CARMN is an Evolutionarily Conserved Smooth Muscle Cell-specific LncRNA that Maintains Contractile Phenotype by Binding Myocardin

Abstract

Background:

Vascular homeostasis is maintained by the differentiated phenotype of vascular smooth muscle cells (VSMCs). The landscape of protein coding genes comprising the transcriptome of differentiated VSMCs has been intensively investigated but many gaps remain including the emerging roles of non-coding genes.

Methods:

We re-analyzed large-scale, publicly available bulk and scRNA-seq datasets from multiple tissues and cell types to identify VSMC-enriched lncRNAs. The in vivo expression pattern of a novel SMC expressed lncRNA, Carmn (CARdiac Mesoderm Enhancer-associated Non-coding RNA) was investigated using a novel Carmn GFP knock-in reporter mouse model. Bioinformatics and qRT-PCR analysis were employed to assess CARMN expression changes during VSMC phenotypic modulation in human and murine vascular disease models. In vitro, functional assays were performed by knocking down CARMN with antisense oligonucleotides and over-expressing Carmn by adenovirus in human coronary artery SMCs. Carotid artery injury was performed in SMC-specific Carmn knockout mice to assess neointima formation and the therapeutic potential of reversing CARMN loss was tested in a rat carotid artery balloon injury model. The molecular mechanisms underlying CARMN function were investigated using RNA pull-down, RNA immunoprecipitation and luciferase reporter assays.

Results:

We identified CARMN, which was initially annotated as the host gene of the MIR143/145 cluster and recently reported to play a role in cardiac differentiation, as a highly abundant and conserved, SMC-specific lncRNA. Analysis of the Carmn GFP knock-in mouse model confirmed that Carmn is transiently expressed in embryonic cardiomyocytes and thereafter becomes restricted to SMCs. We also found that Carmn is transcribed independently of Mir143/145. CARMN expression is dramatically decreased by vascular disease in humans and murine models and regulates the contractile phenotype of VSMCs in vitro. In vivo, SMC-specific deletion of Carmn significantly exacerbated, while overexpression of Carmn markedly attenuated, injury-induced neointima formation in mouse and rat, respectively. Mechanistically, we found that Carmn physically binds to the key transcriptional cofactor myocardin, facilitating its activity and thereby maintaining the contractile phenotype of VSMCs.

Conclusions:

CARMN is an evolutionarily conserved SMC-specific lncRNA with a previously unappreciated role in maintaining the contractile phenotype of VSMCs and is the first non-coding RNA discovered to interact with myocardin.

INTRODUCTION

Smooth muscle cells (SMCs) are the major contractile component of blood vessels and most hollow organs, such as bladder, intestine, and colon. Fully differentiated SMCs are characterized by the presence of a unique repertoire of contractile proteins.^1–8 In response to vascular injuries, VSMCs switch from a contractile to a synthetic phenotype, characterized by increased migration and proliferation as well as reduced expression of contractile pa SMC-specific lncRNA that promotes the controteins.⁹ It has been reported that serum response factor (SRF), the transcription factor binding to CArG elements, plays a central role in regulating SMC phenotypes by interacting with a variety of cofactors.¹⁰ As one of SRF-associated proteins, myocardin (MYOCD) is specifically expressed in cardiomyocytes and SMCs and is a potent activator of SM-specific genes.^11–13 However, the underlying molecular mechanism for regulation of SMC phenotypic switch and SRF-MYOCD interaction is not fully understood.

Although initially regarded as “junk DNA”, >80% of human and mouse genomes is transcribed into a variety of classes of RNA.¹⁴ For example, long non-coding RNAs (lncRNAs), defined as transcripts larger than 200 nucleotides and with no apparent protein-coding potential, outnumber all protein coding genes and have been shown to play critical roles in many physiological and pathological conditions.¹⁵ Several lncRNAs have been implicated in SMC biology.^16–26 However, many of these lncRNAs are not conserved between species, not exclusively expressed in SMCs, and their expression has not been carefully tracked by in vivo reporter systems. These limitations collectively hinder our understanding of the importance of lncRNAs in SMCs and therefore a better exploration of lncRNA landscape in SMCs is required.

In this study, we used an in silico approach to probe unbiased proprietary and diverse publicly available bulk RNA-seq and scRNA-seq datasets to search for SMC-specific lncRNAs. This search identified CARMN (CARdiac Mesoderm Enhancer-associated Non-coding RNA) as a highly abundant, highly conserved SMC-specific lncRNA. CARMN was recently reported to play roles in cardiac differentiation,^{27, 28} and was initially annotated as a host lncRNA for the MIR143/145 cluster, the best characterized microRNAs in regulating SMC differentiation and phenotypic modulation.^29–34 In this study, we confirmed the expression specificity of Carmn using a novel GFP knock-in (KI) reporter mouse model and discovered that CARMN is down-regulated in various vascular diseases. We further found that Carmn is critical for maintaining VSMC contractile phenotype both in vitro and in vivo by directly binding to MYOCD and potentiating MYOCD function. These findings collectively suggest that CARMN is a key regulator of VSMC phenotype and represents a potential therapeutic target for treatment of SMC-related proliferative diseases.

METHODS

Detailed methods are provided in the Supplemental Material. The data, analytic methods including R scripts used for analyzing transcriptome datasets, and related materials will be made available from the corresponding author upon reasonable request. The amended mouse Carmn V2 and V3 cDNA sequences have been deposited into GenBank under accession number MN904529 and MN904530, respectively. Bulk RNA-seq, scRNA-seq, ChIP-seq and ATAC-seq datasets used in this study are fully described in the Supplemental Methods. All animal studies have been approved by the Institutional Animal Care and Use Committee of Augusta University and conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. There are no human subjects in this study.

Statistical Analysis

GraphPad Prism (Version 9.2.0) was used for the statistical analysis. All data are expressed as mean ± SEM of at least three independent experiments. Tests used for statistical significance evaluations are specified in figure legends. An unpaired two tailed t-test was used for data involving two groups only. Two-way analysis of variance (ANOVA) or two-way repeated measures ANOVA followed by a Bonferroni correction was used for data involving more than two groups. Values of P ≤ 0.05 were considered statistically significant, except for analysis of bulk RNA-seq, scRNA-seq, microarray and ATAC-seq data which used FDR (false discovery rate)-adjusted P < 0.05 as the threshold for statistical significance.

