Primed smooth muscle cells acting as first responder cells in disease

Rationale Vascular smooth muscle cell (VSMC) dysregulation is a hallmark of vascular disease, including atherosclerosis. In particular, the majority of cells within atherosclerotic lesions are generated from pre-existing VSMCs and a clonal nature has been documented for VSMC-derived cells in multiple disease models. However, the mechanisms underlying the generation of oligoclonal lesions and the phenotype of proliferating VSMCs are unknown. Objective To understand the cellular mechanisms underlying clonal VSMC expansion in disease. Methods and Results Here we analyse clonal dynamics in multi-color lineage-traced animals over time after vessel injury. We demonstrate that VSMC proliferation is initiated in a small fraction of VSMCs that initially expand clonally in the medial layer and then migrate to form the oligoclonal neointima. Selective activation of VSMC proliferation also occurs in vitro, suggesting that this is a cell-autonomous feature. Mapping of VSMC trajectories using single-cell RNA-sequencing reveals a continuum of cellular states after injury and suggests that VSMC proliferation initiates in cells that have downregulated the contractile phenotype and show evidence of pronounced phenotypic switching. We show that proliferation is associated with induced expression of stem cell antigen 1 (SCA1) and the expression signature previously identified in SCA1+ VSMCs in healthy arteries. A remarkably increased proliferation of SCA1+ VSMCs, directly validated in functional assays, indicates that SCA1+ VSMCs act as “first responders” in vascular injury. Early atherosclerotic lesions also had clonal VSMC contribution and we show that the proliferation-associated injury response is conserved in plaque VSMCs, extending these findings to atherosclerosis. Finally, we identify VSMCs in healthy human arteries that correspond to the SCA1+ state in mouse VSMCs and show that genes identified as differentially expressed in this human VSMC subpopulation are enriched for genes showing genetic association with cardiovascular disease. Conclusions We show that cell-intrinsic, selective VSMC activation drives clonal proliferation after injury and in atherosclerosis. Our study suggests that healthy mouse and human arteries contain VSMCs characterised by expression of disease-associated genes that are predisposed for proliferation. Targeting such “first responder” cells in patients undergoing vascular surgery could effectively prevent injury-associated VSMC activation and neoatherosclerosis.


INTRODUCTION
Cardiovascular disease remains the leading cause of death worldwide according the World Health Organization. Current treatment strategies focus on lipid lowering and blood pressure regulation. However, recent genetic, cellular and molecular studies highlight an important role for vessel wall cells in disease development 1 . Vascular smooth muscle cell (VSMC) contractility controls vessel tone, but the cells display a remarkable phenotypic plasticity characterized by downregulation of the contractile machinery (e.g. MYH11, ACTA2 and CNN1), concomitant increased expression of extracellular matrix (ECM) components (e.g. Matrix gla protein, MGP, and collagen) and exit from quiescence 2 . Such classical VSMC phenotypic switching ensures tissue homeostasis and enables physiological vessel remodeling, but deregulated VSMC plasticity in cardiovascular disease contributes to lesion development and arterial remodeling after surgical intervention 1,3,4 .
Multi-color lineage tracing studies in mouse models show that lesional VSMCs are oligoclonal and generated from very few pre-existing VSMCs [6][7][8][9][10] . Similar clonal VSMC contribution has also been observed in aortic dissection 11 and after injury 7 . The oligoclonal nature of intimal cells appears at odds with observations suggesting widespread VSMC proliferation after injury [12][13][14] , calling for analysis of whether lesional clonality of VSMC-derived cells results from selective activation of a small number of VSMCs or if this is due to differential survival of VSMC clones following a general proliferative response (discussed in Liu&Gomez 9 ).
Using single cell-RNA sequencing (scRNA-seq), we and others have delineated VSMC heterogeneity in atherosclerotic lesions, including the demonstration of substantial variability in expression of genes associated with cardiovascular disease [14][15][16][17][18][19] . However, the diversity of VSMCs in lesions does not inform about how VSMC investment in plaques is initiated, particularly with regard to the resulting clonal outcome. We previously identified a small subset of VSMCs in healthy mouse vessels marked by stem cell antigen 1 (SCA1), that have reduced contractile gene expression and show increased expression of genes associated with cell activation 15 . SCA1 is also induced in VSMCs in atherosclerosis and other disease models [15][16][17] , suggesting that this VSMC subset is relevant for disease development. Whether these atypical cells are functionally distinct is not known and the presence an equivalent human cell population in healthy arteries remains to be demonstrated.
By studying clonal dynamics and acute changes in scRNA-seq profiles in mouse models we here provide evidence that cell-intrinsic mechanisms, associated with a primed cellular state characterized by SCA1 expression, result in selective proliferation of a small number of VSMCs in both atherosclerosis and after injury. We further demonstrate an 8-fold increased proliferative capacity of SCA1-expressing VSMCs from healthy animals, suggesting that these act as "first responder" cells that will rapidly start proliferating after insults. An equivalent VSMC subpopulation displaying the primed gene expression signature is identified through scRNA-seq analysis of healthy human arteries. Collectively, our data suggests that selective activation of predisposed VSMCs could underlie the development of human atherosclerosis and that targeting these cells could represent novel therapeutics in atherosclerosis prevention.

