Hif-1a suppresses ROS-induced proliferation of cardiac fibroblasts following myocardial infarction

SUMMARY

We report that cardiac fibroblasts (CFs) and mesenchymal progenitors are more hypoxic than other cardiac interstitial populations, express more hypoxia-inducible factor 1α (HIF-1α), and exhibit increased glycolytic metabolism. CF-specific deletion of Hif-1a resulted in decreased HIF-1 target gene expression and increased mesenchymal progenitors in uninjured hearts and increased CF activation without proliferation following sham injury, as demonstrated using single-cell RNA sequencing (scRNA-seq). After myocardial infarction (MI), however, there was ~50% increased CF proliferation and excessive scarring and contractile dysfunction, a scenario replicated in 3D engineered cardiac microtissues. CF proliferation was associated with higher reactive oxygen species (ROS) as occurred also in wild-type mice treated with the mitochondrial ROS generator MitoParaquat (MitoPQ). The mitochondrial-targeted antioxidant MitoTEMPO rescued Hif-1a mutant phenotypes. Thus, HIF-1α in CFs provides a critical braking mechanism against excessive post-ischemic CF activation and proliferation through regulation of mitochondrial ROS. CFs are potential cellular targets for designer antioxidant therapies in cardiovascular disease.

Graphical abstract

In brief

HIF-1α is a master regulator of tissue responses to reduced oxygen concentration. Janbandhu and colleagues explored the role of HIF-1α in cardiac fibroblasts after myocardial infarction. HIF-1α provides a critical braking mechanism against damaging fibroblast expansion, which limits heart recovery, via modulating a class of molecules called reactive oxygen species.

INTRODUCTION

Adaptation to hypoxic microenvironments is primarily mediated through hypoxia-inducible factor 1 (HIF-1), which belongs to the bHLH-PAS (basic-helix-loop-helix-Per-Arnt-Sim) family of transcription factors (Semenza, 2014b). Under hypoxic conditions, the HIF-1α subunit is stabilized, allowing translocation to the nucleus, heterodimerization, and activation of gene pathways that minimize oxygen consumption, reduce reactive oxygen species (ROS), and restore oxygen delivery.

HIF-1 pathways play an essential role in cardiovascular (CV) biology (Semenza, 2014a; Sousa Fialho et al., 2019). In the developing heart, gestational hypoxia triggers regional HIF-1-dependent pathways essential for chamber and septal morphogenesis (Guimarães-Camboa et al., 2015; Menendez-Montes et al., 2016). The adult heart experiences periodic hypoxia, both physiologically (such as at high altitude and during exercise) and during ischemia, cardiomyocyte (CM) hypertrophy, inflammation, and fibrosis, conditions that can be associated with increased levels of ROS and oxidative damage.

Research into the roles of HIF-1 in the heart has focused mainly on CMs, although key functions in endothelial cell (EC) remodeling and ischemic preconditioning, and in macrophages during atherosclerosis, have been demonstrated (Semenza, 2014a; Sousa Fialho et al., 2019). However, HIF-1 function in cardiac fibroblasts (CFs) has not been addressed. CFs are a heterogeneous population of stromal cells that contribute to the heart’s biomechanical integrity through control of extracellular matrix (ECM) deposition and turnover (Frangogiannis, 2019a; Tallquist and Molkentin, 2017). Whereas CFs mediate adaptive fibrosis in heart repair and regeneration (Kikuchi et al., 2011; Tallquist and Molkentin, 2017), they specialize to more ECM secretory and contractile phenotypes in virtually all forms of CV pathology, leading to fibrosis, scar formation, noncompliant chambers, and heart failure (Tallquist and Molkentin, 2017).

Hematopoietic and other stem cell and progenitor cell populations have been reported to reside in an hypoxic niche, and HIF-1α expression has been suggested to be critical for stem cell homeostasis and deployment after injury (Schieber and Chandel, 2014). Given the progenitor characteristics of CFs (Chong et al., 2011), HIF-1α may have similar roles in CFs, which could include maintaining their quiescence (Bergmann et al., 2015) or stem and progenitor-like qualities, metabolic and redox control, state heterogeneity, and the complex dynamics of proliferation and differentiation that occur after injury (Farbehi et al., 2019). These issues have been challenging to study in CFs due to poorly characterized cell states, lack of specific cell markers, pathway pleiotropy (Chen et al., 2019; Monden et al., 2007; Wollert and Drexler, 2001), and the heavy reliance on a linear in vitro model of CF activation and differentiation (Tallquist and Molkentin, 2017). Single-cell genomics has begun to revolutionize our understanding of tissue structure and function, and recently, we and others have reported the first comprehensive maps of CF heterogeneity and flux in healthy and diseased hearts, revealing previously uncharacterized CF subsets and states, nonlinear cell dynamics, and intercellular signaling pathways (Alexanian et al., 2021; Farbehi et al., 2019; Forte et al., 2020; Hesse et al., 2021; McLellan et al., 2020), paving the way for a more nuanced understanding of CF biology and therapeutic interventions into fibrosis.

Here, we report that a progenitor-like subpopulation of CFs resides in a hypoxic niche, expresses Hif-1a, and exhibits a distinct metabolic profile. We investigated the function of HIF-1α in CFs in homeostasis and after myocardial infarction (MI) using conditional gene targeting and single-cell genomics. HIF-1α provides a critical braking mechanism for post-ischemic CF activation and proliferation through regulation of metabolism, ROS buildup, and intracellular signaling. Dysregulated HIF-1α function in CFs alone leads to excessive proliferative fibrosis after MI.

RESULTS

Adult CFs are hypoxic and display a distinct metabolic profile

We analyzed the hypoxia status of freshly isolated SCA1⁺ PDGFRα⁺CD31⁻ CFs (termed S⁺P⁺ cells) from uninjured mouse hearts using pimonidazole hydrochloride (Pim) staining and flow cytometry. S⁺P⁺ cells represent a sub-fraction of quiescent CFs that show progenitor features (Chong et al., 2011; Noseda et al., 2015; Soliman and Rossi, 2020). S⁺P⁺ cells showed ~3-fold higher Pim mean fluorescence intensity (MFI) compared to the total cardiac interstitial (non-CM) population (TIP) (Figure 1A), contributing mostly to the Pim^high fraction (Figure 1A, right panel). Among TIP subsets, S⁺P⁺ cells were by far more hypoxic than ECs, as well as SCA1⁻PDGFRα⁺CD31⁻ (S⁻P⁺) CFs, corresponding to the other major quiescent CF population present in uninjured hearts (Farbehi et al., 2019) (Figure S1A).

Figure 1. Metabolic profile of adult cardiac S⁺P⁺ cells.

(A) Flow cytometry analysis of Pim staining (left) and the relative percentage of cells in different Pim fractions (right). MFI, median fluorescence intensity (middle) (n = 4).

(B) Oxygen consumption rate (n = 3).

(C) Intracellular ATP levels (n = 3).

(D) Flow cytometry analysis of 2-NBDG (2-(N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino)-2-deoxyglucose) fluorescence (n = 5).

(E) Intracellular lactate levels (n = 3).

(F) Flow cytometry analysis of MitoTracker fluorescence (n = 3).

(G) Flow cytometry analysis of HIF-1α staining (n = 3).

(H) WB analysis of freshly sorted S⁺P⁺ and TIP cells (n = 3).

(I) qRT-PCR of HIF-1 target genes. Values are normalized to TIP cells (n = 3).

(J) FACS sorting strategy of MitoTracker fractions (Mito^low, Mito^medium, and Mito^high).

(K) Crystal violet (CV)-stained colonies in each Mito fraction at the first passage (passage 0, P0). Scale bar, 1 cm. (L and M) Colony and cumulative cell numbers, respectively, in different Mito fractions (n = 3).

See also Figure S1 and Table S1.

In Pdgfra^nGFP/+ mice (Chong et al., 2011), Pim staining was predominantly in GFP⁺ cells, located in the epicardium, sub-epicardium, myocardial interstitium, and endocardium (Figures S1B and S1C). Consistent with relative hypoxia, isolated S⁺P⁺ cells had a lower mean rate of mitochondrial respiration and intracellular ATP (Figures 1B and 1C), together with higher glycolytic flux and intracellular lactate, as compared to TIP (Figures 1D and 1E). Mean mitochondrial mass assessed using MitoTracker dye was also lower in S⁺P⁺ cells than TIP (Figure 1F), as confirmed by qPCR (Figure S1D). These data suggest that S⁺P⁺ cells have a higher reliance on glycolysis over mitochondrial oxidative phosphorylation (OXPHOS) for energy production, similarly to hematopoietic stem cells (HSCs) (Simsek et al., 2010). Mitochondrial mass in CFs was determined developmentally, becoming established as early as postnatal week 1 and remaining constant into adulthood (Figure S1E), whereas CF hypoxia was established later during adolescence (Figure S1F).

Consistent with hypoxic status, HIF-1α protein levels were significantly higher in S⁺P⁺ CFs compared to TIP, as assessed by flow cytometry and western blotting (WB) (Figures 1G and 1H). HIF-1 target genes were upregulated in S⁺P⁺ CFs (Figure 1I), a subset of which are known to promote glycolysis over OXPHOS (Semenza, 2014b). Among TIP subsets, S⁺P⁺ cells express higher levels of HIF-1α protein than ECs or S⁻P⁺ cells (Figure S1G), consistent with their relative hypoxia (Figure S1A). S⁺P⁺ cells preferentially expressed Hif-1a and Phd2 among genes encoding isoforms (Figures S1H and S1I) (Sousa Fialho et al., 2019). To confirm the hypoxic status of S⁺P⁺ cells using independent approaches, we compared our recently published single-cell RNA sequencing (scRNA-seq) data from purified Pdgfra-lineage⁺ CFs and isolated TIP cells from sham hearts (Farbehi et al., 2019) to a previously published HIF-1 target gene set (Guimarães-Camboa et al., 2015). Analysis revealed enrichment of known HIF-1 targets among genes upregulated in Pdgfra-lineage⁺ CFs relative to TIP cells in sham hearts (184/1,094; ~17% overlap; Fisher’s exact test p = 2.05 × 10⁻⁵). Among Pdgfra-lineage⁺ CF subsets previously defined by scRNA-seq (Farbehi et al., 2019), F-SH (fibroblast-Sca1^high, enriched in S⁺P⁺ cells) showed the strongest over-representation of HIF-1 targets (Fisher’s exact test p = 3.1 × 10⁻⁴). The transcription factor MEIS1 is a positive transcriptional regulator of Hif-1a expression (Simsek et al., 2010), and Meis1 transcripts were also upregulated in CFs compared to other TIP populations in sham hearts (MAST [Model-based Analysis of Single-cell Transcriptomics]; p_adjusted = 7.5 × 10⁻⁹⁰).

Cardiac mesenchymal progenitors have low mitochondrial mass

We and others have previously shown that cardiac S⁺P⁺ cells are enriched in fibroblasts with MSC-like properties (seen as cardiac colony-forming units-fibroblasts [cCFU-Fs]) (Chong et al., 2011; Soliman and Rossi, 2020). The long-term self-renewal and multi-lineage differentiation potential of cardiac MSCs in vitro and in vivo suggests a mesenchymal hierarchy whereby immature progenitors provide a proliferative reservoir for resident stromal and connective tissue cell types in homeostasis and disease. To assess the metabolic properties of cCFU-Fs, we fractionated S⁺P⁺ cells based on mitochondrial mass (Figure 1J) and assessed colony numbers and self-renewal properties in vitro. Mito^low and Mito^medium fractions produced a similar number of colonies at the first passage (passage 0, P0); however, only Mito^low cells showed long-term proliferation equivalent to total S⁺P⁺ cells (Figures 1K–1M). Mito^high cells did not form colonies, suggesting that they are more differentiated, and accordingly, they showed exaggerated fibrogenesis in vitro, and lacked adipogenic potential, in contrast to Mito^low and Mito^medium cells, which could engage both differentiation pathways (Figures S1J and S1K). Bulk RNA-seq on the different Mito cell fractions revealed a continuum of transcriptional states (Figures S1L and S1M). Gene Ontology (GO) analysis of the 355 differentially expressed genes (DEGs) between Mito^high and Mito^low (DESeq2; p_adj < 0.05; log2 fold-change > 0.5) showed that Mito^low CFs downregulate genes involved in protein biosynthesis, consistent with a progenitor-like state (Figures S1M and S1N; Table S1). Overall, the above results indicate that S⁺P⁺ cells are metabolically heterogeneous, with mesenchymal progenitors maintaining lower mitochondrial mass and lower protein biosynthetic activity.

We compared the qualities of CFs from atria and ventricles. Cardiac chambers have different biomechanical properties, and their fibroblasts show distinct growth, signaling, and fibrogenic behaviors in vitro (Burstein et al., 2008; Soliman and Rossi, 2020). We found a higher density of S⁺P⁺ CFs and a lower density of CD31⁺ ECs (relative to TIP) in atria compared to ventricles (Figure S1O). Hoechst dye uptake experiments (Parmar et al., 2007) suggested that atrial S⁺P⁺ CFs were overall less perfused (Figure S1P). They were also more hypoxic than ventricular S⁺P⁺ cells (Figure S1Q) and showed increased numbers of cCFU-Fs with an increased rate of self-renewal in vitro (Figures S1R and S1S). These data suggest that hypoxia is the preferred niche for cCFU-Fs with long-term proliferative capacity.