RESULTS

CARMN is a highly abundant SMC-specific lncRNA in human

To identify SMC-enriched lncRNAs in human, we analyzed 2 independent bulk RNA-seq datasets generated from different human cell types. This unbiased analysis revealed CARMN is the only VSMC-enriched lncRNA identified from both datasets (Figure 1A and B, Table I in the Supplement). Furthermore, CARMN locus in human VSMCs is enriched with active histone mark H3K27ac and lacks repressive mark H3K27me3, in contrast to human umbilical vein endothelial cells (Figure 1C). Examination of two additional promoter-related histone modifications, H3K4me2 and H3K4me3, identified two potentially independent promoters upstream of MIR143/145 and CARMN in SMCs (Figure I A in the Supplement). CARMN is mostly expressed in vascular tissues, followed by SMC-enriched gastrointestinal (GI) tissues as revealed by GTEx database (Figure 1D). Strikingly, CARMN is consistently within the top 15 most abundant lncRNAs expressed in different human arteries and is the most abundant lncRNA among those with features of evolutionary conservation and artery-specific expression pattern. In contrast, the other previously reported lncRNAs such as MALAT1 and NEAT1 lack specific expression for SMC lineages, or are low level non-conserved lncRNAs (Figure 1E).^{16–24, 26} Re-analyzing a scRNA-seq dataset generated from coronary arteries of four human cardiac transplant recipients (GSE131778) ³⁵ suggested CARMN expression is enriched in SMCs and pericytes which also express many SMC markers,³⁶ and is sporadically distributed in some fibroblast cells (Figure 1F and Figure I B–C in the Supplement). The expression pattern of CARMN is consistent with the known SM-specific markers such as MYH11, ACTA2, CNN1 and TAGLN (Figure 1F and Figure I D in the Supplement). This analysis also revealed that CARMN was down-regulated in phenotypically modulated SMCs compared to normal VSMCs (Figure 1F). Taken together, these unbiased large-scale bioinformatics analyses demonstrate that CARMN is an abundantly and specifically expressed lncRNA in SMCs and in SM-enriched tissues in human.

Figure 1. Transcriptome analysis reveals CARMN is a highly abundant SMC-specific lncRNA in human.

(A) Heatmap showing 15 VSMC-enriched lncRNAs identified by the publicly available human RNA-seq dataset (GSE100242). EPC: endothelial progenitor cells; CHCT: chondrocytes; HUAEC: human aortic endothelial cells; HUVEC: human umbilical vein endothelial cells; LCL: lymphoblastoid cell line. FPKM: Fragments Per Kilobase of transcript per Million mapped reads. (B) Heatmap showing 4 VSMC-enriched lncRNAs in 22 primary human cell types retrieved from the ENCODE database. A: artery; V: Vein; ECs: endothelial cells; Ske. M.: Skeletal Muscle. (C) ChIP-seq tracks of histone 3 modifications at the human CARMN locus in aortic VSMC vs HUVEC as revealed by the ENCODE project. Y-axis scale: 20 for H3K27ac and 5 for H3K27me3. (D) CARMN expression in human tissues was revealed by the GTEx database. Th: thoracic aorta; Co: coronary artery; Ti: tibial artery; Ce: cerebellum; Co: cortex; Hi.: hippocampus; At: atrium; LV: left ventricle. TPM: Transcripts Per kilobase Million. (E) The lncRNA landscape in human thoracic aorta (left), coronary artery (middle) and tibial artery (right) revealed by bulk RNA-seq from GTEx database. Annotated dots indicate lncRNAs which have been investigated in VSMCs. (F) t-SNE (t-distributed Stochastic Neighbor Embedding) visualization of cell types isolated from the right coronary artery of four human patients (left), and the expression of CARMN (middle) and MYH11 (right) revealed by the scRNA-seq analysis (GSE131778). The color scale on the right indicates the expression levels of both CARMN and MYH11. Fibro: fibroblast; EC: endothelial cell; PC: pericyte cell; Macro: macrophage; NK: natural killer cell; Pla: plasma cell.

Carmn is a highly conserved SMC-specific lncRNA in mouse

To identify lncRNAs enriched in mouse aorta, we re-analyzed bulk RNA-seq of adult mouse aortic tissues ³⁷ and 5 other non-SM tissues.³⁸ This analysis revealed that Carmn is the most abundant aorta-enriched lncRNA (Figure 2A) and stands out from previously reported SMC-related lncRNAs ^{16, 18, 21, 22, 24–26} as having high abundance, evolutionary conservation, and a tissue-specific expression pattern in mouse aorta (Figure 2B). Subsequent qRT-PCR analysis validated the preferential expression of Carmn in thoracic aorta in mice (Figure 2C). Analysis of scRNA-seq data of normal adult mouse thoracic aorta ³⁹ showed that Carmn is selectively expressed in aortic SMCs (Figure 2D and Figure II A–B in the Supplement), which is further confirmed by an independent scRNA-seq dataset of mouse aorta (GSE117963) (Figure II C–E in the Supplement).⁴⁰

Figure 2. CARMN is a SMC-specific lncRNA in mouse.

(A) Heatmap showing the aorta-enriched lncRNAs identified by RNA-seq data of different mouse tissues. (B) The lncRNA landscape in mouse thoracic aorta revealed by RNA-seq. Annotated dots indicate lncRNAs which have been implicated in VSMCs. (C) Carmn expression was assessed in different mouse tissues by qRT-PCR. The expression of Carmn in mouse thoracic (T.) aorta was set to 1 (dashed line). Ske.: skeletal. N=6. (D) t-SNE visualization of cell types present in normal mouse thoracic aorta (left) and the expression of Carmn (middle) and Myh11 (right) revealed by scRNA-seq (Broad Institute Single Cell Portal). Mono: monocyte. (E) Carmn transcripts (V1–3) were identified by RACE (middle panel) in comparison with the annotated Carmn variants in UCSC genome browser (upper panel). The Sashimi plot is to visualize the splice junctions of Carmn in mouse aorta (bottom panel; N=3; S1–3). (F) qRT-PCR validated that V2 is the most abundant transcript in mouse thoracic aorta. N=6. (G) In vitro transcription and translation of sense or antisense mouse Carmn transcript V2. Plasmid encoding luciferase is used as a positive control and the reaction with no DNA template is used as an additional negative control. (H) Determination of rat Carmn ortholog. The mouse Carmn transcript V2 sequence was used as a query sequence to perform BLAST against rat genome. A putative rat Carmn ortholog positionally conserved to miR143/145 locus, was obtained (mouse sequence in red). A pair of primers (F: forward; R: reverse) specifically targeting two different putative exons of rat Carmn gene were designed and used for PCR amplification with rat aorta cDNA as template. The obtained PCR product was subjected to Sanger sequencing for mapping back to rat genome by BLAST (rat sequence, green). The conservation tracks of 20 vertebrates in the putative rat Carmn gene locus are shown at the bottom panel. (I) Gel picture of PCR product showing that PCR with rat specific primers indicated in “H” (F and R) specifically generated a product from rat thoracic (T.) aorta cDNA, but not from mouse or water. Primers for both rat and mouse Acta2 were used as an internal control for PCR. Water was used as a negative control.

