Cardiac Pericytes Acquire a Fibrogenic Phenotype and Contribute to Vascular Maturation After Myocardial Infarction

Abstract

Background: Pericytes have been implicated in tissue repair, remodeling, and fibrosis. Although the mammalian heart contains abundant pericytes, their fate and involvement in myocardial disease remains unknown.

Methods: We used NG2^Dsred;PDGFRα^EGFP pericyte:fibroblast dual reporter mice and inducible NG2^CreER mice to study the fate and phenotypic modulation of pericytes in myocardial infarction. The transcriptomic profile of pericyte-derived cells was studied using polymerase chain reaction arrays and single-cell RNA sequencing. The role of transforming growth factor-β (TGF-β) signaling in regulation of pericyte phenotype was investigated in vivo using pericyte-specific TGF-β receptor 2 knockout mice and in vitro using cultured human placental pericytes.

Results: In normal hearts, neuron/glial antigen 2 (NG2) and platelet-derived growth factor receptor α (PDGFRα) identified distinct nonoverlapping populations of pericytes and fibroblasts, respectively. After infarction, a population of cells expressing both pericyte and fibroblast markers emerged. Lineage tracing demonstrated that in the infarcted region, a subpopulation of pericytes exhibited transient expression of fibroblast markers. Pericyte-derived cells accounted for ~4% of PDGFRα+ infarct fibroblasts during the proliferative phase of repair. Pericyte-derived fibroblasts were overactive, expressing higher levels of extracellular matrix genes, integrins, matricellular proteins, and growth factors, when compared with fibroblasts from other cellular sources. Another subset of pericytes contributed to infarct angiogenesis by forming a mural cell coat, stabilizing infarct neovessels. Single-cell RNA sequencing showed that NG2 lineage cells diversify after infarction and exhibit increased expression of matrix genes, and a cluster with high expression of fibroblast identity markers emerges. Trajectory analysis suggested that diversification of infarct pericytes may be driven by proliferating cells. In vitro and in vivo studies identified TGF-β as a potentially causative mediator in fibrogenic activation of infarct pericytes. However, pericyte-specific TGF-β receptor 2 disruption had no significant effects on infarct myofibroblast infiltration and collagen deposition. Pericyte-specific TGF-β signaling was involved in vascular maturation, mediating formation of a mural cell coat investing infarct neovessels and protecting from dilative remodeling.

Conclusions: In the healing infarct, cardiac pericytes upregulate expression of fibrosis-associated genes, exhibiting matrix-synthetic and matrix-remodeling profiles. A fraction of infarct pericytes exhibits expression of fibroblast identity markers. Pericyte-specific TGF-β signaling plays a central role in maturation of the infarct vasculature and protects from adverse dilative remodeling, but it does not modulate fibrotic remodeling.

Keywords: fibroblasts; intercellular signaling peptides and proteins; myocardial infarction; pericytes.

Figure 1:. Apoptosis and proliferation underlie changes in pericyte density in the infarcted heart.

NG2^Dsred pericyte reporter mice underwent non-reperfused MI protocols. Dsred (RFP) fluorescence was combined with immunofluorescence for α-SMA in order to identify macrovascular mural cells (NG2+/α-SMA+ with vascular localization), microvascular mural cells (NG2+/α-SMA-) and NG2+ myofibroblasts (NG2+/α-SMA+ cells located outside the vascular media. As previously reported by our group, in normal mouse myocardium (A-C), the majority of NG2+ cells are α-SMA-negative perivascular cells, identified as microvascular pericytes. NG2+/α-SMA+ arteriolar vascular smooth muscle cells are also noted (arrows). Representative images show the changes in pericyte localization during the inflammatory (3 days, D-F), proliferative (7 days, G-I) and maturation phase (28 days, J-L) of infarct healing. During the proliferative phase, there is a marked increase in the density of NG2+/α-SMA+ cells in the infarct (I) located outside the vascular media (G-I, arrows; M). These cells could represent NG2+ myofibroblasts. During the maturation phase of cardiac repair, NG2+/α-SMA+ cells form neoarterioles with a thick media (J-L, arrows) and their density is significantly increased (N). A reduction in the density of NG2+/α-SMA- cells is first noted during the inflammatory phase of cardiac repair (O). Dual fluorescence for Dsred and the proliferation marker Ki67 showed significant numbers of proliferating pericytes 7 days after coronary occlusion (P). TUNEL staining identified significant numbers of apoptotic pericytes at the 3-day timepoint (Q) (*p<0.05, ***p<0.001, ****p<0.0001 vs sham, n=4–6/group). Scalebar=80μm (A-L), Scalebar=15μm (P-Q).