To assess the hypoxia status of cCFU-Fs in vivo, we determined their sensitivity to tirapazamine (TPZ), which is toxic to hypoxic cells (Parmar et al., 2007). TPZ treatment caused a decrease in the weight of the thymus, a severely hypoxic tissue, due to increased apoptosis (Figure S1T and data not shown) (Parmar et al., 2007). In heart, TPZ also led to a pronounced reduction in cCFU-F number and total cell yield after culture at P0, without any noticeable change in Pdgfra^nGFP/+ CF density, CD31⁺ EC numbers, or cell death in vivo (Figures S1U–S1W). There was, in fact, a slight compensatory increase in total S⁺P⁺ cells. Thus, TPZ selectively depletes cCFU-Fs, highlighting their relative hypoxic status.

Generation of CF-specific Hif-1a knockout (KO) mice

To explore the role of HIF-1α in CFs in homeostasis and disease, we generated CF-specific Hif-1a conditional KO (cKO) mice by breeding Pdgfra-MerCreMer (Pdgfra^MCM/+) Cre driver mice (Ding et al., 2013) with Hif-1a-floxed mice (Hif-1a^flox/−) (Ryan et al., 2000). In Hif-1a^fl/−;Pdgfra^MCM/+ progeny, Hif-1a was conditionally deleted in adult CFs after administration of tamoxifen (tam). We also introduced a Cre-dependent R26^tdTomato reporter allele (tdTomato), allowing lineage tracing and purification of CFs and derivative cells. Lineage tracing was precise, with ~95% and ~98% overlap of tdTomato with cell-surface PDGFRα (Figures S2A and S5D) or SCA1/PDGFRa expression (Figure S2B), respectively. Efficient Hif-1a disruption was confirmed by PCR, immunofluorescence (IF), and WB of fluorescence-activated cell sorting (FACS)-sorted tdTomato⁺ CFs from tam-treated adult cKO relative to control Hif-1a^+/+;Pdgfra^MCM/+ mice (hereafter wild type [WT]) (Figures 2A–2C).

Figure 2. Proliferation defects in Hif-1a-deficient CFs after MI.

(A) PCR products showing Hif-1a and Pdgfra^MCM alleles in tdTomato⁺ cells (n = 5).

(B) IF staining of tdTomato⁺ cells at P0. Scale bar, 50 mm.

(C) WB analysis of tdTomato⁺ cells at P0, cultured in indicated conditions.

(D) Stained colonies and colony numbers in S⁺tdTomato⁺ cells at P0 (n = 5). Scale bar, 1 cm.

(E) Cumulative cell number of S⁺tdTomato⁺ cells (n = 3).

(F) Schematic of the experimental design.

(G) Quantification of EdU⁺ cells in IF images shown in (H) (n = 4–8).

(H) F staining of heart sections post-MI and EdU injection. Arrows and arrowheads indicate tdTomato⁺EdU⁺ and tdTomato⁻EdU⁺ cells, respectively. Scale bar, 50 mm.

(I–L) Flow cytometry analysis of EdU⁺ staining (n = 3–5), Ki67⁺ staining (n = 4–5), CD31⁺ cells (n = 4–8), and total tdTomato⁺ cells (n = 4–8), respectively, in sham or MI WT and cKO hearts.

See also Figures S2 and S3 and Table S1.

Adult cKO mice were healthy, fertile, and indistinguishable from WT or HET (heterozygous) (Hif-1a^fl/+;Pdgfra^MCM/+) littermates. Furthermore, S⁺P⁺ cells from uninjured cKO hearts showed no overall changes in mitochondrial mass or glycolytic flux (Figure S2C). We performed scRNA-seq on isolated tdTomato⁺ CFs from uninjured cKO and HET hearts, with CF subpopulations assigned as described previously (Farbehi et al., 2019). There were no significant changes in population proportions using a permutation-based statistical test (Farbehi et al., 2019) (Figures S2D and S2E). However, bulk RNA-seq of tdTomato⁺ CFs revealed 562 DEGs between uninjured cKO versus HET mice (n = 2) (Figure S2F), and the majority (93%) of these genes were downregulated in cKO cells (p_adj < 0.05; log2 fold-change > 0.5; Figure S2G; Table S1). Among these, 100 out of 524 (19%) overlapped with the previously published cardiac HIF-1 target gene set (Fisher’s exact test p = 7.82 × 10⁻⁶) (Guimarães-Camboa et al., 2015). GO term analysis showed targets were involved in regulation of proliferation and other cellular functions (Figure S2I), and a STRING protein-protein interaction network also highlighted proliferative regulators, including c-JUN, CCND3, and β-CATENIN-1 (Figure S2H). Colony assays on isolated SCA1⁺tdTomato⁺ (CD31⁻CD45⁻) CFs showed a ~40% increase in cCFU-Fs in cKOs (Figure 2D), suggesting proliferative priming. However, long-term growth assays showed no difference in proliferation capacity in cKO and WT cCFU-Fs (Figure 2E). The few upregulated DEGs were related to inflammation, proteolysis, and stress pathways (Figure S2J). However, detailed immune cell profiling by flow cytometry and IF showed no differences between cKO and WT hearts (Figures S3A–S3K).

Loss of Hif-1a in CFs leads to increased fibroblast proliferation post-MI

To test whether CFs were primed for proliferation in vivo, we performed sham or MI surgery on WT and cKO mice and quantified EdU incorporation into tdTomato⁺ CFs (Figures 2F, S2K, and S2L). We confirmed no difference in infarct size between WT and cKO hearts at day 1 post-MI (data not shown). However, at day 3, tdTomato⁺ cKO CFs displayed a ~50% increase in EdU incorporation in infarct and border zones (Figures 2G and 2H) and an increase in the border zone only at day 7. Results were confirmed by flow cytometry (Figure 2I) and Ki67 staining (Figure 2J). There was no change in EdU incorporation in lineage-negative (tdTomato⁻) cells or ECs (Figures S2M and 2K). Proliferation was no longer evident at day 14 or day 28 (Figures S2N and S2O), time points well beyond the CF proliferative peak post-MI demonstrated previously (Tallquist and Molkentin, 2017). Notably, the increased CF proliferation in cKO hearts at early time points resulted in increased total CF numbers at day 14 (Figure 2L), which was also confirmed at day 7 using a different Cre reporter (mT/mG) (Figure S2P).

Sensitivity of CFs to activation in sham Hif-1a cKO hearts

We performed scRNA-seq on tdTomato⁺CD31⁻CD45⁻ CFs from dissected cKO and HET hearts at day 3 post-sham/MI (Figures 3A and 3B). In sham hearts, we found a significant increase in the early activated CF subpopulation, F-Act (fibroblast-activated) (Farbehi et al., 2019) (Figures 3A–3C), with the increase confirmed by qRT-PCR for F-Act signature genes (Figures 3D, S4A, and S4B). F-Act cells express the progenitor cell markers Pdgfra and Ly6a (Sca1), along with the activation marker Postn (encoding PERIOSTIN). They are present in low numbers in uninjured and sham WT hearts but expand significantly after MI and thus likely represent an early tier of fibroblast activation (Farbehi et al., 2019). The increase in F-Act cells appeared to occur at the expense of F-SH cells, which, as indicated above, overlap with S⁺P⁺ cells and possess progenitor properties. Notably, the reciprocal changes in F-SH and F-Act cells in sham cKO hearts were not seen in uninjured cKO hearts (Figures S2D and S2E). Furthermore, the increased prevalence of F-Act cells was not associated with an increase in CF proliferation markers (Figures 2F–2J) or the presence of an actively cycling CF subpopulation (F-Cyc) (fibroblast-cycling) (Figures 3A–3C), nor was there an increase in cCFU-Fs over and above that seen in uninjured cKO hearts (Figures 2D and 2E).

Figure 3. Hif-1a null CFs are primed for cell cycle entry after MI.

(A) UMAP (uniform manifold approximation and projection) plot of aggregate tdTomato⁺CD31⁻CD45⁻ single-cell data with identified subpopulations.

(B) UMAP plots according to condition.

(C) Percentages of cells in each population according to experimental condition. *p < 0.01.

(D) qRT-PCR of indicated genes in tdTomato⁺CD31⁻CD45⁻ cells from WT-sham mice. Values are normalized to the WT-healthy (uninjured) sample (n = 3).

(E) CytoTRACE analysis of scRNA-seq data.

(F) Top GO BP terms for genes downregulated in tdTomato⁺CD31⁻CD45⁻ cells from cKO-sham mice.

(G) Analysis of RNA velocity projected onto the HET UMAP plot.

(H) Top GO BP terms for genes downregulated in tdTomato⁺CD31⁻CD45⁻ cells from cKO-MI mice.

Also see Figure S4 and Table S2.

The reciprocal change in F-SH and F-Act cells suggests a precursor-product relationship. To probe this further we analyzed scRNA-seq data from sham hearts using cytoTRACE, a platform that estimates stem cell characteristics without prior knowledge of trajectory origin or intermediate states (Gulati et al., 2020), with higher cytoTRACE scores being a surrogate for more open chromatin and multi-lineage priming. We found that among resting CF populations, F-Act and F-SH had the highest scores (Figure 3E), suggesting that they are the most closely related and share progenitor properties. The other major CF subset in uninjured hearts, F-SL (fibroblast-Sca1^low; corresponding to S⁻P⁺), showed low cytoTRACE scores. We calculated DEG sets for each of the sham CF subpopulation relative to remaining cells and overlapped lists with Mito^low and Mito^high transcriptome data. Only F-SH and F-Act segregated with the Mito^low fraction, which, as shown above, are enriched in cCFU-Fs with long-term proliferation potential (Figures 1J–1M and S4C).

An additional finding from scRNA-seq data was that all major fibroblast subpopulations in sham, but not uninjured cKO hearts, showed significant downregulation of genes associated with protein synthesis, compared to sham HET hearts (Figures 3F, S4D, and S4E; Table S2). Sham surgery is a form of remote injury, and muscle and HSCs have been shown to enter a reversible activated (G-Alert) state in response to remote injury via systemic factors (Rodgers et al., 2017). However, lineage-purified CFs from sham WT and cKO mice did not show an increase in cell size, mitochondrial mass, or mTOR signaling, defining features of G-Alert (Figures S4F–S4H), nor did we see changes in G-Alert parameters, colony-forming ability, or expression of F-Act signature genes in skeletal muscle fibro-adipogenic progenitors isolated from sham WT and cKO mice (Figures S4I–S4M) (Contreras et al., 2020). Thus, the sham injury response may be local, whereby F-SH cells have a decreased threshold for conversion to F-Act cells, which nonetheless maintain a low protein biosynthetic state and do not progress to the proliferative or fibrotic states typically associated with MI.

Fibroblast heterogeneity and dynamics in MI Hif-1a cKO hearts

At day 3 post-MI, scRNA-seq showed that in cKO hearts, there was a significant increase in the F-CI (fibroblast-cycling intermediate) population, again at the expense of F-SH (Figures 3A–3C). F-CI is another form of activated CFs showing strong upregulation of protein synthesis genes compared to quiescent fibroblast populations and is distinct from F-Act (Farbehi et al., 2019). Unlike F-Act, F-CI is barely present in uninjured and sham hearts (Figures S2D and S2E) but significantly accumulates post-MI (Figures 3B and 3C). We previously proposed that F-CI represents pre-proliferative fibroblasts (Farbehi et al., 2019), and this is supported by the fact that both F-CI and F-Cyc undergo transcriptome-wide mRNA 3^′ end shortening, a defining feature of dividing cells (Patrick et al., 2020). RNA velocity analysis (La Manno et al., 2018) of scRNA-seq data suggested that F-CI cells generate and are generated from F-Cyc (Figure 3G). However, the RNA velocity profiles were grossly similar for HET and cKO CFs (data not shown). It is noteworthy that proliferating fibroblasts (F-Cyc) were also increased in MI cKO hearts, albeit nonsignificantly, which is not surprising given the relatively low sampling rate of cells transiting S/G₂/M. Overall, these data are consistent with increased EdU incorporation in CFs and total CF accumulation in cKO hearts.

GO-term analysis of DEGs between cKO and HET CF subsets at day 3 post-MI revealed that F-SH and F-Act in cKO showed downregulation of genes associated with diverse signaling processes (Figures 3H and S4D; Table S2). F-Act uniquely showed downregulation of genes encoding negative regulators of cell signaling, including for CF proliferation pathways involving mitogen-activated protein kinase (MAPK)-ERK1/2 and fibroblast growth factor (FGF) (Figures S4N and S4O), potentially contributing to proliferative priming. Pre-/post-proliferative (F-CI) and proliferating (F-Cyc) CFs also showed downregulation of HIF-1 targets (combined 218/1,990 [11%]; Fisher’s exact test p_adj < 0.001).

