INSERM, U955, Créteil, France and Université Paris-Est, Créteil, France
Correspondence: Anne-Marie Rodriguez, Ph.D., INSERM, U955, 8 rue du Général Sarrail, Créteil F-94010, France. Telephone: +33-1-49-81-37-31; Fax: +33-1-49-81-36-42; e-mail: firstname.lastname@example.org
Mesenchymal stem cells (MSC) are known to repair broken heart tissues primarily through a paracrine fashion while emerging evidence indicate that MSC can communicate with cardiomyocytes (CM) through tunneling nanotubes (TNT). Nevertheless, no link has been so far established between these two processes. Here, we addressed whether cell-to-cell communication processes between MSC and suffering cardiomyocytes and more particularly those involving TNT control the MSC paracrine regenerative function. In the attempt to mimic in vitro an injured heart microenvironment, we developed a species mismatch coculture system consisting of terminally differentiated CM from mouse in a distressed state and human multipotent adipose derived stem cells (hMADS). In this setting, we found that crosstalk between hMADS and CM through TNT altered the secretion by hMADS of cardioprotective soluble factors such as VEGF, HGF, SDF-1α, and MCP-3 and thereby maximized the capacity of stem cells to promote angiogenesis and chemotaxis of bone marrow multipotent cells. Additionally, engraftment experiments into mouse infarcted hearts revealed that in vitro preconditioning of hMADS with cardiomyocytes increased the cell therapy efficacy of naïve stem cells. In particular, in comparison with hearts treated with stem cells alone, those treated with cocultured ones exhibited greater cardiac function recovery associated with higher angiogenesis and homing of bone marrow progenitor cells at the infarction site. In conclusion, our findings established the first relationship between the paracrine regenerative action of MSC and the nanotubular crosstalk with CM and emphasize that ex vivo manipulation of these communication processes might be of interest for optimizing current cardiac cell therapies. Stem Cells2014;32:216–230
Heart failure occurring after acute myocardial infarction (MI) is among the main causes of death in Western countries. Cell therapies, particularly those based on mesenchymal stem cells (MSC), represent one of the most promising approaches to repair damaged heart tissues. Studies performed during the past decade strongly indicate that intramyocardial delivery of MSC from various tissue origins ameliorates heart function after infarction [1-3]. Nevertheless, the mechanism(s) by which MSC exert their therapeutic action is far from being understood, and further investigations are required for improving the modest efficiency of existing cardiac cell therapies.
MSC have been reported to ameliorate heart function primarily through the release of soluble factors promoting new vessel formation, protection of cardiomyocytes (CM) against apoptosis, induction of beneficial extracellular matrix remodeling [4-7], recruitment of bone marrow derived stem cells (BMSC) to the injury site, and activation/differentiation of resident cardiac progenitor cells [4-7]. Recently, a new concept has emerged in which the secretome expression of infused MSC might depend on the surrounding microenvironment in which they were engrafted through the release of physiological cues from injured cells . A series of in vitro studies showed that stress signals emitted by suffering cells could be transmitted to MSC or progenitor cells through several pathways including gap junctions , apoptotic bodies , soluble factors  but also membrane thin channels, commonly referred to as tunneling nanotubes (TNT) which are able to ensure transient membrane continuity between remote cells. Although TNT connections and their roles in vivo remain to be fully characterized due to difficulties in assessing the occurrence of these phenomenons in living animals, a limited number of in vitro studies suggest that these cell-to-cell communication pathways are engaged by MSC to rescue astrocytes from oxidative stress-mediated injury  as well as endothelial cells against premature senescence . Likewise, we and others have observed that CM can also communicate with MSC or endothelial progenitors through transient TNT-like connections made of f-actin and microtubules [14-16]. This process contributes to prolong the survival of adult dying CM through the transfer of stem cell-derived functional mitochondria . Beside the apparent role of TNT in transferring MSC components into suffering somatic cells, it remains unknown whether TNT exchanges in the opposite way might affect MSC behavior. Here, we hypothesized that cell-to cell communication of MSC with CM and in particular those engaged by TNT are important processes by which the microenvironment controls the secretome of MSC and thereby their beneficial effects.
To address this question, we developed a species mismatch coculture system consisting of terminally differentiated CM from mouse in a distressed state and human multipotent adipose derived stem cells (hMADS). In this setting, more than 90% of primary CM were characterized by high oxidative stress associated to either necrosis or apoptosis (Supporting Information Fig. 1). Since similar events occur in vivo after MI [17, 18], this in vitro approach is thought to mimic the microenvironment of an injured heart.
