Mesenchymal Stem Cells Promote Matrix Metalloproteinase Secretion by Cardiac Fibroblasts and Reduce Cardiac Ventricular Fibrosis After Myocardial Infarction§


  • Author contributions: C.M.: Conception and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing; O.L.: Collection and/or assembly of data, data analysis and interpretation; E.T.: Collection and/or assembly of data, data analysis and interpretation; J.R.: Conception and design; D. Calise: Collection and/or assembly of data, data analysis and interpretation; M.-H.S.: Collection and/or assembly of data; C.O.: Collection and/or assembly of data; M.-D.P.-M.: Collection and/or assembly of data; N.A.: Collection and/or assembly of data; A.N.S.: Collection and/or assembly of data; P.B.: Conception and design, provision of study material or patients; A.P.: Conception and design, data analysis and interpretation, manuscript writing, final approval of manuscript, financial support; D. Cussac: Conception and design, data analysis and interpretation, manuscript writing, final approval of manuscript.

  • Disclosure of potential conflicts of interest is found at the end of this article.

  • §

    First published online in STEM CELLS EXPRESS 2009.


Recent studies showed that mesenchymal stem cells (MSCs) transplantation significantly decreased cardiac fibrosis; however, the mechanisms involved in these effects are still poorly understood. In this work, we investigated whether the antifibrotic properties of MSCs involve the regulation of matrix metalloproteinases (MMPs) and matrix metalloproteinase endogenous inhibitor (TIMP) production by cardiac fibroblasts. In vitro experiments showed that conditioned medium from MSCs decreased viability, α-smooth muscle actin expression, and collagen secretion of cardiac fibroblasts. These effects were concomitant with the stimulation of MMP-2/MMP-9 activities and membrane type 1 MMP expression. Experiments performed with fibroblasts from MMP2-knockout mice demonstrated that MMP-2 plays a preponderant role in preventing collagen accumulation upon incubation with conditioned medium from MSCs. We found that MSC-conditioned medium also decreased the expression of TIMP2 in cardiac fibroblasts. In vivo studies showed that intracardiac injection of MSCs in a rat model of postischemic heart failure induced a significant decrease in ventricular fibrosis. This effect was associated with the improvement of morphological and functional cardiac parameters. In conclusion, we showed that MSCs modulate the phenotype of cardiac fibroblasts and their ability to degrade extracellular matrix. These properties of MSCs open new perspectives for understanding the mechanisms of action of MSCs and anticipate their potential therapeutic or side effects. STEM CELLS 2009;27:2734–2743


Despite progress in pharmacological and surgical approaches, postinfarction heart failure remains the major cause of cardiovascular morbidity and mortality in developed countries. Myocardial infarction (MI) results in a significant loss of cardiomyocytes, which are replaced by a fibrotic and akinetic tissue that forms a permanent scar. Structural changes associated with infarct healing involve various cell populations such as neutrophils, mononuclear cells, lymphocytes, mast cells, vascular cells, and, more specifically, fibroblasts [1]. In normal heart, cardiac fibroblasts regulate extracellular matrix (ECM) homeostasis through two major mechanisms: the synthesis and secretion of matrix molecules, and the secretion of the ECM-degrading enzyme metalloproteinases (MMPs) and their endogenous inhibitors (TIMPs). After acute cardiac injury, inflammatory mediators released by blood-derived cells promote the proliferation and phenotypic differentiation of fibroblasts into myofibroblasts. Myofibroblasts infiltrating the infarct area participate in the healing of the damaged ventricle through sequential matrix degradation by MMPs and synthesis of ECM proteins. Termination of the fibrotic response to the injury appears to be associated with myofibroblast apoptosis. However, in the case of extensive heart damage or the presence of comorbid conditions, myofibroblasts maintain their profibrotic properties and actively participate in the pathological cardiac remodeling and progressive function decay [2–5]. Therapeutic strategies that regulate fibroblast activity and avoid maladaptive processes could be a promising approach to prevent progression towards heart failure. For this purpose, cell-based therapy has been proposed as an alternative to classic pharmacological treatments. Among the different cell types available for cell therapy, bone marrow (BM) mesenchymal stem cells appear to be particularly interesting because of some of their peculiarities, including the multilineage potential, the ability to elude detection by the host immune system, the immunomodulatory properties, and the ease of expansion in culture [6–9]. Several recent studies have shown that MSC administration can improve cardiac function in models of MI [10, 11], dilated cardiomyopathy [12], and acute myocarditis [13]. The beneficial effects of MSC graft have been related in part to their paracrine activity. In fact, MSCs are able to secrete angiogenic, antiapoptotic, and anti-inflammatory cytokines that contribute to the recovery of cardiac function [14–22]. Recent studies showed that MSC transplantation significantly decreased fibrosis in the heart [12, 23], lung [24, 25], kidney [26], and liver [27, 28]. At the time of this writing, the mechanisms whereby MSCs reduce tissue fibrosis are still poorly understood. Some studies suggest that the antifibrotic effect of MSCs could be related in part to their ability to produce MMPs [29, 30]. On the other hand, a recent report suggested that the antifibrotic effect of MSCs may be indirect and may involve a decrease of collagen expression by cardiac fibroblasts mediated by MSC paracrine factors [17].