RACE (Rapid Amplification of cDNA Ends) assay using adult mouse aorta cDNA failed to detect the isoform V1 annotated in UCSC database that encompasses Mir143 and Mir145, while revealed the presence of amended V2 and V3 transcripts (Figure 2E, top panel). Our V2 (1228 bp) transcript had an extended exon 4 (GenBank accession number: MN904529), and V3 (759 bp) had a shorter exon 3 (GenBank accession number: MN904530) relative to annotated versions. None of them contain Mir143 or Mir145 (Figure 2E, middle panel). Consistently, bulk RNA-seq of mouse aorta ³⁷ revealed splicing junctions covering V2 and V3, but not exon 4 and exon 5 that are unique to the annotated transcript V1 (Figure 2E, bottom panel). qRT-PCR data using primers specifically targeting different isoforms (Figure II F in the Supplement) suggested that V2 transcript is the most abundant one in mouse aorta (Figure 2F).

In silico prediction and in vitro transcription and translation assay demonstrated that the novel transcripts V2 and V3 are bona fide lncRNA (Figure II G in the Supplement and Figure 2G). Interestingly, a homologous lncRNA of Carmn is present in pig (LOC100514340) that is independently annotated from MIR143/145 (Figure II H in the Supplement). Furthermore, examination of zebrafish genome (ENSDART00000194758) shows a predicted transcript of Carmn based on expressed sequence tag (EST) distancing from the microRNA cluster (Figure II I in the Supplement). Although there is no Carmn gene annotated in rat genome, we cloned a portion of rat Carmn cDNA sequence that is positionally conserved to Mir143/145 by using primers located within the homologous region of mouse (Figure 2H and I). Taken together, we discovered that Carmn is a highly conserved SMC-specific lncRNA and validated it is transcribed independently of Mir143/145 cluster at least in mouse aorta.

GFP KI reporter mouse model confirms the SMC-specific expression of Carmn in vivo

To visualize the expression of Carmn in vivo, we generated a KI mouse model by insertion with a promoterless, reversed splicing acceptor (SA)-membrane bound GFP gene trap cassette, which is flanked by two pairs of oppositely orientated Lox2272 sites ⁴¹ and LoxP sites into intron 2 (referred to as conditional, PFG) (Figure 3A). Upon Cre-mediated deletion of exon 2 and inversion of the reversed GFP cassette (PFG to GFP), the inserted SA will prematurely terminate Carmn expression by splicing of exon 1 to the GFP cassette, while turning on GFP expression under control of the endogenous Carmn promoter (referred to as GFP). GFP expression should therefore faithfully display the endogenous Carmn expression in vivo while at the same time disrupting Carmn expression. We first crossed the female Carmn^PFG/WT (PW) mice with male mice expressing ubiquitous CMV-Cre to invert the GFP cassette in all mouse tissues (Figure 3B). Genotyping results confirmed Cre-mediated recombination within the Carmn locus only in the GW heterozygous mouse (Figure 3C). Western blotting indicated GFP expression is restricted to the aorta, while absent in heart, skeletal muscle and brain of GW heterozygous mice (Figure 3D). Consistently, direct visualization of GFP fluorescence revealed specific expression in aorta and coronary arteries (Figure 3E), as well as in other hollow organs such as bladder, gallbladder, intestine, and stomach, and in regions of SMCs in brain and quadriceps muscles, bronchus of lung, skeletal muscle of GW heterozygous mice (Figure III in the Supplement). Data from co-immunostaining demonstrated that GFP is colocalized with SMC marker ACTA2 in medial layer, but not in endothelial or adventitia cells of GW mouse aorta (Figure 3F). RNA FISH (Fluorescent in situ hybridization) assays further revealed that Carmn expression is exclusively confined to SMC nuclei of mouse aorta (Figure 3G) and cultured mouse VSMCs transduced with Carmn adenovirus (Figure 3H). Collectively, these results provide direct in vivo evidences showing that Carmn is specifically expressed in the nuclei of SMCs.

Figure 3. Visualization of SMC-specific expression of Carmn in vivo.

(A) Strategy used to generate the Carmn GFP KI mouse model showing the inverted GFP KI cassette in the Carmn gene locus. SA: splicing acceptor site. (B) The breeding strategy to generate global Carmn GFP KI mice. (C) PCR analysis of genomic DNA showing recombination of the Carmn gene in tail DNA extracted from WW, PW and GW mice. (D) Western blot showing GFP expression specifically in thoracic aorta (T) of GW mice. Ske: skeletal. (E) Heart and aorta were dissected from adult WW, PW and GW mice and photographed under bright field or GFP channel. T. aorta: thoracic aorta. Scale bar: 5 mm and 1 mm in upper and bottom panel, respectively. The arrowheads point to the representative coronary arteries in the GW mouse heart. (F) Immuno-staining was performed to examine the specific cellular localization of GFP in thoracic aorta dissected from WW, PW and GW mice. L: lumen. Ad: adventitial layer. Scale bar: 10 μm. (G) RNA-FISH was carried out to visualize Carmn (red) in adult mouse thoracic aorta. The sections hybridized with hybridization buffer only (left), human-specific antisense (AS) SENCR probe (middle) were used as negative controls. Cell nuclei were counter stained with DAPI (blue). Dashed lines indicate external or internal elastic lamina. The boxed areas (a and b) are magnified to the right. Arrows point to endothelial cells and arrowheads indicate Carmn signal, respectively. L: lumen; Ad: adventitial layer. Scale bar: 10 μm. (H) Carmn subcellular localization in mouse VSMCs transduced with Carmn adenovirus was revealed by RNA-FISH using Carmn AS probe (red). Carmn sense probe (S) was used as negative control. Cell nuclei were stained with DAPI (blue). The boxed area is magnified below. Arrowheads indicate Carmn signal. Scale bar: 20 μm. (I) Visualization of GFP expression in E16.5 GW embryos showing that GFP is specifically expressed in arteriolar SMCs in right ventricle (RV) and medial SMCs (arrowheads) of left common carotid artery (LCA) but not in cardiomyocytes, endothelium (arrows) or adventitial (Ad.) cells. The boxed area is magnified below. RA: right atrium; L: lumen; Ad: adventitial layer. Scale bar: 20 μm.

Carmn is transiently expressed in cardiomyocytes during embryogenesis

Because CARMN was previously reported to play roles in regulating cardiac specification,^{27, 28} we hypothesized that it may be transiently expressed in cardiomyocytes during development. Re-analysis of public scRNA-seq datasets showed that Carmn expression is present in 8.6% of cardiomyocytes at E10.5 ⁴² (Figure IV A in the Supplement) and is absent in cardiomyocytes but confined to SMCs/pericytes in adult mouse heart (Figure IV B in the Supplement).^{43, 44} Further IF staining analysis for GFP on sections of GW embryos revealed that Carmn expression is primarily co-localized with SMC/early cardiac marker ACTA2 in dorsal aorta or left carotid artery while slightly overlapped with cardiac marker TNNT2 in both left and right ventricles at E11.5 and E13.5 (Figure IV C–F in the Supplement). The percentage of cardiomyocytes expressing Carmn is slightly higher in left ventricle than that in right ventricle and exhibits a declining trend with embryonic development from E11.5 (7.1% in left ventricle and 4.9% in right ventricle) to E13.5 (4.7% in left ventricle and 3.8% in right ventricle) in both ventricles (Figure IV C–F). Strikingly, at E16.5, Carmn expression is restricted to SMCs of carotid artery or in the arteriolar SMCs of the heart, but not in endothelium or cardiomyocytes (Figure 3I and Figure IV G in the Supplement). Similarly, CARMN expression is observed to some extent in cardiomyocytes in human developmental 6.5–7 post conception weeks heart (Figure V A in the Supplement),⁴⁵ while restricted to SMCs and pericytes in adult human heart (Figure V B in the Supplement).⁴⁶ Taken together, these results suggest that CARMN is transiently expressed in cardiomyocytes during early embryonic development but becomes confined to SMCs as development progresses in both mouse and human.

