Loss of the Coronary Artery Disease Risk Gene Leiomodin1 in Vascular Smooth Muscle Cells Triggers Rapid Onset Coronary Atherosclerosis

Abstract

Background:

Atherosclerosis is the primary underlying cause of coronary artery disease (CAD). Leiomodin1 is a CAD risk gene whose role in coronary artery pathophysiology is unknown. Whole-body loss of Leiomodin1 (Lmod1) causes a lethal neonatal visceral myopathy in mice, necessitating unique approaches for the study of vascular smooth muscle cell (VSMC) phenotypes.

Methods:

Control (Lmod1^WT carrying a causes recombination [Cre] allele) and VSMC-restricted Lmod1 knockout (Lmod1^SMKO) mice were subjected to various atherogenic regimens. Atherosclerosis and LMOD1 expression in mouse and human coronary arteries were assessed by histopathology and confocal immunofluorescence microscopy. Coronary arteries from Lmod1^WT and Lmod1^SMKO mice were analyzed with assorted stains and antibodies, immunogold lineage tracing, and spatial metabolomics. Mouse aortic SMCs from Lmod1^WT and Lmod1^SMKO mice were subjected to lipid loading ± lentivirus carrying wild type or actin-nucleation-deficient Lmod1 (Lmod1^ND). Mice harboring an intronic deletion of Lmod1 or Lmod1^ND, were engineered using clustered regularly interspaced short palindromic repeats (CRISPR).

Results:

A lethal neonatal visceral myopathy occurred in Lmod1^SMKO mice using Myh11-CreER^T2, prohibiting further investigation. In contrast, Lmod1^SMKO mice generated with Itga8-CreER^T2 survived and were therefore used in all subsequent studies. Under atherogenic conditions, Lmod1^SMKO mice displayed little vascular disease in several organs but developed diffuse, occlusive coronary atherosclerosis with fibrous caps. No such disease was observed in Lmod1^WT mice. Time-course studies documented lipid insudation and VSMC migration into the intima of coronary arteries of Lmod1^SMKO mice as early as eight days post-regimen. Immunogold lineage tracing revealed that 46% of coronary plaque cells were of VSMC origin. Spatial metabolomics uncovered multiple lipid species within coronary atheromata of Lmod1^SMKO mice. In vitro studies demonstrated elevated lipid accumulation in Lmod1^SMKO VSMCs, which was rescued by viral-mediated Lmod1^WT or Lmod1^ND expression. An intronic deletion of Lmod1, comprising conserved orthologous sequence where the single nucleotide variant associated with CAD exists, showed attenuated LMOD1 expression. Heterozygous Lmod1^SMKO mice, with a comparable reduction in LMOD1, displayed no CAD. Similarly, VSMC-restricted expression of Lmod1^ND resulted in negligible coronary atherosclerosis.

Conclusions:

Under atherogenic conditions, Lmod1^SMKO mice present with rapid onset coronary atherosclerosis. LMOD1 safeguards coronary homeostasis, apparently in an actin nucleation-independent manner.

Ischemic heart disease remains the leading cause of age-standardized death in the world, driven predominantly by the progressive burden of coronary artery disease (CAD).¹ CAD develops from coronary atherosclerosis, initiated by subendothelial retention, modification, and aggregation of low-density lipoprotein (LDL) particles.² These early events promote phenotypic changes in endothelial cells, vascular smooth muscle cells (VSMCs) and an array of immune cells, leading to complex signaling events linked to inflammation, migration, proliferation, extracellular matrix remodeling, and fibrous cap and necrotic core formation.^3–5 Progressive plaque expansion and fibrous cap thinning predispose lesions to erosion or rupture, precipitating thrombosis and myocardial infarction.^6,7 While modifiable risk factors such as smoking, hypercholesterolemia, and hypertension can be managed, non-modifiable factors, such as age and sex, cannot. Further, genome wide association studies (GWAS) have identified heritable single nucleotide variants (SNVs) in and around >340 risk alleles for CAD.⁸ A small number of GWAS risk alleles carry missense mutations directly implicated in CAD (eg, LDLR, APOE, PCSK9); however, the vast majority of CAD risk alleles harbor noncoding variants that are poorly understood.^8–10 The extensive historic and ongoing literature related to GWAS for CAD will require enormous efforts to experimentally elucidate individual and combined effects of SNVs on a patient’s risk of CAD.

Multiple lines of evidence now establish that VSMCs undergo dynamic fate specification and state transitions that critically contribute to atherogenesis.^11–14 Notably, three CAD risk genes (FHL5, LMOD1, and MYH11) are highly restricted to SMCs, but their role in atherosclerosis has largely been limited to expression-based studies.^15–17 Coding mutations in LMOD1 and MYH11 are linked to visceral SMC diseases of the intestine and bladder.^18,19 Predictably, global inactivation of Lmod1 or Myh11 in mice results in neonatal death due to visceral myopathy.^18,20 Conditional inactivation of these CAD risk genes using conventional SMC Cre mouse lines would likely be problematic due to broad Cre activity in both vascular and visceral SMC lineages.^21,22 Recently, a VSMC-restrictive Itga8-CreER^T2 mouse line was developed that circumvents lethal gastrointestinal phenotypes, thus allowing for the unambiguous study of vascular phenotypes.^23–26

Here, we describe a mouse model of rapid coronary atherosclerosis, enabled by Itga8-CreER^T2-mediated inactivation of Lmod1 under various atherogenic conditions. No such phenotype was observed in similarly-treated mice carrying only the Itga8-CreER^T2 allele. Notably, inducible loss of Lmod1 with the popular Myh11-CreER^T2 driver²⁷ resulted in acute intestinal distention and rapid death following tamoxifen administration. Several assays – some never previously applied in mouse models of coronary atherosclerosis – underscore a critical role for normal LMOD1 function in VSMCs to safeguard the coronary vasculature under hypercholesterolemic conditions.

Methods

A more detailed description of the Materials and Methods is available in the Supplemental Materials.

Data availability

The Itga8-CreER^T2 mouse is available through the Mutant Mouse Resource & Research Center (University of North Carolina, Chapel Hill; https://www.mmrrc.org/catalog/sds.php?mmrrc_id=69930). The floxed Lmod1 mouse will be made available to the research community upon written request. The data shown here represents a much larger dataset; we will make available to the public and the research community additional details and/or examples upon reasonable request.

Human specimens and data analysis

Formalin-fixed paraffin-embedded (FFPE) de-identified human coronary arteries were obtained from the University of Miami Health System (IRB exempt) and used to probe LMOD1 and ACTA2 protein expression. Several publicly available genomic/transcriptomic datasets of human coronary or carotid artery were interrogated using a suite of bioinformatic tools.

Animal models

A disease-causing SNV (R365*) in the second exon of Lmod1 was created with three-component CRISPR.²⁸ A newly floxed Lmod1 mouse was crossed with either Itga8-CreER^{T2 23} or Myh11-CreER^T2.²⁷ A segment of intron 1 in Lmod1, where an SNV associated with CAD exists in the orthologous human LMOD1 gene, was excised using two-component CRISPR.²⁸ An inducible five amino acid knockin of Lmod1, deficient for actin nucleation, was generated with an extreme genome editing system (Biocytogen). An Apoe null mouse was crossed with floxed Lmod1 and Itga8-CreER^T2 to generate a double knockout. The membrane tomato/membrane green fluorescent protein (mTmG) reporter mouse was obtained from Jackson Laboratories and crossed with the conditional Lmod1 knockout mouse for lineage tracing studies. All animal procedures were carried out in accordance with institutional guidelines and approved by Institutional Animal Care and Use Committees.