In this study we investigated whether MSCs may directly regulate the balance of MMP/TIMP production by cardiac fibroblasts and prevent collagen accumulation in vitro and in a rat model of postinfarction heart failure.


Isolation and Culture of MSCs

BM was obtained from Lewis rats (Harlan, Gannat, France, weighing 180–200 g. BM from femurs cavity was flushed with MEM medium (ABCYs; Paris, France, containing 10% fetal calf serum and 1% penicillin/streptomycin (Invitrogen; Carlsbad, CA,, and the cell suspension was centrifuged (400g, 5 min). Cells were then plated in culture flasks (200,000 cells/cm2). Nonadherent cells were removed after 72 hours, and MSCs were recovered by their capacity to adhere highly to plastic culture dishes. MSCs were then routinely cultured and were used for the experiments after the third passage.

Collection of MSC-Conditioned Medium

Conditioned medium was collected from MSCs after the third passage of cells cultured in standard medium for 24 hours (1 ml standard medium per 200,000 cells).

Ex vivo Treatment of MSCs with Melatonin and Labeling with Quantum Dots

As previously reported, MSCs were treated with melatonin (5 μM) (Sigma-Aldrich; Saint Quentin Fallavier, France, for 24 hours and extensively washed with phosphate-buffered saline (PBS) one time to improve cell viability after grafting [31]. Then MSCs were trypsinized and resuspended in MEM containing fluorescent quantum dot (QD) nanocrystals (10 nM/1 × 106 cells per 200 μl, Qtracker 655 cell labeling kit (Invitrogen). After incubation for 60 min at 37°C, MSCs were centrifuged (400g, 10 min) and resuspended in MEM prior to injection.

Induction of MI in Rats and Transplantation of MSCs

Experiments were performed in Lewis congenic rats (180–200 g, n = 50) (Harlan) and anesthetized with a mix of isoflurane/oxygen inhalation (3%/97%). After left thoracotomy, the heart was accessed through the fourth intercostals space. The interventricular artery was ligated with a 6-0 polyester suture. Intramyocardial injection of melatonin-treated or untreated MSCs (3 × 2.106 MSCs into 50 μl culture medium) or medium alone (3 × 50 μl) was performed 2 weeks after MI. Sham-operated animals were subjected to the same surgical procedure without coronary artery ligation and MSC transplantation. In all experiments, we used ≥5 rats in each group.