Animals and procedures
Animal experiments were approved by the local ethics committee and performed according to UK Home Office regulation under project license P452C4595. All alleles have been described previously; Myh11-CreERt2 (Myh11) confers expression of a tamoxifen-inducible Cre recombinase in smooth muscle cells 18,19 , Rosa26-Confetti (Confetti) 20 and Rosa26-EYFP (EYFP) 21 are Cre-recombination reporter alleles, Ki67/RFP is an insertion in the Mki67 locus resulting in expression of a KI67/RFP fusion protein 22 and the mutant Apoe allele sensitizes mice to high fat diet (HFD)-induced atherosclerosis development 23 . VSMC lineage labeling was achieved by intraperitoneal tamoxifen injections (10x 0.1 mg tamoxifen over 2 weeks) followed by at least 1 week rest period for tamoxifen metabolism. Only male animals were used as the Myh11-CreERt2 transgene is Y-linked. The left carotid artery was ligated under the bifurcation with a silk suture under anesthesia (2.5-3% isofluorane by inhalation) with subcutaneous pre-operative analgesic (~0.1 mg/kg body weight, Buprenorphine) as described 7 . HFD (Special Diets Services, containing 21% fat and 0.2% cholesterol) was administered for 9-24 weeks.

Tissue analysis
Arteries were fixed, processed for whole mount and confocal imaging followed by cryosectioning, or directly cryo-sectioned and stained as described 7 . Aortic tissue explants were injured using a forceps pinch and embedded in Matrigel. Single cell suspensions were generated by enzymatic digestion (Collagenase IV, Invitrogen and Elastase, Worthington) for scRNA-seq library preparation, ImageStream analysis, flow cytometry and in vitro culture.
Antibodies used for staining are described in Table I of the Online-only Data Supplement.

Single cell-RNA sequencing
Mouse scRNA-seq datasets were generated from VSMC-lineage labeled cells isolated by flow-assisted cell sorting (FACS) from ligated left carotid arteries of Myh11-EYFP-Ki67/RFP animals 5 (D5) or 7 days (D7) after surgery, using the 10x chromium system (mouse D5, mouse D7). The Smart-seq2 protocol was used to process index-sorted cells from Myh11-EYFP-Ki67/RFP animals 7 days after surgery and control unligated animals. Human scRNAseq data were generated from medial cells isolated from a healthy aorta (65-year-old male).
Differential expression analysis was done using DEseq2 26 . Trajectory-inference was done using the R package slingshot (v.1.4.0) 27 . Summarized expression level of gene subsets was calculated and displayed as described 15 . Scripts used for data analysis are available upon request.