Atherogenic regimen

Initial cohorts of mice were subjected to retro-orbital injection of adeno-associated virus, serotype 8, carrying a gain-of-function mutation in proprotein convertase subtilisin/kexin type 9 (PCSK9),²⁹ followed immediately by a high fat diet (HFD). Two weeks later, minipumps were implanted for a four-week infusion of angiotensin II (AngII; 1,000ng/kg/min). Subsequent studies were performed with only the PCSK9/HFD regimen for varying times as indicated. The Apoe/Lmod1 double knockout mouse model was subjected to a Western diet (WD) for 11 weeks.

Blood assays

Total cholesterol, HDL, LDL, triglycerides, and Lipopolysaccharide (LPS) were measured with commercial kits. Complete blood counts were determined with an automated cell counter. Electrolyte, glucose (non-fasting), total protein, and kidney/liver enzyme levels were measured with clinical grade kits.

Blood pressure

Systemic blood pressure was measured in VSMC conditional Lmod1 knockout mice using a tail cuff method.

Mouse vascular tissue analyses

Coronary arteries in the hearts of Lmod1^WT or Lmod1^SMKO mice were processed for frozen sectioning and underwent either histochemical staining with Oil-Red-O (ORO) or confocal immunofluorescence microscopy (CIFM) imaging for various antigens, including ACTA2, CD68, MKI67, LGALS3, LMOD1, MYH11, and SPP1. Paraffin-embedded hearts of each Lmod1 genotype were sectioned for staining with hematoxylin and eosin (H&E), Masson trichrome, Picrosirius Red, or Alizarin Red. The vasculature of the brain, kidney, liver, lung, mesentery, and spleen was stained with H&E, ORO, and antibodies to LMOD1 and ACTA2. The aorta was stained with ORO and carotid arteries were stained with H&E. In some experiments, Lmod1^WT or Lmod1^SMKO mouse coronary arteries were processed for conventional transmission electron microscopy (TEM) or immunogold electron microscopy to assess ultrastructure and lineage traced VSMCs, respectively.

Quantitative gene and protein expression

Total RNA and protein were isolated from the aorta of wild type or conditional Lmod1 knockout/knockin mouse models for quantitative RT-PCR or Western blotting, respectively, using established primers and antibodies as indicated in the results and Tables S1 and S2.

Mass spectrometry imaging (MSI)

Molecular lipid profiling was conducted in the coronary arteries of wild type and conditional Lmod1 knockout mice by sectioning hearts for MSI as generally described.³⁰

Cultured mouse aortic smooth muscle cells

Primary isolates of Lmod1^WT or Lmod1^SMKO mouse aortic SMCs were prepared with an established protocol³¹ for quantitative measures of DNA recombination efficiency (using PCR), LMOD1 protein (Western blotting), and oxidized LDL uptake (fluorescence staining).

Statistical analysis

Over 500 mice were studied over the course of four years in 13 separate cohorts. Data normality was assessed with D’Agostino–Pearson omnibus test with Q–Q plot inspection and Shapiro-Wilk test. Homogeneity of variance was assessed in SPSS using Levene’s test. Where multiple groups were analyzed, one- or two-way ANOVA was performed followed by post-hoc testing. Unless otherwise indicated, when only two groups were compared, statistical significance was determined using a two-tailed unpaired t-test. Where assumptions of normality and equal variances were not satisfied, non-parametric testing was conducted using Kruskal-Wallis, Mann-Whitney U or Welch’s t-test, as indicated in the legends. Kaplan-Meier survival data were analyzed by the Mantel-Cox test. An a priori statistical measure of significance was set at 0.05. Analyses were performed using SPSS^® software V.29 (Chicago, IL) and GraphPad Prism (versions 9–10; GraphPad Software, La Jolla, CA, USA). The latter was also used for creating statistical illustrations. Where continuous data are shown, scatterplot bar graphing was used to show the mean and standard deviation (SD).

Results

LMOD1 is enriched in human coronary artery SMCs and reduced with atherosclerosis

An analysis of previously published scRNA-seq data¹¹ revealed abundant expression of LMOD1 mRNA in SMC and pericyte clusters of human coronary artery, with much lower levels in modulated SMCs and fibroblasts, and barely detectable expression in endothelial cell and immune cell clusters (Figure 1A, 1B; Figure S1A). An analysis of scATAC-seq data³² disclosed a similar trend in chromatin accessibility, with most LMOD1 gene activity clustered in human coronary artery SMCs (Figure 1C, 1D; Figure S1B). Spatial expression data³³ showed a reduction of LMOD1 mRNA in carotid atherosclerotic lesions (Figure 1E, 1F). Consistent with these genomic findings, CIFM demonstrated LMOD1 protein in ACTA2-positive medial SMCs of human coronary arteries, with much lower levels in overlying atheromatous plaque (Figure 1G; Figure S2). Collectively, these results are consistent with the preferential expression of LMOD1 in differentiated SMCs^34,35 and its attenuated expression in human atherosclerotic lesions.^16,32,36

Figure 1: LMOD1 expression in human coronary artery atherosclerosis.

(A) UMAP and attending feature plot (B) of LMOD1 derived from scRNA-seq study¹¹ of human coronary atherosclerosis. SMC, smooth muscle cells; Mod, phenotypically modulated SMCs; PC, pericytes; Fibro, fibroblasts; EC, endothelial cells; Macro, macrophages; T, T cells; NK, natural killer cells; B, B cells; Pla, plasma cells. (C) UMAP and attending feature plot (D) of LMOD1 derived from scATAC-seq study³² of human coronary atherosclerosis. (E) H&E image and (F) LMOD1 expression derived from indicated spatial transcriptomics study of human carotid atherosclerosis.³³ (G) H&E image of human coronary artery from a male subject with adjacent high magnification CIFM images of LMOD1 and ACTA2 from the boxed regions labeled i and ii in H&E panel. Scale bars are 1 mm in the H&E image and 20 μm in the CIFM images. Int, intima; Med, media.

Itga8-CreER^T2 circumvents an otherwise lethal intestinal myopathy with loss of Lmod1