CARMN is downregulated in vascular diseases

Quantification analysis of scRNA-seq data of human diseased coronary artery (Figure 1F) showed that both the percentage of CARMN positive cells and CARMN expression level are lower in modulated SMCs compared to normal SMCs (Figure 4A and B). Targeted re-analysis of published transcriptomic datasets revealed that CARMN is significantly down-regulated in human atherosclerotic arteries and cerebral arteries with aneurysms (Figure 4C and D).^{47, 48} Although not statistically significant, the intensity of multiple ATAC-seq peaks that represent chromatin accessibility of the CARMN gene locus, similar to the MYH11 gene, is remarkably decreased in human atherosclerotic coronary arteries compared to normal coronary arteries (Figure 4E and Figure VI A–C in the Supplement).⁴⁹ To extend these findings to rodent vascular disease models, we re-analyzed scRNA-seq data (GSE131776) of ascending aorta from Myh11-CreER^T2 lineage tracing mice on the ApoE^−/− background at multiple time points, including baseline before a high-fat diet (HFD), as well as after 8 and 16 weeks of HFD feeding.³⁵ This analysis revealed the majority of cells expressing Carmn (>80%) are derived from lineage traced SMCs (Figure 4F and Figure VII A–C in the Supplement) and confirmed Carmn is predominantly expressed in SMCs and pericytes while minimally expressed in fibroblast cells (Figure 4F and Figure VII D in the Supplement). Compared to normal SMCs, phenotypically modulated SMCs exhibit a marked reduction in the percentage of Carmn positive cells and Carmn expression levels over the course of disease progression that reaches statistical significance at 16 weeks after HFD (Figure 4G), which parallels the SM marker Myh11 but is opposite to the Ly6a (aka Sca-1) that has been reported to be up-regulated during SMC phenotypic switching ^{35, 40} (Figure VII E in the Supplement). About 4.6% of modulated SMCs are SMC-lineage negative (Figure VII F in the Supplement), which likely reflects cells originating from non-SMCs such as adventitia/media progenitor cells and endothelial cells which undergo myogenic differentiation to become SMC-like cells during atherogenesis.^{50, 51} The percentage of Carmn positive cells and Carmn expression levels are statistically lower in both SMC lineage positive and negative cells as compared to normal contractile SMCs at 16 weeks after HFD (Figure VII G in the Supplement), suggesting that Carmn is down-regulated in phenotypically modulated SMCs, regardless of their origin. It has been reported that a rare population of medial MYH11 positive SMCs expresses Ly6a and progressively downregulates contractile SMC genes and upregulates genes associated with SMC responses to inflammation and growth factors.⁴⁰ Consistently, our scRNA-seq analysis revealed that some normal SMCs express Ly6a (Figure VII E in the Supplement) and some of these medial Myh11⁺/Ly6a⁺ cells express Carmn (Figure VII E and H in the Supplement), which accounts for 0.05% of the total MYH11 positive contractile SMCs at baseline with an increase to 0.75% at 16 weeks after HFD (Figure VII H in the Supplement). Since it has been postulated that medial Ly6a positive SMCs can contribute to atherosclerotic lesion cells,^{40, 52} it will be interesting to investigate whether Carmn regulates this process in the future. The enrichment of Carmn in SMCs and downregulation in modulated SMCs (statistically significant at 16 and 26 weeks after HFD) was further confirmed in an additional scRNA-seq dataset of lineage-traced aortic SMCs from the atherosclerotic Ldlr KO mouse model (GSE155513) (Figure VIII in the Supplement). ⁵³ In an additional, independent scRNA-seq dataset of lineage-traced mouse aortic SMCs (GSE117963), a remarkable decrease in Carmn expression was observed in phenotypically modulated SMCs at 18 weeks after HFD feeding although it was not statistically signficant (Figure IX in the Supplement). ⁴⁰ The decrease of Carmn expression in mouse atherosclerotic aortic tissues was independently validated by qRT-PCR (Figure 4H). Additional RNA-seq and qRT-PCR analyses indicated that CARMN expression is decreased in balloon-injured carotid arteries in rat (Figure 4I), human saphenous vein SMCs (HSVSMCs) stimulated with PDGF-BB or IL-1α (Figure 4J),^{19, 54} HCASMCs induced by switching the culture medium from a differentiation to a growth medium (Figure 4K), and rat vascular SMCs treated with PDGF-BB (Figure 4L and M). Collectively, these data suggest CARMN expression is decreased during VSMC phenotypic modulation in human diseased artery, in rodent models of vascular wall diseases and in cultured VSMCs in response to growth and inflammatory stimuli.

Figure 4. CARMN expression is down-regulated in human VSMC-related diseases, rodent vascular disease models and in VSMCs in response to stimuli of phenotypic modulation.

(A) Quantitative analysis of the percentage of CARMN positive cells (non-zero) and (B) expression level of CARMN in normal (Nor) and modulated (Mod) SMC clusters revealed by scRNA-seq of diseased human (H.) coronary (co.) artery as shown in Figure 1F. *FDR < 0.05; differential analysis with Seurat package. (C) Re-analysis of previously reported microarray or bulk RNA-seq datasets showing CARMN expression significantly decreased in diseased human arteries such as atherosclerotic arteries (GSE43292, N=32) and (D) cerebral arteries with aneurysm (Aneu.) (GSE66240, N=6–11), respectively. Ctrl: control. *FDR < 0.05; differential analysis with the GEO2R online tool for microarray data and DESeq2 package for bulk RNA-seq data. (E) Re-analysis of ATAC-seq dataset (GSE72696) showing intensity of ATAC-seq peaks (P1–5) at CARMN gene locus exhibits a notable decrease in human atherosclerotic coronary artery as compared to normal coronary arteries (representative picture of 1 out of 3 samples). N=3; differential analysis with DESeq2 package. (F) t-SNE visualization of cell types (upper), SMC lineage-traced cells (middle), and Carmn expression (bottom) of scRNA-seq data from aortic root and ascending aorta of Rosa26^tdTomato+; ApoE^−/−; Myh11-CreER^T2 mice at baseline, 8 and 16 weeks after HFD feeding (GSE131776). Mod: modulated SMCs. (G) Quantitative analysis of percentage of Carmn positive cells (non-zero, upper) and expression level of Carmn (bottom) in normal and modulated SMC clusters (including both tdTomato positive and negative cells). *FDR < 0.05; differential analysis with Seurat package. (H) qRT-PCR analysis of expression of Carmn and SM-specific gene Tagln in lesion-laden aortic arch in ApoE^−/− mice following western-diet for 12 weeks or in chow diet-fed control WT mice. N=6; *P < 0.05; unpaired Student’s t-test. (I) qRT-PCR analysis of Carmn and Tagln expression in balloon-injured rat left carotid artery (LCA). Uninjured right CA (RCA) served as control (set to 1). N=6; *P < 0.05; unpaired Student’s t-test. (J) Re-analysis of public RNA-seq data showing decreased expression of CARMN upon stimulation of PDGF-BB or IL-1α in human saphenous vein (HSV) SMCs. N=4; *FDR < 0.05; differential analysis with DESeq2 package. (K) qRT-PCR to show CARMN and TAGLN expression in HCASMCs cultured with differentiation medium (D.M., set to 1) or growth medium (G.M.) for 48 hours. N=6; *P < 0.05; unpaired Student’s t-test. (L) Re-analysis of public RNA-seq data showing reduced expression of Carmn in rat VSMCs treated with PDGF-BB for 24 hours. N=2–3; *FDR < 0.05; differential analysis with DESeq2 package. (M) qRT-PCR analysis of Carmn expression in rat VSMCs treated with PDGF-BB (50 ng/ml) for 24 hours. N=5; *P < 0.05; unpaired Student’s t-test.