Data accessibility
The scRNA-seq datasets generated in this study (mouse D5 10x, mouse D7 10x, mouse D7 Smart-seq2, human 10x) have been deposited in the Gene Expression Omnibus (GEO) repository; accession number will be made available prior to publication. The scRNA-seq dataset of VSMC-lineage-labeled plaque cells from high fat diet-fed Myh11-Confetti-Apoe animals is available from GEO (accession number: GSE117963).

Statistical analysis
Statistical analysis was performed in R, the Shapiro-Wilk test was used to ascertain normal distribution, equal variance assessed using Bartlett or Levine tests. Tests used to assess statistical significance are indicated in figure legends. Local regression analysis was used to fit a LOESS curve of patch number. To assess statistical significance of SCA1 expression status in the clonal proliferation assay, a generalized linear model was fitted for patch number whereas multiple linear regression used for patch area, as the data showed equal variance and linearity and the residuals were approximately normally distributed.

Proliferation is restricted to a small subset of VSMCs after injury
To determine whether the observed oligoclonal VSMC contribution to vascular disease 6,7,11 results from selective cell activation or clonal competition following general activation of VSMC proliferation 9 we employed the carotid ligation injury model. We previously showed  Figure 1E). Double positive (EYFP+RFP+) cells were almost absent in healthy arteries, increased in frequency from D5, peaked at D8 and never constituted more than 5% of all EYFP+ VSMCs.
Collectively, these data suggest that activation of VSMCs proliferation occurs at a low frequency and that only a fraction of medial VSMC clones migrate across the inner elastic lamina to form oligoclonal lesions.

VSMC investment in atherosclerotic plaques mimics the injury response
Almost all medial and intimal VSMC patches were restricted to arterial segments with increased medial diameter. These remodeled segments or "bulges", also displayed disorganized cellular arrangement ( To assess whether selective VSMC proliferation also underlies oligoclonal VSMC contribution observed in atherosclerosis 6,7 , we analyzed Myh11-Confetti animals on an Apoe -/background (Myh11-Confetti-Apoe) with tamoxifen-mediated VSMC-lineage-labeling prior to feeding an atherogenic diet for 9-15 weeks. Analysis of early stage plaque (<50 Confetti+ cells per section) revealed VSMC investment to lesions from carotid arteries, aortic root, arch and the descending aorta ( Figure 2E). Like late stage lesions 7 , VSMC-derived cells were typically only of a single color and where several Confetti colors were detected, cells were arranged in a non-random manner ( Figure 2E). Examples of individual VSMC clones contributing exclusively to either cap or core was observed, and VSMC investment was found in lesions that lacked an obvious fibrous cap structure ( Figure 2E). The medial layer generally remained mosaic with respect to Confetti protein expression, however, evidence of medial VSMC patches expressing the Confetti color observed in lesion VSMC clones was often observed underneath the plaque, where medial cell disarray and breaks in the elastic lamina were also detected ( Figure 2E). Collectively, this analysis suggests that selective activation of VSMC proliferation and other hallmarks of VSMC injury-responses are also found in early atherogenesis.