Histology and Immunohistochemistry

Heart sections were collected 48 hours or 2 months after MSC injection. Paraffin sections (6 μm) of hearts were stained with H&E, Sirius red (collagen synthesis), Alizarin red (calcium deposit), Alcian blue (ECM accumulation), or Red Oil solution (adipocyte staining) using standard methods. For MSC detection, sections were successively incubated (90 min, room temperature) with mouse monoclonal anti-CD90 antibody diluted 1:100 (Tebu-bio; Le Perray en Yvelines, France, and with goat monoclonal anti-mouse antibody HRP-conjugated, diluted 1:500 (Tebu-bio). For hepatocyte growth factor-α (HGF-α) detection, sections were, successively, incubated (120 min, room temperature) with rabbit polyclonal anti-HGF-α antibody diluted 1:100 (Tebu-bio) and with goat monoclonal anti-rabbit antibody horseradish peroxidase (HRP)-conjugated, diluted 1:500 (Tebu-bio). Quantification of histological sections were performed with computerized image analysis carried out on 12 fields of injection sites (QD fluorescence and HGF labeling) or on whole cardiac sections (Sirius red staining) of three different histological preparations obtained from each animal (n ≥ 5 animals per group). For connexin 43 detection, sections were successively incubated (120 min, room temperature) with mouse monoclonal anti-connexin 43 antibody diluted 1:50 (Millipore; Billerica, MA, and with Oregon green 488 goat anti-mouse antibody, diluted 1:200 (Invitrogen). For MMP2 detection, sections were successively incubated (120 min, room temperature) with mouse monoclonal anti-MMP2 antibody diluted 1:100 (Calbiochem; San Diego, CA; and with goat monoclonal anti-mouse antibody HRP-conjugated, diluted 1:500 (Tebu-bio). Proliferating cell nuclear antigen (PCNA) expression was used as a marker for cell proliferation. Immunostaining was performed with the streptavidin/biotin immunoperoxidase method (Zymed Laboratories Inc; South San Francisco, CA, The deparaffinized sections were immersed in 3% H2O2 in methanol for 10 min to quench the endogenous peroxidase activity, and then the sections were incubated with biotinylated rat PCNA (1 hour, room temperature). After being washed with PBS one time, the sections were incubated with peroxidase-labeled streptavidin for 10 minutes. Peroxidase staining was carried out using 3,3′-diaminobenzidine as the chromogen. Sections were counterstained using hematoxylin. The number of labeled nuclei per section was counted (magnification ×400), and the labeling index [labeled nuclei/total nuclei × 100(%)] was calculated on 12 microscopic fields in three level cuts from each animal (n ≥ 4 animals per group). Apoptosis was evaluated with the DeadEnd Fluorometric TUNEL system according to the manufacturer's instructions (Promega; Madison, WI, Briefly, the deparaffinized kidney sections were incubated in a proteinase K solution (20 μg/mL) to permeabilize the tissues, rinsed, and fixed in 4% paraformaldehyde. The sections were then incubated with terminal deoxynucleotidyl transferase (25 U/μL) and fluorescein-12-dUTP (1 hour, 37°C). After rinsing in PBS one time, the slides were immersed in propidium iodide solution (1 μg/μL, 15 min).

For angiogenesis analysis, tissue sections were stained with polyclonal rabbit anti-human von Willebrand factor diluted 1:30 (DAKO; Glostrup, Denmark, Antigen retrieval (EDTA) was used for this immunolabeling. For angiogenesis quantification, 14 microscopic fields (×400 magnification) of three different histological preparations obtained from each animal (n ≥ 5animals per group) were analyzed. Capillaries or isolated labeled cells were counted from injected area.

Isolation of Adult Cardiac Fibroblasts and MMP2-Knockout Cell Line

Cardiac cells were collected from adult rat hearts by incubation with collagenase IV (0.1 mg/ml) and pancreatine (0.5 mg/ml) every 10 min until the entire tissue was digested. Then, cells were plated on dishes in culture medium (DMEM/F12 medium supplemented with 1% penicillin/streptomycin and 10% fetal bovine serum). After 2 hours, nonadherent cells were removed and fresh medium was added. Passage 0 was used in all experiments. Cardiac fibroblasts were identified with positive staining for vimentin and negative staining for von Willebrand factor and troponine-T. MMP2-Deficient fibroblasts issued from MMP2-knockout mice [32] were grown in DMEM supplemented with 10% fetal calf serum.

Cardiac Fibroblast Viability Assay

The viability of cardiac fibroblasts was measured using CellTiter-Glo® luminescent cell viability assay (Promega), a homogeneous method of determining the number of viable cells in culture on the basis of quantification of the ATP present, an indicator of metabolically active cells. Fibroblasts were plated in six-well plates and incubated with medium of each condition for 24 hours. Plates and its contents were equilibrated for 30 min at room temperature, and the solution assay was added (v/v). After 10 min of incubation at room temperature, the luminescence was measured with a microplate reader (Mithras LB940, Berthold Technologies; Bad Wildbad, Germany,

Real-Time Polymerase Chain Reaction, Western Blot, and Zymography Analysis

For real-time polymerase chain reaction (PCR), total RNA was isolated from fibroblasts using a nucleospin kit (Macherey Nagel; Düren, Germany,, and cDNA was synthesized from 1 μg total RNA using SuperScript II reverse transcriptase (Invitrogen). Real-time PCR analysis was performed in 96-well plates using SYBR Green PCR Master Mix (ABI Prism 7000 HT Sequence Detection System, Applied Biosystems; Foster City, CA, Amplification reactions (25 μl) were carried out in triplicate with 5 μl of 1:5 diluted template cDNA according to manufacturer's protocol. Each assay was normalized by amplifying the housekeeping cDNA 36B4 from the same cDNA sample. Real-time PCR (membrane type 1 MMP (MT1-MMP), TIMP-2, 36B4) was carried out using the following primers (Eurogentec; Seraing, Belgium,




