Causal Gene Confusion – The Complicated EDN1/PHACTR1 locus for Coronary Artery Disease

Genome wide association studies (GWAS) have identified thousands of genetic loci associated with vascular disease. For coronary artery disease (CAD) the list has grown to over 300 significant associations, based on meta-analyses of the Cardiogram and Million Veterans’ Project^1,2. Most of these loci are not linked to genes that regulate known vascular risk factors, and therefore represent potential new mechanisms of disease. Yet very few variants have been mechanistically characterized. The major challenge is that the relevant cell type and causal gene for a disease-associated locus are difficult to determine. As a result, GWAS loci are named for the closest gene to the associated single nucleotide polymorphism (SNP), and though this is thought be accurate about half the time³, it underestimates the role of variants with distal regulatory effects.

This causal gene confusion persists for the majority of CAD loci, including the chromosome 6p24 association first reported in 2009⁴. Since then, this same SNP has been associated with multiple vascular disorders including migraine headache⁵, coronary calcification⁶, hypertension⁷, fibromuscular dysplasia⁸, microvascular angina⁹, and arterial dissection^8,10. This extensive pleiotropy suggests an important effect on the biology of the arterial wall for genetic variants in this locus. This was confirmed with our identification of a vascular-specific non-coding regulatory element located at the lead SNP at chr6p24¹¹. In iPSC-derived endothelial cells (ECs) the major regulatory effect is on a gene 600kB upstream of the locus—Endothelin-1 (EDN1). This potent vasoconstrictive peptide has been previously linked to both atherosclerosis and hypertension¹². Given the strong effects of EDN1 on vascular tone and vascular smooth muscle cell proliferation, it is reasonable to expect a role in the other associated vascular diseases such as fibromuscular dysplasia and arterial dissection. Our finding of a regulatory effect for the lead SNP on has been confirmed in studies of patients with microvascular coronary disease, with patients carrying the risk allele (rs9349379, G) having higher risk of disease and higher EDN1 protein levels⁹. As with all common variants, the effect size of this regulation is small, and there remains controversy whether these changes in EDN1 are sufficient to cause all 8 associated vascular disease. Many studies also link the closest gene to the GWAS SNP—PHACTR1—to vascular effects in cultured ECs^13,14. Therefore it remains an open question whether PHACTR1 is the true causal gene for the chr6p24 locus.

In this issue of ATVB, Rubin et. al. provide an important in vivo analysis of PHACTR1 in the vasculature, and find no significant effects¹⁵. Three separate knockout mice assessed the role of Phactr1 deletion in adult ECs and vascular smooth muscle cells (VSMCs), and the embryologic effect of deletion during embryogenesis in ECs. All mice were generated with loxP-Cre deletion of exon 6, which is common to all Phactr1 transcripts. Even after 8 months of observation and vascular perturbations such as Angiotensin-II infusion and hind limb ischemia, there was no difference in disease phenotypes or outcomes compared with littermate controls. Importantly, the authors assessed the role of Phactr1 knockout in actin organization and EC polarization, and even these microscopic features previously associated with Phactr1 in cultured cells showed no difference with Phactr1 knockout in vivo. The primary data supporting a role for PHACTR1 remains a strong expression-quantitative trait locus (eQTL) in post-mortem vascular tissue, and its position as the closest gene to the GWAS SNP. Though these are often good predictors for causality, they are by no means perfect. Disease associated SNPs often regulate multiple genes, resulting in a high false positive rate for eQTLs¹⁶.

Therefore, despite the eQTL for PHACTR1, the lack of a vascular phenotype for the Phactr1 knockout mice strongly suggests it is not a major driver of vascular function, in stark contrast to Endothelin-1 (Figure). EDN1 transgenic mice develop increased atherosclerosis, and Endothelin Receptor A (ET-A) antagonist-treated mice show regression of plaques^12,17. Edn1+/− mice paradoxically have increased blood pressure, despite lower levels of the vasoconstrictive peptide¹⁸. This paradox is also seen in the GWAS associations at chr6p24, with the allele associated with increased EDN1 and CAD surprising associated with lower blood pressure. These counterintuitive effects can be explained in both humans and mice by the opposing effects of the ET-A and ET-B receptors. In coronary arteries where Endothelin-1 mostly binds the ET-A receptor, the effect is vasoconstriction and VSMC proliferation. However, in the kidneys the ET-B receptor predominates, where it promotes natriuresis and lower systemic blood pressure. The directionally consistent effects of EDN1 with the GWAS findings establishes a common link between known EDN1 physiology and the 8 genetically associated vascular diseases.

Figure:

Summary of evidence implicating either PHACTR1 or EDN1 as the causal gene at the chromosome 6p24 GWAS locus associated with multiple vascular diseases.

It is difficult to dismiss entirely the role of PHACTR1 based on these data alone. Though ECs and VSMCs are the obvious first cells to study given the vascular disease associations, PHACTR1 in monocytes, macrophages, and/or adventitial fibroblasts may contribute to disease phenotypes as well. A recent study identified Phactr1 knockout macrophages had decreased efferocytosis in the plaques of Ldlr−/− mice¹⁹. This suggests a role in myeloid cells, but does not explain the other chromosome 6p24 associations for vasoactive diseases without plaque such as migraine headache and microvascular angina. A second possible explanation could be that PHACTR1 and EDN1 play synergistic roles in vascular disease, each affecting different cells in the arterial wall. New efforts to identify the relationship between genetic variation and cell-type specific gene expression (single cell eQTLs) will help shed light on each of these possibilities.

The evolving story at the chromosome 6p24 locus provides important insights for future variant-to-function studies for CAD. The small effect of non-coding GWAS variants remains a challenge, and therefore numerous studies are required to establish a causal link to a gene. Hypotheses generated in genome-wide transcriptional databases are a reasonable place to start, but in vivo models of vascular disease are invaluable for causal gene function. For the EDN1/PHACTR1 locus at chr6p24 the story still requires more experiments in tissue-specific models of disease, to determine if one, both, or even other distal genes explain the effects of genetic variation. Ultimately, a definitive link between variant and biologic function will shed light on the shared pathophysiology of multiple vascular diseases.