Selective initiation of VSMC proliferation in vitro
To investigate whether the selective clonal expansion observed in vivo is intrinsic to VSMCs, we cultured aortic tissue explants in vitro after introducing a forceps-pinch "injury" to promote VSMC proliferation ( Figure 3A). After culture, tissue explants displayed persistent mosaic labeling in most regions but developed monochromatic patches similar to those observed in vivo along tissue edges and at forceps pinch-injuries ( Figure 3B). Quantification of contiguous surfaces expressing the same Confetti protein ( Figure  whereas these were rare at day 0 (2.2±1.3, Figure 3D). Confirming the idea that VSMC patches resulted from selective proliferation of a small number of cells, similarly sized patches were observed in tissue explants from animals with reduced labeling density ( Figure   II in the Online-only Data Supplement).
To test whether selective proliferation also occurred in freshly isolated, enzymatically dissociated VSMCs, we designed an assay that allowed detection of emerging VSMC clones while maintaining the cell-cell contacts required for VSMC survival. Single cell suspensions of lineage-labeled VSMCs from Myh11-Confetti animals were mixed with wild-type VSMCs and live cell imaging performed periodically over a 3-week period ( Figure 3E). Most Confetti+ cells in these cultures remained as "singlets" isolated by wild-type cells but occasionally a small patch of lineage-labeled VSMCs of one color formed ( Figure 3F Table III in the Online-only Data Supplement) confirming that rather than being formed by discrete cellular subsets, lineage-labeled cells represented a continuum of varying cell states. We merged this unselected, but proliferation-enriched, dataset with profiles of cells index-sorted for the VSMC-lineage label (EYFP) and the Ki67/RFP reporter from injured vessels, or EYFP+ cells from control animals ( Figure 4D).
This suggested that VSMCs post-injury display a spectrum of phenotypes from a contractile Myh11-positive state similar to that seen in healthy vessels (cell clusters 1, 2, 3 and 6), to a proliferative state characterized by high S and G2M scores ( Figure 4A, B). In addition to increasing levels of classical markers of a synthetic VSMC state, the total number of genes detected also increased gradually along this axis ( Figure 4B) indicative of increasing activation level. These analyses show that the transcriptional signatures of proliferating VSMCs substantially overlap with that of non-proliferating VSMCs after injury, suggesting that cells adopt states along a trajectory from a quiescent-contractile to a proliferative state.

Evidence of segregated injury responses at the onset of VSMC proliferation
Our analysis suggests that VSMC proliferation does not result from activation of a distinct subpopulation of VSMCs, but rather derives from cells displaying extensive phenotypic switching. To investigate this idea further, we profiled cells 5 days after injury at the onset of VSMC proliferation ( Figure 1E). Similar to day 7, VSMCs formed a continuous population displaying anticorrelated gradual changes in contractile and synthetic markers, with Mki67 expression restricted to cell cluster 9 ( Figure 5A). Surprisingly, trajectory inference using the Slingshot algorithm 27 , suggested the existence of two distinct VSMC injury-responses, of which only one included proliferating cells in cluster 9 ( Figure 5B), and partition-based graph abstraction (PAGA) 30 analysis confirmed these trajectories ( Figure IV in the Online-only Data Supplement). As shown in Figure 5B, the pseudotime axes defining these two paths shared a common origin in the Myh11-positive cell clusters. Genes showing significant changes in expression (p-adj<0.05, log(fold change)>0.5) along pseudotime for Path1 or Path2 were identified using generalized additive models (GAMs) and organized into gene clades based on Pearson correlation ( Figure 5B, lower panels, and Table IV in Table IV in the Online-only Data Supplement). Genes with increased expression along the trajectories also showed substantial overlap as well as some significant differences ( Figure   5C and Table IV in  To evaluate the relevance of proliferation-associated, injury-induced genes in atherosclerosis, we assessed the expression of Path1-induced genes in VSMC-derived plaque cells 15 . This demonstrated anti-correlation with Myh11 levels and overlap between the injury-response genes and markers of phenotypically modulated VSMCs in lesions (Chad, Ly6a, Figure 5E). Additionally, genes showing increased expression along the proliferation-associated Path1 included factors previously associated with vascular disease, such as Lum and Tnfrsf11b [15][16][17][32][33][34] . We also detected FBLN2-expression (a Path1-induced gene) in a subset of VSMC-derived plaques cells in Myh11-Confetti-Apoe animals, in particular in the lesion core ( Figure 5F). Importantly, FBLN2 was detected in aSMA-stained cells in human carotid artery plaques ( Figure 5G), indicating that this signature is also relevant for human disease.
Taken together, we identify two related but distinct injury-responses in VSMCs. Interestingly, Path2 is mainly characterized by enrichment for genes associated with protein folding, which has been linked to cholesterol responses 35 . In contrast, genes defining Path1, which represents a transition in cellular state associated with injury-induced VSMC proliferation, are also expressed in human atherosclerosis.