Hif-1a deficiency enhances post-MI fibrosis through increased CF numbers

Myofibroblasts are the predominant source of ECM that stabilizes the infarct zone (Frangogiannis, 2019b). In day 14 post-MI left ventricles, qRT-PCR analysis of genes encoding ECM components (COL1a1, COL3a1, and FN1), remodeling proteases (TIMP1, MMP2, and MMP9), and fibrosis mediators (ACTA2, CTGF/CCN2, TGFB1, and NOX4) confirmed their significant upregulation in cKO hearts, suggesting an enhanced fibrotic response (Figure 4A). Most tdTomato⁺ CFs at day 28 post-MI were myofibroblasts, as defined by their large size, intracellular expression of α-SMA, and deposition of COL I and COL VI in the extracellular space (Figures 4B, S5A, and S5B). Marked deposition of collagen-rich ECM was also visualized by multiphoton second harmonic generation in the scar region of WT and cKO hearts after MI (Figure S5C).

Figure 4. Hif-1a deficiency aggravates post-MI fibrosis.

(A) qRT-PCR of indicated genes in the left ventricles after sham/MI surgeries. Values are normalized to the WT-sham sample (n = 3).

(B) IF staining of heart sections at day 28 after surgeries. Scale bar, 50 µm.

(C) Phase contrast images of tdTomato⁺ cells with or without TGFb1 treatment. Scale bar, 50 µm.

(D) IF staining of tdTomato⁺ cells with or without TGFb1 treatment. Scale bar, 100 µm.

(E) Images of collagen matrices and quantification of area (n = 3). Scale bar, 1 cm.

Also see Figure S5.

We analyzed cell-surface PDGFRα levels, the downregulation of which indicates fibroblast differentiation (Farbehi et al., 2019; Tallquist and Molkentin, 2017). Live tdTomato⁺ cells were gated and the levels of tdTomato versus PDGFRα-APC (antibody) staining were plotted (Figure S5D). At day 14 post-MI, the percentage of PDGFRα-APC⁺ (Rα⁺) cells within the tdTomato⁺ fraction was reduced (quadrant 2, Q2; Figure S5D), whereas the percentage of Rα⁻ cells (quadrant 1, Q1) was increased, reflecting differentiation to myofibroblasts. However, there was no change in the relative proportion of Rα⁺ and Rα⁻ sub-fractions between WT and cKO hearts (Figure S5D), suggesting a self-limiting process. Indeed, when we reanalyzed ECM/fibrosis gene expression in FACS-isolated tdTomato⁺ cells (day 14), normalizing for cell number, there was no difference between WT and cKO hearts (Figure S5E). Furthermore, in 2D cultures of tdTomato⁺ CFs, transforming growth factor β1 (TGFβ1) increased myofibroblast-like cells expressing α-SMA⁺ stress fibers similarly in WT and KO cells (Figures 4C and 4D), and WT and cKO CFs developed identical contractile force in 3D collagen gel assays (Figure 4E).

Pro-fibrotic response of cKO CFs in working 3D engineered cardiac microtissues

We also used 3D engineered CM microtissues (‘‘cardiac bundles’’) to further explore the behavior of cKO CFs in vitro (Figure 5A; Video S1) (Li et al., 2020). In neonatal rat ventricular myocyte (NRVM) bundles, CMs align along the bundle length, while co-purified CFs primarily locate peripherally, where they deposit COL I (Jackman et al., 2016; Li et al., 2017, 2020). However, co-cultured tdTomato⁺ WT (+WT) and cKO (+cKO) CFs became distributed throughout the bundle (Figure 5B), with cKO CFs undergoing more significant expansion (Figures 5C and 5D), leading to enhanced COL I deposition centrally (Figures 5B and 5E), with no effect on CM area (Figure 5F). We optically mapped action potential propagation in cardiac bundles during 2-Hz point stimulation applied at one end of the bundle (Figures 5G and 5H) (Li et al., 2017). +cKO bundles exhibited significantly slower conduction velocity (CV), longer action potential duration (APD), and lower maximum capture rate (MCR) compared to +WT and NRVM-only bundles (Figures 5I–5K). Contractile force development and force-length relationships were also compromised (Figures 5L and 5M). The addition of adult WT CFs to NRVM considerably reduces CM active contractile force (Li et al., 2017); however, the force amplitude was further reduced in +cKO bundles (Figures 5M and 5N), and twitch kinetics were slower, as evidenced by longer rise and decay times (Figures 5O and 5P). Thus, the impact of cKO CFs in this simplified 3D CM microtissue (lacking vascular, immune, and other cardiac cell types) is consistent with our in vivo studies and supports the notion that pathological fibrosis driven by the proliferation of CFs has scalable negative cell non-autonomous effects on CM function.

Figure 5. Hif-1a null CFs significantly deteriorate electrical and mechanical functions of engineered cardiac bundles in vitro.

(A) Schematic showing fabrication of 3D engineered cardiac tissue bundles.

(B) IF staining of bundle cross sections. Scale bar, 50 µm.

(C and D) Total nuclei and tdTomato⁺ nuclei counts per bundle cross-section (n = 3, 4–7 bundles per group).

(E and F) Collagen I area and F-actin⁺ area per bundle cross-section (n = 3, 4–7 bundles per group).

(G) Isochrone maps of action potential propagation in response to 2-Hz point stimulation in bundles.

(H) Optical action potential traces from bundles stained with di-4 ANEPPS and electrically stimulated at 2Hz.

(I–K) Conduction velocity (CV), action potential duration (APD), and maximum capture rate (MCR) in bundles (n = 3, 10–11 bundles per group).

(L) Contractile force traces recorded from bundles electrically stimulated at 1 Hz by field electrodes at optimal tissue length.

(M) Active force curves as a function of tissue stretch in 4% increments (n = 3, 18–23 bundles per group).

(N) Maximum contractile forces recorded during 1-Hz pacing at optimal tissue length (n = 3, 18–23 bundles per group).

(O) Twitch rise time measured between 10% and 90% of peak amplitude at the resting length (n = 3, 18–23 bundles per group).

(P) Twitch decay time measured between 90% and 10% of peak amplitude at the resting length (n = 3, 18–23 bundles per group).

Also see Video S1.

Hif-1a suppresses ROS-induced proliferation of CFs post-MI

In response to hypoxia, Hif-1a regulates metabolism by ensuring efficient use of available oxygen and energy substrates while minimizing mitochondrial oxidative stress (Figure 6A). At day 1 post-MI, qRT-PCR analysis of purified tdTomato⁺ cells showed a significant reduction in expression of HIF-1 targets involved in glycolysis (Slc2a1 and Hk2), angiogenesis (Pgf, Vegfa, and Cxcl12), and redox homeostasis (Ldha, Pdk1, Bnip3, Cox4i2, and Lonp1) in cKO CFs (Figure 6B). However, whereas glycolytic flux was increased in CFs at day 2 post-MI, there was no difference between cKO and WT cells (Figure S6A). Furthermore, there were no differences in capillary density, adaptative CM hypertrophy, or CM cell death between genotypes post-MI (Figures S6B–S6F).

Figure 6. HIF-1α-mediated redox regulation controls proliferation of CFs after MI.

(A) Schematic showing regulation of mitochondrial metabolism by HIF-1α.

(B) qRT-PCR of indicated genes in tdTomato⁺ cells from cKO-MI hearts. Values are normalized to the WT-MI sample (n = 3).

(C) Flow cytometry analysis of DHE fluorescence in PDGFRα⁺ cells post-surgery. Values are normalized to vehicle-treated WT-sham mice (n = 3–5). Also see (I).

(D) Schematic of experimental design for MitoPQ treatment.

(E) Flow cytometry analysis of DHE fluorescence in PDGFRα⁺ cells from WT hearts after surgery and treatment with vehicle or MitoPQ. Values are normalized to vehicle-treated WT-sham mice (n = 3–5).

(F) Flow cytometry analysis of EdU⁺ staining in tdTomato⁺ fraction from WT hearts after surgery and vehicle or MitoPQ treatment (n = 4–6).

(G) Flow cytometry analysis of tdTomato⁺ cells from WT hearts after surgery and vehicle or MitoPQ treatment (n = 4–6).

(H) Schematic of experimental design for MitoT treatment.

(I) Flow cytometry analysis of DHE fluorescence in PDGFRα⁺ cells after surgery and treatment with vehicle or MitoT. Values are normalized to WT-sham mice (n = 3–4).

(J) Flow cytometry analysis of EdU⁺ staining in tdTomato⁺ fraction after surgery and vehicle or MitoT treatment (n = 3–6).

(K) Flow cytometry analysis of tdTomato⁺ cells after surgery and vehicle or MitoT treatment (n = 3–6).

(L) Dot immunoblot analysis of tdTomato⁺ cells after surgery and vehicle or MitoT treatment (n = 2, 3 technical replicates per group).

(M) Signal intensity quantification for pERK1/2 dot blot shown in (L). Values are normalized to WT-sham mice.

(N) Signal intensity quantification for pAKT dot blot shown in (L). Values are normalized to WT-sham mice

Also see Figure S6.

Given that the HIF-1α targets governing redox homeostasis downregulated in cKO hearts post-MI serve to reduce mitochondrial ROS, and knowing that physiological ROS can act as ‘‘go signals’’ for proliferation and differentiation of stem/progenitor cell proliferation (Schieber and Chandel, 2014), we hypothesized that increased ROS underlies the increased CF proliferation and fibrosis in cKO CFs. We fractionated S⁺P⁺ cells from uninjured hearts based on endogenous ROS levels and found that the ROS^high fraction yielded greater cCFU-F numbers, colony size, and total cell yield (Figures S6G–S6J), consistent with ROS acting as positive regulators of CF proliferative potential. Furthermore, at day 2 post-MI, WT CFs exhibited a substantial increase in ROS levels compared to sham levels, and in cKO CFs, there was a further marked increase (Figure 6C). There was no difference in ROS levels between WT and cKO ECs (Figure S6K) or in purified tdTomato⁺ CF lineage cells at day 7 post-MI (data not shown). Whereas CFs mounted a cytoprotective anti-oxidative stress response driven by the redox-sensitive transcription factor NRF2 (nuclear factor erythroid-2-related factor 2) after MI, this was no greater in cKO CFs (Figure S6L). Findings in CFs were replicated using an alternative ROS dye, CM-H2DCFDA (DCF), and a mitochondria-targeted version of dihydroethidium (hydroethidine) (DHE) termed MitoSOX, suggesting that the ROS increase in cKO CFs occurred in mitochondria (Figures S6M and S6N).

We treated WT mice with the mitochondria-targeted oxidant MitoParaquat (MitoPQ), which was previously reported to elevate mitochondrial ROS in cell lines and isolated mitochondria without affecting mitochondrial function (Robb et al., 2015) (Figure 6D). At day 2 post-MI, there was a modest (~25%) increase in ROS in CFs from MitoPQ-treated versus vehicle-treated mice, with no change in treated sham hearts, which must buffer oxidants appropriately (Figure 6E). Consistent with mitochondrial ROS being a key driver of CF proliferation in vivo, there was a substantial increase in EdU incorporation into tdTomato⁺ CFs at day 3 and total tdTomato⁺ CF numbers at day 14 in treated hearts (Figures 6F and 6G).

We next treated WT and cKO mice with the mitochondria-targeting antioxidant, MitoTEMPO (MitoT), a mitochondrial superoxide dismutase mimetic and peroxyl scavenger (Zielonka et al., 2017) (Figure 6H). The excessive ROS seen in cKO CFs at day 2 was reduced by MitoT treatment to a level seen in treated or untreated WT-MI hearts (Figure 6I). This was accompanied by a pronounced reduction in EdU incorporation and total CF number in cKO hearts (Figures 6J and 6K), without affecting ECs (Figure S6O).

Previous studies have suggested that ROS can activate phosphatidylinositol 3-kinase (PI3K)/AKT/mTOR and MAPK pathways through reversible inactivation of PTEN and ERK1/2-directed phosphatases (Zhang et al., 2016). Indeed, treatment of cultured WT CFs with hydrogen peroxide (H₂O₂) led to a significant increase in active serine-473-phosphorylated AKT (pAKT), which was blunted by pretreatment with antioxidant N-acetyl-L-cysteine (NAC) (Figure S6P). Dot immunoblot analysis of freshly isolated tdTomato⁺ CFs revealed increased pAKT and activated phosphorylated ERK1/2 (pERK1/2) at day 2 post-MI relative to sham CFs, which was further enhanced in cKO CFs post-MI (Figures 6L–6N). MitoT treatment decreased pERK1/2 and pAKT in cKO-MI CF to levels seen in treated WT-MI hearts. These data provide a mechanistic link between increased ROS and AKT and ERK1/2 signaling pathways and their role in CF proliferation.