For Western blot analysis, proteins (5 μg) were extracted from fibroblasts. Western blot analyses were performed with samples normalized for protein concentration. Membranes were probed with anti-mouse α-smooth muscle actin (α-SMA) (1:7,000; Sigma-Genosys; Cambridge, United Kingdom, After several washes in TBS-Tween (0.2%), membranes were incubated with HRP-conjugated anti-mouse secondary antibody (1:10,000; Santa Cruz Biotechnology, Inc.; Santa Cruz, CA,

For zymography, supernatants from fibroblasts or MSCs were collected and centrifuged (500g, 5 min) to remove cells and debris. Protein extract was electrophoresed in 8% SDS-PAGE containing 0.1% gelatin. After migration and washing, gels were incubated (16 h, 37°C) in activation buffer (50 mM Tris-base (pH 7.5), 5 mM CaCl2, 0.02% NaN3, and 1 μM ZnCl2). Gels were stained with Coomassie staining solution (0.5% Coomassie, 50% MeOH, 10% acetic acid, and 40% H2O) for 90 min, followed by destaining (0.5% Coomassie, 50% MeOH, 10% acetic acid, and 40% H2O). Quantification of Western blot and zymography was performed with densitometry (TotalLab v1.10, Nonlinear Dynamics; Durham, NC,

In vitro Sirius Red Staining and Spectrophotometric Analysis

As described previously [33], fibroblasts cultured up to confluency in six-well plates were fixed in methanol overnight at −20°C. After washing with PBS one time, cells were incubated in Sirius red solution staining (0.1%) at room temperature for 60 min. Staining solution was removed, and the cells were washed three times with acetic acid (0.1%). For photography, cells were dehydrated in absolute EtOH (three times, 5 min) and toluene (three times, 10 min). Then, the total nodule number was counted. For spectrophotometric analysis, Sirius red was eluted in sodium hydroxide (0.1 N) at room temperature for 60 min on a rocking platform, and optical density was measured at 540 nm.


Left ventricular (LV) function was assessed in anesthetized animals with two-dimensional echocardiography with a General Electric Vivid 7® (GE Medical System; Milwaukee, WI, equipped with a 13-MHz linear probe immediately before and 2 weeks after MI, and 2, 4, and 8 weeks after cells transplantation. For anesthesia, rats were induced with isoflurane and received continuous inhaled anesthetic (2%) for the duration of the imaging session. The animals were placed in the supine or lateral position on a warming pad. Numeric images of the heart were obtained in both parasternal long-axis and short-axis views. Two-dimensional end-diastolic and end-systolic long-axis views of the LV were standardized as follows: inclusion of the apex, the posterior papillary muscle, the mitral valve, and the aortic root. Two-dimensional echocardiographic measurements were performed with the cine-loop feature to retrospectively catch true end-diastolic and end-systolic phases, defined as the phases in which the largest and the smallest LV cavity size was obtained, respectively. End-diastolic and end-systolic areas (A) were obtained by hand-tracings of the LV endocardial contours, according to the American Society of Echocardiography leading edge method. On these frames, end-diastolic and end-systolic lengths (L) of the LV were obtained by tracing a line connecting the more distal part of the apex and the center of a line connecting the mitral annular hinge points. End-diastolic and end-systolic volumes (LVEDVs and LVESVs, respectively) were then calculated by means of the single-plane area-length method (volume = 8×A2/3×π×L). LV ejection fraction (LVEF, %) was calculated as ((LVEDV−LVESV)/LVEDV)×100. All measurements were averaged on three consecutive cardiac cycles and analyzed by a single observer who was blinded to the treatment status of the animals. Only rats with infarct size of such magnitude that LVEF was <45% at baseline were included in the study so as to maximize possible treatment effects.

Statistical analysis

Data are represented as mean ± SEM. Statistical comparison of the data was performed using t test for comparison between two groups or one-way ANOVA and post hoc Tukey's test for comparison of more than two groups. A value of p < .05 was considered statistically significant.


Effects of MSC-Conditioned Medium on Cardiac Fibroblasts In Vitro

To investigate the effects of MSCs on cardiac fibroblasts, primary cultures of cardiac fibroblasts were maintained in fibroblast standard medium, MSC standard medium, or MSC-conditioned medium for 24 hours. As shown in Figure 1A, MSC-conditioned medium decreased the viability of cardiac fibroblasts compared with fibroblasts cultured in MSC standard medium (0.89 vs. 0.62 arbitrary units, p < .001). This effect was concomitant with a decrease in α-SMA expression, a marker of myofibroblast differentiation (0.83 vs. 1.02 arbitrary units, p < .01; Fig. 1B). The decrease in the number of viable and activated cardiac fibroblasts was associated with a significant reduction in the ECM production. Indeed, MSC-conditioned medium strongly reduced the quantity of type I and type III collagen nodules by fibroblasts, as detected with Sirius red staining and optical density measurement (p < .01; Fig. 1C).

