Aging induces cardiac mesenchymal stromal cell senescence and promotes endothelial cell fate of the CD90 + subset

Abstract Aging is a major risk factor in the development of chronic diseases, especially cardiovascular diseases. Age‐related organ dysfunction is strongly associated with the accumulation of senescent cells. Cardiac mesenchymal stromal cells (cMSCs), deemed part of the microenvironment, modulate cardiac homeostasis through their vascular differentiation potential and paracrine activity. Transcriptomic analysis of cMSCs identified age‐dependent biological pathways regulating immune responses and angiogenesis. Aged cMSCs displayed a senescence program characterized by Cdkn2a expression, decreased proliferation and clonogenicity, and acquisition of a senescence‐associated secretory phenotype (SASP). Increased CCR2‐dependent monocyte recruitment by aged cMSCs was associated with increased IL‐1ß production by inflammatory macrophages in the aging heart. In turn, IL‐1ß induced senescence in cMSCs and mimicked age‐related phenotypic changes such as decreased CD90 expression. The CD90+ and CD90‐ cMSC subsets had biased vascular differentiation potentials, and CD90+ cMSCs were more prone to acquire markers of the endothelial lineage with aging. These features were related to the emergence of a new cMSC subset in the aging heart, expressing CD31 and endothelial genes. These results demonstrate that cMSC senescence and SASP production are supported by the installation of an inflammatory amplification loop, which could sustain cMSC senescence and interfere with their vascular differentiation potentials.

senescence-associated secretory phenotype (SASP). Cell cycle inhibition is mainly carried out by cyclin-dependent kinase inhibitors (CDKIs) from the INK4 family, such as p16, and from the Cip/ Kip family such as p21, which inhibit S-phase entry and cell cycle progression (He & Sharpless, 2017). In the context of cardiac aging, elimination of p16 senescent cells using INK-ATTAC mice or senolytic treatments prevented age-related cardiac remodeling and SASP factors production (Anderson et al., 2019;Baker et al., 2016;Lewis-McDougall et al., 2019;Walaszczyk et al., 2019). SASP is characterized by the production of several bioactive factors such as growth factors, pro-inflammatory cytokines, chemokines and proteases but the fine composition of the secreted factors appears to be cell-type-dependent (Hernandez-Segura et al., 2017).
Whereas transient SASP production during morphogenesis and wound healing allows the coordination of several stromal cellular partners to facilitate immune infiltration, angiogenesis and limitation of scar fibrosis (Demaria et al., 2014;Storer et al., 2013), the installation of a chronic pro-inflammatory microenvironment during aging has been shown to be deleterious with loss of tissue homeostasis and progressive organ dysfunction (Baar et al., 2017;Jurk et al., 2014). This chronic inflammatory state, named inflammaging, is associated with age-related pathologies such as type 2 diabetes, as well as chronic kidney and cardiovascular diseases (Franceschi, Garagnani, Parini, Giuliani, & Santoro, 2018).
In the adult heart, cardiac mesenchymal stromal cells (cMSCs) participate in stromal cardiac tissue renewal by their potential to differentiate into vascular smooth muscle cells (SMC) and endothelial cells (EC) and by their ability to produce a variety of paracrine factors with trophic, angiogenic, and pro/anti-inflammatory effects.
We hypothesized that aging could impact cardiac microenvironment homeostasis by inducing senescent programs in cMSCs and promoting their SASP.
In the present study, by performing transcriptional expression analysis of cell-sorted cMSCs, we show that aged cMSCs acquire a senescent program including the expression of selective cell cycle regulators from the INK4 family and SASP factors involved in the regulation of the immune response. This senescence program is associated with modification of cMSC subset diversity and functional changes in vascular differentiation potential and monocyte chemoattraction. During aging, IL-1ß, produced by cardiac macrophages (cMPs), could mediate paracrine senescence of cMSCs and account for CD90+ cMSC rarefaction. We show that aging is associated with specific changes in cardiac stroma microenvironment and coincides with the installation of a deleterious amplification loop promoting paracrine senescence of cMSCs and modifying their endothelial differentiation potential.

| RE SULTS
2.1 | Gene expression profiling of cMSCs from young and aged C57BL/6 mice revealed specific ageassociated gene expression programs To assess the main biological pathways modified by aging in cMSCs, transcriptomic analysis by microarray was conducted on native cells after cell sorting based on the expression of Sca-1 and CD140a (PDGFRα) and the lack of CD31 and CD45 markers ( Figure S1). We identified 195 genes significantly up-regulated and 140 genes downregulated in the aged group compared with young group (p value ≤ 0.01 and absolute log 2 fold change > 0.5) (Figure 1a). The maximum log 2 FC value (4.1) was for haptoglobin (Hp), a protein of the acute inflammatory phase, with hemoglobin scavenger activity and immunomodulatory functions (Serrano, Luque, & Aran, 2018) (Figure 1a; Table S1).