MitoT attenuates cardiac fibrosis and improves ventricular function in Hif-1a cKO hearts post-MI

To evaluate the functional effects of increased scar formation, we subjected WT and cKO mice to sham/MI surgery and analyzed scar volume and cardiac function using serial transthoracic echocardiography. Kaplan-Meier analysis indicated that there was no significant difference in cohort survival (Figure S7A). However, in cKO-MI hearts at day 28, there was a significant proportional increase in scar volume and decrease in viable myocardial volume, as shown using micro-computed tomography (micro-CT) and a semi-automated segmentation (Figures 7A–7E; Video S2; STAR Methods). Notably, whereas there was an absolute increase in scar volume (p = 0.0022), viable myocardial volume remained constant (p = 0.99), which, in the absence of CM hypertrophy (Figure S6E) or changes in inflammation (Figure S3), suggests that the adverse chamber remodeling in cKO-MI hearts is exclusively the result of CF/ECM expansion or a change in ECM properties. MitoT treatment significantly rescued these changes (Figures 7D and 7E).

Figure 7. MitoT treatment rescues cardiac anatomical and functional defects in cKO post-MI.

(A) Micro-CT analysis of hearts after surgery. Myocardium and scar are shown in red and blue, respectively. Scale bar, 1 mm.

(B) The area enclosed by the white box in (A) is shown at higher magnification depicting semi-automated segmentation of myocardium (red line) and scar (blue line). Scale bar, 0.5 mm.

(C) Pixel intensity profile along the yellow lines shown in (B). Red and blue lines represent the average intensity value of the whole myocardium and the whole scar, respectively. Corresponding position of green dot in (B) is shown on the graph.

(D) Micro-CT analysis of hearts of WT and cKO mice after surgery and vehicle or MitoT treatment. Scale bar, 1 mm.

(E) Volumetric quantification of myocardial or scar tissue shown in (D) (n = 5–6).

(F) Quantification of (i) fractional area change (FAC), (ii) left ventricular end-diastolic volume (LVEDV), (iii) left ventricular end-systolic volume (LVESV), (iv) left ventricular stroke volume (LVSV), (v) left ventricular ejection fraction (LVEF), (vi) cardiac output (CO)/body weight (BW), (vii) heart weight (HW) and BW, and (viii) HW/BW ratio of WT and cKO mice after surgery and vehicle or MitoT treatment. Values are normalized to untreated WT-sham mice (n = 10–13).

Also see Figure S7 and Video S2.

At days 7 and 28 post-MI, there was a significant worsening of function in cKO compared to WT-MI hearts, including decreased fractional area change, left ventricular (LV) ejection fraction, stroke volume, and cardiac output and increased LV end-systolic and diastolic volumes (Figure 7F, i–vi; and Figures S7B and S7C). cKO-MI mice also had a significantly higher heart weight (HW)/body weight (BW) ratio (Figure 7F, vii and viii; and Figure S7B), reflecting the CF/ECM expansion. Strikingly, MitoT treatment reversed these adverse cardiac functional parameters and HW/BW ratio to levels seen in MitoT-treated or untreated WT-MI hearts (Figures 7F, S7B, and S7C).

DISCUSSION

In contrast to CMs, which become highly oxidative in the transition to air breathing (Puente et al., 2014), S⁺P⁺ cells and mesenchymal progenitors acquire a more hypoxic niche postnatally, a state maintained through to adulthood. How this hypoxic niche is established in the highly vascularized postnatal myocardial environment is unknown, although the unique cardiac ECM, which develops postnatally (Del Monte-Nieto et al., 2020), may limit oxygen diffusion, or oxidative CMs may constitute an oxygen sink.

Uninjured Hif-1a cKO mice appeared normal despite downregulation of HIF-1α targets in CFs, and they showed no change in CF proportions, mitochondrial mass, or glycolytic flux, demonstrating that these metabolic states are not set by HIF-1α alone, as suggested for HSCs (Ito and Suda, 2014; Liu et al., 2015). Metabolic parameters may be set developmentally, or HIF-2 may be involved. However, the increase in cCFU-Fs in uninjured cKO CFs suggest more subtle metabolic and/or epigenetic shifts that prime them for proliferation.

In sham conditions, which can be regarded as a remote injury, Hif-1a deletion reduced the threshold for activation of F-SH to F-Act, the latter present in low numbers in uninjured hearts. However, F-Act increases to 30%–50% of total CFs at day 3 post-MI (Figure 3C), suggesting a major role in cardiac repair (Farbehi et al., 2019). In the sham cKO context, F-Act cells did not progress to higher metabolic, proliferative, or fibrotic states, which occurs only after injury. How the threshold is altered by Hif-1a deletion is not known. However, our findings suggest a sentinel function for the F-SH subfraction, whereby they respond to physiological stimuli and acquire a poised activated state, as shown for other stem cells (Rodgers et al., 2017).

After MI, cKO hearts showed increased fibrosis, leading to substantially increased absolute scar volume and more severe cardiac systolic dysfunction. Our results suggest that the increased CF accumulation and ECM deposition do not invoke feed-forward amplification of fibrosis, such as via ECM-CF mechanosensing (Herrera et al., 2018) or ROS-mediated TGFβ1 activation (Liu and Desai, 2015), perhaps because HIF-1α has its major impact on the initiation rather than progression phase of fibrosis. However, as seen also in 3D engineered microtissues, the increased CF numbers and ECM deposition were associated with a significantly diminished CM contractile function, likely through altered CM-ECM interactions and/or CM-CF signaling cross-talk.

Our findings suggest that the increase in mitochondrial ROS in cKO CFs post-MI leads to increased CF proliferation. An increase in ROS in other stem/progenitor populations has been shown to be essential for tissue homeostasis and regeneration (Gauron et al., 2013; Hamanaka et al., 2013; Han et al., 2014; Le Belle et al., 2011; Love et al., 2013; Morimoto et al., 2013). ROS in this context likely acts within physiological parameters as essential second messengers for pathways regulating stem cell activation and proliferation (Schieber and Chandel, 2014; Sun et al., 2020). The increase in ROS in cKO CFs post-MI occurred without a change in the NRF2-dependent antioxidant response, suggesting that cKO CFs did not experience increased oxidative stress.

Our studies using mitochondria-targeted redox reagents confirm that the increased ROS drives pathological CF proliferation in the cKO model and suggest that ROS buildup occurs principally in mitochondria. Interestingly, there was no impact of MitoT on ROS levels in WT CFs after MI, perhaps because ROS accumulation in WT CFs occurs mainly outside of mitochondria, including at the membrane due to activation of NADPH oxidases via TGFβ (Cucoranu et al., 2005; Liu and Desai, 2015) or outside-in integrin signaling (Kong et al., 2018), or is acquired as a result of an inflammatory environment. Our model is that in the context of myocardial ischemia, ROS increases in cKO CFs due to compromised HIF-1α-dependent adaptive pathways regulating mitochondrial metabolism and contributes to proliferative fibrosis via AKT and ERK activation. It is likely that the cardiac functional decline in WT-MI hearts is due mostly to the impact of oxidative damage in CMs, noting that the dose and duration of antioxidant treatment used here in Hif-1a cKO mice are far lower than required to reverse oxidative stress in CMs (Puente et al., 2014). However, our data show that HIF-1α function in CFs alone provides a critical braking mechanism that protects the heart against post-ischemic CF activation and proliferation driven by ROS buildup and physiological signaling. HIF-1α-driven injury-related adaptive responses have also been described in developing and adult CMs and ECs (Guimarães-Camboa et al., 2015; Semenza, 2014a; Sousa Fialho et al., 2019; Wei et al., 2012); however, in these settings, loss of HIF-1α represses proliferation and adaptive responses. By contrast, loss of Hif-1a in CFs is more akin to responses in stem cells, which leads to excessive proliferation and exhaustion. In the heart, uncontrolled CF proliferation leads to excessive fibrosis and functional decline. HIF-1α likely performs a protective bridging function in the early temporal window between the development of oxidative stress in CFs post-MI and their ability to mount robust antioxidant responses via NRF2 and other pathways (Finkel, 2011).

Oxidative stress is a component in virtually all CV diseases, including heart failure. Whereas established frontline drugs have been found subsequently to reduce ROS, clinical trials of dietary antioxidants have failed to produce evidence of long-term benefits (Schmidt et al., 2015). Nonetheless, redox control in CV disease remains a promising frontier, potentially requiring precision drugs targeting specific molecules, cell types, or intracellular compartments. Our study suggests that ROS in CFs will be an essential target, especially in the context of human Hif-1a genetic variants (Semenza, 2014a) or chronic CV disease, where HIF-1α levels diminish (Sousa Fialho et al., 2019). Our study offers important new insights into disturbed cardiac homeostasis and metabolism in a genetic model and the complex cell dynamics underpinning cardiac fibrosis at the single-cell level. A deeper understanding of these cellular and molecular processes will be critical to identifying new therapeutic targets in CV disease.

Limitations of the study

Our study has not directly measured cellular compartmentalization of ROS or the molecular species involved. Dyes that measure hypoxia and ROS, while widely used, may not accurately measure differences between cell types. Likewise, reagents used to alter redox balance in CFs may also have unanticipated consequences elsewhere. We do not know which local or systemic factors induce activation of F-SH to F-Act after sham injury or why the threshold for activation is altered by Hif-1a deletion. Although we have identified AKT and ERK signaling as mediators of the proliferative response to increased mitochondrial ROS in cKO CFs post-MI, other pathways are likely to be involved. Finally, our findings are exclusively based on mouse data and may not fully reflect responses to reduce HIF-1α function in CFs in human hearts.

STAR★METHODS

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Richard P. Harvey (r.harvey@victorchang.edu.au).

Materials availability

Transgenic mouse lines used in this study are available upon request from the lead contact, Richard P. Harvey (r.harvey@victorchang.edu.au), with a completed Materials Transfer Agreement.

Data and code availability

Sequencing data have been deposited in the ArrayExpress database at EMBL-EBI (https://www.ebi.ac.uk/arrayexpress) under accession codes E-MTAB-9675 (Mito fractions), E-MTAB-9674 (Hif-1a cKO RNA-seq) and E-MTAB-9583 (Hif-1a cKO scRNA-seq). This paper does not report original code. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Ethics Statement

All experimental procedures were approved by the Garvan Institute/St. Vincent’s Hospital Animal Experimentation Ethics Committee (No. 13/02, 16/10, 19/14) and performed in strict accordance with the National Health and Medical Research Council (NHMRC) of Australia Guidelines on Animal Experimentation. All efforts were made to minimize suffering.

Mice

Mice were bred and housed in the BioCORE facility of the Victor Chang Cardiac Institute. Rooms were temperature and light/dark cycle controlled, and standard food was provided ad libitum. Hif-1a^+/−;Pdgfra^MCM/+ males were bred with Hif-1a^flox/+ females, to generate: Hifa1a^+/+;Pdgfra^MCM/+ (WT), Hif1a^flox/+;Pdgfra^MCM/+ (HET) and Hif-1a^Δflox/−;Pdgfra^MCM/+ (cKO). For irreversible labeling and fate mapping of PDGFRα⁺ cells, males were mated with either Hif-1a^flox/+;tdTomato^+/+ or Hif-1a^flox/+;mT/mG^+/+ females as indicated. Hif-1a deletion and reporter activation was induced by 3 intraperitoneal injections (i.p.) of tamoxifen on consecutive days. Both male and female mice aged 8–12 weeks were used for the experiments, and each experiment was carried out with sex- and age-matched controls. Sham or MI surgeries were performed 1 or 2 weeks after the tamoxifen treatment, and hearts were subjected to cells or tissue analysis at indicated time-points post-surgery. Experiments involving mice were performed with the operator blinded to genotype.

METHOD DETAILS

Drug administration

• Tamoxifen was dissolved in 90% peanut oil+10% ethanol, aliquoted and stored at −20 °C. Adult mice were i.p. injected @ 100 mg/kg body weight. All downstream treatments and/or experiments were performed 1–2 week after the last dose of tamoxifen.

• Pimonidazole hydrochloride (Hypoxyprobe) was dissolved in PBS and adult mice were i.p. injected @ 120 mg/kg body weight and were sacrificed after 2 hr.

• Tirapazamine (1,2,4-benzotriazine-3-amine 1,4-dioxide) was dissolved in sterile PBS and adult mice were i.p. injected once a day for 3 days @ 30 mg/kg body weight and were sacrificed 24 hr after the last dose.

• Ethynyl-2^′-deoxyuridine (EdU) was dissolved in PBS and adult mice were i.p. injected on the indicated day after MI surgery @ 10 µg EdU/g body weight and were sacrificed after 24 hr. Cells or tissues were isolated and processed as described.

• Mito-Paraquat (MitoPQ) stock was made in ethanol. Stock was diluted in PBS and adult mice were i.p. injected with 10 µM once a day for 8 days starting from the day of MI surgery @ 0.1 µg/kg body weight.

• MitoTEMPO (MitoT) was dissolved in PBS, and adult mice were i.p. injected with 245 mM once a day for 8 days starting from the day of MI surgery.

• Verapamil was dissolved in PBS and adult mice were i.p. injected @ 120 µg/mouse.

