The interaction between enhancer variants and environmental factors as an overlooked aetiological paradigm in human complex disease

The interactions between genetic and environmental risk factors contribute to the aetiology of complex human diseases. Genome-wide association studies (GWAS) have revealed that most of the genetic variants associated with complex diseases are located in the non-coding part of the genome, preferentially within enhancers. Enhancers are distal cis -regulatory elements composed of clusters of transcription factors binding sites that positively regulate the expression of their target genes. The generation of genome-wide maps for histone marks (e.g., H3K27ac), chromatin accessibility and transcription factor and coactivator (e.g., p300) binding profiles have enabled the identification of enhancers across many human cell types and tissues. Nonetheless, the functional and pathological consequences of the majority of disease-associated genetic variants located within enhancers seem to be rather minor under normal conditions, thus questioning their medical relevance. Here we propose that, due to the prevalence of enhancer redundancy, the pathological effects of many disease-associated non-coding genetic variants might be preferentially (or even only) manifested under environmental stress.

cell type, tissue and/or developmental stage in which enhancers manifest their regulatory activity, but not their putative target genes.In fact, due to the long-range regulatory capacity of enhancers, assigning these elements to their putative target genes is not straightforward.
[13] Therefore, 3D chromatin maps generated in different cell types/tissues can be used to globally assign enhancers to their putative target genes.However, neither gene reporter assays nor physical proximity can demonstrate whether an enhancer is necessary for the expression of a given gene, which requires the genetic or epigenetic disruption of the candidate enhancers. [14,15]Importantly, recent advances in CRISPR-based technologies and single-cell genomics enable the functional assessment of thousands of enhancers in specific cellular contexts. [16,17]erall, technological advances during the last decade, particularly in the genomics and (epi)genome editing fields, have dramatically improved the identification and characterisation of human enhancers.
Furthermore, these human enhancer catalogues have revealed that genetic variants associated with a broad set of human disorders are often located within enhancers, suggesting that disruption of these regulatory elements might represent a major cause of the human disease (i.e., enhanceropathies). [18,19]However, as we will describe below, the functional and medical relevance of many enhancers and of the disease-associated variants located within them remains dubious.

Enhancers are highly enriched for genetic variants associated with human complex diseases: Correlation or aetiological significance?
Common human diseases have a polygenic and multifactorial aetiology resulting from the complex interaction between genetic, epigenetic and environmental factors.Genome-wide association studies (GWAS) have allowed the identification of a large number of genetic variants associated with complex human disorders and traits, [20][21][22] by statistically associating single nucleotide polymorphisms with a particular trait or disease across many individuals.Notably, the majority (∼90%) of these disease-associated variants are located outside of the coding genome, within the vast non-coding genomic space. [23,24]Until relatively recently, our understanding of the non-coding genome was rather rudimentary, and thus, evaluating the relevance of diseaseassociated variants was challenging.However, as non-coding regulatory elements (namely enhancers) were mapped and characterised in an increasing number of human cell types and tissues, it became evident that a large fraction (75%-80%) of disease-associated variants are located within putative enhancers. [23,24]These observations lead to the general hypothesis that a significant fraction of the genetic variants associated with human complex diseases may disrupt enhancers, which can then lead to cell type-specific and quantitative changes in gene expression with downstream pathological consequences. [19,25,26][29][30][31] However, more global assays interrogating the overall functional and medical relevance of GWAS SNPs or the enhancers in which they are located suggest that only a minor fraction (<10%) of disease-associated variants might have an impact on gene expression. [16,32,33]Therefore, despite major efforts and technological advances, our understanding of the genetic basis of complex human diseases in general and of the aetiological significance of non-coding variants in particular remains incomplete.Furthermore, human complex disorders are not simply caused by particular combinations of genetic variants but also by environmental risk factors (e.g., diet, pollution, smoking, ageing, etc.).Nonetheless, the interactions between the environment and our genetic make-up and how these interactions might ultimately influence disease risk are only starting to be elucidated. [34]