SCA1 expression marks "first responder" VSMCs with increased proliferative capacity
We previously identified a small subset of SCA1-expressing VSMCs in healthy arteries, which express a "Response Signature" suggestive of cell activation 15 . The frequency of SCA1+ cells increase in vascular disease 15,16,36 and the kinetics of SCA1 induction after injury correlates with emergence of KI67+ cells ( Figure 6A, Figure 1E). Further indicating an association between SCA1 expression and VSMC proliferation, the Ly6a/SCA1 transcript was detected in Path1-associated cells in the D5 scRNA-seq dataset and Ly6a/SCA1+ cells were juxtaposed to Mki67+ cells, preceding those in pseudotime ( Figure 6B). Interestingly, the "Response Signature" expressed by SCA1-positive VSMCs in healthy arteries 15 was also induced in Path1-specific cells ( Figure 6B), suggesting that SCA1 expression may mark VSMC that have undergone partial transition towards a proliferative state.
To test whether functional differences are associated with the SCA1-expression in healthy arteries, we analyzed FACS-isolated SCA1-positive and SCA1-negative lineage-labeled EYFP+ VSMCs from non-injured Myh11-EYFP animals. SCA1-positive VSMCs had remarkably reduced F-actin levels, demonstrated by significantly reduced phalloidin staining compared to the SCA1-negative counterparts ( Figure 6C, D, Figure VI in the Online-only Data Supplement). Reduced F-actin in SCA1+ cells was accompanied by significantly lower levels of ROCK1, consistent with reduced Rock1 transcript levels along Path1 during the injury response ( Figure 5C). To compare cell proliferation, we adapted the in vitro clonal proliferation assay; mixing 500 SCA1+ or SCA1-EYFP+ cells from aortas of healthy, VSMClineage traced animals with medial cells from wild-type animals and periodic live imaging over 3 weeks of culture ( Figure 6E). There was no difference in cell numbers 2 days after seeding, demonstrating equal survival. However after 1 week of culture, significantly more cells were detected in SCA1+ compared to SCA1-samples and this difference persisted throughout the experiment ( Figure 6E, F). The increasing cell number resulted from emergence of coherent patches of EYFP+ cells ( Figure 6E). In SCA1+ samples, 1-3 patches were observed per well 1 week after seeding (2.4 patches per well on average). Patch number remained approximately constant, whereas the size of individual patches increased over time (Figure 6G, H). In contrast, wells containing SCA1-EYFP+ cells did not contain patches after 1 week of culture ( Figure 6G) and patches were observed at low frequency in SCA1-cultures at later timepoints (4/18 wells).
This analysis demonstrates that SCA1-expressing cells in healthy vessels are phenotypically and functionally distinct from the bulk of VSMCs. The 8-fold increased patch number and faster kinetics of clone formation for SCA1+ VSMCs suggests that these cells might act as "first responder" cells in healthy arteries exposed to disease-inducing stimuli. SCA1 does not have an obvious human orthologue, preventing direct translation to human disease. Therefore, to assess whether healthy human arteries contain similarly primed cells, we performed scRNA-seq of cells from the medial layer of a histologically normal human aorta. Cells formed a single population that was split into 4 clusters without clearly defined borders ( Figure 7A). Most cells expressed a contractile MYH11+ signature, consistent with the absence of signs of vascular disease, but reduced levels of contractile genes were observed in cell cluster 3 and a subset of cells in cluster 0. Anti-correlating with MYH11, gradually higher expression of COL8A1, MGP and other synthetic genes was detected through clusters 2, 0 and 3, suggesting that different extents of phenotypic switching exist in human vessels ( Figure 7A). Transcripts for orthologues of injury-induced proliferationassociated genes, including FBLN2 and LUM, were also detected in Cluster 0 and 3 ( Figure   7A). To assess whether the healthy human aorta contains cells corresponding to those we identified in mouse vessels, we first tested whether genes associated with Path1 in the mouse D5 dataset show differential expression in cluster 0 versus cluster 1 ( Figure 7B).