• Hoechst 33342 dye was i.v. injected via tail vein @ 1.5 mg/mouse under isoflurane anesthesia for 1 min prior to sacrifice.

Genotyping

Genomic DNA was extracted from ear punch biopsies for genotyping, and PCR amplification assay was performed using REDEx-tract-N-Amp Tissue PCR Kit (Sigma-Aldrich) and appropriate primer sets for each mouse line (Table S3).

Myocardial infarction (MI)

Adult mice were randomized, anaesthetised (ketamine: 100 mg/kg, xylazine: 20 mg/kg) by i.p. injection, placed on a heating pad, intubated and ventilated (120 breaths/min). Depth of anesthesia was monitored by the color of mucous membranes and skin, and toe pinch reflex. The heart was accessed by an incision of the left wall of the chest at the fourth intercostal space. Mice were subjected to MI by permanent ligation of the left anterior descending coronary artery (8/0 Prolene suture) at about 2 mm below the edge of the left auricle. The chest was closed, and the pneumothorax reduced. Sham surgery was performed that was identical, but without ligation of the coronary artery. Mice were monitored continuously during recovery until the righting reflex was regained and housed overnight half on a heating pad. Post-operative analgesia (buprenorphine, 0.075 mg/kg) was administered subcutaneously twice daily for first three days post-surgery.

Transthoracic echocardiography

Echocardiography was performed using Vevo 3100 ultrasound imaging system (FUJIFILM VisualSonics). Animals were anesthetized with 4% isoflurane in oxygen and placed in a supine position on a temperature-controlled imaging platform under a continuous supply of 1.5% isoflurane. After shaving the chest and washing with 70% ethanol, ultrasound gel was applied to the chest wall and B-mode images were taken of LV long axis (LAX) followed by four short-axis (SAX) views at the mid-papillary level, and then at the levels 1, 2 and 3 mm toward the apex. LV length was measured from the apical dimple to the base of the aortic valve leaflets from the LAX images at end-diastole and end-systole. Each measurement reported was an average of ten cardiac cycles. Left ventricular (LV) end-diastolic diameter (LVIDED) and LV end-systolic diameter (LVIDES) were obtained from M-mode tracings from parasternal short-axis views at the mid and apical levels at each time point. Formulae: LVEF = [(LVEDV-LVESV)/LVEDV] × 100%; LVSV = (LVEDV – LVESV); FAC = (EDA-ESA)/EDA, CO/BW = [(LVSV x HR/1000)]/BW (LVSV, left ventricular stroke volume; FAC, fractional area change; EDA, end-systolic area; ESA, LV end-systolic area; CO/BW, cardiac output/ body weight; HR, heart rate).

Cell isolation

Total interstitial cell population (TIP) were isolated as previously described (Chong et al., 2011). Briefly, hearts were minced and incubated in collagenase type II (Worthington) at 37°C before filtering through 40 µm strainers. Cells were resuspended in red cell lysis buffer, followed by dead cell removal, immunostaining for 15 min on ice with indicated fluorophore-conjugated antibodies and two times wash with FACS buffer (1× DPBS, 2% FBS, 2 mM EDTA pH 7.9) before the acquisition.

FACS and flow cytometry

We employed stringent gating strategies to exclude doublets in the FACS analysis. FSC-H versus FSC-A, FSC-H versus FSC-W and SSC-H versus SSC-W cytograms were used to discriminate and gate out doublets/cell aggregates during sorting or analysis. DAPI was used to distinguish live (DAPI⁻) and dead (DAPI⁺) cells. For each sample, at least 10,000 final gate events were collected and stored for the later analyses. All FACS or flow cytometry experiments were performed on BD FACSAria III or Canto II (BD Biosciences), respectively using the FACSDiva software (BD Biosciences). FACS analysis was performed using the FlowJo analysis program (Treestar). Gating parameters were kept identical between samples whenever comparisons were made.

Cell culture and colony formation (CFU-F) assay

FACS sorted SCA1⁺PDGFRα⁺CD31⁻ (S⁺P⁺) or SCA1⁺tdTomato⁺CD31⁻ (S⁺tdTom⁺) cells were seeded at a clonogenic density of 50 cells/cm² (500 cells per well of 6-well plate) and were cultured in α-Minimal Essential Medium (α-MEM) containing 20% FBS+1% Pen/Strep in H35 Hypoxystation (Don Whitley Scientific, United Kingdom) at 1% O₂, 5% CO₂, nitrogen balance, 37°C with humidification, with medium changes every 2–3 days. Medium was pre-equilibrated in Hypoxystation every time before medium-change until the end of the assay/culture. Normoxic cells were maintained in a nitrogen-controlled cell-culture incubator (Thermo Fisher Scientific). After 8-day culture (passage 0, P0), colonies were rinsed with PBS, fixed with 2% paraformaldehyde (PFA) and stained with 0.05% (v/v) crystal violet dye in water. Averages obtained from a minimum of 3–5 technical replicates/sample were used for all the quantitative analysis.

Histology, immunostaining, confocal microscopy and quantification

All samples were processed in parallel under identical conditions for comparisons. Hearts were arrested at diastole by injection of potassium chloride. Tissues were harvested and immediately immersed in fixation buffer (4% PFA in PBS, pH 7.4). Following overnight fixation at 4°C, hearts were immersed in 30% sucrose in PBS for 16 hr, embedded in tissue-embedding medium (O.C.T.) (Tissue-Tek, Sakura) and stored at −80°C. Transverse sections were cut on a Leica CM1950 clinical cryostat (Leica Biosystems) at 8 mm-thick and collected onto SuperFrost PLUS charged glass slides (Thermo Scientific). For immunofluorescence (IF) staining of tissue sections (or cells), blocking and permeabilization was achieved by incubation of the sections in 5% BSA and 0.1% Triton X-100 in PBS for 1 hr at RT. Primary antibodies were diluted in blocking/permeabilization solution and incubated overnight at 4°C. Sections were washed three times (10 min at RT each) with PBS and counterstained with suitable species-specific secondary antibodies coupled to Alexa Fluor dyes (1:500, all from Invitrogen) and/or directly conjugated fluorophores (WGA or IB4) for 1 hr at RT in blocking/permeabilization buffer. Sections were washed three times, counterstained with DAPI for 5 minutes and mounted with MOWIOL mounting medium. IF staining of cells was performed using same protocol. Images were acquired using a Zeiss LSM780 confocal laser scanning microscope. Settings for confocal scanner detection and laser excitation were kept identical between samples whenever comparisons were made.

Co-acquired Second Harmonic Generation (SHG) and confocal microscopy

Fibrillar collagen was imaged within the scar region of hearts taken from mice 28 days post-MI, prepared as above. Two-photon excitation microscopy was used to acquire SHG images with a Zeiss LSM 880 microscope using a Mai Tai Insight DeepSee 2Ptunable laser, tuned to 840nm operating at 80MHz repetition rate and Plan-Apochromat 20.0 × 0.8 N.A. Air objective (Zeiss-GmbH, Germany). Forward propagated SHG fibrillar collagen signal was collected in the GaAsP PMT Non-Descanned Detector with a 390–440nm bandpass filter. Samples were sequentially excited by two laser lines (DPSS: 561nm; HeNe: 633nm) was and fluorescent signal was collected using internal PMT detectors for the acquisition of TdTomato lineage-labeled cells and TO-PRO3 iodide-labeled nuclei.

Quantification of confocal images

Automated segmentation and quantification of confocal images for nuclei, tdTomato⁺ cells, immune cells, EdU incorporation, cardiomyocytes size and vessel calibre were performed using CellProfiler version 3.0. Briefly, images were pre-processed using ImageJ by partial smoothing, contrast enhancement and the image brightness was corrected. Hoechst⁺ and EdU⁺ nuclei, IB4⁺ capillaries, tdTomato⁺ cells and CD45⁺ immune cells were segmented using an adaptive, two-class Otsu thresholding method, filtering by size and brightness. While WGA-labeled CM images were inverted and the inner CM area (minus cell border) was quantified, using an adaptive, three-class Otsu threshold and filtered stringently by object brightness. For all the quantitative analysis based on confocal microscopy images, averages obtained from a minimum of 2–6 comparable fields of view per sample were used.

TUNEL staining

According to the manufacturer’s instructions, apoptosis in heart sections was detected by In Situ Cell Death Detection Kit, TMR red (Sigma-Aldrich). DNase-treated sections served as positive control while sections incubated with reaction mix without labeled nucleotides served as negative control. Sections were counterstained with WGA and DAPI to visualize cell boundary and the nucleus, respectively.

EdU incorporation assay

Adult mice were i.p. injected with EdU 24 hr prior to killing and cells or tissues were isolated and processed as described. EdU incorporation assay was performed with the Click-iT EdU Assay Kits (Thermo Fisher Scientific) according to the manufacturer’s instructions.

Measurement of intracellular lactate

Intracellular lactate (L-Lactate) concentration was measured in lysates prepared from FACS sorted cells using L-Lactate Assay kit (Abcam) according to the manufacturer’s instruction.

Measurement of mitochondrial mass

TIP cells were isolated as described and stained with indicated antibodies. After washing with FACS buffer, cells were then incubated for 15 min with 200 nM MitoTracker Green FM (Thermo Fisher Scientific). Cells were washed twice with PBS, resuspended in PBS and kept on ice till acquisition on Canto II. At least 10000 live cell counts of the population of interest were recorded for analysis using FlowJo.

PCR based determination of mitochondrial DNA copy number

Total DNA was isolated from FACS sorted cells and relative mitochondrial DNA content was measured with qPCR targeting mitochondrial gene, mtCo-1 and normalization against genomic gene, Rn18s using the primers enlisted in Table S3.

Measurement of mitochondrial immaturity

TIP cells were isolated as described and stained with indicated antibodies. After washing with PBS, cells were then incubated for 15 min with 200 nM TMRM (Thermo Fisher Scientific). Cells were washed twice with FACS buffer, resuspended in PBS and kept on ice till acquisition on Canto II. At least 10000 live cell counts of the population of interest were recorded for analysis using FlowJo.

Measurement of cellular respiration

The OCR in intact FACS sorted cells was determined using a respirometer (OROBOROS Oxygraph-2k) according to the manufacturer’s protocol. The rate of oxygen consumed was recorded, and respiration rates were normalized to cell number.

Measurement of ATP levels

ATP levels in FACS sorted cells were quantified using ATP Bioluminescence Assay Kit HS II (Sigma-Aldrich) in accordance with the manufacturer’s recommendations.

Glycolytic flux assay

The fluorescently labeled glucose analog, 2-(N-(7-nitrobenz-2-oxa-1, 3-diazol-4-yl) amino)-2-deoxyglucose (2-NBDG, Thermo Fisher Scientific), was used to measure glucose transport in cells. TIP cells were isolated, stained with indicated antibodies and kept on ice. Cells were then incubated at 37°C for 15 minutes with 15 µM 2-NBDG, washed once with ice-cold PBS, resuspended in PBS and kept on ice till acquisition.

Analysis of intracellular ROS levels

TIP cells were isolated, stained with indicated antibodies and kept on ice. Cells were incubated with 1 mM Dihydroethidium (DHE) or 1 mM 5-(and-6)-carboxy-2^′,7^′-dichlorofluorescein diacetate (DCF) or 5 µM MitoSOX for 30 min at 37°C (all dyes-Thermo Fisher Scientific). Fluorescence of oxidized DHE or DCF or MitoSOX was determined flow cytometrically using CantoII. For DHE and MitoSOX assays, PDGFRA-APC⁺ (antibody) cells were analyzed and for DCF assays, tdTomato⁺ cells were analyzed.

Flow cytometric profiling based on mitochondrial mass and ROS

Flow cytometric profiling and separation of S⁺P⁺ cells based on their mitochondrial content was performed with MitoTracker green-FM. Flow cytometric separation and colony assays of high, mid and low MP cells were carried out by gating and sorting 15%–20% cells in each gate and an equivalent number of total S⁺P⁺ cells. Similarly, flow cytometric profiling and separation of S⁺P⁺ cells based on ROS levels was performed with DHE or DCF in independent experiments.

Measurement of Hoechst perfusion in vivo

Adult Pdgfra^nGFP/+ mice were injected with Verapamil, and Hoechst 33342 was used to determine tissue perfusion level based on Hoechst fluorescence intensity following systemic administration, as described previously (Parmar et al., 2007). Cell isolation and flow cytometry was performed as described.

Flow cytometric profiling of immune cells

Flow cytometric profiling of immune cells based on the indicated cell surface markers was performed using Canto II. The protocol, gating strategy and classification of immune cells were adapted (Nahrendorf et al., 2007).

Collagen gel contraction assay

WT and KO CFs were isolated, and FACS sorted as described and cultured for 1 passage (P0). At P0, cells were trypsinized and mixed equal numbers with 400 µL of neutralized rat-tail type I collagen (Corning). Suspension was loaded into each well of a 24-well plate and plate was incubated at 37°C for 1 hr to allow collagen polymerization. To initiate gel contraction, the gels were floated with 500 µL 500 µL α-MEM containing 1% FBS in the absence or presence of TGFβ1 (50 ng/mL). The plate was incubated at 37°C for 48 hr. Gel contraction was assessed through the measurement of the gel surface area every day.