Whole-body knockout of Lmod1, through CRISPR-mediated disruption of exon 1 or installation of pathologic substitutions that result in a premature termination codon in exon 2, results in neonatal death due to visceral myopathy of the intestine and bladder (Figure S3A, S3B).¹⁸ Thus, defining chronic vascular phenotypes in global Lmod1 knockout mice is impossible. Accordingly, the Lmod1 promoter containing two SRF-binding CArG boxes³⁵ and its first exon were floxed for conditional gene loss experiments (Figure S3C). Homozygous floxed Lmod1 mice showed no change in expression of Lmod1 mRNA or LMOD1 protein indicating the LoxP sequences did not disrupt regulatory elements in the Lmod1 promoter or intronic region (Figure S3D, S3E). A comparable loss of aortic LMOD1 protein was observed in Tamoxifen (Tmx)-treated Lmod1^KO-Myh11 and Lmod1^KO-Itga8 mice (Figure 2A, 2B), with an estimated recombination efficiency of ~80% using Itga8-CreER^T2 (Figure S3F). Quantitative RT-PCR revealed the conditional knockout of Lmod1 to be a true null allele (Figure S3G).³⁷ Lmod1^KO-Myh11 mice showed loss of LMOD1 protein in intestinal SMCs (Figure 2A, 2C) and distension of the intestine, leading to rapid death within a week of the last dose of Tmx (Figure 2D, left; Figure S3H). Moreover, all oil-treated Lmod1^KO-Myh11 mice died by 12 weeks of age, likely from cumulative leakiness of this specific Cre driver (Figure 2D, left).^23,38 In contrast, Tmx-treated Lmod1^KO-Itga8 mice showed near normal levels of LMOD1 protein in the intestine (Figure 2A, 2C), and the majority of mice lived for at least six months (Figure 2D, right). LMOD1 protein levels were reduced by ~90% in the aorta of Lmod1^KO-Itga8 mice after six months of Tmx administration with no significant change in levels of TAGLN or ITGA8 protein (Figure 2E, 2F) and little evidence of intestinal pathology (Figure S3H). Female Lmod1^KO-Itga8 mice exhibited similar body weight as oil-control mice over a six-month period, but Lmod1^KO-Itga8 males began to slow in weight gain at 31 weeks of age (Figure S3I). Both male and female Lmod1^KO-Itga8 mice showed a reduction in systolic blood pressure versus oil-administered controls at six months post-Tmx (Figure S3J). Interestingly, serum levels of LPS were elevated in Lmod1^KO-Myh11 mice, but not in Lmod1^KO-Itga8 mice, suggesting sepsis-mediated death in the former animals (Figure S3K). Taken together, these findings extend previous reports demonstrating circumvention of lethal visceral myopathies using Itga8-CreER^T2, thus allowing for exploration of vascular phenotypes that otherwise would not be possible.^23–26

Figure 2. Conditional knockout of Lmod1 with Myh11-CreER^T2 or Itga8-CreER^T2.

(A) CIFM of LMOD1 and ACTA2 in aorta and intestine of indicated mouse genotype. Scale bars are 20 μm. Western blotting for LMOD1 in aorta (B) and intestine (C) of indicated mouse genotype. (D) Kaplan-Meier survival curves for Oil-treated versus Tmx-treated floxed Lmod1/Myh11-CreER^T2 (left) and floxed Lmod1/Itga8-CreER^T2 (right) mice. The black arrow indicates the time of Oil or Tmx administration. Survival curves were compared by the log-rank (Mantel-Cox) test. Western blotting (E) and quantitation (F) of indicated proteins after six months of Oil or Tmx administration; n=5 independent mice per genotype (Lmod1^WT here is Lmod1^fl/fl). Data were analyzed by two-tailed unpaired Student’s t-test and are shown as the mean ± SD.

Lmod1^SMKO mice have enlarged blood vessels and reduced blood pressure at baseline

After 10 days of Tmx administration, the aorta of Lmod1^KO-Itga8 mice (hereafter abbreviated as Lmod1^SMKO) was measurably larger than Lmod1^WT controls (carrying Itga8-CreER^T2 and receiving Tmx), from ascending to abdominal aorta (Figure S4A–S4C). Compared with Lmod1^WT mice, the carotid artery of Lmod1^SMKO mice was enlarged (Figure S4D) with a significant increase in circumference and an attending decrease in medial thickness (Figure S4E). An analysis of hearts from apex to base (Figure S4F) revealed a significant increase in the circumference of coronary arteries of Lmod1^SMKO mice (Figure S4G, S4H). The attenuated systolic blood pressure seen at six months post-Tmx (Figure S3J) was observed as early as 10 days post-Tmx in both male and female Lmod1^SMKO mice (Figure S4I). Despite the above morpho-physiological changes in the vasculature and lowered LMOD1 protein in the aorta of Lmod1^SMKO mice, there was no significant reduction in several SMC contractile proteins, including MYH11, ACTA2, and TAGLN (Figure S4J, S4K). Further, although there is functional loss of one Itga8 allele in Lmod1^SMKO mice, levels of ITGA8 protein were unchanged across genotypes (Figure S4J, S4K). These findings establish a baseline hypotensive phenotype with outward vascular remodeling and normal SMC contractile protein expression in Lmod1^SMKO mice.

Lmod1^SMKO mice exhibit AngII-independent occlusive CAD with an atherogenic regimen

Based on the enlarged vessel phenotype, we surmised Lmod1^SMKO mice would be susceptible to aneurysm formation following a PCSK9/HFD/AngII regimen (Figure S5A). All mice showed similar hypercholesterolemia (Figure S5B) and body weights (Figure S5C) after two weeks of PCSK9/HFD, but only Lmod1^WT mice displayed evidence of abdominal aneurysm following AngII infusion (Figure S5D). Although there was some death in the Lmod1^WT control arm, far more Lmod1^SMKO mice died with the combined atherogenic/AngII regimen (Figure S5E). Surprisingly, Lmod1^SMKO mice did not exhibit cardiac hypertrophy (Figure S5F) or cardiac fibrosis (Figure S5J). However, gross inspection of the heart of Lmod1^SMKO mice revealed a diffuse, tofu-like appearance of the coronary vasculature (Figure S5G, arrows). Histological analysis of Lmod1^SMKO mouse hearts demonstrated occlusive CAD with fibrous caps (Figure S5H, S5I) and lipid droplet formation (Figure S5K). In no instance were any of these coronary artery phenotypes manifested in Lmod1^WT mice.

To clarify whether the CAD phenotype in Lmod1^SMKO mice required AngII, we conducted a study in the same manner as above only with no AngII administration (Figure 3A). All mice had comparable hypercholesterolemia at 11.5 weeks of age, or about two weeks post-PCSK9/HFD (Figure 3B left), and the levels of cholesterol, triglycerides, LDL, and HDL were similar between genotypes at the termination (16 weeks of age) of the study (Figure 3B right, 3C). Lmod1^SMKO mice had ~35% mortality over the course of the study (Figure 3D), considerably less than that seen with inclusion of AngII (Figure S5E). Blood cell and blood chemistry measures revealed higher circulating lymphocyte and lower neutrophil counts as well as a decrease plasma chloride level in Lmod1^SMKO mice (Table S3). ORO staining of the aorta was greater in Lmod1^SMKO mice than Lmod1^WT mice (Figure 3E, 3F). Similar to PCSK9/HFD/AngII, the PCSK9/HFD regimen triggered a tofu-like appearance of coronary arteries in Lmod1^SMKO mice, with no such appearance in any of the Lmod1^WT mice (Figure 3G). Trichrome staining revealed occlusive coronary lesions in Lmod1^SMKO mice (Figure 3H) with evidence of a fibrous cap as shown by ACTA2 staining (Figure 3I). CIFM consistently demonstrated virtually no detectable LMOD1 staining in the coronary arteries of Lmod1^SMKO mice, indicating efficient recombination of the floxed Lmod1 alleles with Itga8-CreER^T2 (Figure 3I; Figure S6). Collectively, these findings demonstrate an AngII-independent, occlusive CAD phenotype in Lmod1^SMKO mice under an atherogenic regimen.

Figure 3. AngII-independent CAD phenotype in Lmod1^SMKO mice.