| CCR2-dependent chemoattraction of monocytes by aged cMSCs associated with increased frequencies of cardiac CCR2+ macrophages in aged mice
The expression of chemokines (Ccl2, Ccl8, Cxcl12, and Cx3cl1) and cytokines (Il33, Il34 and Il7) associated with the acquisition of a SASP by aged cMSCs could confer an increased ability of cMSCs to interact with immune cells (Figure 1g). Conversely, aged vascular endothelial cells (EC) did not up-regulate Ccl2 and Ccl8 with aging ( Figure S3a). As these chemokines are CCR2 ligands and known to play a major role in monocyte recruitment, we compared the ability of young and aged cMSCs to recruit monocytes using a chemotaxis assay (Figure 2a,b). Aged cMSCs attracted more monocytes than young cMSCs, and the inhibition of CCR2 by RS504393 prevented this increase (Figure 2a,b). These results showed that aged cMSCs increased monocyte recruitment through CCR2 activation, supporting a key role of these cells in cardiac monocyte recruitment.
To determine whether aging was associated with increased monocyte recruitment in the heart, blood monocytes and cardiac monocytes were analyzed by flow cytometry. Aged mice had higher percentages of circulating CCR2+ monocytes (Ly6C+ CD62L+ CX3CR1 low ) in the blood (Figure 2c, Figure S3b-d) and of CCR2+ monocytes (Ly6C+ MHCII-CD64-) in cardiac stroma compared with young mice (Figure 2d, Figure S3i). While the absolute number of cardiac macrophages (cMPs) per mice was not consistently modified with aging ( Figure S3i), the frequency of CCR2+ cells in cMP population increased with age ( Figure 2e; Figure S3e). The CCR2+ cMPs from aged mice had higher expression of MHCII and Ly6C compared with young ( Figure S3f,g), revealing that aging favored the conversion of CCR2+ monocytes into activated CCR2+ cMPs in the heart.
To assess the potential impact of the cMP population shift on the cardiac microenvironment during aging, we cell-sorted total cMP population and confirmed higher Ccr2 gene expression in aged cMP   Figure S3k), confirming the higher propensity of aged cMPs to produce this cytokine in the cardiac microenvironment. These results show that with aging, the cMP pool is modified with increased frequencies of CCR2+ cMPs that could account for the observed M1/M2 mixed profile and the increased IL-1ß production.

| Treatment of cMSCs by IL-1ß mimicked the phenotypic changes associated with aging
We hypothesized that some pro-inflammatory mediators produced by cMPs during aging could in turn affect surrounding stromal cells and contribute to the phenotypic changes observed in aged cMSCs.
Indeed, one of the GO biological processes enriched by our differential expression analysis was the "cellular response to cytokine stimulus" (Table 1, Figure S4a).

| Loss of CD90 expression is a hallmark of cMSC aging
To examine the phenotypic changes associated with aging in cMSCs, In conclusion, aging induced major changes in cardiac stromal cell diversity with a specific decrease of the CD90 subset in cMSC population.

| CD90 expression identified a cMSC subset more prone to differentiate toward the endothelial cell lineage
To verify that both the CD90-and CD90+ cMSCs belonged to the same cell population, we analyzed the expression of genes related to the cardiac mesenchymal lineage. Both CD90-and CD90+ cMSCs subsets from young mice expressed higher levels of Tcf21, Tbx20, Tbx5, Hand2 and Gata4 transcription factors compared with EC ( Figure 5a), confirming the pro-epicardial origin of both cMSC subsets (Chong et al., 2011;Noseda et al., 2015).
We then tested the expression of classical genes related to bone- with EC (Figure 5b). CD90+ cMSCs had higher Lepr expression than the CD90-subset. In contrast, expression of the Angpt1 gene was higher in the CD90-cMSC subset (Figure 5b), suggesting a role of this subset in the regulation of angiogenesis through cross talk with ECs and stabilization of quiescent vessels (Carmeliet & Jain, 2011).
When the entire cMSC population from young and aged mice was stimulated with angiogenic factors for EC differentiation, Thy1 mRNA (CD90) was up-regulated ( Figure S2i). However, these cells still ex- These data showed that CD90 expression helps to delineate a specific subset of cMSCs, with higher Lepr expression and more prone to acquire EC markers in response to VEGFA and FGF2.

| Senescence of the cMSC CD90+ subset favored the acquisition of an endothelial-like cell fate and correlated with the appearance of a new CD31+ cMSC subset in the aging heart
As we showed that aging decreased CD90+ cMSC frequencies, we asked whether aging could also impact the vascular differentiation potential of this subset. Young and aged CD90+ cMSC populations were cell-sorted, cultured for 10 days without differentiation factors, and then tested for the acquisition of endothelial and smooth muscle cell markers. As shown in Figure 6a, aged CD90+ cMSCs expressed higher mRNA levels of Pecam1, Kdr, Cadh5 and Vwf compared with young CD90+. No modulation of these genes could be detected for the CD90-subset (Figure 6a). On the contrary, aged CD90− cMSCs expressed higher mRNA levels of smooth muscle cell markers compared with young CD90−. No modulation or decreased expression of these markers was observed for the CD90+ subset ( Figure S6e). Hence, these results showed that, with aging, the CD90+ and CD90-cMSC subsets spontaneously shifted in vitro toward two different cell fates, CD90+ cMSCs more prone to acquire EC markers and CD90-cMSCs, SMC ones.
Moreover, induction of Cdkn2a but not Cdkn2b was superior in aged CD90+ cMSCs than in aged CD90-cMSCs (Figure 6b). We previously observed that expression of Cdkn2a was up-regulated during in vitro vascular differentiation of the entire cMSC pool and, notably, in aged cMSCs in response to endothelial differentiation medium ( Figure S2j).
These results suggested that the preferential up-regulation of