Western and dot blot Analysis

For WB, cells were lysed in ice-cold modified RIPA buffer (10 mM Tris, pH 7.4; 150 mM NaCl; 1 mM EDTA; 1 mM EGTA; 1% NP-40; 10% glycerol; 0.1% SDS; 0.5% sodium deoxycholate; 1 mM DTT) containing EDTA-free Complete Protease Inhibitor cocktail and PhosSTOP (Sigma Aldrich). Lysates were cleared by centrifugation and resolved on NuPAGE Bis-Tris 4%–12% gels (Thermo Fisher Scientific) followed by transfer onto a PVDF membrane (Bio-Rad). After incubation with primary and appropriate secondary antibodies, the blots were developed by chemiluminescence using SuperSignal West Pico PLUS Chemiluminescent Substrate (Thermo Fisher Scientific) and scanned using a Chemidoc MP imaging system (Bio-Rad).

For dot blot analysis, cells were FACS sorted and lysed directly in hydroxyurea lysis buffer (6.7 M Urea; 10% glycerol; 10 mM Tris, pH 6.8; 1% SDS; 1 mM DTT) containing EDTA-free Complete Protease Inhibitor cocktail and PhosSTOP. Cell lysates with an equal amount of proteins were applied directly on a nitrocellulose membrane (Bio-Rad). The membrane was left at room temperature for 30 mins to dry membrane completely. The blots were processed and developed as described above. Images were processed using Image lab software (Bio-Rad), and densitometry was performed with ImageJ software (NIH).

For HIF-1α WB, lysates of freshly sorted cells were loaded on a 4%–15% GTX Stain-Free acrylamide gels (Bio-Rad) in denaturing conditions with protein ladder. The Stain-Free gel was activated using Chemidoc Stain-Free gel activation protocol (Bio-Rad). Proteins were transferred to a nitrocellulose membrane (Trans-Blot Turbo Transfer Pack, Bio-Rad) using the Trans-blot Turbo Transfer system (Bio-Rad). Transfer of protein was confirmed by imaging the membrane using the Stain-Free blot protocol of the Chemidoc. Rabbit-anti HIF-1α antibody (Novus Biologicals) was used to detect HIF-1α protein using WesternBreeze Chemiluminescent detection kit (Thermo Fisher Scientific) according to the manufacturer’s instructions. Quantification of band intensity was performed by normalizing bands to total protein in each lane.

RNA isolation and real-time quantitative RT-PCR (qRT-PCR)

Cells or tissues were washed with PBS, and total RNA was isolated using TRIzol (Thermo Fisher Scientific) or QIAzol Reagent (QIAGEN) following the manufacturer’s instructions. RNA was resuspended in nuclease free water and RNA concentration and purity determined spectrophotometrically. 1 ug RNA were used to prepare cDNA using the Quantitect reverse transcription kit (QIAGEN) followed by real-time PCR using SYBR Green Supermix (Bio-Rad). The PCR protocol on a BioRad CFX96 Real-Time Detection System was as follows: an initial denaturation for 5 min at 95°C, followed by 35 cycles of 10 s denaturation at 95°C, annealing for 10 s at 60°C and extension for 15 s at 72°C. Melting curve analyses and sequencing of the amplification products were performed to verify the specificity of the amplification. The threshold cycle (Ct) was determined, and the relative quantitative expression of mRNAs was calculated using DDCT method and normalized to Hprt as an internal control. Primers were synthesized by IDT (Integrated DNA Technologies) and are listed in Table S3.

FACS and assays on skeletal muscle fibro-adipogenic progenitors

One-step digestion of skeletal muscle tissue for fibro-adipogenic progenitor isolation was performed as described before with few modifications (Contreras et al., 2020). Briefly, skeletal muscles from both hindlimbs of female mice were carefully dissected, washed with ice-cold DMEM, and cut into small pieces with blades until a homogeneous, paste-like slurry was formed. Seven ml of digestion solution containing collagenase type II (295 U/ml), 2.5 mM CaCl₂, and 1% BSA in DMEM was added to two hindlimbs and the preparation was placed on a water bath with constant rotation at 37°C for 45 min and intermittent mixing every 15 min. Tissue preparations were gently pipetted up and down 5–10 times to enhance muscle dissociation with a 10 mL stripette on ice. Ice-cold FACS buffer was added to make the final volume up to 30 mL volume and samples were then passed through 100 µm and 70 µm cell strainers sequentially after gentle mixing. Following centrifugation at 600 g for 5 min at 4°C, the pellet was resuspended in 3–4 mL of ice-cold FACS buffer. Cell preparations were filtered again using 5 mL polystyrene round-bottom tubes with cell-strainer caps before proceeding to FACS. Single cells that were DAPI negative and tdTomato⁺ were sorted using FACS Aria II.

CFU-F assay was performed as described above. Briefly, freshly sorted tdTomato⁺ cells were seeded at a density of 30/cm² in high–glucose DMEM with L-glutamine, and sodium pyruvate (GIBCO) supplemented with 10% heat-inactivated FBS (HyClone) in a 6-well plate. Initially, the media was changed after 72 hr of culture, and then changed every two days. CFU-F averages were obtained from 2 technical replicates/sample. Flow cytometry for Mitotracker and phospho-S6 fluorescence and qRT-PCR were performed as described above.

Adipogenic and fibrogenic differentiation of Mito fractions

Mito fractions were FACS sorted and 3000 cells/well were seeded in a 12-well plate. After 5 days of cell culture, adipogenic differentiation was induced for 6 days with MesenCult Adipogenic Differentiation Kit (Mouse) according to the manufacturer’s instructions (STEMCELL Technologies). Cells were fixed in 4% PFA for 10 min at room temperature, washed with PBS and permeabilized in permeabilization/blocking buffer (1% BSA, 2.5% donkey serum, 0.1% TX-100 in PBS) for 30 min. Cells were incubated overnight with anti-perilipin A antibody (1:200 dilution, Cat. no. ab3526, Abcam) in permeabilization/blocking buffer at 4°C. Alexa 488-conjugated secondary antibody incubation and nuclear staining were performed as described above. Images were acquired on Nikon Eclipse Ti2E microscope fitted with an Andor Zyla 4.2+ camera using Nikon NIS-Elements AR software (4.00.00 [build 764] LO, 64 bit) and 20 × objective (Nikon, CFI S Plan Fluor LWD 20XC, Air immersion, 0.7 NA). In brief, tiled images were acquired (2.65 mm² area at the center point of the well) and the total cell number and the percentage of perilipin⁺ cells were quantified using Fiji software. To calculate the number of cells (nuclei), we utilized the analyze particles function in Fiji. Briefly, numbers of nuclei were determined using Hoechst signals following Huang thresholding, binarization, conversion to mask and applying watershed algorithm. Perilipin⁺ cells were counted manually, and the values expressed as the % of perilipin⁺ cells.

For fibrogenic assay, freshly sorted cells were cultured for 5 days and culture medium was then replaced by high–glucose DMEM supplemented with 2% heat-inactivated FBS and TGFβ1 (50 ng/mL) for 2 days. Following staining with anti-α-SMA antibody and Hoechst, tiled images (3 mm² area) were acquired using Thunder widefield fluorescence microscope (Leica Microsystems, Germany) using Leica Application Suite X (LAS X) software version 3.7.3 and a 20x objective (PL FL L, NA 0.4). 390nm and 475nm LED lines were used to detect Hoechst (nuclei) and α-SMA (Alexa 488-conjugated secondary antibody), respectively. Images were exported in TIFF format for analysis. Total cell number and the percentage of differentiated cells (myofibroblasts) were quantified manually using Fiji software. Myofibroblasts, identified as large, flat cells with incorporation of α-SMA in stress fibers, were counted and the values expressed as the % of myofibroblasts.

Micro-CT

Sample preparation

Hearts arrested at diastole were isolated and fixed in 4% PFA+1% Glutaraldehyde for 1 day at 4°C. Afterward, the fixed hearts were incubated in hydrogel (4% acrylamide, 0.05% BIS, 4% PFA, 0.25% VA044, 0.05% Saponin) for 3 days at 4°C. The hearts were then polymerized in hydrogel using the X-Clarity Polymerization System (Logos Biosystems) (–90 kPa, 37°C, 3 hr 10 min). Hydrogel was then removed from the hearts, which were then incubated in Lugol solution (Sigma-Aldrich) for 5 days before micro-CT scanning.

Scanning

Skyscan1272 (Bruker Corporation) was used for scanning heart samples using the following settings: resolution 2452 X 1640, filter 0.5 µm aluminum + filter 0.038 µm copper, rotation 360 degrees, rotation step 0.4. Image pixel size was 5 µm. Scans were reconstructed using NRecon (v1.7.1.0) and optimized for post-alignment, smoothing, ring artifacts, and beam hardening.

Analysis

Reconstructed scans were visualized via CTvox (v3.3.0) and CT Analyzer (v1.17.7.2+) and resized using DataViewer (v1.5.4.0) (Bruker Corporation). Segmentation and quantification were performed using AMIRA software (v2019.1) (Thermo Fisher Scientific). Semi-automatic 3D segmentation was achieved using the watershed algorithm in Amira’s segmentation editor. In the first step, two separate materials were created: ‘‘myocardium’’ and ‘‘scar.’’ The brush tool was used to accurately select greyscale values in the regions of interest for seeding input in the ‘‘myocardium’’ and ‘‘scar’’ material in every 10–20 slices in all planes (axial, coronal and sagittal). Next, watershed algorithm was applied to separate the myocardium region from the scar, using the user input as seeds. Contours were refined and semi-automatic segmentation was curated manually where necessary. The 3D volume was created using the iso-surface rendering tool. Volume measurements of segmented myocardium and scar tissue were extracted from the material statistics table and plotted as the ventricular volume percentage.

Cardiac tissue bundles

Fabrication of 3D engineered co-cultured cardiac tissue bundles consisting of neonatal rat ventricular myocytes (NRVMs) and mouse CFs, and their histological and functional analyses were performed as previously described (Li et al., 2017, 2020).

Fabrication of co-cultured cardiac tissue bundles

WT and KO CFs were isolated, FACS sorted, and cultured as described and were used at passage 2–4. NRVMs were isolated as previously described (Jackman et al., 2016). NRVMs and CFs (WT and KO) were combined at a ratio of 10:3 and encapsulated in cylindrical hydrogel constructs (Li et al., 2017, 2020). Each bundle was made using a solution of 2.25 × 10⁵ NRVMs with or without 0.675 × 10⁵ CFs (WT and KO), 20 µL of culture media (DMEM, 10% (v/v) horse serum, 1% (v/v) chick embryo extract, 100 U/ml penicillin G, 1 mg/ml Aminocaproic Acid and 50 µg/ml Ascorbic Acid), 8 µL of 10 mg/ml Fibrinogen (Akron), 4 mL of Matrigel, 8 µL of2 × media, and 0.4 µL of 50 unit/ml thrombin in 0.1% BSA in PBS). The cell-hydrogel mixture was injected into polydimethylsiloxane (PDMS) molds cast from Teflon masters placed in 12-well plates. The molds were pre-treated with 0.2% (w/v) pluronic F-127 (Invitrogen) and fitted with laser-cut Cerex frames (9.2 × 9.5 mm outer dimensions, 6.8 × 8.3 mm inner dimensions). The cell-hydrogel mixture was polymerized within PDMS molds for 45 min at 37°C followed by addition of 2 mL culture media per well. Frames with polymerized cardiac bundles were removed from the molds the next day and cultured dynamically at a rocking platform in suspension for 14 days. Culture media were changed every other day.

Measurement of electrical propagation in cardiac tissue bundles

Action potential propagation in cardiac bundles was optically mapped using our previously established methods (Jackman et al., 2016). Briefly, bundles were stained with a voltage-sensitive dye, Di-4 ANEPPS (10 µM), for 6 min and then incubated in Tyrode’s solution supplemented with 10 µM blebbistatin to prevent motion artifacts. Electrical activity was stimulated at 2 Hz with a point electrode at a bundle end and was recorded in microscopic mode at 4x magnification using a 504-channel photodiode array (RedShirt Imaging). Data analysis for conduction velocity (CV) and action potential duration (APD) were performed by customized MATLAB software (Badie et al., 2009).

Measurement of contractile force generation by cardiac tissue bundles

Passive tension and contractile force generation in response to electrical stimulation were recorded by mounting cardiac bundles onto a custom-made setup with a force transducer and a computer-controlled linear actuator, as previously described (Jackman et al., 2018). Briefly, frames cut to contain a single cardiac bundle were transferred to a chamber with Tyrode’s solution maintained at 37°C. The side of the frame was cut to allow cardiac bundle to stretch by linear actuator to 24% above the resting culture length in 4% increments. One Hz electrical stimulation was applied by a pair of platinum electrodes and generated isometric contractile force was measured after tissue was equilibrated for 5 min at each stretch increment. Contractile force traces were analyzed for maximum peak contractile force, twitch rise time and decay time, and passive force using a custom MATLAB program (Jackman et al., 2018).