Evidence for priming of VSMCs in human arteries
Path1 downregulated genes generally showed lower expression in cluster 0 versus cluster 1, whereas most Path1-upregulated genes were detected at higher levels in Cluster 0. The "Response Signature" defining SCA1+ VSMCs in healthy mouse arteries 15 also showed pronounced differences across the human VSMC dataset, with low levels in cluster 1 and high expression in some cells from cluster 0 and 3 ( Figure 7C). This suggests heterogeneity of VSMC in the medial layer of human aorta, with phenotypic modulation similar to that found in SCA1+ mouse VSMCs in healthy arteries and after vascular injury. Interestingly, only few genes from injury Path1-associated gene clade 2, that included most cell-cycle genes, showed differential expression in the human dataset, consistent with the notion that VSMCs in healthy arteries are largely quiescent ( Figure 7B, yellow dots).
The analysis above suggested that medial cells in human arteries are defined by a phenotypic spectrum, similar to what has been suggested 29 . To identify genes defining this spectrum in an unbiased manner, trajectory-inference was used to generate a pseudotime axis ( Figure VII in the Online-only Data Supplement), which showed strong correlation with the Response Signature (R 2 =0.66, Figure 7D). As expected, pseudotime-dependent genes with reduced expression along the trajectory were detected at lower levels in cluster 0 compared to 1 ( Figure 7E, blue dots), and were enriched for "muscle contraction" and "actinbinding" GO-terms (Table VII in the Online-only Data Supplement). Trajectory-induced genes showed increased expression in cluster 0 vs cluster 1 ( Figure 7E, red dots) and were generally restricted to cells in cluster 3 and a subset of cluster 0 cells ( Figure 7F). Enrichment for GO-terms related to extracellular matrix modification, cell adhesion, tissue development and response to TGF-beta ( Figure VIII and Table VII in the Online-only Data Supplement) suggested that a VSMCs in healthy human vessels displaying evidence of phenotypic activation. In line with this idea, pseudotime-induced genes included growth factor binding proteins (LTBP2, HTRA1) and CXCL12 that encodes stromal cell-derived factor 1 (SDF1) and is linked genetically to cardiovascular disease 37 . This analysis suggested that human arteries contain a subpopulation of VSMCs in a state corresponding to that defined by SCA1 expression in mouse. To verify this idea and add positional information for such primed VSMCs, we stained healthy human aorta sections for FBLN2, which is also pseudotime-induced in the human dataset ( Figure 7F and Table VII in the Online-only Data Supplement). FBLN2+ cells were detected in the medial layer ( Figure 7G) -in addition to adventital and endothelial staining -but at lower frequency compared to in lesions ( Figure 5G). FBLN2+ medial cells did not cluster to specific regions that could represent pre-atheromas not detected in histological examination, instead, we find that FBLN2+ cells are dispersed in the medial layer of human arteries.
Interestingly, of the 23 genes associated with the SNP-27 coronary artery disease risk score panel 38 that were included in our dataset, 11 showed differential expression along the trajectory; including TCF21, CXCL12 and ADAMTS7 (p<0.005; Figure 7H). This statistically significant enrichment strongly indicates that the changes along the pseudotime axis are relevant for human cardiovascular disease.