Enhancer redundancy as a buffering mechanism against environmental stress
There could be several reasons explaining why the disruption of most enhancers does not have a major impact on the expression of their putative target genes: (1) since enhancers display cell typespecific activities, their regulatory function might only become evident when evaluated in the appropriate cell types; (2) enhancers might be assigned to the incorrect target genes.Although this possibility should be minimised by the use of 3D chromatin data, the contacts between genes and enhancers seem to be rather transient, and in some cases might not involve close physical proximity. [35,36]Consequently, enhancer-gene contacts might not be easily detectable with some 3Cbased methods such as Hi-C [12] ; (3) both invertebrate and vertebrate organisms display complex enhancer regulatory landscapes, whereby multiple enhancers with totally or partially redundant regulatory activities can control the expression of a given gene. [37,38]Here we will focus on enhancer redundancy and how it might mask the functional dissection of non-coding GWAS SNPs (Figure 1).
The term shadow enhancer was first coined to describe secondary (distal) enhancers with overlapping activities with respect to primary (proximal) enhancers controlling the expression of a few developmental genes (e.g., Snail, Shavenbaby, Brinker) in Drosophila. [39,40]Individual deletion of either the primary or secondary enhancers does not lead to significant changes in the expression of their target genes, while when deleted simultaneously gene transcription is severely compromised and phenotypic defects become evident.Subsequent work using Drosophila mesoderm development as a model showed that for >60% of the evaluated loci, gene expression was putatively controlled by multiple and partially redundant enhancers. [37]Moreover, similar observations were also made in mammals, suggesting that enhancer redundancy is a pervasive feature of developmental regulatory networks. [38,41,42]Further support for the prevalence of enhancer redundancy comes from quantitative trait locus (QTL) and F I G U R E 1 Enhancer redundancy as a buffering mechanism against genetic variation.The expression of gene A (left) and gene B (right) is controlled by a single enhancer (Enh1) or multiple enhancers (Enh1, Enh2 and Enh3), respectively.When controlled by a single enhancer, an enhancer variant can have deleterious effect on the expression of gene A, decreasing its expression under a tolerable level (red dashed line).On the other hand, enhancer redundancy confers robustness, as alterations in several enhancers are required to down-regulate the expression of gene B.This property is especially important during development, when gene expression has to be tightly controlled in time and space.[45] Similarly to GWAS statistically linking SNPs with a particular trait or disease, eQTLs and chromatin-state QTLs studies link SNPs with either gene expression or chromatin features (e.g., chromatin accessibility, histone modifications, etc.), respectively.
Identification of chromatin-state QTLs in large cohorts of human lymphoblastoid cell lines (LCLs) revealed thousands of genetic variants locally influencing chromatin state, preferentially at putative enhancer elements. [43,45]Similarly, gene expression QTLs (eQTLs) studies in LCLs revealed that the majority (∼80%) of expressed genes display local (within 1Mb) eQTLs. [46]However, while the majority of eQTLs overlap chromatin-state QTLs, the opposite is not true and most chromatin-state QTLs, especially those located within enhancers, are not associated with expression changes in nearby genes. [44,45]Interestingly, the number of variable enhancers associated with one gene seems to be positively correlated with gene expression variability, suggesting that changes in multiple enhancers are required to alter gene expression. [45]Overall, these observations support the prevalence of enhancer redundancy in humans and its potential relevance for human disease. [47], what could be the function of this widespread enhancer redundancy?Interestingly, pioneer work in Drosophila showed that when embryos with single enhancer deletions (either a primary or a shadow enhancer) were exposed to genetic (e.g., reduced Dorsal dosage) or environmental (e.g., sub-optimal temperatures) perturbations, obvious gene expression changes and downstream phenotypic consequences were observed. [39,40]Mechanistically, it has been proposed that functionally redundant enhancers could confer their target genes with an activatory input higher than the minimum required to induce and/or maintained gene expression (i.e., 'supra-threshold' activatory input). [48,49]This might ensure that genes are robustly expressed even if the activatory inputs provided by the enhancers decrease due to genetic or environmental perturbations.Therefore, enhancer redundancy is thought to confer transcriptional and phenotypic robustness against genetic or environmental stress.Although examples of redundant enhancers conferring robustness against environmental stress are still missing in humans, it is tempting to speculate that quantitatively small changes in gene expression due to the disruption of individual enhancers (e.g., due to GWAS SNPs) might be magnified by certain environmental factors, ultimately contributing to the emergence of phenotypic defects.This scenario resembles the gene-environmental (GxE) interactions described in various human mendelian disorders and heterozygous mouse models in which haploinsufficiency for certain genes becomes exacerbated (i.e., increased penetrance and/or expressivity) upon exposure to various environmental and teratogenic insults (e.g., alcohol, retinoic acid, nicotine, folic acid deficiency, etc.). [50,51]ese interactions between gene dosage and environmental factors suggest that considering environmental stress might be also important when studying GWAS SNPs located within enhancers, as their functional relevance may remain cryptic under normal conditions.

