Cardiac Mesenchymal Stem Cells Promote Fibrosis and Remodeling in Heart Failure: Role of PDGF Signaling

Abstract

Heart failure (HF) is characterized by progressive fibrosis. Both fibroblasts and mesenchymal stem cells (MSCs) can differentiate into pro-fibrotic myofibroblasts. MSCs secrete and express platelet-derived growth factor (PDGF) and its receptors. We hypothesized that PDGF signaling in cardiac MSCs (cMSCs) promotes their myofibroblast differentiation and aggravates post-myocardial infarction left ventricular remodeling and fibrosis. We show that cMSCs from failing hearts post-myocardial infarction exhibit an altered phenotype. Inhibition of PDGF signaling in vitro inhibited cMSC-myofibroblast differentiation, whereas in vivo inhibition during established ischemic HF alleviated left ventricular remodeling and function, and decreased myocardial fibrosis, hypertrophy, and inflammation. Modulating cMSC PDGF receptor expression may thus represent a novel approach to limit pathologic cardiac fibrosis in HF.

Keywords: CCL, C-C motif chemokine ligand; CCR2, C-C chemokine receptor 2; DDR2, discoidin domain receptor 2; DMEM, Dulbecco’s modified Eagle medium; EDV, end-diastolic volume; EF, ejection fraction; ESV, end-systolic volume; HF, heart failure; IL, interleukin; INF, interferon; LV, left ventricular; Lin, lineage; MI, myocardial infarction; MSC, mesenchymal stem cell; PBS, phosphate-buffered saline; PCR, polymerase chain reaction; PDGF, platelet-derived growth factor; PDGFR, platelet-derived growth factor receptor; TGFβ, transforming growth factor beta; WGA, wheat germ agglutinin; cDNA, complementary DNA; cMSC, cardiac mesenchymal stem cell; cardiac remodeling; fibrosis; heart failure; mRNA, messenger RNA; mesenchymal stem cells; myocardial inflammation; myofibroblasts; platelet-derived growth factor receptor; siRNA, small interfering RNA; α-SMA, alpha smooth muscle actin.

Graphical abstract

Figure 1

Characterization of Mouse Resident cMSCs (A) Left panel: Representative histograms representing expression of mesenchymal stem cell (MSC) markers in minimally expanded primary cardiac MSC (cMSC) isolates (passages 2-4) analyzed by using flow cytometry. Shown also are the histograms (blue shaded) generated by using isotype controls for each marker analyzed. cMSCs expressed Sca1, CD90, CD49, CD44, CD29, and CD105; were negative for c-Kit expression; and exhibited low expression of discoidin domain receptor 2 (DDR2) and CD31. Right panel: Representative flow cytometry histograms depicting surface marker expression from cMSCs depleted for CD45, lineage markers (Lin), CD31, and DDR2 expression, and enriched for Sca1 expression using antibody-conjugated magnetic beads. (B) Top: Results of baseline gene expression of fibroblast markers collagen-I, collagen-III, fibronectin, alpha smooth muscle actin (α-SMA), and tenascin-C in sorted (Lin^–CD45^–CD31^–DDR2^–Sca1⁺) cMSCs compared with expression in cardiac fibroblasts (Card Fib) isolated from wild-type adult mouse hearts. 18s ribosomal rRNA expression was used to normalize mRNA expression. Data were analyzed by using unpaired Student’s t-test. n = 3 to 4 per group. Bottom: Representative images of the collagen gel contraction assay after overnight detachment of the collagen plugs after 4 days of myofibroblast differentiation using cMSCs (left) or commercially available WPMY-1 human prostrate stromal myofibroblast cell line (right). ∗∗P < 0.01, ∗∗∗P < 0.001. ns = not significant.