DISCUSSION
Using an acute model of VSMC proliferation, we demonstrate that oligoclonal VSMC contribution to lesion formation results from selective activation of proliferation in very few pre-existing VSMCs and provide evidence that this mechanism is shared at early stages of atherosclerotic plaque formation. Immediately after injury, VSMCs form a continuous spectrum of phenotypes where a pseudotime axis connects quiescent cells with a contractile signature to proliferative cells. Cells along this trajectory include SCA1-expressing cells that share characteristics of the atypical VSMC we previously identified in healthy arteries 15 . The increased proliferative capacity observed for SCA1+ VSMCs further supports the notion that these cells are predisposed to react to activating signals. The demonstration that healthy human arteries also display transcriptional heterogeneity for genes linked genetically to cardiovascular disease, and contain cells displaying significant similarities to the SCA1+ cells in mouse arteries, suggests that VSMC priming could also underlie vascular pathologies in humans.
In addition to the pseudotime axis resulting in cell proliferation (Path1), we find evidence for another VSMC response at early timepoints after injury (Path2). Both response trajectories show increased expression of ECM-related factors and reduced contractile gene expression suggesting that both represent induction of a "synthetic state". Increased expression of structural components and regulators of the ECM is consistent with the observation that VSMC proliferation is observed in arterial segments showing pronounced remodeling of the vessel wall ( Figure 2). Yet, only a fraction of VSMCs within the arterial "bulges" exit quiescence to initiate proliferation and, while substantial VSMC loss is observed, most cells within these remodeled arterial segments remain as singlets. This selective VSMC activation and oligoclonality of lesional VSMCs appears at odds with the gradually changing cells states observed in scRNA-seq where SCA1 expressing cells partially overlap proliferating cells. We suggest that, rather than being a dedicated progenitor population, some cells are predisposed, or primed, for proliferation. In accordance with this idea, SCA1+ VSMCs show an 8-fold increased proliferation frequency in vitro, although proliferation was also observed in SCA1-cells, albeit with slower kinetics. We speculate that formation of patches in SCA1-sorted samples result from the induction of a Sca1 signature previously seen in cultured VSMCs 15 . An alternative idea is that activation of VSMC proliferation induce negative feedback mechanisms to prevent neighboring cells from exiting quiescence akin to lateral inhibition. Experimental testing of these ideas is not trivial. Firstly, SCA1 is expressed by other cell types in the vasculature, necessitating a dual lineage labeling approach 16,39 .
Secondly, current SCA1-Cre drivers are not sufficiently highly expressed in medial cells to yield recombination-induced cell labeling 40 , probably due to the relatively lower expression level of Ly6a transcripts in VSMCs compared to, for example, adventitial and endothelial cells 15 .
The transcriptional signature defined by the contractile-to-proliferative axis in post-injury VSMCs shares considerable overlap with transcriptional states of VSMC-derived cells in other vascular disease models, including atherosclerosis and aneurysm [15][16][17][32][33][34] . We therefore propose that the mechanisms acting at the onset of VSMC proliferation after injury also regulate early steps of atherosclerotic plaque development. Consistently, we observe clonal VSMC contribution at early stages of plaque development, even before formation of the fibrous cap, in contrast to a study suggesting that VSMC investment results from migration along the fibrous cap and that VSMC-derived cells in the plaque core are derived from expanding aSMA+ cells 10 . Despite these apparent discrepancies, our findings are in accordance with the idea of a phenotypically modulated, plastic cell state that underpins atherosclerotic lesion VSMC infiltration 15 . Such a state has been defined by expression of Lgals3, which is present prior to cap formation 32 and also SCA1 16,17,33 , consistent with the observation of proliferation-associated SCA1+ cells in our dataset. We did not observe VSMC-derived cells expressing an osteochondrocytic signature present in atherosclerotic lesions 15,33 . Whether this is due to model-specific differences or the time point of analysis remains to be determined. However, we note that the osteochondrocytic phenotype is more pronounced at late time points and was not observed in studies limited to early-mid stage disease 32,33 .
Our study identifies additional phases of activation where VSMCs that could be subject to regulation, including VSMC priming, cell cycle activation, VSMC loss and migration across the intimal layer. Understanding how documented regulators of VSMC function in disease 1,36,41 and novel pathways -such as retinoic acid signaling and efferocytosis identified by scRNA-seq analysis of atherosclerotic plaque cells 16,33 -impact on these mechanisms will provide important insight into how targeting of vessel wall cells could be achieved to limit cell accumulation and disease severity. The existence of cells in human arteries that correspond to the murine SCA1+ VSMCs and the genetic evidence linking variable expression to cardiovascular disease highlight this cell population a promising starting point.