Enhancers as potential hotspots of GxE interactions implicated in human disease
In summary, (1) human enhancer landscapes are complex, particularly around major developmental and cell identity genes; (2) enhancer redundancy is a widespread feature across invertebrate and vertebrate regulatory networks; (3) enhancer redundancy might represent an important mechanism to buffer gene expression changes in the face of environmental insults; (4) environmental risk factors and GxE interactions play important roles in human disease aetiology.Taken altogether, we postulate that functionally redundant enhancers are likely to represent a pervasive feature of human regulatory networks ensuring that tolerable gene expression levels are robustly maintained under environmental stress conditions (Figure 2).Consequently, we hypothesize that, due to this enhancer redundancy, the pathological effects of many disease-causative genetic variants (i.e., GWAS SNPS) occurring within enhancers might be preferentially revealed under environmental stress (Figure 2).In addition, another possibility is that certain GWAS SNPs can be located within enhancers that, under normal conditions, are found in a primed state, and that only get activated in response to certain stimuli (e.g., IFNγ). [52,53]In this case, the functional F I G U R E 2 Enhancers as potential hotspots of GxE interactions implicated in human disease.Disease-associated SNPs (black cross located in Enh1) or environmental insults (red star) might not cause any phenotypic defects by themselves.However, when the SNP is combined with environmental stress (bottom right of the table), then the SNP might manifest its functional relevance and cause gene expression changes and phenotypic defects.In other words, GxE interactions might be necessary to reveal the functional and pathological relevance of disease-associated genetic variants located within enhancers.consequences of these SNPs might be also preferentially manifested under environmental stress (e.g., infection).
To test these hypotheses, it will be important to consider the human cell types or tissues in which the candidate enhancers harbouring GWAS SNPs are active and that are relevant for a particular disease (e.g., smooth muscle or endothelial cells in cardiovascular disease) as well as the environmental risk factors associated with the investigated disorder (e.g., high cholesterol levels in cardiovascular disease).Human cell lines in general and human embryonic stem cell (hESC) differentiation systems in particular (including organoids technology [54] ) might be well suited to identify enhancers that only or preferentially manifest their regulatory functions under environmental stress.To do so, these cellular systems can be combined with genome-wide CRISPRi screens and scRNA-seq technologies [16] to globally assess enhancer function in the presence or absence of environmental risk-factors.Similarly, CRISPR-Cas technology can be used in the cellular systems described above to edit GWAS SNPs located within enhancers and derive clonal lines containing either risk or protective alleles.Then, the resulting clonal lines can be exposed to environmental risk-factors to evaluate whether GxE in the context of enhancers contribute to pathological changes in gene expression.If these assays confirm that GxE within enhancers contribute to the aetiology of complex disorders, additional work would be needed in order to dissect the molecular and mechanistic basis of such interactions.