Figure 1.

Effects of MSC conditioned-medium on cultured cardiac fibroblast viability and activity. Effect of MSC standard medium or MSC-conditioned medium on cardiac fibroblasts cell viability (A), α-SMA expression (B), collagen type I and type III deposit (C). All results are expressed as a value relative to the value obtained with fibroblast standard medium. ∗∗∗, p < .001 vs. MSC medium; ∗∗, p < .01 vs. MSC medium. Abbreviations: MSC, mesenchymal stem cells; OD, optical density; SMA, smooth muscle actin.

To define whether the decrease in ECM was related to secretion of MMPs, we studied the effect of MSC-conditioned medium on MMP2 and MMP9 activity in cardiac fibroblasts. As shown in Figure 2A, MMP2 was found in MSC supernatants while MMP9 was not detectable. Incubation with MSC-conditioned medium increased MMP2/MMP9 activities in cardiac fibroblast supernatants (Fig. 2A). RT-PCR Analysis also showed that addition of the MSC-conditioned medium to culture fibroblasts increased MT1-MMP expression and decreased the expression of the metalloproteinase inhibitor TIMP-2 (Fig. 2C). These results showed that conditioned medium from MSCs increased the degradation properties of cardiac fibroblasts by stimulating the amount of MMPs and decreasing the expression of the endogenous metalloproteinase inhibitor.

Figure 2.

Effects of MSC-CDT medium on cardiac fibroblast MMP activities and MT1-MMP/TIMP-2 expression. Effect of fibroblast standard medium, MSC standard medium, MSC-CDT medium, or MSC-CDT medium incubated with anti-HGF antibody (MSC-CDT medium + HGF-Ab) on cardiac fibroblasts MMP2 and MMP9 activities. MMP2 and MMP9 activities in MSC supernatant are also represented (A). Effect of fibroblast standard medium, MSC standard medium, MSC-CDT medium, or MSC-CDT medium + HGF-Ab on collagen type I and type III deposit of MMP-2 wild-type or MMP-2−/− cell line (B). Effect of fibroblast standard medium, MSC-CDT medium, or MSC-CDT medium + HGF-Ab on cardiac fibroblasts MT1-MMP and TIMP-2 expression (C). All results are expressed as a value relative to the value obtained with fibroblast standard medium. ∗∗∗, p < .001 vs. fibroblast standard medium; ∗, p < .05 vs. fibroblast standard medium, ###, p < .001 vs. MSC-CDT medium, #, p < .05 vs. MSC-CDT medium. Abbreviations: MMP-2−/−, MMP-2 knockout; CDT, conditioned; HGF-Ab: hepatocyte growth factor antibody (and not human growth factor); MMP, metalloproteinase; MT1, membrane type 1; ND, not detectable.

To identify the role of MMP2 in the antifibrotic effects of MSC-conditioned medium, we performed additional experiments in fibroblasts obtained from wild-type or MMP2-knockout mice. These cells are particularly interesting to investigate the involvement of MMP2 in collagen accumulation as we showed that, in MMP2-knockout fibroblasts, expression of MMP9, TIMP-1, and TIMP-2 was not modified by MSC-conditioned medium (data not shown). As observed in Figure 2B, incubation of wild-type mouse fibroblasts with MSC-conditioned medium reduced collagen accumulation. This effect was abolished when experiments were performed in MMP2-knockout fibroblasts. These results support the predominant role of MMP2 in the reduction of extracellular collagen accumulation. Nevertheless, it is conceivable that the decrease in fibroblast viability by the MSC supernatant may also contribute to the reduction of ECM deposits.

Recent studies have shown that HGF is one of the factors that reduce fibrosis in different organs. In addition, we demonstrated that MSCs express HGF. To determine whether HGF is involved in MSC-mediated stimulation of MMP activities in cardiac fibroblasts, we incubated MSC-conditioned medium with anti-HGF antibody. Preincubation of MSC-conditioned medium with anti-HGF antibody partially prevented MMP activation and TIMP-2 inhibition. These results suggested that HGF is one of the MSC paracrine factors that mediate regulation of MMPs and TIMP in cardiac fibroblasts.