Figure 2:. Emergence of cells expressing both pericyte and fibroblast markers in the infarcted heart.

NG2^Dsred;PDGFRα^EGFP pericyte:fibroblast dual reporter mice were used to study changes in the cellular composition of the infarct after MI. A-D: NG2+ cells and PDGFRα+ cells were identified using flow cytometry in cells harvested from sham (A-B) and infarcted hearts (C-D, 7 days permanent occlusion/PO). E: Quantitative analysis showed that 7 days after infarction, there was a marked increase in the number of CD31-/CD45- cells as a percentage of the viable (7AAD-), metabolically active Calcein+ cells harvested from the infarcted heart (****p<0.001, n=9–16/group). These non-hematopoietic, non-endothelial cells are predominantly fibroblasts and pericytes. Although the number of NG2+ (F) and PDGFRα+ (G) cells (as a percentage of the CD31-/CD45- cells) did not change after infarction, an NG2+/PDGFRα+ cell population emerged with characteristics of both fibroblasts (PDGFRα expression) and pericytes (NG2 expression) (H; ****p<0.0001, n=9–16/group). I: A significant reduction in the percentage of NG2+/PDGFRα- cells was also noted (p=0.01, n=9–16/group). Immunofluorescent staining of sham (J) and infarcted pericyte:fibroblast dual reporter mouse hearts (K, L: 7 days PO, M: 28 days PO) showed the emergence of a small double positive NG2+/PDGFRα+ cell population (arrows) 7 days after infarction. Quantitative analysis shows a significant increase in the density of NG2+/PDGFRα+ cells at the 7-day timepoint (N) (**p<0.01, n=4–5/group). O-Q: High magnification images show NG2+/PDGFRα+ cells (arrows) in infarcted hearts. Scalebar=30μm (J, K, M), Scalebar=12μm (L), Scalebar=10mm (O-Q).

Figure 3:. Pericyte to fibroblast conversion in the healing infarct.

Lineage tracing studies were performed in order to track NG2+ pericytes in the infarcted myocardium using the inducible NG2CreER;R26^tdTom line (iNG2Cre). These animals were bred with PDFRα^EGFP fibroblast reporter mice, for reliable identification of fibroblasts (based on their nuclear labeling through the use of the reporter). Abundant PDGFRα+ fibroblasts were identified in the infarcted heart after 7 and 28 days of coronary occlusion (green nuclei: A-C, E-G). In contrast, there is no significant expansion of the fibroblast population in the remote remodeling myocardium (D, H). High magnification images (I-L) show that at the 7-day timepoint, a small fraction of the PDGFRα+ cells is derived from NG2+ pericytes (arrows), Quantitative analysis showed no statistically significant changes in the density of NG2-derived cells in healing infarcts (M). In contrast, the density of PDGFRα+ fibroblasts markedly increased in the infarcted segments [I] after 7–28 days of coronary occlusion (N). A significant increase in the density of pericyte-derived fibroblasts was noted 7 days after infarction (O). Although the number of pericyte-derived fibroblasts is reduced at the 28-day timepoint (E-G), it remains higher than in sham myocardium. There is a modest, but significant increase in the density of pericyte-derived fibroblasts in the remote remodeling myocardium [R], 28 days after coronary occlusion (O). The short arrows (A, E, F) show pericyte-derived vascular mural cells. *p<0.05, ***p<0.001, ****p<0.0001, n=6–8/group). Scalebar=50μm (A-H), Scalebar=10μm (I-L).