CARMN promotes VSMC contractile phenotype in vitro

To determine the causality of CARMN down-regulation in vitro, we knocked down endogenous CARMN with phosphorothioate modified antisense oligonucleotide (ASO) in HCASMCs. Knocking down CARMN significantly enhanced cell proliferation (Figure 5A and B) and migration (Figure 5C), while attenuated the expression of SM-contractile proteins at both mRNA and protein levels (Figure 5D–F), as well as MIR145 expression. MIR143 expression was slightly but not significantly reduced (Figure 5D). Conversely, over-expression of CARMN promoted a contractile phenotype as indicated by increased numbers of spindle shaped HCASMCs (Figure 5G and H), decreased VSMC proliferation and migration (Figure 5I–K), and enhanced expression of SM-contractile genes at both mRNA and protein levels, as well as both MIR143 and MIR145 expression (Figure 5L–N). To assess functional roles of Carmn on individual cells, we performed IF staining for MYH11 in HCASMCs transduced with GFP or Carmn adenovirus and randomly selected 50 cells from each group to quantify IF signal intensity. Positive staining for MYH11 confirmed the SMC identity of these cells and demonstrated that Carmn over-expression generally increases MYH11 abundance in individual cells (Figure X in the Supplement). Collectively, these results suggest that CARMN is critical in maintaining a contractile phenotype of VSMCs in vitro.

Figure 5. CARMN is required for maintaining the VSMC contractile phenotype in vitro.

(A) CARMN or control phosphorothioate-modified antisense oligonucleotide (ASO) (ASO-CTRL) were transduced into human coronary artery SMCs (HCASMCs) for 5 days to knock down (KD) endogenous CARMN expression. EdU incorporation assays were performed to assess cell proliferation following depletion of CARMN in HCASMCs. Arrows point to the representative EdU positive cells. (B) Quantitative analysis of EdU positive cells following CARMN depletion. N=6; *P < 0.05; unpaired Student’s t-test. (C) Boyden chamber assays were performed to examine cell migration after knocking down CARMN in HCASMCs. N=5; *P < 0.05; unpaired Student’s t-test. (D) Expression of SM-contractile genes and MIR143/145 was analyzed by qRT-PCR or (E) Western blot. (F) Densitometric analysis of protein expression as shown by Western blot (WB) in “E” with normalization to the loading control HSP90. Relative signal in ASO-CTRL group was set to 1. N=3; *P < 0.05; unpaired Student’s t-test. (G) Carmn or control GFP adenovirus were generated and (H) transduced into HCASMCs for 48 hours. An increased number of spindle shaped cells (arrowheads) was observed in Carmn over-expressed HCASMCs. Scale bar: 50 μm. (I) Cell proliferation was measured by cell proliferation WST-1 kit and (J) cell number counting at the indicated time points. N=4 for WST-1 assay and N=3 for cell counting assay; *P < 0.05; unpaired Student’s t-test for WST-1 assay; two-way ANOVA followed with a post-hoc testing within day for cell counting assay. (K) Over-expression of Carmn attenuated the migration of HCASMCs cultured in the medium containing 0.1% or 5% FBS as revealed by Boyden chamber assays. N=5; *P < 0.05; unpaired Student’s t-test. (L) Expression of SM-contractile genes and MIR143/145 was analyzed by qRT-PCR or (M) Western blot after over-expressing Carmn for 48 hours. (N) Densitometric analysis of protein levels as shown in “M”. N=3; *P < 0.05; unpaired Student’s t-test.

SMC-specific deletion of CARMN exacerbates neointima formation in mice

To investigate a functional role of Carmn in SMCs in vivo, we generated SM-specific Carmn inducible KO (iKO) mice by crossing Carmn KI mice (PFG mice) with Myh11-CreER^T2 mice (Figure 6A). SM lineage tracing mice (Carmn^W/W; Myh11-CreER^T2; mTmG^+/−) that express GFP specifically in SMCs upon tamoxifen-induced Cre recombinase activation, served as control mice to exclude potential cytotoxicity caused by the expression of GFP or Cre in SMCs of Carmn iKO mice (Figure 6B). In adult mice, tamoxifen was injected (10x) to initiate Cre activation and 2 weeks later, vascular injury was induced with left carotid artery (LCA) ligation. Neointima formation was assessed 14 days after injury (Figure 6C). Specific deletion of Carmn in the arteries of SM-specific Carmn iKO mice was validated by detection of the GFP allele (Figure 6D). qRT-PCR analysis further demonstrated SM-specific Carmn deletion leads to complete loss of Carmn expression and significant down-regulation of SM contractile genes including Mylk and Acta2 in thoracic aorta (Figure 6E). Direct visualization of GFP in control right carotid arteries (RCA) revealed that GFP expression is restricted to the medial layer of RCAs in both groups (Figure 6F). In the injured LCA of SM lineage tracing control mice, the GFP signal was comparable between the neointima and medial layer, as GFP labels SMCs-derived cells prior to injury and independent of their altered phenotypes.⁵⁵ However, in Carmn iKO mice, GFP signal that represents endogenous Carmn expression was dramatically decreased in the neointima as compared to medial layer (Figure 6F), supporting our observations that Carmn expression is decreased in phenotypically modulated SMCs (Figure 4). HE staining further showed that the formation of neointima in injured LCA of Carmn iKO mice is markedly exacerbated versus control mice (Figure 6G and H). Furthermore, co-staining of ACTA2 and MKI67 revealed that Carmn deletion in SMCs significantly increases the total number of MKI67-postive proliferating VSMCs in injured LCA (Figure 6I and J). These data suggest that loss of Carmn expression in SMCs exacerbates neointima formation due to increased VSMC proliferation.