(A) Experimental study starting at seven weeks of age. The end point of 16 weeks indicates the time of serum lipid measurements. Most studies below were extended beyond this time point. Beginning here, all Lmod1^WT mice carried the Itga8-CreER^T2 allele and were treated with Tmx and the same atherogenic regimen as Lmod1^SMKO mice. (B) Total serum cholesterol in Lmod1^WT (black dots) and Lmod1^SMKO (red dots) mice at indicated ages. (C) Total serum triglycerides, HDL and LDL. (D) Kaplan-Meier survival curves for Lmod1^WT and Lmod1^SMKO mice were compared by log-rank (Mantel-Cox) test. (E) ORO staining of aortae of Lmod1^WT (n=5) and Lmod1^SMKO (n=9) mice after 103 days of PCSK9/HFD treatment. (F) Quantitation of percent ORO staining in panel E using Mann-Whitney U test (mean ± SD). (G) Gross cardiac images showing tofu-like appearance of intramyocardial coronary arteries (yellow arrows) in Lmod1^SMKO mice treated with PCSK9/HFD for 103 days. Masson trichrome staining (H) and CIFM imaging of LMOD1 and ACTA2 protein (I) in sections of coronary artery from Lmod1^WT (top panels) and Lmod1^SMKO (bottom) mice under the PCSK9/HFD regimen for 52 days. Scale bars, 50 μm. Comparisons between Lmod1^WT and Lmod1^SMKO in each indicated panel were analyzed by two-tailed unpaired Student’s t -test.

Apoe/Lmod1 double knockout mice exhibit CAD

PCSK9 has pleiotropic effects in the vessel wall that are independent of the low-density lipoprotein receptor (LDLR), including VSMC phenotypic modulation to an osteogenic state, endothelial cell oxidative stress and consequent reduction of nitric oxide, and generalized inflammation.^39–41 Such effects are likely amplified with the gain-of-function PCSK9 adeno-associated virus. To test whether LDLR-independent effects of PCSK9 underlie the CAD phenotype observed in Lmod1^SMKO mice, we crossed Lmod1^SMKO mice with Apoe^−/− mice to generate Lmod1^SMKO/Apoe^−/− double knockouts and subjected these mice to either Tmx or Oil treatment, followed by a WD for 11 weeks. Oil-treated Lmod1^SMKO/Apoe^−/− mice showed no evidence of tofu-like coronary arteries or atherosclerotic disease upon sectioning (Figure 4A, 4B). Further, these control animals showed colocalization of LMOD1 and ACTA2 within coronary arteries (Figure 4C, 4D). In contrast, Tmx-treated Lmod1^SMKO/Apoe^−/− mice displayed tofu-like coronary arteries with similar occlusive lesions seen in the PCSK9/HFD model (Figure 4E, 4F). As expected, vanishingly low levels of LMOD1 protein were observed in coronary arteries of Tmx-treated Lmod1^SMKO/Apoe^−/− mice (Figure 4G, 4H). These findings demonstrate CAD in an independent Lmod1^SMKO atherogenic model and suggest the CAD phenotype is a function of the loss in LMOD1 rather than LDLR-independent actions of PCSK9.

Figure 4. CAD phenotype in Lmod1/Apoe dKO mice.

Control (A-D) and Lmod1/Apoe dKO (E-H) mice were subjected to a WD for 11 weeks and hearts imaged and processed for staining with H&E (A, B, E, F) or CIFM imaging (C, D, G, H) of ACTA2 and LMOD1. Note tofu-like appearance of coronaries (black arrows, insets) in dKO hearts. Comparable results were found in two additional pairs of mice. Scale bars are 50 μm for H&E and 20 μm for CIFM. Lu, lumen.

CAD lesions in Lmod1^SMKO mice are manifested early following PCSK9/HFD

Time course studies were performed to define the onset of CAD and the physicochemical properties of lesions in Lmod1^SMKO mice under a PCSK9/HFD regimen. Intimal accretion and ORO positivity were readily detected in coronary arteries of Lmod1^SMKO mice after only eight days of the PCSK9/HFD regimen (Figure 5A–5C; Figure S7). At 16 days, H&E, trichrome, and ORO positive lesions were larger (Figure 5D–5F; Figure S7) and by 63 days, lipid-laden, nearly occluded coronary lesions were evident with a prominent fibrous cap as revealed by trichrome staining (Figure 5G–5I; Figure S7). The latter finding was further substantiated by quantitative picrosirius red staining (Figure S8A–S8E). Quantitative measures of ORO area (Figure 5J), plaque area (Figure 5K), and coronary circumference (Figure 5L) were highly significant in Lmod1^SMKO mice. Of note, occlusive lesions were frequently observed in small (40 μm) coronary arterioles, but were notably infrequent in larger coronary arteries (>200 μm) at the base of the heart (Figure S9). The latter may explain why no evidence of myocardial infarction was observed in >100 Lmod1^SMKO mice. The necrotic core of advanced lesions occupied ~35% of plaque area (Figure S10A–S10D); however, while human coronary lesions showed obvious calcification, no such change was observed in Lmod1^SMKO mice under the experimental conditions here (Figure S10E, S10F). Surprisingly, despite clear evidence of a loss in LMOD1 staining, no occlusive atherosclerosis was observed in the brain, kidney, liver, lung, mesentery, and spleen vasculature of Lmod1^SMKO mice (Figure S11). However, variable frequencies of very mild ORO positive vessels could be observed in these organ beds, ranging from 0% (kidney) to 26% (liver) (Figure S11, Table S4).

Figure 5. Time course of coronary lesion development in Lmod1^SMKO mice.

Representative staining of coronary vessels from Lmod1^WT (n=5 female and n=5 male) and Lmod1^SMKO (n=5 female and n=5 male) mice after 8 (A-C), 16 (D-F), and 63 (G-I) days of PCSK9/HFD. Scales bars (located in upper right corner of each panel) are 20 μm save lower panel H (50 μm). Quantitative measures of ORO staining (J), plaque area (K), and circumference (L) of coronary arteries from Lmod1^WT and Lmod1^SMKO mice after 63 days of PCSK9/HFD (mean ± SD) were analyzed by two-tailed unpaired Welch’s t-test. Data in panel J represents 10 and 12 coronary arteries from 5 Lmod1^WT and 5 Lmod1^SMKO mice, respectively. Data in panels K and L represent 26 and 19 coronary arteries from 5 Lmod1^WT and 5 Lmod1^SMKO mice, respectively. TEM showing (M) a presumptive medial SMC migrating through a fenestra (yellow arrows) of the internal elastic lamina (marked in red asterisks) at 8 days; (N) two intimal foam cells (black arrows), two recently migrated intimal SMC (yellow arrows, one of which has a lipid droplet), and a medial SMC foam cell (white arrow) at 16 days; and (O) a more complex plaque with lipid rich foam cell core and foam cells of the fibrous cap at 63 days. See Figures S14–S16 for additional electron micrographs of Lmod1^WT and Lmod1^SMKO coronary arteries, including an enlarged image of panel M. Lu, lumen.

There was no obvious bias for atheromatous lesions between right, left, or septal coronary artery of Lmod1^SMKO mice and no sexual dimorphism after nine weeks of PCSK9/HFD (Figure S12; Table S5). Using the Stary classification of atherosclerotic lesion types,⁴² a comprehensive and blinded evaluation of >500 coronary sections across 116 mice of equal sex revealed a predominance of Type I and Type II lesions at early time points, with a transition to Type IV and Type V lesions at later stages of the PCSK9/HFD regimen, in Lmod1^SMKO mice (Figure S13; Table S6). 3/161 (1.9%) of Lmod1^WT coronary artery profiles showed very mild Type I lesions with no other advanced lesions recorded, even as late as 203 days of PCSK9/HFD (Figure S13; Table S6). On the other hand, 311/340 (91.5%) of Lmod1^SMKO coronary artery profiles showed lesions of Types I-V (Figure S13; Table S6). Notably, there were no complex Type VI lesions, consistent with the absence of myocardial infarction. Collectively, these findings define a highly penetrant, coronary-restricted, occlusive atherosclerotic phenotype in both male and female Lmod1^SMKO mice, with lesion development evident by eight days of PCSK9/HFD exposure.