Effects of MSCs on Cardiac Fibrosis In Vivo

To evaluate the effects of MSCs on cardiac fibrosis, we performed MI by permanent ligation of the interventricular artery 15 days before MSC injection (supporting information Fig. 1A). Two weeks after MI, the ischemic area displayed morphological features of a fibrotic scar characterized by a massive loss of cardiomyocytes (supporting information Fig. 1B, panel c) associated with collagen I and type III deposition (supporting information Fig. 1B, panel d). As previously described, we used an ex vivo pretreatment of MSCs with the pineal hormone melatonin (5 μM, 24 hours) to improve the number of viable MSCs after intraparenchymal injection. In our experiments we injected a total of 6 × 106 melatonin-treated or untreated MSCs, previously labeled with fluorescent nanocrystal QDs, in three points of the infarcted area (Fig. 3A, arrows). Forty-eight hours after administration of MSCs, injection sites were detected with CD90 labeling (MSC surface marker) and QD fluorescence. After injection of untreated MSCs, few QD could be detected in the heart (Fig. 3B, panels c and e). In contrast, pregraft melatonin treatment of MSCs induced an increase of more than seven-fold (74.5 vs. 9.3 arbitrary units, respectively, p < .001) in the number of viable MSCs (QD-positive cells) (Fig. 3B, panels d and e). The viability of QD-positive cells was further supported by the negative TUNEL labeling (supporting information Fig. 2A). To define the mitogenic activity of grafted cells, we performed PCNA staining. Our results showed an intense PCNA labeling within the injection site of animals grafted with melatonin-treated MSCs. These results show that melatonin-treated MSCs undergo proliferation after injection.

Figure 3.

Engraftment and survival of MSCs 48 hours after graft in infarcted heart. Melatonin-treated MSC engraftment (A): CD90 staining and QD fluorescence allows identification of grafted and implanted melatonin-treated MSCs (magnification ×100). Hematoxylin staining of histological sections obtained from hearts 48 hours after injection of untreated or melatonin-treated MSCs (B, a and b). QD fluorescence demonstrates presence of MSCs within infarct area (B, c and d) (magnification ×400). Quantification of QD-labeled MSC fluorescence in heart grafted with melatonin-treated or untreated MSCs (B, e). ∗∗∗, p < .001 vs. MSC (n ≥ 5 animals per group). Abbreviation: QD, quantum dots.

Two months after injection, few MSCs were detectable (data not shown). In contrast, a strong increase in the number of MSCs was observed when MSCs were pretreated with melatonin (Fig. 4A). Some of the QD-positive cells still expressed the MSC surface antigen CD90. On the other hand, this antigen was not detectable in other QD-positive cells, suggesting that some of the grafted MSCs underlie transdifferentiation in different cell types (Fig. 4B, panel a). These cells were negative for immunodetection of connexin 43, a marker of cardiomyocytes (Fig. 4B, panels b, c, and d). In addition, Alcian blue, Alizarin red S, and Red Oil staining of cardiac tissues did not reveal morphological features indicating chondrogenic, osteogenic, or adipogenic activity of grafted cells (Fig. 4C).

Figure 4.

Long-term engraftment of melatonin-treated MSCs in infarcted heart. (A, a) HES staining of histological section obtained from heart 2 months after injection of melatonin-treated MSCs (∗, MSCs; $, collagen accumulation; #, cardiomyocytes). An injection site (A, b). QD-Labeled cells (magnification ×200) (A, c). Higher magnification of (A, b and c), respectively (magnification ×400) (A, d and e). CD-90 labeling (B, a). An injection point stained with hematoxylin or labeled with an anti-connexin 43 antibody, respectively (B, b and c). Positive control (cardiac tissue) of connexin 43 immunodetection (B, d). Alizarin red S staining of MSC injection site (C, a); human femur as positive control (C, b); Alcian blue staining on MSC injection site (C, c) and human chondrosarcoma as positive control (C, d); Red Oil staining on MSC injection site (C, e) and rat adipose tissue (C, f) (n ≥ 5 animals per group). Abbreviation: HES, hematoxylin/eosin/safran.