Figure 6. SM-specific deletion of CARMN exacerbates injury-induced neointima formation in mice.

(A) Strategy for generating SM-specific Carmn inducible KO (iKO) mice and (B) Carmn^W/W; Myh11-CreER^T2; mTmG^+/− SM lineage tracing control mice. (C) Experimental design: tamoxifen (Tam) was injected intraperitoneally (i.p.) to both control and SM-specific Carmn iKO mice (8-week-old, male mice only) for 10 days (2 days’ break after the first 5 injections) to activate Cre recombinase. Two weeks after the last injection, left carotid artery (LCA) ligation was performed with the uninjured right carotid artery (RCA) serving as the contralateral control. Both LCA and RCA were harvested 14 days post-injury to assess neointima formation. (D) Representative agarose gel picture of PCR genotyping using DNA extracted from brachiocephalic artery (BCA) or heart as templates. (E) qRT-PCR analysis of Carmn and SMC contractile gene expression in thoracic aorta of control or SM-specific Carmn iKO mice. N=6–8; *P < 0.05; unpaired Student’s t-test. (F) Direct visualization of GFP (green) in normal RCA and ligation-injured LCA of SM lineage tracing control mice or SM-specific Carmn iKO mice. Nuclei were counterstained with DAPI (blue). Yellow and white dashed lines denote external/internal elastic lamina and neointima border, respectively. L: lumen. Scale bar: 50 μm. (G) Representative pictures of HE staining in ligation-injured LCA of control or SM-specific Carmn iKO mice. Yellow dashed lines denote external/internal elastic lamina. L: lumen; Scale bar: 100 μm. (H) Quantification of neointima/medial layer ratio of injured LCA. N=8–9; *P < 0.05; unpaired Student’s t-test. (I) IF staining of MKI67 (red) and ACTA2 (green) in ligation-injured LCA of control or SM-specific Carmn iKO mice. Nuclei were counterstained with DAPI (blue). Scale bar: 20 μm. (J) Quantification of numbers of ACTA2 and MKI67 double positive cells (white arrows) in medial and neointimal areas of ligation-injured LCA in both control and SM-specific Carmn iKO mice. L: lumen; N=5; *P < 0.05; unpaired Student’s t-test.

Restoration of Carmn expression attenuates neointima formation and SM dedifferentiation in rats

We next sought to test whether restoration of Carmn expression is capable of reversing VSMC phenotypic change in vivo. We transduced Carmn adenovirus in rat LCA immediately following balloon injury. Injured LCA infected with GFP virus served as control, in addition to the contralateral control of the intact RCA. Arteries were then harvested 14 days post injury for analyzing neointima size and SM-specific gene expression (Figure 7A). qRT-PCR first confirmed that endogenous Carmn expression is significantly down-regulated in GFP-infused injured LCA compared to control RCA, while local delivery of Carmn adenovirus dramatically restored the expression of Carmn (Figure 7B). Histological analysis revealed over-expression of Carmn significantly attenuates injury-induced neointima formation as indicated by decreased neointima/media layer area ratio (Figure 7C and D). IF staining of proliferation marker MKI67 and SM marker MYH11 further demonstrated that restoration of Carmn expression inhibits VSMC proliferation while alleviating injury-induced downregulation of MYH11 (Figure 7E and F). Moreover, over-expression of Carmn partially, but significantly, inhibited injury-induced downregulation of SM-specific genes at both mRNA (Figure 7G) and protein levels (Figure 7H and I). Taken together, these data demonstrate that lncRNA CARMN plays a critical role in promoting VSMC contractile phenotype in vivo.

Figure 7. Restoration of Carmn expression attenuates neointima formation in rat carotid artery balloon injury model.

(A) Schematic diagram of the experimental design. Adenoviruses expressing GFP or Carmn were infused into balloon-injured left carotid artery (LCA) after injury. Fourteen days post injury, intact right carotid artery (RCA) and injured LCA were harvested for histology, qRT-PCR or Western blotting analysis. (B) qRT-PCR analysis of Carmn expression in rat control RCA or balloon-injured LCA infused with either GFP or Carmn adenovirus. Endogenous rat Carmn expression in RCA and LCA of GFP adenovirus-infected rats was measured with the primers specific for rat Carmn. Both RCA and LCA of rats that were locally delivered with mouse Carmn adenovirus were measured with the primers specific for mouse Carmn. N=4–5; *P < 0.05; unpaired Student’s t-test. (C) HE staining of carotid artery sections from either control RCA or balloon-injured LCA that were transduced with GFP or Carmn adenovirus. M: media; NI: neointima; Scale bar: 100 μm (D) Quantification of neointima/media layer ratio of LCA sections shown in “C”. N=6; *P < 0.05; unpaired Student’s t-test. (E) Co-staining of proliferative marker MKI67 (green) and SM-specific marker MYH11 (red) was performed in the rat control RCA or balloon-injured LCA that was infused with either GFP or Carmn adenovirus. Nuclei were counterstained with DAPI (blue). Arrowheads point to representative MKI67 positive SMCs. Scale bar: 50 μm. (F) Quantification of MKI67 positive SMCs shown in “E”. N=3; *P < 0.05; two-way repeated measures ANOVA. (G) qRT-PCR analysis of Myh11 and Tgfb1i1 expression in rat control RCA or balloon-injured LCA that was infused with either GFP or Carmn adenovirus. N=4; *P < 0.05; two-way repeated measures ANOVA. (H) Western blotting analysis of SM-specific contractile proteins in rat (R) control RCA or balloon-injured LCA with either GFP or Carmn adenoviral transduction. (I) Densitometric analysis of protein expression as shown in “H”. Signal of control RCA from the rat transduced with GFP adenovirus group was set to 1. N=6; *P < 0.05; two-way repeated measures ANOVA.

CARMN directly binds to MYOCD to potentiate MYOCD function

The observations that CARMN is a nuclear-restricted lncRNA and over-expression of exogenous Carmn by adenoviral system is functional (Figure 5 and Figure 7), a mechanism that is not observed for cis-acting lncRNAs,^{56, 57} indicate a potential trans mechanism of action of Carmn. We performed an in silico prediction and identified 55 putative Carmn-interacted proteins (Table V in the Supplement). MYOCD and SRF, two most potent mediators of SMC differentiation,^{10, 12} were captured (Figure 8A) and their potential interaction with Carmn was further supported by an additional prediction tool RPIseq (Figure XI A in the Supplement).⁵⁸ In vitro RNA immunoprecipitation (RIP) assays demonstrated that Carmn was specifically precipitated by MYOCD instead of SRF (Figure 8B). MYOCD and Carmn interaction was further confirmed by in vivo RIP using anti-MYC antibody to immunoprecipitate aortic lysates extracted from MYC/HA-tagged Myocd mouse (Figure 8C and Figure XI B in the Supplement).⁵⁹ In a complementary approach, biotin-labeled in vitro transcribed Carmn sense transcript, but not antisense transcript, directly pulled down MYOCD, rather than SRF (Figure 8D). Carmn retrieved SRF only in the presence of MYOCD, which is attributed to the known physical interaction of MYOCD and SRF,¹² instead of direct Carmn/SRF binding (Figure 8D). Further RNA pull-down assays using a series of truncated MYOCD mutants ⁶⁰ (Figure XI C in the Supplement) suggested that Carmn specifically interacts with MYOCD at the regions containing the basic (++), glutamine-rich (Q) and SAP domains where SRF binds (Figure 8E and Figure XI C in the Supplement).

