex ID: W4316618124