| D ISCUSS I ON
In diverse organs, a better understanding of stroma heterogeneity and locally produced microenvironmental factors has revealed F I G U R E 4 Loss of CD90 expression is a hallmark of cMSC aging. (a-b) Representative histograms (a) and percentages (b) of CD34 and CD90 in young (blue) and aged (orange) cMSCs, and isotype control (gray) (n = 8-10 mice per group).
We hypothesized that MP-derived IL-1ß could in turn re-enforces cMSC senescence. Microarray analysis supported this concept, as "response to cytokine" was one of the main biological process identified in aged cMSCs. We demonstrated that stimulation of cMSCs by IL-1ß but not IFN-ß mimicked several phenotypic and functional changes associated with aging in murine and human primary cell cultures. These results strongly support the concept that IL-1ß production in the aging heart contributes to paracrine senescence of cMSCs and modulation of their subsets.
The DECyt method enabled us to visualize several age-dependent modifications in the composition of the cardiac stroma, impacting diverse populations and, in particular, decreasing the frequencies of CD90+ cMSCs. CD90 expression allowed us to delineate a specific cMSC subset, expressing Lepr and Vegfa, which preferentially acquired EC markers in response to differentiation medium. At the molecular level, the CD31+ cMSC subset had strong similarities with endovascular progenitors (EVPs) involved in developing vasculature during wound healing (Patel et al., 2017) and tumor growth (Donovan et al., 2019). Similar to CD31+ cMSCs, Understanding how aging leads to modulation of the Notch pathway in CD31+ cMSCs, prohibiting their full conversion to the EC fate, requires further investigation. We currently hypothesize that the reduction of CD90+ cMSCs is due to both IL-1ß-dependent paracrine senescence and conversion to non conventional CD31+ cMSCs, blocked at a transitional stage with EC-like features.
The age-related modifications that take place within the cardiac microenvironment include alterations in cMSC subset dynamics and senescence-associated functional changes.
Modifications of the cardiac microenvironment are known to impact parenchymal cell function and contribute to the incidence of numerous cardiac diseases. We have shown here that aging leads to negative effects, exerted by the stroma, on cardiac homeostasis.
Genetic and environmental factors likely exacerbate these natural tendencies leading to a higher frequency of cardiac disease among the elderly.

| E XPERIMENTAL PROCEDURE S
A more detailed account of experimental procedures can be found in the Supplementary Material accompanying this article.

| Primary cardiac cell isolation
For murine cardiac cells, hearts were harvested from young mice (4 months ± 1.47) and aged mice (20 months ± 2.55) and enzymatically processed as indicated in Appendix S1. Cardiac stromal cells were stained with specific fluorochrome-conjugated mAbs and cell-sorted by high speed sorting with BD InfluxTM cell sorter (BD Biosciences).
Human MSCs (hMSCs) were extracted from apex biopsies by enzymatic digestion and selected by overnight adhesion and cultures as indicated in Appendix S1. Control medium consisted of EGM2 with 2% FCS but without growth factors, supplements, or anti-TGFß. These assays were started directly after cell sorting, at P0.

| Cytokine treatments
cMSCs were treated three times (every 2 days) with recombinant IL-1β or IFN-β at 10 ng/ml (Peprotech) in αMEM 10% HIFCS. Eight days after the first treatment, cells were either lysed for mRNA extraction or stained for flow cytometry. These assays were started at P0 or P1, and the flow cytometry analysis was performed after cell trypsinization.
RNA was extracted from cMSCs (pool of 3 mice), and quality and dosage were controlled with a Fragment analyzer (Advance Analytical).
After Feature Extraction (Agilent), data were analyzed in R using the

| Statistics
Results are expressed as means ± SEM. The statistical significance between two groups was estimated using either unpaired Student's two-tailed t test or with the nonparametric Mann-Whitney U test, when indicated in the figure legend. Interactions between the effect of age and the subset of cMSCs used for vascular differentiation were evaluated with 2-way ANOVA. Multi-group comparisons were performed using either 1-way ANOVA with Bonferroni's post hoc test or, when indicated, with the Kruskal-Wallis test with Dunn's post hoc test, for samples with a nongaussian distribution.
Differences between groups were tested using GraphPad Prism software (version 7) and considered statistically significant for p < 0.05.