Figure 8. CARMN directly binds to MYOCD to enhance MYOCD function.

(A) Circos plot showing 55 genes that encode proteins predicted to interact with Carmn. The outer rings indicate chromosome numbers. The grey lines indicate predicted Carmn-interacted protein genes that are aligned to their chromosome locations, with SRF and MYOCD in red. (B) In vitro RIP to examine the direct interaction between Carmn and bacterially expressed MYOCD (MYO) or SRF. qPCR was performed to measure the relative Carmn binding to MYOCD or SRF using primers designed for Carmn. Samples immunoprecipitated with IgG served as control. The upper gel picture represents pooled PCR products from each group. N=4; *P < 0.05; unpaired Student’s t-test. (C) RIP assay was performed using lysates extracted from aortic tissues of MYC/HA-tagged Myocd mouse with anti-MYC antibody. qPCR was performed to examine relative binding between MYOCD and Carmn or Gapdh that served as a negative control. Samples immunoprecipitated with IgG served as additional control. The upper gel picture represents pooled PCR products from each group. N=3; *P < 0.05; unpaired Student’s t-test. (D) RNA pull-down assay was performed using biotinylated sense (S) Carmn and bacterially expressed full-length of MYOCD (MYO) and SRF, followed by Western blotting. Pull-down assays performed with biotinylated antisense (AS) Carmn served as negative control. (E) RNA pull-down assay was performed using biotinylated Carmn sense probe and bacterially expressed full-length or truncated MYOCD mutants to map the interaction regions of MYOCD with Carmn. (F) Luciferase reporter assays were performed to examine the effects of Carmn on MYOCD-induced promoter luciferase activity of wildtype (WT) or CArG box mutated SM-specific genes Cnn1, Lmod1, and Tgfb1i1. N=4; *P < 0.05; two-way ANOVA. (G) qRT-PCR and (H) Western blotting analysis showing Carmn enhances MYOCD-induced SM-specific gene expression in 10T1/2 fibroblast cells. (I) Densitometric analysis of protein expression as shown in “H”. N=3; *P < 0.05; two-way ANOVA. (J) VSMCs isolated from the aortae of Carmn^PFG/PFG mice were infected with Cre adenovirus to delete Carmn in vitro (Carmn-null). Cells infected with LacZ adenovirus served as control (Con.). Cells were then infected with either GFP or Myocd adenovirus and harvested for qRT-PCR or (K) Western blotting analysis of SM-specific genes. F2/R2 and F3/R3 represent primers specifically detecting Carmn transcript V2 and V3, respectively. N=4; *P < 0.05; two-way ANOVA (J). (L) Densitometric analysis of protein expression as shown in “K”. N=3; *P < 0.05; two-way ANOVA. (M) Proposed working model for CARMN function in maintaining contractile phenotype of VSMCs. SMC nucleus-restricted lncRNA CARMN acts in-trans by directly binding to MYOCD (solid arrow), which promotes MYOCD/SRF transcriptional activity to induce expression of CArG box-dependent SM-specific genes such as MIR143/145, CNN1, TAGLN and TGFB1I1, leading to a contractile phenotype of VSMCs in healthy artery. It is not clear whether in cis-mechanism contributes to the regulation of MIR143/145 expression by CARMN in VSMCs (dashed arrow). Conversely, in response to stimuli of SMC phenotypic modulation in diseased artery, CARMN expression is down-regulated, leading to diminished transcriptional activity of MYOCD/SRF complex. As a result, expression of SM-specific genes is attenuated, thereby driving SMC into dedifferentiation phenotype.

We next tested the functional consequence of the physical interaction between Carmn and MYOCD. Data from reporter assays demonstrated that Carmn alone has no effect on the basal activity of SM-specific gene promoters but significantly promotes MYOCD-induced transactivation activity in a CArG-dependent manner (Figure 8F). Consistently, over-expressing Carmn significantly enhanced MYOCD-activated expression of SM-specific gene CNN1 in 10T1/2 cells at both mRNA and protein levels. Expression of several additional SM markers such as Myh11 and Acta2, although not statistically significant, exhibited a trend towards an increase following MYOCD over-expression in the presence Carmn (Figure 8G–I). Furthermore, inactivation of endogenous Carmn mediated by Cre adenovirus in VSMCs isolated from Carmn^PFG/PFG mice significantly abrogated MYOCD-induced SM-specific gene expression at both mRNA and protein levels (Figure 8J–L, Figure XI D in the Supplement). Interestingly, deletion of Carmn or overexpression of Myocd had no effect on the expression of non-CArG-dependent SM-specific genes such as Notch3, Apeg and Eln (Figure XI E in the Supplement). Furthermore, over-expression of MIR143/145 was unable to rescue the Carmn deficiency-induced down-regulation of SM markers including Myh11 and Mylk in mouse VSMCs (Figure XI F in the Supplement). Together, these mechanistic results suggest that Carmn is critical for VSMC differentiation by directly binding and potentiating MYOCD function in trans, but independent of MIR143/145.

In summary, our study demonstrates a previously unappreciated role of the nuclear lncRNA CARMN in maintaining the contractile phenotype of VSMCs in healthy artery by directly binding to MYOCD, thereby potentiating MYOCD function in trans (Figure 8M, top panel). CARMN also mediates the expression of its neighboring gene MIR143/145 cluster in VSMCs through an unclear mechanism. Conversely, in diseased arteries, expression of CARMN is down-regulated, thereby attenuating the transactivation activity of MYOCD/SRF complex on SMC-specific gene expression and triggering dedifferentiation of VSMCs, leading to exacerbated neointima formation (Figure 8M, bottom panel).

DISCUSSION

Distinct from lncRNAs previously implicated in SM biology,^16–26 we identified CARMN as a highly abundant and conserved SMC-enriched lncRNA through a series of in silico analysis. Using a novel GFP KI reporter mouse model, we demonstrate that Carmn is transiently expressed in embryonic cardiomyocytes and thereafter becomes restricted to adult SMCs in vivo. Future investigations are necessary to unveil the mechanisms underlying the transcriptional regulation of the SMC-specific expression of CARMN. Additionally, Carmn expression is occasionally observed at low levels in fibroblasts. It will be of interest to test potential functional roles of Carmn in fibroblasts, especially in Ly6a positive fibroblast cells in future studies.