To assess the effect of MSCs on cardiac fibrosis, we examined type I and type III collagen deposits with red Sirius staining before and after MSC transplantation. As shown in Figure 5, cardiac fibrosis was detectable 15 days after MI and significantly increased after 2 months (27% vs. 39.1%, p < .01). Morphological and quantitative analysis of collagen accumulation demonstrated that the extent of fibrosis was significantly lower in the heart injected with untreated MSCs as compared with the ungrafted ischemic heart (24.5% vs. 39.1%, respectively, p < .001). This effect was amplified when the initial number of surviving MSCs was increased by the melatonin pretreatment (13.9%, p < .001). Interestingly, the percentage of fibrosis was lower than that found after 15 days in untreated animals, suggesting that grafted MSCs induced a degradation of ECM. In accordance with these results, we observed that immunodetection of MMP2 was almost undetectable in untreated animals, whereas there was a large increase in MMP2 in hearts injected with melatonin-treated MSCs (Fig. 6A). The increase in MMP2 was concomitant with the increase in HGF immunodetection (10,601 vs. 4,060 arbitrary units, respectively, p < .001). (Fig. 6B, panels d and e).

Figure 5.

Effect of MSC injection on myocardial fibrosis. Photomicrographs of myocardial sections stained with Sirius red (magnifications ×200) in sham-operated animals (a), 15 days after MI (b) or 2 months after MI (c), 2 months after injection of untreated MSCs (MI + MSC) (d), and 2 months after injection of melatonin-treated MSCs (MI+MSC/melatonin) (e). Quantitative analysis of the fibrotic area in the five different groups (f). $$, p < .01 Vs. MI at 15 days; ∗∗∗, p < .001 vs. MI at 2 months; #, p < .05 vs. MI+MSC at 2 months (n ≥ 5 animals per group). Abbreviation: MI, myocardial infarction.

Figure 6.

Effect of MSC injection on MMP-2 and HGF expression in the infarcted area. MMP-2 staining in the scar area of animals ungrafted (A, a) or grafted with melatonin-treated MSCs (A, b). HGF staining in sham-operated animals (B, a), in the scar area of ungrafted rats (B, b), or grafted with MSCs alone (B, c) or melatonin-treated MSCs (B, d). Quantification of HGF staining (B, e). ∗∗∗, p < .001 vs. MI; #, p < .05 vs. MI+MSC (n ≥ 5 animals per group).

We next investigated whether MSC graft promoted cardiac angiogenesis 2 months after injection. Immunolabeling of histological sections with anti–von Willebrand antibody revealed a higher number of vascular structures in areas injected with melatonin treated MSCs as compared with zones injected with untreated cells (supporting information Fig. 2B).

Effects of MSCs on Ventricular Remodeling and Cardiac Function

Ventricular remodeling was investigated with the use of morphometric analysis of cardiac tissue. Two months after MSC administration, ischemic hearts exhibited a large decrease in infarcted LV anterior wall thickness as compared with noninfarcted heart (0.43 mm vs. 2.14 mm, p < .001) (Fig. 7A). In contrast, LV wall thickness was higher in hearts receiving MSCs (0.86 mm). This effect was significantly (p < 0.05) incremented in hearts injected with melatonin-treated MSCs (1.24 mm).

Figure 7.

Effect of MSCs injection on left ventricular anterior wall thickness and cardiac function. (A): Left ventricular anterior wall thickness 2 months after graft in the four groups: sham-operated animals, MI rats, MI+MSC rats, and MI+MSC/melatonin rats. $$$, p < .001 vs. sham; ∗∗, p < .01 vs. MI; #, p < .05 vs. MI+MSC. (B): Left ventricular ejection fraction, assessed with echocardiography just before and 14, 30, 45, and 75 days after MI in sham-operated animals, MI rats, MI+MSC rats, and MI+MSC/melatonin rats. ∗, p < .01 vs. MI (n ≥ 5 animals per group).

Cardiac function was evaluated by measurement of LVEF with echocardiography (Fig. 7B). Bidimensional echocardiographic images acquired 2 weeks after MI showed a significant decrease in LVEF after MI. LVEF was significantly improved in animals injected with MSCs, as compared with ungrafted rats (36.9% and 23.2%, respectively, p < .05). This effect was higher when MSCs were pretreated with melatonin (39.4%, p < .05). Interestingly, the partial recovery of LVEF was only maintained for the duration of the study when MSCs were previously treated with melatonin (38.3%, p < .05). In contrast, LVEF gradually decreased up to 8 weeks (27.7%) after injection of untreated-MSCs and reached values similar to those of noninjected animals.

These data suggest that the increase in the number of surviving and engrafted MSCs significantly improved the long-term recovery of cardiac function after MI.