Coronary ultrastructure in Lmod1^SMKO mice reveals phenotypically diverse VSMCs

Surprisingly, few comprehensive studies exist on the ultrastructure of mouse coronary arteries.^43–45 The ultrastructure of PCSK9/HFD fed Lmod1^WT coronary arteries was indistinguishable from that of chow-fed controls, with abundant myofilaments, prominent dense bodies, intact myoendothelial junctions, and an absence of intracellular lipid droplets (Figure S14). In contrast, coronary arteries of Lmod1^SMKO mice showed presumptive SMCs migrating through the fenestra of the internal elastic lamina (IEL) into a thickened intima as early as eight days post PCSK9/HFD (Figure 5M; Figure S15). By 16 days, the intima was thicker with medial and intimal SMCs (defined by dense bodies) containing lipid droplets; the intimal SMCs were frequently found at the base of lesions, below presumptive macrophage-derived foam cells (Figure 5N). At later stages of coronary atherosclerosis, plaques were progressively larger with extracellular lipid, cholesterol crystals, and an increasing number of foam cells in the intima and media (Figure 5O; Figure S16). Extensive profiling demonstrated considerable heterogeneity of lipid droplets in SMCs, suggesting active lipid metabolism (Figure S16). Indeed, MSI profiling revealed a pronounced increase in an array of lipid-derived molecules within atheromatous coronary arteries of Lmod1^SMKO mice, including sulfated cholesterol, ceramides, and sphingomyelin species (Figure S17; Table S7). Taken together, these results suggest the presence of metabolically active, migrating coronary SMCs in evolving coronary atheromata of Lmod1^SMKO mice.

Coronary artery SMCs of Lmod1^SMKO mice contribute to atheromatous plaque formation

Previous lineage tracing studies have documented SMC-derived cells in atherosclerotic lesions of the aorta and brachiocephalic artery.^{11–14,46–48} Here, Lmod1^WT and Lmod1^SMKO mice (both carrying the mTmG reporter) underwent the PCSK9/HFD regimen and coronary arteries were analyzed by CIFM for the presence of SMC-derived GFP+ cells in atheromatous plaques. As expected, Lmod1^WT coronaries showed no evidence of disease and all GFP+ SMCs were confined to the media (Figure 6A, 6B). In contrast, SMC-derived GFP+ cells were found in coronary plaques of Lmod1^SMKO mice (Figure 6D, 6E, 6G, 6H). To accurately quantify the percentage of GFP+ plaque cells, we developed a novel immunogold electron microscopy lineage tracing (IEMLT) assay. Extensive testing of IEMLT in control mice showed unambiguous, specific labeling of medial coronary artery SMC membranes (Figure 6C; Figure S18). The IEMLT method demonstrated obvious distinction between GFP+ and GFP− cells (Figure 6F, 6I, 6J; Figure S19A–S19F), with GFP+ SMCs migrating through the IEL as early as six days post-PCSK9/HFD (Figure S19E, S19F). Rigorous quantitation of cells within advanced atheromatous lesions revealed that ~46% of plaque cells were of coronary SMC origin (Figure 6K). Interestingly, 75% of GFP+ cells in the lower core of lesions were foam cells (Figure 6K; Figure S19A–S19C) and a surprising 50% of GFP+ fibrous cap cells also contained lipid droplets (Figure 6I–6K; Figure S19D). These results of mouse coronary atherosclerosis further corroborate an important role for VSMCs in atherogenesis.^49–52

Figure 6. Coronary artery SMC lineage tracing in Lmod1^SMKO mice.

Lmod1^WT (A-C) and Lmod1^SMKO (D-J) coronary arteries from mice carrying the mTmG reporter. Red fluorescence (A, B, D, E, G, H) indicates surrounding cardiomyocytes, endothelial cells, immune cells, and fibroblasts whose mTmG reporter did not undergo recombination (i.e., Itga8-CreER^T2 was inactive in these cells). Conversely, green fluorescence represents medial and plaque cells of SMC origin (i.e., Itga8-CreER^T2 was active in these cells and recombined out the tomato reporter allowing for membrane GFP expression). Note GFP+ cells beginning to populate early atheroma (D, E; see also, Figure S19E–S19F) and comprising a large portion of advanced plaques (G, H). White asterisks indicate necrotic core. IEMLT of (C) an Lmod1^WT coronary artery following PCSK9/HFD; (F) an Lmod1^SMKO coronary artery with two intimal core GFP+ cells (labeled 1 and 2); and (I) an advanced lesion showing several GFP+ cells in the fibrous cap. (J) Higher magnification of red boxed region in panel I shows three fibrous cap cells (labeled 1–3), two of which (1 and 2) are GFP+. (K) Percentage of GFP+ cells in plaques of seven independent Lmod1^SMKO coronaries (red bar) and the percentage of Core (green bar) and Cap (blue bar) derived foam cells (FC) that were GFP+. Replicates represent the percentage of cells from single coronary arteries in seven independent animals. Shown are the means ± SD. Panels A-C, female Lmod1^WT mice treated six weeks with PCSK9/HFD; panels D-F, female Lmod1^SMKO mice treated six weeks with PCSK9/HFD; panels G-J, male Lmod1^SMKO mice treated nine weeks with PCSK9/HFD. Adv, adventitia; CM, cardiomyocyte; EC, endothelial cell; Int, intima; Med, media; RBC, red blood cell. Scale bars are 50 μm for panels A, B, D, E, G, H and as indicated in panels C, F, I, J. See Supplemental Figures S18 and S19 for additional IEMLT images.

VSMCs undergo diverse fate and state transitions during atherogenesis.^50,51,53,54 CIFM showed a very small percentage of coronary plaque-derived SMCs (GFP+) expressing CD68, SPP1, or LGALS3 (Figure S20). Further, despite carefully controlled studies, no DNA synthesizing or apoptotic GFP+ cells were found within coronary plaques at early or late stages of atherogenesis; however, as expected, TUNEL positive (GFP-) cells were found in the necrotic core of some plaques (Figure S21). These results indicate that, within the context of experiments performed here, coronary SMC-derived plaque cells failed to show active DNA synthesis or apoptosis.

Reduced expression of LMOD1 does not confer CAD in mice

Two SNVs within the first intron of human LMOD1 have been associated with vascular disease, including CAD,^9,55–59 and one of them (rs34091558) has been validated in vitro.⁶⁰ We used a two-component CRISPR approach²⁸ to delete a portion of intron 1 in Lmod1 encompassing the orthologous sequence where the human rs34091558 resides (Figure 7A). Positive founders were bred for germline transmission, and mice homozygous for the 11 kilobase deletion were overtly normal despite a ~50% reduction in LMOD1 protein, as measured by Western blotting (Figure 7B) and CIFM (Figure 7C). A similar reduction in LMOD1 protein was observed in Lmod1^SM-Het mice lacking one Lmod1 allele following Itga8-CreER^T2-mediated recombination (Figure 7D, 7E). Interestingly, despite a clear reduction in coronary SMC LMOD1 staining (Figure 7G vs. 7I), Lmod1^SM-Het mice showed little evidence of coronary atherosclerosis after seven weeks of PCSK9/HFD treatment. The minimal ORO staining observed was comparable to that occasionally detected in coronaries of Lmod1^WT mice (Figure 7F vs. 7H). Thus, we infer that a 50% reduction in LMOD1 protein is insufficient to cause CAD in mice, at least under the experimental conditions reported here.