Figure 4:. Pericyte-derived fibroblasts exhibit increased expression of genes associated with matrix synthesis and remodeling, and high levels of integrins.

Pericyte-derived fibroblasts (P-F), fibroblasts derived from other cellular sources (Non-P F) and pericytes (P) were sorted from NG2^CreER;R26^tdTom;PDGFRα^EGFP mice 7 days after MI. Sorted cells were used for RNA extraction and PCR array to assess levels of fibrosis-associated genes. When compared with pericytes and with fibroblasts derived from other cellular origins, pericyte-derived fibroblasts were overactive, exhibiting much higher expression of structural collagens (A, Col1a2), and matrix-remodeling genes (B-C, Mmp2 and Timp3). Levels of Thbs1 (D), Ccn2 (E), Postn (F), SerpinE1 (G), and Timp2 (H) were much higher in pericyte-derived fibroblasts than in pericytes. Pericyte-derived fibroblasts also expressed much higher levels of integrins than pericytes and fibroblasts from other sources (I-O). (*p<0.05, **p<0.01, ***p<0.001, ****p<0.0001, n=3/group).

Figure 5:. Pericytes contribute to vascular maturation, coating microvessels in the healing infarct.

Lineage tracing studies were performed in order to track NG2+ pericytes in the infarcted myocardium using the inducible NG2CreER;R26^tdTom line (iNG2Cre). Sections were stained for NG2 (tdTom) and α−SMA. A: In the non-infarcted myocardium pericyte-derived cells are noted in α-SMA+ arterioles (arrows) and microvascular capillaries. Normal myocardial arterioles have a thin media, comprised of NG2+/α-SMA+ mural cells. B: 7 days after MI, remodeling arterioles are identified in the infarcted myocardium (arrows). These vessels exhibit active incorporation of clusters of NG2-derived α-SMA-negative cells into the media. The short arrows show NG2-derived α-SMA+ cells, located outside the vascular media, likely representing pericyte-derived myofibroblasts. C: After 28 days of coronary occlusion, remodeling arterioles have acquired a thick media comprised of NG2-derived α-SMA+ cells (arrows). There is limited new incorporation of α-SMA-negative pericytes into the media. An immature vessel with low pericyte coverage is also shown (yellow arrow). D: Quantitative analysis of the NG2+/α-SMA-negative area as a fraction of the area of the vessel documents the active recruitment of α-SMA-negative pericytes into the vascular wall during the proliferative phase of cardiac repair (7 days after coronary occlusion). E: Quantitative analysis of the α-SMA+ area as a fraction of the area of the vessel demonstrates the progressive thickening of the vascular media in remodeling infarct neovessels (***p<0.001, ****p<0.0001, n=5–8/group). These findings document the active recruitment of α-SMA-negative pericytes in remodeling infarct neovessels, which ultimately results in formation of mature scar arterioles with a thick α-SMA+ mural cell coat. Scalebar=100μm.

Figure 6:. scRNA-seq demonstrates diversification and fibrogenic activation of NG2 lineage cells after infarction.