Human Primary Cells, Cell Lines, and Culture Conditions
Human mesenchymal stem hMADS cells were isolated using previously described procedures . Human BMSC (hBMSC) were generously given by Dr Hélène Rouard (Etablissement Français du Sang [EFS], Creteil, France). Human primary human adult heart fibroblasts and progenitors were from PromoCell Heidelberg, Germany, http://www.promocell.com/ and Innoprot, Bizkaia, Spain, http://www.innoprot.com, respectively. Cell culture conditions were detailed in online Supporting Information.
Coculture Between hMADS and Mouse Adult CM
Adult ventricular CM were isolated from hearts of 2–5 months-old male mice from C57BL/6J (Janvier France, http://www.janvier-labs.com) or GCAG-green fluorescent protein (GCAG-GFP) transgenic strains  as previously described  and cocultured with hMADS in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% Fetal Bovine Serum (FBS) at a density of 3,500 cells per centimeter square each (ratio 1:1). Single cultures of CM or hMADS were seeded at the same concentration. Indirect cocultures were performed with cell culture inserts containing polycarbonate membrane (0.4 µm or 1 µm size pore, Millicell, Millipore, Billerica, MA, http://www.millipore.com).
Collection and Biological Activities of Culture Conditioned Media
For collection of conditioned media from single or cocultures, hMADS and CM were seeded at 105 cells per milliliter, in DMEM containing 0.8% FBS during 24 hours. Cytokines from culture supernatants were measured by luminex using MILLIPLEX MAP kits (Millipore) and the Bio-plex 200 system (Bio-Rad, Hercules, CA, http://www.bio-rad.com). Supernatant angiogenic activities were evaluated on human umbilical vein endothelial cells (HUVEC) by 2D and 3D angiogenesis (Promocell, GmbH) assays. Chemotactic activities of culture supernatants were assessed by using µ-slide chemotaxis (Ibidi, Biovalley Marne La Vallée, France, http://www.biovalley.fr/) seeded with 18 × 103 human Bone Marrow Derived Cells (hBMDC) per channel. Neutralizing antibodies against human VEGF165 human Vascular Endothelial Growth Factor (hVEGF) (0.08 µg/mL), human HGF human Hepatocyte Growth Factor (hHGF) (0.4 µg/mL), human MCP-3 human Monocyte Chemotactic Protein-3 (hMCP-3) (20 µg/mL), and human SDF-1 human Stromal cell-Derived Factor-1 (hSDF-1) (3 µg/mL) were from R&D systems (Minneapolis, MN, http://www.rndsystems.com). Antiapoptotic effects of supernatants were evaluated on mouse neonatal cardiomyocytes isolated from 1 to 3 days old C57BL/6J mice  by PE-Annexin V (PhycoE rythrin) staining (BD Pharmingen, San Diego, CA, http://www.bdbiosciences.com/index[lowenus.shtml]) and flow cytometry analysis.
Methyl Thiazol Tetrazolium (MTT) assays (Sigma-Aldrich, St. Louis, MO, http://www.sigmaaldrich.com) were performed on cardiac fibroblasts or progenitors initially seeded at 104 cells per centimeter square as previously reported . More details on this methodology are available online (see Supporting Information).
Real-Time Polymerase Chain Reaction Assays
Total RNA was extracted using the Qiagen RNeasy Mini Kit (Qiagen, Hilden, Germany, http://www1.qiagen.com) and then reverse-transcribed using the Superscript First-Strand Synthesis System (Invitrogen, Carlsbad, CA, http://www.invitrogen.com) and Oligo(dT)20. Quantitative reverse transcriptase polymerase chain reaction (qRT-PCR) reactions were performed in triplicate on a 7900 real-time PCR detection system (Applied Biosystems, Foster City, CA, http://www.appliedbiosystems.com) using Platinium SYBR Green qPCR SuperMix (Invitrogen). PCR conditions were 50°C for 2 minutes, 95°C for 2 minutes, 45 cycles at 95°C for 15 seconds, and 60°C for 45 seconds, using glyceraldehyde 3-phosphate deshydrogenase (GAPDH) as the reference gene. Primer sequences are described in Supporting Information Table 1. Results are reported as mean ± SD.