Figure 7. Reduced LMOD1 expression does not confer a CAD phenotype.

(A) UCSC genome browser screenshot of human LMOD1 locus and position of an annotated SNV, rs34091558 (top arrow). The approximate position of sgRNAs used to generate an 11 kb intronic deletion in orthologous mouse sequence is shown (orange horizontal arrows). This deletion removes the conserved, orthologous region in human LMOD1 where the rs34091558 resides and the nearly overlapping conserved SRF-binding consensus CArG box, both of which overlap H3K4Me1 and H3K27Ac chromatin marks (bottom). Two downstream conserved CArG boxes remained intact, indicated by ChIP-seq peaks in human coronary artery SMCs (red vertical arrows) and the “CArGs” track (top). Western blot (B) and CIFM (C) of aortic LMOD1 in control (C57BL6/J) and Lmod1^Int1Δ (homozygous for the intronic deletion) mice. Western blot (D) and quantitation (E) of aortic LMOD1 and ITGA8 in the indicated genotypes (n=3 mice). Statistical analysis was done with one-way ANOVA followed by Tukey’s multiple comparisons test. Adjusted P values are shown. ORO staining (F, H) and CIFM of LMOD1 and ACTA2 (G, I) in coronary arteries from Lmod1^WT (F, G) and Lmod1^SM-Het (H, I) mice following seven weeks of PCSK9/HFD. Scale bars, 50 μm.

The CAD phenotype is lost in mice expressing an actin-nucleation-deficient LMOD1

To date, actin nucleation remains the only known and validated function of LMOD1.^61,62 Amino acid residues 313–600 of human LMOD1, comprising the leucine rich repeat (LRR) domain and the C-terminal WH2 domain-containing extension, account for most of its nucleation activity.^61,62 To begin testing whether loss of this function is involved in the Lmod1^SMKO CAD phenotype, we made a recombinant LMOD1 in which five basic amino acids in the LRR were substituted with five acidic residues; the five amino acid charge-reversal in mouse LMOD1 (K444D, H448D, R455D, R470D, R473D) targets highly conserved residues on the actin-binding surface of the LRR domain (Figure 8A, left). This recombinant protein (LMOD1^ND) was purified in E. coli and found to be expressed equally with LMOD1^WT (Figure 8A, right). LMOD1^ND was anticipated to disrupt LMOD1 nucleation activity, as confirmed by polymerization assays where the nucleation rate of LMOD1^WT was over six-fold higher than that of LMOD1^ND (Figure 8B). Consistent with the accumulation of lipids in coronary VSMCs of Lmod1^SMKO mice, cultured MASMCs lacking LMOD1 exhibited oxLDL uptake (Figure 8C, left). Transducing such cells with lentivirus carrying Lmod1^WT restored LMOD1 expression (Figure 8D) and significantly reduced oxLDL uptake (Figure 8C, middle). Notably, lentiviral transduction with Lmod1^ND similarly reduced oxLDL uptake in Lmod1^SMKO MASMCs (Figure 8C, right), despite a much lower level of LMOD1 expression (Figure 8D). To explore the basis for this attenuated LMOD1 expression, a cycloheximide (CHX) study was performed and the results showed that the half-life of LMOD1^ND was considerably less than LMOD1^WT (Figure S22A, S22B). We then engineered a Cre-inducible mouse with the same five amino acid substitutions in LMOD1 (Figure 8E). The fidelity of this Lmod1^ND mouse model was validated by Sanger sequencing (Figure S22C) and a qRT-PCR assay (Figure 8F). Homozygous Lmod1^ND mice ± Tagln-Cre⁶³ were subjected to the PCSK9/HFD regimen for 10 weeks. Similar to in vitro findings above, there was a pronounced reduction in LMOD1^ND protein compared to LMOD1^WT in aorta (Figure 8G). Remarkably, despite such low LMOD1^ND protein levels, there was negligible ORO staining in coronary arteries after the 10-week atherogenic regimen (Figure 8H). To confirm these findings, the Lmod1^ND mouse was crossed with Itga8-CreER^T2 and subjected to the same atherogenic regimen as above. Itga8-CreER^T2-induced Lmod1^ND mice were also resistant to CAD (Figure S22D) and showed a similar trend in reduced LMOD1 protein expression (Figure S22E). Mice with Tagln-Cre or Itga8-CreER^T2 induced Lmod1^ND showed comparable hypercholesterolemia (Figure S22F). Collectively, these findings suggest that a low level of LMOD1 protein or its very low actin nucleation activity (or both) is sufficient to impede coronary atherogenesis.

Figure 8. LMOD1 actin nucleation and CAD phenotype in Lmod1^SMKO mice.

(A) LRR domain of mouse LMOD1 bound to an actin subunit at the pointed end of the actin filament, showing the mutated amino acids in LMOD1^ND (K444D, H448D, R455D, R470D, and R473D). Model produced based on superimposition of Protein Data Bank codes 4Z79, 5WFN, and 8F8S.^61,93 Also shown at right is an SDS-PAGE gel of E. coli-expressed human LMOD1^WT and LMOD1^ND (K449D, H453D, R460D, R475D, R478D). (B) Time course of polymerization of 2 μM actin (6% pyrene-labeled), measured as the fluorescence increase upon incorporation of pyrene-actin into filaments. Curves represent the average of three independent experiments, color-coded by protein construct and concentration. Shown at right is the mean fluorescence ± SD. (C) Lmod1^SMKO MASMCs incubated with oxLDL in the presence of Lentivirus carrying each indicated construct, with quantitation of oxLDL area/nuclei by Kruskal-Wallis test (mean ± SD shown, n=5 fields of view from a single batch of MASMCs; representative of two independent batches of MASMCs). (D) Western blot of indicated proteins from similar experiments as shown in panel C. (E) Strategy for engineering the inducible Lmod1^ND mouse model. The wild type mouse (top) expresses normal levels of LMOD1 from the minigene placed in first intron which is recombined out upon Cre-mediated excision (bottom), thus enabling expression of the mutated second exon (E2-ND) containing the five amino acid substitutions. Validation studies demonstrate the loss of Lmod1^WT and induction of Lmod1^ND mRNA (F) and protein (G) with Tagln-Cre. Levels of each Lmod1 mRNA quantitated by unpaired (two-tailed) t-test and shown as the mean ± SD (n=3 or 4 mice). (H) CIFM of LMOD1 (Hi, Hiii) and ORO staining of coronary artery (Hii, Hiv) in Lmod1^WT and Lmod1^ND (induced with Tagln-Cre) mice treated for 10 weeks with PCSK9/HFD. n ≥ 5 mice per genotype. Scale bars, 100 μm for CIFM and 20 μm for ORO staining. (I) Graphical abstract.