Paracrine factors secreted by MSCs play an important role in the beneficial effects of cell therapy [12, 21, 34-36]. Most of the available data on the paracrine activity of MSCs concern the stimulation of vascular cells [16, 20, 21]. Much less is known about the MSC cytokines involved in the regulation of ECM [37]. In this study, we have reported for the first time that MSCs regulate the phenotype of fibroblasts and their ability to cleave ECM. Indeed, we have shown that conditioned medium from MSCs decreased the viability of cardiac fibroblasts and reduced expression of α-SMA, indicating that factors secreted by MSCs act on survival and differentiation of fibroblasts. These effects were associated with a decrease in collagen secretion by cardiac fibroblasts in culture. We hypothesized that the antifibrotic potential of MSCs could be related not only to a decrease in collagen synthesis [17] but also to an increase in the ability of cardiac fibroblasts to degrade ECM. We have shown that MSC-conditioned medium stimulated MMP2 and MMP9 activity as well as MT1-MMP expression in cardiac fibroblasts. This effect was associated with a decrease in the expression of TIMP-2. Experiments performed with fibroblasts from MMP2-knockout mice demonstrated that MMP2 plays a preponderant role in preventing collagen accumulation after incubation with conditioned-medium from MSCs.

Converging evidence suggests that HGF may be involved in the effects of MSCs on the antifibrotic properties of cardiac fibroblasts: first, HGF behaves as an antifibrotic factor in different tissues [23, 38-42]; second, HGF is produced and released by MSCs; third, we showed that anti-HGF antibody prevented the effects of MSC-conditioned medium on MMP2/MMP9 activities and MT1-MMP/TIMP-2 expression in cardiac fibroblasts; finally, an increase in MSCs survival after intraventricular injection was associated with the increase in HGF immunodetection.

To define the relevance of in vivo ECM regulation by MSCs, we used a model of postischemic cardiac fibrosis. Results showed that an extensive fibrotic scar was evident in the ischemic area 2 weeks after MI. The fibrotic response increased with time up to 2 months after myocardial injury. Transplantation of MSCs 2 weeks after injury was associated with a significant decrease in fibrosis 2 months after graft. Furthermore, this antifibrotic effect was amplified by pretreatment of MSCs with melatonin. Interestingly, the percentage of fibrosis in the group with melatonin pretreatment was lower than that found after 15 days in ungrafted animals. These results indicate that injected MSCs induced a degradation of ECM, in accordance with in vitro studies.

As we have previously shown in a rat model of ischemic renal injury [31], MSC administration stimulated long-term angiogenesis after MI. Although angiogenesis may contribute to the decrease in scar extension and improve the recovery of ventricular function, it seems unlikely that this could be relevant in term of ECM degradation and regression of fibrosis.

On the basis of our previous results, which show that melatonin stimulates cytokine secretion by MSCs, it was conceivable that melatonin could increase the production and release of MSC paracrine factors regulating the antifibrotic properties of cardiac fibroblasts. However, we found that the effects of conditioned media from MSCs on fibroblast viability, fibroblast differentiation, collagen accumulation, MMP2/MMP9 secretion, and MT1-MMP/TIMP-2 expression were not increased by MSC pretreatment with melatonin (data not shown). These results suggest that the increase in the antifibrotic activity observed after in vivo injection of melatonin-treated MSCs was related to the improvement of MSC survival rather than to the oversecretion of MSC paracrine factors directly acting on cardiac fibroblasts.

Reduction of ventricular wall thickness and decrease in LVEF are markers of cardiac damage and decline of ventricular function, respectively. Our results showed that MSCs significantly improved these morphological and functional cardiac parameters 2 weeks after injection. However, the partial recovery of ventricular ejection fraction was maintained up to 2 months only when MSC survival was increased by melatonin treatment. These data indicate that the increase in the number of viable cells is critical for the amplification of the beneficial effects of MSCs on cardiac injury and the recovery of ventricular function.

In conclusion, we have shown that MSCs modulate survival, differentiation, and antifibrotic activity of cardiac fibroblasts. These properties of MSCs open new perspectives for understanding the mechanisms of action of MSCs and anticipate their potential therapeutic or side effects.


The authors acknowledge Serge Estaque (Service d'anatomie et cytologie pathologiques, CHU Rangueil, Toulouse, France) for his assistance in tissue embedding and processing. The authors also acknowledge the Service de Zootechnie (INSERM, IFR31, Toulouse, France).

This work was supported in part by INSERM and by grants from the National Research Agency (ANR) (Grant under program Physiopathologie des Maladies Humaines, project SYNMESCARI), the Région Midi-Pyrénées, and the Fondation pour la Recherche Médicale (FRM).