Cells were fixed with 4% paraformaldehyde (PFA) and stained with antibodies against GATA-4 (R&D Systems) and phospho Histone H3 (pH3, Abcam, Cambridge, U.K., http://www.abcam.com). Fluorescent secondary antibodies were from Jackson ImmunoResearch Laboratories Inc. (West Grove, PA, http://www.jacksonimmuno.com). Wheat germ agglutinin (WGA) conjugated to Alexa 647 (10 µg/mL) was from Invitrogen. For costaining with phalloidin-rhodamine (5 µg/mL, Sigma-Aldrich) and Fluorescein Isothiocyanate (FITC)-conjugated α-tubulin (Abcam), hMADS were fixed with 4% PFA followed by cold acetone. Nuclei were stained with Hoechst 33342 (Sigma-Aldrich). Fluorescence was analyzed with a Zeiss Axioplan 2 Imaging microscope.
Intercellular Dye Exchanges and Inhibition of Cell-to-Cell Communication Pathways
Prior to coculturing, CM were labeled with MitoTracker Red FM (1 µM) or calcein AM (1 µM) (Molecular Probes, Eugene, OR, http://probes.invitrogen.com). Intercellular exchanges were examined by flow cytometry or conventional microscopy (Zeiss Axioplan 2 Imaging microscope).
To inhibit gap junctions as well as f-actin or α-tubulin polymerization, fresh cocultures were treated during 24 hours with 100 µM 18α-glycyrrhetinic acid (18α-GA, Sigma-Aldrich), 2.5 × 10−8 M latrunculin A (Invitrogen) or 5 × 10−8 M nocodazole (Sigma-Aldrich), respectively. Analysis of cell-to-cell communication processes by flow cytometry and transmission electron microscopy are detailed online (Supporting Information).
Mouse MI and Cell Injections
All animal procedures were in accordance with the regulations issued by the French Ministry of Agriculture and were approved by the prefecture of our administrative district (Prefecture du Val de Marne, France; Licenses to conduct animal research: B 94-028-7 and A94-028-245). MI was induced in 8–10-week-old male C57BL6/J mice by occluding the left coronary artery for 90 minutes and reperfusing the artery for 10 minutes. Just after surgery, approximately 2.0 × 105 cocultured cells in a total volume of 20 µL were injected into the myocardium surrounding the infarcted site. Control mice received Hanks' balanced saline solution (HBSS) or similar amount of hMADS or CM.
To obtain blood chimeric mice, BM cells from GCAG-GFP transgenic mice were injected retro-orbitally into 8-week-old C57BL/6 mice (3 x 106 cells per mouse) previously irradiated at 9 Gy. Recipient mice were treated with 10 mg/kg per day ciprofloxacin for 14 days. Blood chimerism of >90% was controlled at 8 weeks post-transplantation.
Echocardiography was performed before MI, 5 and 20 days postinfarction using a 13-MHz linear transducer (VIVID 7 Echocardiogram, GE Medical System) as previously described . Details on this methodology are available online (Supporting Information).
Heart Histology and Immunohistochemistry
After mouse sacrifice, hearts were either directly snap-frozen in isopentane precooled with liquid nitrogen for GFP detection, perfused, and fixed with 4% PFA followed by overnight saturation in 20% sucrose. Fibrosis was determined by quantification of Masson's trichrome staining of cryostat serial sections (8 µm) separated by 300 µm from the apex to the base (Sigma-Aldrich) (ImageJ 1.42q, NIH). Capillary density of peri-infarct area was determined by counting microvessels stained with isolectin B4 (40 µg/mL, Sigma-Aldrich) from at least 20 randomly selected fields in border areas.
After fixation with 4% PFA, frozen heart sections were incubated with primary antibodies against GATA-4 (R&D systems), PH3 (Abcam), human Lamin AC (Novocastra), GFP (Gene Tex), caspase-3 (Cell Signaling, Beverly, MA, http://www.cellsignal.com), human mitochondria (Abcam). Fluorescent secondary antibodies were from Jackson Immunoresearch laboratories. Fluorescence was analyzed by conventional (Zeiss Axioplan 2 Imaging microscope) or confocal microscopy (Zeiss LSM 510 Meta).
Statistical analysis was performed using Prism 5.04 Software (GraphPad Software), and results are reported as mean ± SD. Comparison between more than two experimental groups were analyzed by one-way ANOVA followed by Bonferroni's post hoc test. For cardiac function measurements, left ventricular ejection fraction (LVEF) differences between groups and time (day 5–21) were evaluated by analysis of repeated measures followed by post hoc Student's t test for each treatment group. Other comparisons between two groups were performed using a two-tailed Student's t test. p values smaller than .05 were considered significant.