Discussion

A lethal neonatal visceral myopathy affecting the gastrointestinal tract has stalled efforts to elucidate loss-of-function phenotypes of the CAD risk gene Leiomodin1 in the vessel wall.¹⁸ Recently, whole-body heterozygous Lmod1 knockout mice were shown to have enhanced neointimal formation following acute injury to the carotid artery, providing the first reported evidence of Lmod1 haploinsufficiency.⁶⁴ Here, the lethal visceral myopathy accompanying whole-body loss of both Lmod1 alleles was overcome by breeding the Itga8-CreER^T2 mouse to a newly floxed Lmod1 strain. VSMC-restricted Lmod1 knockout mice (Lmod1^SMKO) displayed a rapidly manifested, fully penetrant, and non-sexually dimorphic CAD phenotype under a PCSK9/HFD regimen. Importantly, a similar CAD phenotype was observed in WD-fed Lmod1^SMKO/Apoe^−/− mice. In sharp contrast, floxed Lmod1 mice carrying the popular Myh11-CreER^T2 driver²⁷ displayed massive intestinal distention and death (probably from sepsis) within a week of Tmx administration. Other SMC Cre drivers, with activity in both vascular and visceral SMC lineages,^21,22 would likely have limited utility here and in other contexts where a targeted gene is of critical importance for intestinal homeostasis. Even in the absence of acute intestinal dysmotility, any gene’s inactivation in both vascular and visceral SMCs may complicate the interpretation of vascular phenotypes given the established role of the gut microbiome in vascular disease.^65,66 Accordingly, when a pronounced vascular phenotype is observed, particularly with pan-SMC Cre drivers, the potential effects of gene loss on intestinal smooth muscle should be recognized and addressed.

MYH11 represents another SMC-specific CAD risk allele,⁹ and global inactivation of Myh11 in mice leads to a lethal visceral myopathy.²⁰ Therefore, accurate assessment of vascular phenotypes following SMC Myh11 loss-of-function will require the use of Itga8-CreER^T2. It will be interesting to determine whether VSMC-restricted loss of Myh11 produces a CAD phenotype. More broadly, we suspect that CAD in the mouse may be underestimated because most studies do not evaluate the coronaries, likely because of the comparatively small number of reports documenting murine CAD. Indeed, coronary atherosclerosis in mice typically requires complex genetic crosses,⁶⁷ superimposed injury,⁶⁸ or prolonged periods of time for lesions to become manifest⁶⁹ (Table S8). Here, using standard histology, lipid staining, coronary lineage tracing, and electron microscopy, coronary VSMCs of Lmod1^SMKO mice were found to populate the base of early and advanced atheromas, with many containing lipid droplets. We also found lipid-containing VSMCs in their native medial milieu and within fibrous caps. These findings align well with older literature using ultrastructural features of lipid droplet-containing VSMCs, such as the presence of a basement membrane and peripherally-oriented dense bodies,^70–72 both of which we observed in medial SMC foam cells of murine coronary arteries (see Figure S5K).

Surprisingly, despite examining both early and late stages of disease, using different assays, we were unable to detect active DNA synthesis in VSMCs of Lmod1^SMKO mice. Moreover, not a single mitotic VSMC was ever observed in our ultrastructural analysis of dozens of vessels and hundreds of individual VSMCs over the course of several years. Early human experiments showed a very low rate of VSMC proliferation in coronary plaques, although the study was conducted primarily in advanced lesions.⁷³ Lineage tracing studies have demonstrated clonal expansion of VSMCs in various disease processes, suggesting some SMCs undergo replication to generate clonal patches in intimal plaques.^{12,48,74–76} Misra et al showed phosphohistone H3 (a cell mitosis marker) in fibrous cap cells of SMC origin during murine atherogenesis as well as human ACTA2 positive fibrous cap cells.⁷⁶ More recently, Pan et al administered EdU for four weeks and demonstrated an accumulation of ~70% of plaque-derived SMCs that underwent DNA synthesis at some point between 12 and 16 weeks of athero- progression.⁷⁷ However, convincing evidence for SMC DNA replication within a clonal patch of an atheroma remains incompletely characterized, possibly reflecting transient bursts of limited DNA synthesis during poorly defined stages of disease progression. Here, using an immunogold lineage tracing assay, ~46% of plaque cells were found to be of coronary SMC origin. The early emergence of these VSMCs in nascent coronary fatty streaks, as early as six days following the atherogenic regimen (see Figure S19E, S19F), suggests an initial migratory wave rather than proliferation of a small number of SMCs. This interpretation is supported by several examples here and elsewhere⁷⁸ of medial VSMCs moving through the fenestra of the IEL. Such “transfenestral migration” was elegantly demonstrated in an endothelial cell-specific knockout of the Mef2c transcription factor.⁷⁹ Further work is needed in the Lmod1^SMKO model to determine whether clonal expansion occurs in mouse coronary arteries and to define the timing and location of VSMC DNA synthesis within the arterial wall, as migrating VSMCs would need to be replaced within the medial layer.

Coronary hemorrhage, plaque rupture or EC erosion, and occlusive thrombosis all contribute to coronary symptomatology and death. Mice lacking Lmod1 in coronary VSMCs exhibited no evidence of these characteristic features of complex plaques, and myocardial infarction was not observed during the study period, although most experiments were limited to only a few months. Therefore, the Lmod1^SMKO model, like most coronary atherosclerosis mouse models (Table S8), currently does not recapitulate the terminal events of CAD found in humans. Moreover, there was negligible atherosclerosis in cerebral vessels of the brain and no evidence of stroke or paralysis. Nevertheless, ~35% of Lmod1^SMKO mice died during the course of these studies. Other than mild neutropenia and lymphocytosis, the latter of which might be expected in this model, there were no indications of systemic toxicity. Although Lmod1^SMKO mice were hypotensive, it is currently unclear how this physiological change could lead to death. On the other hand, the CAD phenotype in Lmod1^SMKO mice was pervasive and included the occlusion of smaller coronary vessels (see Figure S9D). We surmise that Lmod1^SMKO mortality arises from ischemic cardiomyopathy and arrhythmia secondary to the occlusive CAD. Future work in this model is needed including extensive cardiac functional studies, coronary flow and reactivity, as well as chronic instrumentation to monitor cardiac electrical activity.

As shown here and by others,^16,32,36 LMOD1 protein levels are attenuated in human vascular diseases such as atherosclerosis. Further, LMOD1 is a known CAD risk allele, harboring two SNVs located in the 45 kilobase first intron.^10,55,56,59 One of these SNVs (rs34091558) was recently demonstrated to disrupt a nearly 100% conserved FOXO3 binding site, leading to reduced LMOD1 expression in cultured human coronary artery SMCs.⁶⁰ Interestingly, rs34091558 is located three helical turns away from a highly conserved serum response factor (SRF) binding site known as a CArG box (see Figure 7A). SRF-binding CArG boxes are found in most SMC regulatory regions and help direct the normal program of SMC differentiation.⁸⁰ In this report, we generated a mouse model harboring an 11-kilobase intronic deletion encompassing the SRF-binding CArG element and the adjacent orthologous SNV region. This deletion reduced baseline LMOD1 protein levels to those seen in heterozygous Lmod1^SMKO mice. However, despite ~50% reduction in the level of LMOD1 protein, heterozygous Lmod1^SMKO mice failed to develop CAD. These results support the absence of Lmod1 haploinsufficiency for the CAD phenotype. It will be interesting to impose an additional risk factor for CAD, as was done recently in another mouse model of coronary atherosclerosis,⁸¹ to determine whether augmented stress elicits CAD in heterozygous Lmod1^SMKO mice. Further, it remains to be determined whether modeling the rs34091558 variant in humanized mice lowers LMOD1 expression as reported in cultured human SMCs.⁶⁰

