Functional regulatory mechanism of smooth muscle cell-restricted LMOD1 coronary artery disease locus

Abstract

Recent genome-wide association studies (GWAS) have identified multiple new loci which appear to alter coronary artery disease (CAD) risk via arterial wall-specific mechanisms. One of the annotated genes encodes LMOD1 (Leiomodin 1), a member of the actin filament nucleator family that is highly enriched in smooth muscle-containing tissues such as the artery wall. However, it is still unknown whether LMOD1 is the causal gene at this locus and also how the associated variants alter LMOD1 expression/function and CAD risk. Using epigenomic profiling we recently identified a non-coding regulatory variant, rs34091558, which is in tight linkage disequilibrium (LD) with the lead CAD GWAS variant, rs2820315. Herein we demonstrate through expression quantitative trait loci (eQTL) and statistical fine-mapping in GTEx, STARNET, and human coronary artery smooth muscle cell (HCASMC) datasets, rs34091558 is the top regulatory variant for LMOD1 in vascular tissues. Position weight matrix (PWM) analyses identify the protective allele rs34091558-TA to form a conserved Forkhead box O3 (FOXO3) binding motif, which is disrupted by the risk allele rs34091558-A. FOXO3 chromatin immunoprecipitation and reporter assays show reduced FOXO3 binding and LMOD1 transcriptional activity by the risk allele, consistent with effects of FOXO3 downregulation on LMOD1. LMOD1 knockdown results in increased proliferation and migration and decreased cell contraction in HCASMC, and immunostaining in atherosclerotic lesions in the SMC lineage tracing reporter mouse support a key role for LMOD1 in maintaining the differentiated SMC phenotype. These results provide compelling functional evidence that genetic variation is associated with dysregulated LMOD1 expression/function in SMCs, together contributing to the heritable risk for CAD.

Fig 1. Association of rs2820315 with CAD and LMOD1 expression.

LocusZoom plot depicting association results at rs2820315 locus at chromosome 1q32.1 in (A) CAD from CARDIoGRAMplusC4D + UKB GWAS, (B) LDL cholesterol from Global Lipid Genetics Consortium and rs2820315 eQTL association (highlighting LD SNP rs34091558) with LMOD1 and other candidate genes in (C) Tibial Artery and (D) Liver in the GTEx dataset. Circles represent SNPs associated using an additive or recessive model, color-coded for LD (r²) with the lead GWAS SNP rs2820315 (purple diamond), which resides near the LMOD1 gene.

Fig 2. Validation of LMOD1 as a causal gene in GTEx database.

(A) Schematic indicating neighboring genes in the LMOD1 locus using RefSeq database, coding transcripts in blue and non-coding transcripts in green (upper panel). Expression profile of coding and non-coding transcripts at LMOD1 locus in coronary artery tissues (n = 133) from GTEx database (v6p) (lower panel). (B) Tissue expression profile of LMOD1 and SHISA4 across the entire GTEx dataset (v7) ranked according to median transcripts per million (TPM) for each tissue. Boxplots showing rs2820315 allele dosage correlated with LMOD1 (C) and rs34091558 allele dosage correlated with LMOD1 (D) in tibial artery tissue in GTEx dataset (v7). (E) Allelic expression imbalance (AEI) for LMOD1 in HCASMCs heterozygous at rs2820315 and rs34091558, using rs2820312 coding SNP as a proxy (n = 12). Values represent mean ± standard deviation of triplicates of cDNA/gRNA normalized allelic ratios.

Fig 3. Functional mapping of the LMOD1 locus identifies FOXO3 dependent mechanism at rs34091558.

(A) Prediction of rs34091558 non-risk (TA) and risk allele (T) on FOXO3 binding based on JASPAR PWM scores. Human sequence (hg38) is shown aligned to the consensus FOXO3 sequence and mammalian genomic sequences and PhyloP conservation track. (B) Relative expression (RPKM) of candidate Forkhead transcription factors in coronary artery tissues (n = 133) from the GTEx dataset. (C) Luciferase activity examined for each of the LMOD1 enhancer constructs in A7r5 SMCs in the presence of FOXO3. Results were reproduced in n = 3 independent experiments performed in quadruplicates. (D) Chromatin immunoprecipitation (ChIP) assay for LMOD1 and EGFR (as a positive control region) in HCASMC chromatin lysates immunoprecipitated with antibodies to FOXO3 or a negative control rabbit IgG. Results were repeated in n = 3 independent studies. (E) Allele specific ChIP (haploChIP) for FOXO3 protein in DNA derived from cultured HCASMC ChIP experiments. DNA from cell line homozygous for the ancestral allele was used as a positive control and arbitrarily set to 1. Values represent mean ± standard deviation of triplicates. Similar results were observed from n = 4 independent lines for each genotype. (F) Quantitative RT-PCR analysis of LMOD1 in HCASMCs following knock down of endogenous FOXO3 (F) or overexpression of FOXO3 (G). The results were reproduced in n = 3 independent experiments.

Fig 4. LMOD1 expression is mediated by PDGF-BB-FOXO3 signaling cascade.

(A) UCSC Browser screenshot of the LMOD1 CAD locus at chromosome 1q32.1 highlighting the candidate causal variant, rs34091558, overlapping ATAC-seq open chromatin and RNA-seq tracks in HCASMCs treated with PDGF-BB (n = 2 biological replicates). Genomic coordinates refer to hg19 assembly. Quantitative RT-PCR (B) and Western blotting (C) data revealing PDGF-BB and phosphorylated-FOXO3 (P-FOXO3) mediated LMOD1 expression. (D) Co-expression microarray analysis performed in a cohort of carotid atherosclerotic plaques (n = 127) indicating that LMOD1 and FOXO3 are positively correlated in arterial tissues.

Fig 5. LMOD1 deficiency promotes de-differentiated SMC phenotype.

(A) Quantitative RT-PCR and Western blotting analysis confirming endogenous knockdown of LMOD1 in cultured HCASMCs. Cell proliferation measured in HCASMCs transfected with siCtrl or siLMOD1 via (B) trypan blue exclusion and (C) CellTiter96 Non-Radioactive Cell Proliferation Assay (MTT assay). (D) Differences in cell migration assessed via Boyden Chamber assay in HCASMCs transfected with siCtrl or siLMOD1. (E) Images of cell contraction in siCtrl or siLMOD1 transfected HCASMCs at indicated time points and represented via quantification (F-G). All data represent mean ± standard deviation from n = 3 independent experiments.

Fig 6. LMOD1 expressing cells associate with atherosclerotic fibrous cap.

(A and D) Immunofluorescence staining for LMOD1, Tomato and DAPI in the brachiocephalic artery of 24 week old Myh11-Cre-ER^T2 ROSA26-STOP-tdTomato Apoe^-/- mice fed high fat diet for 18 weeks. (B and C) LMOD1 and tomato expressing cells were identified in the fibrous cap (white arrows). (E and F) Another lesion showing LMOD1 and tomato expressing cells in the media (blue arrows). Tomato positive cells in the lesion stained negative for LMOD1 in the lesion. Images were captured using a 20x objective and are representative of n = 3 independent mice.

Fig 7. Proposed mechanism for how rs34091558 impairs FOXO3 mediated LMOD1 expression therefore altering CAD risk.

Individuals having the rs34091558-TA ancestral, protective allele would be at reduced CAD risk due to greater LMOD1 expression levels, whereas individuals having the rs34091558-T derived, risk allele would be at greater CAD risk due to reduced LMOD1 expression levels, through a FOXO3-dependent mechanism.