Cell-to-Cell Communications with CM Alter the MSC Secretome
To determine whether early cell-to-cell communication between hMADS and distressed adult CM could alter hMADS secretome, cocultures were performed with hMADS and freshly isolated CM in medium with low serum to minimize contamination by exogenous bovine soluble factors. After 24 hours of coculturing (time at which more of 95% of CM were dead), supernatants were collected and the amount of 20 cardioprotective diffusible mediators [25, 26] was measured by using human specific luminex assays. Changes in secretion levels were found for eight of these factors between supernatants from hMADS cultured alone or cocultured with CM, those from cocultures containing higher amounts of VEGF (×1.61 ± 0.28), HGF (×5.3 ± 0.94), SDF-1α (×1.31 ± 0.1), MCP-3 (×1.58 ± 0.13), Interleukin 6 (IL6) (×1.44 ± 0.18), and Growth Regulated Oncogene alpha (GROα) (×2.40 ± 0.31) and decreased concentration of Monocyte Chemotactic Protein-1 (MCP-1) (×0.74 ± 0.07) and Matrix Metalloproteinase-3 (MMP-3) (×0.54 ± 0.07) (Fig. 1A, 1B). Similar changes were found at the transcriptional level in the human genome strengthening the stem cell origin of paracrine activation (Fig. 1C).
In addition, carboxyfluorescein diacetate succinimidyl ester (CFSE) flow cytometry analysis and qRT-PCR experiments demonstrated that changes in secretome cannot be ascribed either to hMADS proliferative changes in the presence of CM or to differences in cell density between coculture and monoculture (data not shown and Supporting Information Fig. 2). Finally, paracrine differences existing between cocultivated and control hMADS, except for MCP-1 and MMP3, were abrogated in cocultures performed with a cell culture insert which prevented physical interaction between hMADS and CM (Fig. 1A–1C). Taken together, these data indicate that preconditioning with CM alters the secretome of hMADS through different communication routes including both paracrine pathways and direct cell-to-cell contacts.
Cell-to-Cell Communications with CM Enhance the Paracrine Repair Function of MSC
Next, we tried to assess the physiological impact of coculture-induced hMADS secretome changes. We found that supernatants collected from cocultures, in comparison to supernatants from control hMADS, CM, or mix of the two ones promoted higher angiogenesis of endothelial HUVEC cells (Fig. 2A, 2B) and a significantly faster migration of BM-derived MSC. It is worth to note that of these BM cells exhibited a GATA-4+/PH3+ cardiac like progenitor phenotype (Fig. 2C). Importantly, supernatants collected from indirect cocultures (i.e., made with transwell insert) failed to stimulate hMADS proangiogenic and prochemotactic properties suggesting that these phenomena were triggered by a direct contact between stem and cardiac cells (Fig. 2B, 2C). In addition, assays performed with neutralizing antibodies against some human-specific growth factors upregulated in cocultures revealed that hVEGF and hHGF triggered the proangiogeneic activity of cocultured hMADS whereas hVEGF, hHGF, hMCP-3, and hSDF-1α participitated to their prochemotactic effects (Fig. 2D, 2E). Finally, coculture preconditioning was found not to alter the paracrine ability of naïve hMADS to stimulate proliferation of human primary cardiac progenitor cells (data not shown), protect CM against serum withdrawal-induced apoptosis (Fig. 2F), and inhibit cardiac fibroblast proliferation and collagen synthesis (Fig. 2G and data not shown).
TNT Can Be Selectively Inhibited by Latrunculin A or Nocodazole Treatments
To assess the involvement of TNT in the alteration of hMADS secretome, we attempted to develop a pharmacological approach to specifically disrupt TNT networks. As TNT connecting stem to cardiac cells are composed of f-actin and microtubules as shown by fluorescent and transmission electron microscopy (Fig. 3A, 3B), we treated cocultures with 2.5 × 10−8 M latrunculin A or 5.0 × 10−8 M nocodazole, to inhibit, respectively, polymerization of these two cytoskeletal components. As expected, the average number of heterologous TNT connections was dramatically decreased after each of both drug exposures (Fig. 3C).