How might the conditional knockout of Lmod1, a SMC-restricted cytoskeletal gene, trigger the rapid onset of CAD under hypercholesterolemic conditions? One possibility is a mechanism involving loss of LMOD1’s only validated function – actin nucleation.⁶¹ To address this possibility, we generated a mutant LMOD1 defective for actin nucleation (LMOD1^ND). Both LMOD1^ND and LMOD1^WT fully reversed the lipid accumulation in cultured Lmod1^SMKO VSMCs. Further, inducible expression of an actin-nucleation-deficient LMOD1 mutant conferred complete protection against the lipid-laden, occlusive CAD phenotype after 10 weeks of PCSK9/HFD. These data suggest that expression of LMOD1^ND, with impaired actin-nucleation activity, is sufficient to prevent coronary atherosclerosis and that loss of an additional unknown functional domain in LMOD1 promotes its development. Interestingly, the resistance to lipid uptake in vitro and coronary atherosclerosis in vivo with LMOD1^ND occurs despite a pronounced decrease in expression of LMOD1^ND protein. CHX studies indicate that reduced LMOD1^ND protein expression is due to a shorter half-life than that seen with LMOD1^WT. In this context, a clinical study found low LMOD1 protein in a compound heterozygous patient with mutations in the vicinity of the five amino acid substitutions generated here for the Lmod1^ND mouse.⁸² Further, compound heterozygous mice with low-level LMOD1 failed to exhibit a visceral myopathy seen in homozygous null mice.¹⁸ Interestingly, conditional knockout of Lmod2 (a cardiac-restricted Leiomodin paralog) showed that as little as 15% of LMOD2 protein was sufficient to prevent a lethal cardiac failure phenotype.⁸³ Thus, it appears that levels of LMOD1 (and LMOD2) protein must fall below a critical threshold before pathology ensues. Of course, we cannot formally exclude residual actin nucleation activity contributing to the rescue of lipid accumulation in cultured VSMCs and the coronary atherosclerosis phenotype following LMOD1 loss. Recent evidence suggests that in addition to actin nucleation, LMOD2 mediates elongation of actin filaments.⁸⁴ Given the substantial sequence conservation between LMOD1 and LMOD2, it is reasonable to propose that LMOD1 may also participate in filament elongation, and that loss of this function mediates the coronary phenotype. Ongoing studies will elucidate the functions of LMOD1 that maintain coronary homeostasis, thereby providing new insight into pathways underlying coronary atherosclerosis.

There are some limitations of this study that should be noted. First, most experiments were restricted to a short time interval of only three or four months. Longer term studies will provide insight into whether lesions evolve into complex plaques, precipitating myocardial infarction and/or heart failure. Second, the underlying cause of death in Lmod1^SMKO mice remains to be determined. Third, the question of SMC clonality remains unresolved, particularly with respect to the time and location of SMC expansion in coronary arteries. Fourth, the Lmod1 intronic deletion encompassed the orthologous region where the GWAS risk variant resides in human LMOD1, an adjacent CArG box, and additional downstream sequences. Thus, this 11-kb deletion model does not reflect what may be seen with a humanized mouse carrying only the risk variant as described in vitro.⁶⁰ Finally, given that atherogenesis is initiated by lipid retention and LDL modifications,² how loss of LMOD1 promotes early lipid and SMC infiltration into the intima is currently unknown and will require in depth cellular and molecular studies.

In conclusion, the findings reported here represent the first SMC-restricted risk allele for CAD to be genetically inactivated and studied under an atherogenic regimen. With the exception of Nos3,⁸⁵ Lmod1 is the only non-lipid related CAD risk gene whose genetic knockout yields a CAD phenotype in mice. Interestingly, a recent in vitro report showed that cholesterol-loading of human SMCs reduced LMOD1, and knockdown of LMOD1 elicited a cholesterol overload-induced gene signature. The authors proposed the idea of elevating LMOD1 as a means of reducing CAD risk.⁸⁶ In this context, the rapid manifestation of coronary atherosclerosis in Lmod1^SMKO mice, while not modeling the more indolent nature of the human disease, offers an opportunity to rapidly test therapeutics that mitigate disease progression or promote plaque stability.

Supplementary Material

Checklist

Expanded Methods

Figures S1–S22

Tables S1–S8

Clinical Perspective.

What is new?

• Loss of Leiomodin1 (Lmod1) in vascular smooth muscle cells (VSMCs) causes diffuse and occlusive coronary atherosclerosis in mice, with little detectable disease in other vascular beds.

• A novel immunogold lineage tracing assay shows VSMC migration into the intima as early as six days following an atherogenic regimen, and quantitative studies demonstrate that 46% of coronary plaque cells are of VSMC origin.

• The coronary phenotype appears to be independent of the loss in LMOD1 actin nucleation activity.

What are the clinical implications?

• LMOD1, an annotated smooth muscle cell-restricted risk gene for human coronary artery disease (CAD), offers new insight into the role of VSMCs in coronary atherogenesis.

• The rapidly manifesting CAD phenotype in Lmod1^SMKO mice enables expedited testing of novel therapeutics to mitigate disease progression.

• Further interrogation of single nucleotide variants in the LMOD1 locus may improve understanding of LMOD1 pathobiology and its contribution to CAD risk.

Nonstandard abbreviations and acronyms

AAV

adeno-associated virus

AngII

angiotensin II

Apoe

apolipoprotein E

ARRIVE

animal research: reporting of in vivo experiments

BSA

bovine serum albumin

CIFM

confocal immunofluorescence microscopy

Cre

causes recombination

CRISPR

clustered regularly interspaced short palindromic repeats

CHX

cycloheximide

EC

endothelial cell

EdU

5-ethynyl-2′-deoxyuridine

ER^T2

estrogen receptor ligand-binding domain^{Tamoxifen, 2nd generation}

FFPE

formalin-fixed, paraffin-embedded

GFP

green fluorescence protein

GWAS

genome wide association studies

HFD

high fat diet

IEL

internal elastic lamina

IEMLT

immunogold electron microscopy lineage tracing

Itga8

integrin alpha 8

LDL

low-density lipoprotein

LDLR

low-density lipoprotein receptor

LPS

lipopolysaccharide

LMOD1

leiomodin1

LMOD^ND

leiomodin1 actin-nucleation-deficient

LMOD1^SMKO

leiomodin1^{smooth muscle knockout}

LMOD1^WT

leiomodin1 wild type

LoxP

locus of X-over P1

MASMC

mouse aortic smooth muscle cell(s)

MSI

mass spectrometry imaging

mTmG

membrane tomato, membrane green fluorescent protein

Myh11

myosin heavy chain 11

ORO

oil-red-o

ox-LDL

oxidized low-density lipoprotein

PCSK9

proprotein convertase subtilisin/kexin type 9

PFA

paraformaldehyde

RT

room temperature

scATAC-seq

single cell assay for transposase-accessible chromatin using sequencing

scRNA-seq

single cell RNA sequencing

SRF

serum response factor

SNV

single nucleotide variant

TEM

transmission electron microscopy

Tmx

tamoxifen

TUNEL

terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling

UMAP

uniform manifold approximation and projection

VSMC

vascular smooth muscle cell(s)

WD

western diet

WT

wild type