NG2^CreER;R26^tdTom (iNG2Cre) mice were used to harvest NG2 lineage cells using FACS. Control (n=2, pooled from 2 animals each) and infarcted hearts (n=3, 7 days coronary occlusion) were studied. A: scRNA-seq analysis identified 12 transcriptionally distinct clusters of NG2 lineage cells in mouse hearts. B: The number of cells/heart suggests expansion of the NG2 lineage cells after infarction. C: The percentage of cells of each cluster in control (blue) and infarcted hearts (brown/green) shows that 4, 1, 6, 5 and 12 were the main clusters in control hearts. After infarction two new clusters that were absent in control myocardium emerged (clusters 2 and 7), and several other clusters found in very low numbers in control myocardium, markedly expanded (clusters 3, 8, 9, 10 and 11), (control: n=2, representing pooled samples from 2 different mice each; infarct: n=3, 2 of which represent pooled samples from 2 mice each). The gene expression profiles of various clusters are illustrated in Figures S13-S15. D: Pseudotime trajectory analysis was performed for single cell gene expression data from infarct NG2 lineage cells to infer paths between clusters of cells. The velocity data show the central role of cluster 8 cells (proliferating cells) in derivation of other clusters, including clusters with fibroblast gene expression (cluster 7) and high matrix gene synthesis (cluster 2), and cells with characteristics of mature vascular smooth muscle cells (cluster 5). Thus, proliferation may drive diversification of NG2 lineage cells after infarction E: Gene expression dotplots (left for control and right for infarct) show fibrogenic activation of resident clusters after infarction. Cluster 5 cells (cells expressing high baseline levels of mature pericyte and vascular smooth muscle cell genes) exhibited significantly increased expression of collagens after infarction, including genes encoding the structural collagen chains Col1a1 and Col3a1, Pcolce, the gene encoding procollagen C endopeptidase enhancer 1, a collagen-processing enzyme, genes associated with a matrix-preserving phenotype, such as Timp1 and Loxl2, Postn and Fap, accompanied by reduced levels of mural cell genes, including Acta2, Myh11, Mylk and Rgs4.

Figure 7:. TGF-β modulates expression of fibrosis-associated genes in human pericytes.

TGF-β1 stimulation induces synthesis of Col1a2 (A), matricellular genes (C-D, Ccn2/Ctgf and Thbs1), genes involved in matrix remodeling (E-G, Mmp2, Timp1, Timp3), and fibrogenic growth factors (H-J, Pdgfa, Tgfb1, Tgfb2) in human placental pericytes. In contrast to its effects on Tgfb1 and Tgfb2, TGF-β1 suppresses expression of Col3a1 (B) and decreases Tgfb3 expression (K). TGF-β1 stimulation also induces pericyte integrin (L-O) and Mmp14 (P) synthesis. (*p<0.05, **p<0.01, ***p<0.001, ****p<0.0001, n=4/group).

Figure 8:. Pericyte-specific TGF-β signaling plays a central role in vascular maturation, protecting from dilative remodeling, but does not significantly contribute to myofibroblast infiltration.

The role of TGF-β signaling pathways in pericytes was studied in vivo using mice with inducible pericyte-specific loss of TGFβR2 (iPTbR2KO), the only type 2 receptor transducing responses to all 3 TGF-β isoforms. Echocardiographic analysis showed that pericyte-specific loss of TGF-β signaling did not affect the severity of systolic dysfunction 7 days after MI (A, C), but significantly increased LVEDV (B, E). LV mass was not significantly different between iPTbR2KO and Cre+ control animals (C, F). Panels D-F show the differences (Δ) in these parameters in comparison to baseline (pre-infarction) levels (*p<0.05, n=14–17/group). Myofibroblast density and pericyte to myofibroblast conversion were assessed in infarcted NG2^CreER;R26^TdTom;Tgfbr2 fl/fl mice (H) and corresponding NG2^CreER:R26^TdTom (Cre+ WT) controls (G). There were trends towards reduced density of pericyte-derived cells in the infarct of iPTbR2KO mice that did not reach statistical significance (I). No significant differences were noted in the density of pericyte-derived myofibroblasts (MF, J), total myofibroblast density (K) and the fraction of pericyte-derived myofibroblasts (L) between groups. The effects of pericyte-specific TGF-β signaling on infarct angiogenesis and vascular maturation were assessed using CD31/α-SMA staining (M-N). There was a significant reduction in the density of CD31+/α-SMA-coated mature vessels (arrows) in iPTbR2KO infarcts, when compared to Cre+ WT controls (O). In contrast, there were no significant effects on total microvascular density between groups (P) (*p<0.05, n=7/group). Scalebar=100μm.