Figure 2

cMSCs and cMSC-Derived Myofibroblasts Are Increased in Chronic Heart failure (A) Gating strategy for flow cytometric quantification of total Sca1⁺ and Sca1^–CD31^–DDR2^– cMSCs and Sca1–DDR2⁺ cardiac fibroblasts in mouse hearts 1 day post–myocardial infarction (MI) (or sham operation). n= 4 to 6 per group. Data are expressed as cells per heart. ∗P < 0.05. (B) Flow cytometric gating and corresponding group data for assessment of cMSCs (Sca1⁺CD31^–DDR2^–) and cardiac fibroblasts (Sca1^–DDR2⁺) in mouse hearts 8 weeks’ post-MI. Data are expressed as cells per heart. n = 5 to 6 per group. ∗∗P < 0.005, ∗∗∗P < 0.001. (C) Flow cytometric gating strategy and group data for quantification of cMSC-derived myofibroblast in mouse hearts 8 weeks’ post-MI. Myofibroblasts were identified based on a-SMA expression and classified as being derived from cMSCs based on their Sca1 expression. Data are expressed as cells per heart. n = 5 to 6 per group. ∗∗∗P < 0.001. Statistical comparisons in all panels were performed by using the unpaired Student’s t-test. HF = heart failure; SSC-H = side scatter–height; other abbreviations as in Figure 1.

Figure 3

HF-Derived cMSCs Exhibit a Pro-inflammatory Phenotype (A) Cell-free conditioned media from untreated sham and HF cMSCs cultured for 24 hours was analyzed for chemokine content using antibody arrays. cMSC media was used as control. Array controls (positive and negative), blanks, and some of the most dramatically up-regulated chemokines are indicated by numbered boxes. (B) Quantitative polymerase chain reaction analysis of baseline gene expression from sham and HF cMSCs for genes involved in monocyte and T-cell recruitment. Data are represented as mean ± SD of 3 to 4 independent experiments done in duplicate or triplicate. Statistical comparisons (unpaired Student’s t-test) in B were performed after logarithmic data transformation to satisfy the normality assumption as described in the Methods. ∗P < 0.05, ∗∗∗P < 0.005. (C) Quantitative polymerase chain reaction analysis of genes involved in macrophage polarization after 24 hours of co-culture of mouse RAW246.7 macrophage cells and conditioned media collected from untreated sham and HF cMSCs. Results are expressed as mean ± SD of 3 independent experiments done in duplicate or triplicate. ∗P < 0.05, ∗∗P < 0.005. (D) Results from Transwell migration assay showing representative images and the group data of the migrated M1-polarized mouse RAW 246.7macrophages when co-cultured with either sham or HF cMSCs (upper panel) or the conditioned (Cond) media collected from untreated sham and HF cMSCs (lower panel) as chemoattractant in the bottom wells of the Transwell plates. cMSC media was used as control. Each image is a stitched composite of 4 separate overlapping images. The yellow dotted circle demarcates the boundary of the Transwell. Four independent areas of observation were counted from each Transwell. Data are expressed mean ± SD of 2 independent experiments with multiple replicates. ∗P < 0.05 vs sham groups and control cMSC media. Statistical comparisons in C and D were analyzed by 1-way analysis of variance followed by Tukey’s post hoc analysis. CCL = C-C motif chemokine ligand; G-CSF =  granulocyte-colony stimulating factor; IL-6 = interleukin-6; iNOS = inducible nitric oxide synthase; SDF-1 = stromal cell–derived factor 1; TNF = tumor necrosis factor; TLR = Toll-like receptor; other abbreviations as in Figures 1 and 2.

Figure 4

HF-Derived cMSCs Exhibit a Pro-fibrotic and Pro-inflammatory Phenotype (A) Hierarchical clustering and heat maps of RNA-sequencing data from CD45^–Sca1⁺ flow sorted from myocardial mononuclear cell isolates from sham and HF hearts (8 weeks’ post-MI; n = 3 per group). Counts were normalized with DeSeq2 and then transformed via log 2 before clustering and visualization using the R platform. Shown is the data set generated using both P < 0.05 and q < 0.05 values (adjusted P value using false discovery rate value <0.05). (B) Upstream regulator analysis of genes identified from RNA-sequencing data (data obtained with P < 0.05 and q < 0.05 cutoffs; 400 genes) using Ingenuity Pathway Analysis software predicted among other pathway, activation of the pro-fibrotic transforming growth factor-β (TGFβ) pathway, and inflammatory pathways including IL-6 and nuclear factor–κB (NF-κB) signaling pathway. Abbreviations as in Figures 1, 2, and 3.