Our study reveals that CARMN expression is markedly reduced in the diseased vascular wall of humans and rodent vascular disease models, and in VSMCs in response to stimuli of phenotypic modulation in vitro. In agreement with the previous studies showing that CARMN depletion is associated with decreased SMC-specific genes during CPC specification ^{27, 28} and in HCASMCs,^{61, 62} data from gain- and loss-of-function assays demonstrate that CARMN is sufficient and required to promote the contractile phenotype of VSMCs in vitro. Moreover, SM-specific deletion of Carmn in mice markedly exacerbates injury-induced neointima formation while restoration of Carmn expression in rats attenuates balloon injury-induced vascular remodeling. In a rat artery injury model, we observed that in contrast to the dramatic inhibition of neointima formation, Carmn over-expression only partially restored injury-induced loss expression of SMC contractile genes. This suggests that Carmn-mediated inhibition of SMC proliferation and migration also play a critical role in attenuating intimal thickening, in addition to enhancing the expression of SMC contractile genes. Furthermore, it has been shown that non-SMCs such as adventitial and medial progenitor cells may be additional contributors to neointima formation following injury.^{50, 51, 63–67} We cannot rule out the possibility that ectopic expression of Carmn in non-SMCs may also play a role in repressing neointima formation, which is also a limitation of two recent studies using Carmn global KO mice ⁶¹ or gapmeR-mediated Carmn knockdown mice ⁶². To address this, we generated a SM-specific Carmn iKO mouse model and we now provide direct evidence for a functional role of Carmn in SMCs in vivo. Collectively, these findings suggest that increasing or maintaining expression of CARMN represents a potentially promising strategy for the treatment and prevention of VSMC-related proliferative diseases.

CARMN was initially known as the host gene for the MIR143/145 cluster, that are derived from a common precursor and have been reported to be highly enriched in SMCs and important for regulation of SMC phenotypic modulation.^{29–32, 34, 68} Our results provide several additional lines of evidence supporting the previous reports that CARMN is transcribed and functionally independent of MIR143/145 in fetal ²⁷ and adult CPCs,²⁸ and in HCASMCs.^{61, 62} First, distinct promoter-associated H3K4me2 and H3K4me3 marks that are known to be enriched at the transcription start site,⁶⁹ exist immediately upstream of both CARMN and MIR143/145 in human VSMCs. Second, all Carmn isoforms do not overlapped with Mir143/145 in mouse aorta and in other species like pig. Finally, the Carmn transcript V2 used for adenovirus construction does not contain Mir143/145 but is sufficient to promote contractile phenotype of human VSMCs and to rescue the injured-induced neointima hyperplasia and SMC dedifferentiation in vivo.

As a neighboring gene of CARMN, we observed that MIR143/145 expression is induced by over-expression of Carmn in human VSMCs while it is inhibited in mouse aortic SMCs with Carmn deletion. Like other SM markers, MIR143/145 expression is enriched in contractile SMCs in a MYOCD/SRF complex-dependent manner and is down-regulated in phenotypically modulated SMCs.³² Therefore, it is difficult to distinguish whether the expression oscillation of MIR143/145 in SMCs following CARMN expression change is through in cis function of CARMN, or due to secondary effect of CARMN-mediated SMC phenotypic switching. Some lncRNAs are reported to act both in cis and in trans.⁷⁰ CARMN may represent one example of a lncRNA using both mechanisms since CARMN is an enhancer-expressed lncRNA which has been reported to act in cis.⁵⁷ It has been reported that MIR143/145 cluster is a key player in promoting SMC differentiation through diverse mechanisms such as directly inhibiting transcription factors that suppress MYOCD, including KLF4, KLF5 and ELK1.³² However, our over-expression assay showed that MIR143/145 has no effect on SM-specific gene expression in control VSMCs, despite the dramatic decrease in Klf4 and Klf5 expression. The reason for the discrepancy is unclear and remains to be determined. In addition, we found restoration of MIR143/145 in Carmn-null VSMCs slightly but not significantly up-regulates the expression of Myh11 and Mylk, suggesting that CARMN-mediated MIR143/145 expression is likely not the predominant casual factor for the functional consequence of CARMN in maintaining SMC contractile phenotype.

SRF and its coactivator MYOCD are potent inducers of the SMC differentiation program.^{10, 12} MYOCD does not bind directly to DNA, but interacts with SRF and transactivates SM-specific genes via SRF directly binding to an evolutionarily conserved CArG element.¹¹ We and others have shown that several factors physically interact with MYOCD to competitively displace MYOCD from SRF which prevents the binding of MYOCD/SRF complex to CArG elements, thereby inhibiting SM-specific gene expression. ^71–75 While these factors are proteins and act as repressors for MYOCD function, we found for the first time, that lncRNA Carmn directly interacts with MYOCD, but not SRF,⁶² and potentiates MYOCD function. More interestingly, Carmn specifically binds to the basic region and glutamine-rich region of MYOCD, the same region that is required for MYOCD interaction with SRF. We speculate that CARMN binding to MYOCD facilitates the binding of MYOCD and SRF, leading to enhanced transcriptional activity on SM-specific gene promoters.

In summary, we discovered that CARMN is a highly abundant and conserved SMC-specific lncRNA that plays a critical role in SMC differentiation through direct binding to MYOCD. Restoring or maintaining expression of CARMN could be a potential therapeutic approach for the treatment of proliferative vascular diseases.

CLINICAL PERSPECTIVE.

What is new?

• CARMN is a SMC-specific lncRNA that promotes the contractile phenotype of SMCs, independent of MIR143/145.

• SMC-specific deletion of Carmn exacerbates while over-expression of CARMN attenuates injury-induced neointima formation and SMC dedifferentiation.

• CARMN is the first non-coding RNA discovered to interact with the SMC-specific transcriptional cofactor, myocardin.

What are the clinical implications?

• Our study provides novel insights into the mechanisms underlying SMC phenotypic modulation that contributes to the development of vascular diseases such as atherosclerosis and restenosis after angioplasty.

• Restoration of CARMN expression in susceptible individuals may be a promising therapeutic approach for the treatment of SMC-driven vascular diseases.

Nonstandard Abbreviations and Acronyms:

ASO

Antisense oligonucleotide

CPC

Cardiac precursor cell

EST

Expressed sequence tag

FISH

Fluorescence in situ hybridization

GFP

Green fluorescent protein

GI

Gastrointestinal

GTEx

Genotype-tissue expression

GW

CMV-Cre⁺/Carmn^GFP/WT

HCASMCs

Human coronary artery smooth muscle cells

HFD

High-fat diet

HSVSMCs

Human saphenous vein smooth muscle cells

HUVECs

Human umbilical vein endothelial cells

KI

Knock in

LncRNA

Long non-coding RNA

miR/miRNA

MicroRNA

PCW

post conception weeks

PW

Carmn^PFG/WT

RACE

Rapid amplification of cDNA ends

SA

Splicing acceptor

scRNA-seq

Single cell RNA-sequencing

RIP

RNA immunoprecipitation

ATAC-seq

Assay for Transposase Accessible Chromatin sequencing

LCA

Left carotid artery

RCA

Right carotid artery