Immunostaining and image analysis of cardiac tissue bundles

Cardiac bundles were fixed with 2% v/v paraformaldehyde on a rocking platform overnight at 4°C. Fixed cardiac bundles were washed in PBS, blocked in antibody buffer (5%w/v chick serum, 0.5%v/v Triton X-100, in PBS) for 1 hr at room temperature, and then incubated with primary antibodies overnight at 4°C in antibody buffer. The following primary antibodies were used at indicated dilutions: α-sarcomeric actinin (Sigma A7811, 1:200), Vimentin (Abcam ab92547, 1:400), Collagen I (Abcam ab34710, 1:200). Alexa Fluor conjugated secondary antibodies (Invitrogen) and DAPI were applied at a 1:400 dilution in antibody buffer for 3 hr at room temperature. Cardiac bundles were washed in PBS, mounted on slides, and imaged using a Leica inverted SP5 confocal microscope. Image analysis was performed using ImageJ software.

Bulk RNA sequencing of Mitotracker fractions, and WT and KO cells

FACS sorted cells were resuspended in 700 ul Qiazol and RNA was extracted using the miRNeasy miniRNA prep kit (QIAGEN). RNA integrity was assessed on RNA pico chip and all samples produced a RIN score greater than 9. RNA libraries were generated using SMARTer low input kit (Takara Bio) using 5 ng of total RNA and following manufacturers protocol. We used 10 PCR cycles for full length cDNA amplification and 8 PCR cycles to amplify fragmented cDNA. Libraries were pooled and sequenced on a HiSeq2500 through an external provider.

Bulk RNA-seq processing and differential expression analysis

Sequencing data was trimmed for adapters using Trimmomatic (Bolger et al., 2014) and then aligned to the mm10 mouse genome using STAR aligner (v2.5.1) (Dobin et al., 2013). Gene counts were generated using the summarizeOverlaps function of the GenomicAlignments R package (Lawrence et al., 2013) and analyzed using DESeq2 (Love et al., 2014). Principal component (PC) analysis was performed on gene expression values normalized using the rlog function in DESeq2, following filtering out of lowly expressed genes (requiring counts per million [CPM] > = 0.25 for > = 2 samples). Differential expression (DE) was calculated using the Wald DE test, with an adjusted p value < 0.05 and an absolute log2 fold change > 0.5 used as thresholds for determining significantly DE genes.

Single-cell RNA sequencing (scRNA-seq)

Cells were isolated and FACS sorted as described. The single cell libraries were prepared according to manufacturer’s instructions of version 2 (v2) 3-prime kit of 10X Chromium platform (10x Genomics), a commercially available droplet method for single cell mRNA encapsulation. Before analysis, cells were diluted to the final concentration of ~600 cells/µl in 2% FBS in 1X PBS to recover the desired number of captured cells for sequencing. For each experiment, the number of desired cells were approximately 6000. The final libraries were pooled together and loaded onto a Novaseq 6000 Illumina S4 (200 cycles) platform and targeted for ~100,000 or ~50,000 reads per cell in the healthy heart and sham versus MI experiments, respectively.

Processing of 10x Genomics Chromium scRNA-seq data

Raw scRNA-seq data was processed using the 10x Genomics CellRanger software (version 2.2.0). The BCL files obtained from the Illumina NovaSeq platform were processed to Fastq files using the Cell Ranger mkfastq program. The Fastq files were then mapped to the mm10 version 1.2.0 reference, downloaded from the 10x Genomics website. The CellRanger count program was run on individual Fastq datasets from the different conditions. The aggr program was run to generate aggregate count matrices for the 2 or 4 conditions in the healthy heart or sham versus MI experiments, respectively. For healthy heart samples, aggr was run with the normalize parameter set to ‘mapped’. For the sham versus MI experiment, aggr was run with the normalize parameter set to ‘none’ due to a greater disparity of reads/cell among the experimental runs.

Filtering, dimensionality reduction and clustering of scRNA-seq data

Bioinformatics processing of the scRNA-seq data was performed in (R Core Team, 2018) using the Seurat package (Butler et al., 2018) (version 3.1.4) and visualization performed with ggplot2 (Wickham, 2009). Initial quality control filtering metrics were applied as follows. Cells with fewer than 200 detected expressed genes were filtered out. Genes that were expressed in less that 10 cells were filtered out. In order to control for dead or damaged cells, cells with over 5% of raw unique molecular identifiers (UMIs) mapping to mitochondrial genes were filtered out. The distribution of expressed genes and unique molecular identifier (UMI) numbers was visualized and cells with clear outliers were filtered out (Healthy: 200 < number of genes < 4,000; 500 < number of UMIs < 15,000. Sham versus MI: 200 < number of genes < 5,500; 500 < number of UMIs < 35,000). UMIs were normalized to counts-per-ten-thousand, log-transformed, and a set of top 2000 highly variable genes was identified by gating for mean expression level and dispersion level. The log-normalized data was scaled, with variation due to total number of UMIs regressed out using a linear model. PC analysis was run on the scaled data for the set of previously-defined highly variable genes.

Cell populations were defined based on previous analysis of enriched fibroblasts in sham or MI-day 3 conditions (Farbehi et al., 2019). Previously defined cell identities were mapped by applying the Seurat FindTransferAnchors and TransferData functions, based on the first 41 PCs as used in the original clustering analysis. The cells and clusters were visualized on a UMAP dimensionality reduction plot generated on the same set of PCs used for cell population identification.

Differential proportion analysis

Differential proportion differences between conditions were calculated using the Differential Proportion Analysis (DPA) method described previously (Farbehi et al., 2019), applying a weight (w value) of 0.1 and with a p value cut-off of 0.05 used to determine significance.

Differential expression in scRNA-seq

For calculating differentially expressed (DE) genes, we used the Seurat (Stuart et al., 2019) FindMarkers program, testing genes expressed in at least 25% of cells for at least one of the conditions being compared and with an absolute log fold-change difference of 0.25 (including a pseudo-count of 1). DE testing was performed using the ‘MAST’ test (Finak et al., 2015) implemented in FindMarkers. A Bonferroni-adjusted p value of 0.01 was used to determine significance.

STRING network analysis

STRING interaction networks were obtained through the STRING web server (Szklarczyk et al., 2019), with interactions visualized using Cytoscape (Shannon et al., 2003).

Gene Ontology testing

Over-representation of GO terms in gene lists was calculated using the PANTHER web-service (Mi et al., 2017), with a Fisher’s exact test. The set of expressed genes from the relevant experiment were used as background gene lists. A false-discovery rate cut-off of 0.05 was used to determine statistical significance. For analysis of antioxidant genes among CFs, a list of mouse genes for the GO term antioxidant activity (GO:0016209) was obtained from AmiGO.

Analysis of single-cell progenitor states

Analysis of potential progenitor states in scRNA-seq was performed using the CytoTRACE R package (Gulati et al., 2020). UMI counts from the HET sham condition were input to the CytoTRACE function.

RNA Velocity

Velocity estimates were obtained using the velocyto software (La Manno et al., 2018). Loom files were generated and then analyzed in R. Data was filtered with the filter.genes.by.cluster.expression function, setting min.max.cluster.average to 0.2 for spliced data and 0.05 for unspliced. Velocity estimates were calculated using the gene.relative.velocity.estimates function, setting fit.quantile to 0.02, deltaT to 1 and kCells to 25. Cell velocity trajectories were visualized on UMAP coordinates using the show.velocity.on. embedding.cor function with neighborhood size set to 200 and velocity scale set to ‘sqrt’.