Figure 5

cMSC–Macrophage Interactions Induce cMSC Myofibroblast Differentiation (A) Overlaid confocal images (left) of sham cMSCs treated with TGFβ (10 ng/mL) for 4 days and immunostained for α-SMA (red), phalloidin (green), and 4′,6-diamidino-2-phenylindole (DAPI) (blue). Middle: quantitative polymerase chain reaction analysis and protein immunoblotting of α-SMA expression in TGFβ-differentiated (and control undifferentiated) sham cMSCs. n = 3 per group. (B) Functional assessment of TGFβ-induced myofibroblast differentiation of sham and HF cMSCs using collagen gel contraction assay and the corresponding quantification. Data are expressed as percent change in the collagen gel area relative to the area of the culture well. (C) Quantitative polymerase chain reaction analysis of fibrotic gene markers in TGFβ-differentiated sham and HF cMSCs isolated from collagen gels by collagenase digestion after contraction assay. n = 2 per group (each pooled from 2 independent wells) run in triplicates. (D) Representative results of collagen gel contraction assay from naive macrophage (mouse RAW 246.7 macrophages) with either sham or HF cMSC co-culture and the corresponding quantification. Results obtained from sham and HF cMSCs cultured with RAW-macrophages were used as control. n = 4 per group. (E) Results from M1- and M2-polarized mouse RAW 246.7 macrophage co-culture on myofibroblast differentiation of sham and HF cMSCs, and National Institutes of Health 3T3 fibroblasts with the corresponding quantification of collagen gel contraction. Results in BandE are representative of 2 to 4 independent experiments done in duplicate or triplicate. Data in AandC were analyzed by Student’s t-test, and data in B, D,andE were analyzed by 1-way analysis of variance followed by Tukey’s post hoc analysis. ∗P < 0.05, ∗∗P < 0.005. Other abbreviations as in Figures 1, 2, and 4.

Figure 6

HF cMSCs Exhibit Increased Expression of PDGFRβ That Promotes Their Myofibroblast Differentiation (A) Platelet-derived growth factor receptor (PDGFR) expression in sham and HF cMSCs by quantitative polymerase chain reaction (qPCR) (left, n = 3-4 per group) and immunoblot analysis (right, n = 3-4 per group). Results are presented as mean ± SD. (B) qPCR analysis for baseline expression of PDGFRβ ligands PDGF-B and PDGF-D in sham and HF cMSCs. n = 3 per group. (C) qPCR analysis of PDGFR ligand expression in sham and HF cMSCs after 24 hours of co-culture with cell-free conditioned media collected from in vitro polarized mouse RAW 246.7 macrophages. n = 3 per group. (D) Upper panel: Representative results from collagen gel contraction assay and corresponding quantitation upon HF cMSCs co-culture with cell-free conditioned media from polarized macrophages (Raw 246.7 cells) for 4 days in the absence or presence of overnight treatment with 10 μM imatinib. Quantitation is from 4 independent experiments. Also shown are results from qPCR and immunoblot assays for α-SMA expression in HF cMSCs treated with imatinib for 4 days. n = 3 per group. P value comparison is vs control untreated cells. Lower panel: Left, Immunoblotting for PDGFRβ in HF cMSCs after 96 hours of transfection with PDGFRβ-specific (or control) small interfering RNA (siRNA) (30 nM). n = 3 per group. Right: Representative results from collagen gel contraction assays, and the corresponding quantification, using siRNA-transfected HF cMSCs and conditioned media from in vitro polarized RAW 246.7 macrophages. Quantitation reflects 2 independent experiments each done in triplicate. Data in A, B,andD (α-SMA and PDGFRβ data) were analyzed by using Student’s t-test, and data in CandD (gel contraction data) were analyzed by 1-way analysis of variance followed by either Tukey’s (D) or Šidák’s (C) post hoc analysis. ∗P < 0.05, ∗∗P < 0.005, ∗∗∗P < 0.001. GAPDH = glyceraldehyde-3-phosphate dehydrogenase; other abbreviations as in Figures 1 and 2.

