A common polymorphism that protects from cardiovascular disease increases fibronectin processing and secretion

Recent large scale bioinformatic analyses have identified common genetic variants within the fibronectin (FN1) gene that predispose to cardiovascular disease, through mechanisms that remain to be investigated. This work explores the underlying mechanisms and identifies a novel process controlling fibronectin secretion. First, we demonstrate that higher levels of FN1 protein in plasma associate with a reduced risk of cardiovascular disease.. Next, cellular models were leveraged to demonstrate that the CAD associated region encompasses a L15Q polymorphism within the FN1 signal peptide that impacts secretion of FN1 both qualitatively and quantitatively. Thus, by reducing FN1 secretion, a variant within the signal peptide contributes to lower circulating FN1 and increased CAD risk. In addition to providing novel functional evidence implicating FN1 in cardiovascular disease, these findings demonstrate that a common variant within a secretion signal peptide regulates protein function.


Introduction
(ECM), where they are deposited as insoluble fibers involved in cell adhesion. The second 73 major form of FN1, pFN, is secreted by the liver into the circulation where it is abundant. Mice 74 deficient in pFN display largely normal hemostasis and wound-healing, consistent with a 75 predominant or exclusive role of cFN in these processes [15]. Interestingly, pFN deficient mice 76 display increased neuronal apoptosis and larger infarct areas following focal brain ischemia, 77 suggesting that pFN plays a protective role, possibly by activating anti-apoptotic mechanisms via 78 integrin signaling [15]. While pFN is not essential to vascular integrity, pFN has been shown to 79 penetrate the vessel wall and to constitute a significant portion of arterial FN where it may 80 participate in tissue remodeling [16,17]. 81 Here, we explore and clarify the mechanisms linking CAD to common GWAS identified 82 variants that map to the FN1gene. Using bioinformatic and molecular approaches we provide 83 evidence that differential post-transcriptional regulation underlies the FN1-CAD association. 84 More specifically a polymorphism within the propeptide of FN1 was found to regulate the ability 85 of FN1 to be secreted. These findings provide a unique portrait of a common coding variant 86 linked to CAD that has functional consequences at the protein level without affecting its mature 87 amino acid sequence. The CAD-linked haplotype harbors several tightly linked SNPs that correlate with the disease 96 (including top SNP rs1250229) that are causal candidates (Figure 1, Table S1, Figure S1). 97 98 Strikingly, the region contains a single coding SNP (rs1250259), central to the CAD associated 99 region, which was prioritized for follow-up. Interrogation of genome-wide association studies 100 using PhenoScanner and Open Targets points to an association between the most common allele 101 (rs1250259-A) and lower pulse pressure, reduced CAD risk, as well as to changes in blood FN1 102 levels ( Table S2, Table S3) [2,[18][19][20][21]. While FN1 levels as a function of the rs1250259 103 genotype are not available, the proximal CAD protective T allele (rs1250258-T), closely linked 104 (R 2 =0.99) to rs1250259-A is associated with increased circulating FN1 and fragments thereof, 105 suggesting that it may play a cardioprotective role [22]. 106 We next performed Mendelian randomization to test a causal role for FN1 per se. In this 107 analysis, all SNPs associated with changes in FN1 protein expression are pooled and tested for 108 association of each of CAD risk and FN1 levels. Consistent with individual SNP contributions, 109 FN1 and CAD were inversely correlated, with higher circulating FN1 linked to lower CAD 110 prevalence ( Table 1). Identification of a missense mutation within FN1 linked to CAD that is predicted to affect 120 secretion 121 Although the above analysis focused on FN1 protein expression, the contribution to FN1 mRNA 122 expression remained to be tested. To examine the impact of this substitution on FN1 secretion, a model fusion protein consisting of 136 amino acids 1-182 of FN1 fused to a GFP-HA tag moiety was generated (Figure 2A). migrates slightly slower than the L15 variant ( Figure 2B). Both fusion variants were present at 144 comparable levels in HEK293T lysates after correcting for transfection efficiency (Figure 2 C). 145 Presence of the secreted protein in the media was tested next (Figure 3).

147
In HEK293T and HeLa cells, transfection resulted in the secretion of FN1-GFP in the media,

148
with the L15 exhibiting greater propensity to be secreted, defined as the signal in the culture 149 media relative to the cellular signal. To examine the impact of this polymorphism on secretion 150 by the liver, which is the major physiological source of pFN1, HuH-7 hepatoma cells, a widely 151 model of hepatocyte function, were transfected next. Although the difference was smaller than 152 observed in the epithelial models above, L15 FN1-GFP was also more readily secreted by HuH-7 153 cells. Finally, FN1-GFP was transduced into several primary cell models with relevance to 154 CAD, i.e., adventitial fibroblasts, endothelial cells, and coronary smooth muscle cells. In all 155 models, the Leu form seemed on average better secreted than the Gln form, although the 156 difference reached statistical significance only in a lot of coronary smooth muscle cells.

Secreted forms of Q15 qualitatively differ in some primary cells 159
Examination of the variants by SDS-PAGE revealed some unexpected findings. Delivery of 160 FN1-GFP demonstrated isoform-specific differences in the secreted forms, in a cell-type 161 different manner (Figure 4).

163
In some cells (fibroblasts, muscle models as well as HeLa cells), transduction of the Q15 form 164 led to enrichment relative to the L15 form of a slower migrating band on SDS-PAGE. By 165 contrast, FN1-GFP from endothelial cells and HEK293T resembled HuH-7 cells in that both 166 secreted forms exhibited qualitatively more similar profiles. Thus, in some cell types, the L15Q 167 polymorphism appears to dictate both quality and quantity of FN1-GFP secreted. 169 We hypothesized that this 3-5 kDa difference was due to variable levels of post-  polymorphism. Additional investigations will be needed to resolve this question.

262
We demonstrate that the cardioprotective allele is linked to increased FN1 secretion 263 indicating that circulating FN1 protects against CAD. One limitation to this interpretation is that One mechanism underlying the role of FN1 in CAD involves the inflammatory 272 compartment. As shown for alveolar macrophages, maturation from monocytes in vitro is linked 273 to increased fibronectin production and secretion, which is in turn reduced upon inflammatory   min) to remove cellular debris, and further cleared at high speed for 5 min (13,000 X g).

374
Recombinant FN1-GFP-HA was isolated from 10 ml of media (corresponding to a 10 cm culture     Table S3. Phenotypic studies associated with variations in FN1 expression. Open Targets Genetics was interrogated for studies associated with FN1. The tool identifies SNPs associated with the FN1 term that were linked to various phenotypical traits. All variants are in tight LD (R 2 > 0.8). A subset of the genome-wide significant (< 5E-8) associations are shown.

FN1
Fig S1. Haploview map with overlapping genes. Region spanning rs1975319 to rs6726337 is shown. LD intensity is proportional to linkage (R 2 ) values. Blocks were defined using the LD spine method. LD Only common SNPs (frequency > 0.05) are shown. Red arrow points to rs1250259.