QUANTIFICATION AND STATISTICAL ANALYSIS

Unless otherwise specified, all results obtained from independent experiments are reported as means ± standard errors of means (SEM) of multiple replicates. Where applicable, normality was estimated using D’Agostino & Pearson or Shapiro-Wilk normality test. Comparisons between two groups of normally distributed and not connected data were performed using unpaired, nonparametric Student’s t test. Multiple group comparisons were performed by one-way analyses of variance analyses (ANOVA, for one independent variable) or two-way ANOVA (for two independent variables), followed by Tukey’s post hoc comparison (GraphPad Prism version 8.0, La Jolla, CA). Survival probabilities were estimated using Kaplan-Meier analysis and significance calculated using the log-rank (Mantel-Cox) test. Unless otherwise indicated, ‘‘n’’ in the Figure Legends represents the number of animals or independent biological samples per group used in the indicated experiments. p values < 0.05 were considered statistically significant. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti-Hypoxyprobe Mab-1	Hypoxyprobe	Cat# HP mAb-1; RRID:AB_2801307
Anti-Hif-1a	Novus Biologicals	Cat# NB100-449; RRID:AB_10001045
Anti-Cleaved caspase 3 (Asp175)	Cell Signaling Technology	Cat# 9661; RRID:AB_2341188
Anti-Tubulin (clone DM1A)	Sigma-Aldrich	Cat#T9026; RRID: AB_477593
Anti-Ki67	Abcam	Cat# ab15580; RRID:AB_443209
Anti-Collagen VI	Abcam	Cat# ab6588; RRID:AB_305585
Anti-αSMA	Sigma-Aldrich	Cat# A2547; RRID:AB_476701
Anti-α-sarcomeric actinin	Sigma-Aldrich	Cat# A7811; RRID:AB_476766
Anti- Collagen I	Abcam	Cat# ab34710; RRID:AB_731684
Anti-Akt (pan-Akt)	Cell Signaling Technology	Cat# 4691; RRID:AB_915783
Anti-pAkt473	Cell Signaling Technology	Cat# 4060; RRID:AB_2315049
Anti-pErk1/2	Cell Signaling Technology	Cat# 3179; RRID:AB_2095853
Anti-GAPDH (14C10)	Cell Signaling Technology	Cat# 2118; RRID:AB_561053
Anti-Collagen IV	Abcam	Cat# ab19808; RRID:AB_445160
Anti-CD31	BD Biosciences	Cat# 553370; RRID:AB_394816
Anti- Sca1 (Ly6A/E)-PE	BD Biosciences	Cat# 553108; RRID:AB_394629
Anti- Sca1 (Ly6A/E)-FITC	BD Biosciences	Cat# 557405; RRID:AB_396688
Anti-CD45- APC/Cy7	BD Biosciences	Cat# 557659; RRID:AB_396774
Anti-CD31-PE/Cy7	Thermo Fisher Scientific	Cat# 25-0311-82; RRID:AB_2716949
Anti-PDGFRa (CD140a)-APC	Thermo Fisher Scientific	Cat# 17-1401-81; RRID:AB_529482
Anti-Ly-6C-PE	BD Biosciences	Cat# 560592; RRID:AB_1727556
Anti-F4/80-FITC	Thermo Fisher Scientific	Cat# 11-4801-81; RRID:AB_2735037
Anti-Anti-Mouse CD11c-FITC	BD Biosciences	Cat# 557400; RRID:AB_396683
I-Ab FITC	BD Biosciences	Cat# 553551; RRID:AB_394918
Anti-Ly-6G (Gr-1)-PerCP/Cy5.5	Thermo Fisher Scientific	Cat# 45-5931-80; RRID:AB_906247
Anti-CD11b-PE/Cy7	Biolegend	Cat# 101215; RRID:AB_312798
Phospho-S6 Ribosomal Protein (Ser235/236) (2F9)-Alexa Fluor 488	Cell Signaling Technology	Cat# 4854; RRID: AB_390782
Anti-Col1a1 (E8I9Z)	Cell Signaling Technology	Cat# 91144; RRID:AB_2800169
Anti-Perilipin-1	Abcam	Cat# 3526; RRID:AB_2167274
Wheat Germ Agglutinin (WGA), Alexa Fluor 488 Conjugate	Thermo Fisher Scientific	Cat# W11261
Isolectin GS-IB4 From Griffonia simplicifolia, Alexa Fluor 568 Conjugate	Thermo Fisher Scientific	Cat# I21412
Anti-rabbit IgG-HRP	Jackson ImmunoResearch Labs	Cat# 111-035-144; RRID:AB_2307391
Anti-mouse IgG-HRP	Jackson ImmunoResearch Labs	Cat# 115-035-062; RRID:AB_2338504
AlexaFluor Donkey Anti-Rabbit 488	Thermo Fisher Scientific	Cat# A-21206; RRID: AB_141708
AlexaFluor Donkey Anti-Mouse 488	Thermo Fisher Scientific	Cat# R37114; RRID: AB_2556542
AlexaFluor Goat Anti-Rat 488	Thermo Fisher Scientific	Cat# A-11006; RRID:AB_2534074
AlexaFluor Donkey Anti-Rabbit 555	Thermo Fisher Scientific	Cat# A-31572; RRID: AB_162543
AlexaFluor Donkey Anti-Mouse 555	Thermo Fisher Scientific	Cat# A-31570; RRID: AB_2536180
AlexaFluor Donkey Anti-Rabbit 647	Thermo Fisher Scientific	Cat# A-31573; RRID: AB_2536183
AlexaFluor Donkey Anti-Mouse 647	Thermo Fisher Scientific	Cat# A-21235; RRID: AB_141693
Chemicals, peptides, and recombinant proteins
MEM α, nucleosides, GlutaMAX Supplement	Thermo Fisher Scientific	Cat# 32571036
DPBS, no calcium, no magnesium	Thermo Fisher Scientific	Cat# 14190250
Penicillin-Streptomycin	Thermo Fisher Scientific	Cat# 15140163
Fetal Bovine Serum (FBS)	Sigma-Aldrich	Batch# 14J211
TrypLE Express Enzyme (1X)	Thermo Fisher Scientific	Cat# 12604013
DMSO	Sigma-Aldrich	Cat# D2650
MitoTracker Green FM	Thermo Fisher Scientific	Cat # M7514
Tetramethylrhodamine, Methyl Ester, Perchlorate (TMRM)	Thermo Fisher Scientific	Cat # T668
CM-H2DCFDA (DCF)	Thermo Fisher Scientific	Cat# C6827
Dihydroethidium (Hydroethidine) (DHE)	Thermo Fisher Scientific	Cat# D11347
MitoSOX	Thermo Fisher Scientific	Cat# M36008
4’,6-Diamidino-2-Phenylindole, Dihydrochloride (DAPI)	Thermo Fisher Scientific	Cat# 62247
Crystal violet	Sigma-Aldrich	Cat# C0775
Peanut oil	Sigma-Aldrich	Cat# P2144
Tamoxifen	Sigma-Aldrich	Cat# T5648
Tirapazamine	Sigma-Aldrich	Cat# SML0552
EdU (5-ethynyl-2′-deoxyuridine)	Thermo Fisher Scientific	Cat# A10044
MitoParaquat	Cayman Chemical	Cat# 18808
MitoTEMPO	Sigma-Aldrich	Cat# SML0737
N-Acetyl-L-cysteine (NAC)	Sigma-Aldrich	Cat# A7250
Lugol’s solution	Sigma-Aldrich	Cat# L6146
40% Acrylamide	Bio-Rad	Cat# 1610140
2% Bis-acrylamide	Bio-Rad	Cat# 1610142
Saponin	Sigma-Aldrich	Cat# S4521
PhosSTOP	Sigma-Aldrich	Cat# 4906845001
cOmplete Protease Inhibitor Cocktail	Sigma-Aldrich	Cat# 11697498001
PVDF membrane	Bio-Rad	Cat# 1620177
Nitrocellulose Membrane	Bio-Rad	Cat# 1620112
NuPAGE 4%-12% Bis-Tris Protein Gel	Thermo Fisher Scientific	Cat# NP0321
4–15% Mini-PROTEAN® TGX Stain-Free Protein Gels	Bio-Rad	Cat# 4568084
Trans-Blot Turbo Transfer Pack	Bio-Rad	Cat# 1704158
Precision Plus Protein Kaleidoscope Prestained Protein Standards	Bio-Rad	Cat #1610375
SuperSignal West Pico PLUS Chemiluminescent Substrate	Thermo Fisher Scientific	Cat# 34577
Bovine Serum Albumin	Sigma Aldrich	Cat# A4503
Tween 20	Merck	Cat# P9416
Triton X-100	Sigma-Aldrich	Cat# X100
Agarose	Merck	Cat# A0169
Paraformaldehyde	Sigma-Aldrich	Cat# P6148
25% Glutaraldehyde	Sigma-Aldrich	Cat# G6257
Sucrose	Sigma-Aldrich	Cat# S0389
Tissue-Tek O.C.T. Compound	Sakura	Cat# 4583
MOWIOL 4–88	EMD Millipore	Cat# 475904
SuperFrost PLUS charged glass slides	Thermo Scientific	Cat# J3800AMNZ
Glycerol	Sigma-Aldrich	Cat# G5516
Ketamine	Provet (NSW) Pty Ltd	N/A
Xylazine	Provet (NSW) Pty Ltd	N/A
Buprenorphine	Provet (NSW) Pty Ltd	N/A
Isoflurane	Abbott Laboratories	Cat# VQISO250
Heparin	Pfizer	Cat# AUST R 49232
Hydrogen Peroxide solution, 30%	Chem-supply	Cat# HA154-500M
GlycoBlue	Thermo Fisher Scientific	Cat# AM9516
QIAzol	QIAGEN	Cat# 79306
TRIzol	Thermo Fisher Scientific	Cat# 15596018
Collagen I, Rat Tail	Corning	Cat# 354236
Collagenase Type II	Worthington	Cat# LS004176
Recombinant Human TGF-beta 1 Protein	R&D Systems	Cat# 240-B
Low glucose DMEM	Sigma	Cat# D6046
High glucose DMEM	Thermo Fisher Scientific	Cat# 11995065
Donor Equine Serum (U.S.) Heat Inactivated	Cytiva	Cat# SH30074.03HI
Chick Embryo Extract, Ultrafiltrate, Liquid	US Biological	Cat# C3999
L-Ascorbic acid 2-phosphate sesquimagnesium salt hydrate	Sigma	Cat# A8960
6-Aminocaproic Acid	Sigma	Cat# A7824
Fibrinogen from bovine plasma	Sigma	Cat# F8630
Thrombin from bovine plasma	Sigma	Cat# T6634-1KU
Matrigel	BD Biosciences	Cat# 356234
Pluronic F-127	Sigma-Aldrich	Cat# P2443
Di-4 ANEPPS	Thermo Fisher Scientific	Cat# D1199
Blebbistatin	Sigma-Aldrich	Cat# B0560
Rhod-2 AM	Thermo Fisher Scientific	Cat# R1244
Verapamil hydrochloride	Sigma	Cat# V4629
Hoechst 33342	Thermo Fisher Scientific	Cat# H3570
Round-Bottom Polystyrene Test Tubes with Cell Strainer Snap Cap	Falcon	Cat# 352235
Cell Strainer, 40 µm	BD Biosciences	Cat# 352340
Cell Strainer, 70 µm	BD Biosciences	Cat# 352350
Cell Strainer, 100 µm	BD Biosciences	Cat# 352360
Critical commercial assays
Click-iT EdU Alexa Fluor 488 Flow Cytometry Assay Kit	Thermo Fisher Scientific	Cat# C10425
Click-iT EdU Alexa Fluor 448 imaging kit	ThermoFisher Scientific	Cat# C10337
2-NBDG (2-(N-(7-Nitrobenz-2-oxa-1,3-diazol-4-yl)Amino)-2-Deoxyglucose)	Thermo Fisher Scientific	Cat# N13195
In Situ Cell Death Detection Kit, TMR red	Sigma-Aldrich	Cat# 12156792910
ATP Bioluminescence Assay Kit HS II	Sigma-Aldrich	Cat# 11699709001
Quantitect reverse transcription kit	QIAGEN	Cat# 205313
L-Lactate Assay kit	Abcam	Cat# ab65331
Hypoxyprobe Plus Kits (FITC-Mab)	Hypoxyprobe	Cat# HP2-100Kit
REDExtract-N-Amp Tissue PCR Kit	Sigma-Aldrich	Cat# XNAT-1000RXN
iTaq Universal SYBR Green Supermix	Bio-rad	Cat# 1725122
Dead Cell Removal Kit	Miltenyi Biotec	Cat# 130-090-101
MesenCult Adipogenic Differentiation Kit (Mouse)	StemCell Technologies, Inc.	Cat# 05507
WesternBreeze Chemiluminescent detection kit	Thermo Fisher Scientific	Cat# WB7106
Chromium Single Cell A Chip Kit	10x Genomics	Cat# 120236
Chromium i7 Multiplex Kit	10x Genomics	Cat# 120262
SMART-Seq v4 Ultra Low input kit for sequencing	Takara Bio	Cat# 634889
Nextera XT DNA Library preparation kit	Illumina	Cat# FC-131-1024
miRNeasy mini kit	QIAGEN	Cat# 217004
High sensitivity DNA kit	Agilent	Cat# 5067-4626
RNA 6000 pico kit	Agilent	Cat# 5067-1513
Agencourt AMPure XP reagent	Beckman Coulter	Cat# A63881
504-channel photodiode array	RedShirt Imaging	NeuroPDA-III custom design
Deposited data
RNA-seq	This study	ArrayExpress: E-MTAB-9675 (Mito fractions) ArrayExpress: E-MTAB-9674 (Hif-1a cKO)
scRNA-seq	This study	ArrayExpress: E-MTAB-9583 (Hif-1a cKO)
Experimental models: Organisms/strains
Mouse: Wild type [Inbred C57BL/6J]	Jackson Laboratory	Stock No: 000664
Mouse: PdgfraMCM/+ [Pdgfratm1.1(cre/Esr1*)Nshk]	RIKEN, Japan	MGI Cat# 5475226
Mouse: PdgfranGFP/+ [B6.129S4 Pdgfratm11(EGFP)Sor/J]	Jackson Laboratory	Stock No: 007669
Mouse: Hif-1aflox/flox [B6.129-Hif1atm3Rsjo/J]	Jackson Laboratory	Stock No: 007561
Mouse: Hif-1a+/ [B6.129-Hif1atm3Rsjo/J]	Jackson Laboratory	Stock No: 007561
Mouse: Rosa26-tdTomato [B6.Cg-Gt(ROSA)26Sortm9(CAG-tdTomato)Hze/J]	Jackson Laboratory	Stock No: 007909
Mouse: Rosa26-mT/mG [Gt(ROSA)26Sortm4(ACTBtdTomato,-EGFP)Luo/J]	Jackson Laboratory	Stock No: 007576
Rat: Sprague Dawley rats	Charles River Laboratories	MGI Cat# 5651135, RRID:MGI:5651135
Oligonucleotides
See Table S3 for all primers	This paper (Integrated DNA Technologies)	N/A
Software and algorithms
FlowJo (v10.3)	FLOWJO LLC, TreeStar	https://www.flowjo.com/solutions/flowjo RRID:SCR_008520
FACSDIVA software	BD Biosciences	https://www.bdbiosciences.com/en-us/instruments/research-instruments/research-software/flow-cytometry-acquisition/facsdiva-software RRID:SCR_001456
ImageJ	Schneider et al., 2012	https://imagej.net/RRID:SCR_003070
Image Lab Software	Bio-rad	https://www.bio-rad.com/en-us/sku/1709690-image-lab-software RRID:SCR_014210
Nikon NIS-Elements AR	Nikon	https://www.microscope.healthcare.nikon.com/products/software/nis-elements
CFX Manager (qPCR) software	Bio-rad	https://www.bio-rad.com/en-eh/product/cfxmanager-software RRID:SCR_017251
CellProfiler (v3.0)	Broad Institute (McQuin et al., 2018)	https://cellprofiler.org/RRID:SCR_007358
Amira Software	Thermo Fisher Scientific	https://www.fei.com/software/amira-3d-for-lifesciences/ RRID:SCR_007353
CTVOX (v3.0)	Bruker Corporation	https://www.bruker.com/en/products-and-solutions/microscopes/3d-x-ray-microscopes.html
NRecon (v1.7.1.0)	Bruker Corporation
DATAVIEWER	Bruker Corporation
Vevo 3100 Software	Fujifilm VisualSonics	https://www.visualsonics.com/product/imaging-systems/vevo-770
GraphPad Prism (v8.0)	GraphPad Software	https://www.graphpad.com:443/ RRID:SCR_002798
CellRanger	10x Genomics	https://support.10xgenomics.com/single-cellgene-expression/software/downloads/latest
STAR	Dobin et al., 2013	https://github.com/alexdobin/STAR; RRID:SCR_015899
Seurat (v3.1.4)	Butler et al., 2018	https://satijalab.org/seurat/; RRID:SCR_007322
PANTHER	Mi et al., 2017	http://www.pantherdb.org; RRID:SCR_004869
Differential Proportion Analysis	Farbehi et al., 2019	https://elifesciences.org/articles/43882#abstract
RNA Velocity	La Manno et al., 2018	http://velocyto.org
CytoTRACE	Gulati et al., 2020	https://cytotrace.stanford.edu
DESeq2	Love et al., 2014	https://bioconductor.org/packages/release/bioc/html/DESeq2.html; RRID:SCR_015687
Cytoscape	Shannon et al., 2003	https://cytoscape.org; RRID:SCR_003032
STRING	Szklarczyk et al., 2019	https://string-db.org; RRID:SCR_005223
Andor Solis software	Li et al., 2017	https://andor.oxinst.com/products/solis-software/
Contractile force	Jackman et al., 2016	Custom design - Bursac lab
Optical mapping	Badie and Bursac, 2009	Custom software – Bursac lab

Highlights.

• The PDGFRα⁺SCA1⁺ fraction of cardiac fibroblasts is hypoxic and expresses HIF-1α

• Hif-1a deletion in cardiac fibroblasts lowers the threshold for their activation

• Hif-1a knockouts show increased ROS and excessive fibrosis after ischemic injury

• HIF-1α provides a brake that limits mitochondrial ROS and fibroblast proliferation