Figure 7

Imatinib Treatment Post-MI Attenuates LV Remodeling and Preserves Systolic Function (A) Top: Schematic of the experiment protocol. Following 2 weeks of permanent coronary artery ligation (or sham surgery), wild-type C57BL/6 mice were randomized to receive daily injections of either 200 μL phosphate-buffered saline (PBS) vehicle or imatinib (30 mg/kg) for 3 weeks before being reassessed by echocardiography and subsequent euthanasia. Bottom: Representative long-axis B-mode serial echocardiographic images at end-diastole from HF mice at 2 and 5 weeks’ post-MI after treatment with either PBS or imatinib as described in A. The dotted line marks the left ventricular (LV) cavity. (B) Corresponding paired echocardiographic group data from HF mice at 2 and 5 weeks’ post-MI after treatment with either PBS or imatinib as described in A. n = 11 to 14 per group. (C) Echocardiographic group data depicting changes (Δ) in end-diastolic volume (EDV) and end-systolic volume (ESV) and ejection fraction (EF) from 2 to 5 weeks’ post-MI with treatment with either PBS or imatinib as described in A. n = 11 to 14 per group. Data were analyzed by using either paired (B) or unpaired (C) Student’s t-test. ∗P < 0.05, ∗∗P < 0.005. Abbreviations as in Figures 1, 2, 3, and 4.

Figure 8

Imatinib Treatment Post-MI Decreases Myocardial Fibrosis and Cardiomyocyte Hypertrophy and Enhances Myocardial Capillary Density (A) LV mass/tibia length ratio from wild-type HF mice at 5 weeks after treatment with either 200 μL PBS or imatinib (30 mg/kg per day) for 3 weeks starting 2 weeks after MI. n = 13 to 15 per group. (B) Representative wheat germ agglutinin (WGA) (green) and 4′,6-diamidino-2-phenylindole (blue) overlaid staining of the left ventricle and quantification of cardiomyocyte hypertrophy (upper panel, top; scale bar = 100 μm), Masson’s trichrome stains and quantification of LV border zone fibrosis (middle panel), and cardiac isolectin (red), WGA (green), and 4′,6-diamidino-2-phenylindole (blue) stained and combined images and corresponding quantification of capillary:cardiomyocyte ratio in remote zone left ventricle from HF mice after treatment with either PBS or imatinib at 2 weeks’ post-MI (bottom panel). n = 4 to 6 per group. Data in A and B were analyzed by using unpaired Student’s t-test. Scale bars = 100 μm. ∗P < 0.05, ∗∗P < 0.005. HW = heart weight; other abbreviations as in Figures 1, 2, and 7.

Figure 9

Imatinib Treatment Attenuates Post-MI Myocardial Inflammation (A) Representative flow cytometry scatter plots and corresponding group quantification of cMSCs (CD45^–Sca1⁺) in HF hearts after treatment with either PBS or imatinib (30 mg/kg per day) for 3 weeks after 2 weeks of MI. (B) Representative flow cytometry gating strategy and quantitation of total leukocytes (CD45⁺), total macrophages (CD45⁺MerTK⁺MHCII⁺Ly6C^–F4/80⁺), infiltrating macrophages (CD45⁺MerTK⁺MHCII⁺Ly6C^–F4/80⁺CCR2⁺), and resident macrophages (CD45⁺MerTK⁺MHCII⁺Ly6C^–F4/80⁺CCR2^–) from experimental groups as in panel A. n = 4 to 7 per group. Data in A and B were analyzed by using unpaired Student’s t-test (statistical comparisons in B were performed after logarithmic data transformation to satisfy the normality assumption as described in the Methods). Data are presented as cells per heart. ∗P < 0.05. FSC-A = forward scatter–area; FSC-W = forward scatter–width; MHC-II = major histocompatibility complex-II; SSC-A = side scatter–area; other abbreviations as in Figures 1, 2, 3, and 4 and 7.