To determine if these two chemical compounds at the concentration used could interfere with other cell-to-cell communication pathways, we then evaluated intercellular exchanges between CM and hMADS of two other compounds: calcein, a small gap junction diffusible molecule and MitoTracker Red, a fluorescent probe that labels mitochondria. Flow cytometry analysis showed that mitochondria and, to a lesser extent, calcein were transferred from CM to hMADS within the first 24 hours of coculture (Fig. 3F; Supporting Information Fig. 3A). A TNT transfer was also confirmed by fluorescence and transmission electron microscopy (Fig. 3D, 3E). The quantity of transferred mitochondria was significantly decreased after coculture treatment with either latrunculin A or nocodazole whereas flow of calcein was sensitive only to latrunculin A (Fig. 3G). In addition, these intercellular exchanges were not altered by the gap junction inhibitor α-18GA indicating that the action of latrunculin A or nocodazole was not attributable to blockade of GAP channel activity (Fig. 3G). Lack of functional gap junction between stem and cardiac cells within the first 24 hours of coculture was further confirmed by data showing that calcein flux was unidirectional and that calcein failed to diffuse from hMADS to CM (Supporting Information Fig. 4A, 4B).
Finally, to exclude a potential interference of these drugs with cell signaling processes mediated by either exosomes or largest microparticules as apoptotic bodies, cocultures were performed with TW insert of 0.4 or 1 µm pore size, respectively. In these conditions, intercellular traffic of calcein and organelles were dramatically decreased but remained unchanged by addition of latrunculin A or nocodazole (Fig. 3F, 3G; Supporting Information Fig. 3A). In agreement with these results, we showed that these drugs do not affect phagocytosis or endocytosis processes in hMADS through exposure with pHrodo red S. aureus bioparticles or pHrodo green dextran, respectively (Supporting Information Fig. 3B, 3C). Together, these data indicate that latrunculin A and nocodazole, primarily inhibit TNT, without apparent effect on other cell-cell communication routes.
TNT Mediate MSC Secretome Changes Improving Angiogenesis and Chemotaxis
To assess the role of TNT in changes of the hMADS secretome during coculture, human specific luminex assays were performed on supernatants collected from latrunculin A- or nocodazole-treated cocultures. We found that treatment with both drugs for 24 hours abrogated the secretory stimulation of hVEGF, hHGF, hMCP-3, and hSDF-1α supporting a role of nanotubular connections. In contrast, these drugs had no effect on coculture-induced secretion of human IL-6, GRO-α, MMP3, and MCP-1 suggesting that production of these factors was triggered by other cell-to-cell communication mechanisms (Fig. 4A).
Accordingly, we found that the paracrine ability of hMADS to promote HUVEC angiogenesis and BM-MSC chemotaxis failed to be induced in cocultures exposed to the drugs (Fig. 4B, 4C), indicating that TNT composed of both f-actin and microtubules play a critical role in the modulation of MSC paracrine regenerative responses to CM-emitted cues.
Finally, we investigated whether TNT might be affected by inflammation, a key component of the infarcted microenvironment, by priming hMADS with Tumor Necrosis Factor alpha (TNF-α) or Interferon gamma (IFN-γ) prior to coculture with CM. Under these conditions, the number of TNT connecting stem to cardiac cells was increased, and the transcription of TNT-dependent cytokines was significantly activated (Supporting Information Fig. 5A, 5B), suggesting that TNT cell-to-cell communication is sensitive to inflammatory stimuli.
Coculture with CM Improves MSC Cardiac Cell Therapy Efficacy
We then investigated whether coculturing hMADS with CM might have an impact on their cardiac cell therapeutic efficacy. Immunocompetent mouse hearts previously subjected to myocardial ischemia/reperfusion were injected with one of the following: (a) HBSS saline solution (control); (b) human cells grown alone; (c) mouse CM grown alone; or (d) coculture of hMADS and CM. Echocardiography follow-up indicated that from day 5 to day 21 postinfarction, mice treated with either HBSS, mouse CM, or hMADS exhibited no significant changes in the LVEF of the heart whereas LVEF in mice treated with cocultured cells was improved by 35% (Fig. 5A). The improved cardiac function following injection with cocultured cells was associated with a significant reduction in infarct size compared to all other conditions (Fig. 5B, 5C). In addition, we found that a similar number of cocultured or naïve hMADS cells were present in mouse hearts at day 3 following infarction and cell delivery, while at day 7, both kinds of human cells were hardly detectable (Supporting Information Fig. 5C). These results indicate that the regenerative capacity of cocultured cells cannot be attributed to a survival advantage over their non cocultured counterparts.
To further investigate why cocultures were superior to naïve hMADS in restoring heart function, we examined angiogenesis, CM apoptosis, and cardiac progenitor mobilization in mouse hearts at days 3 and 7 postinfarction. A higher capillary density, assessed by isolectin B4 staining, was found in the border zone of coculture-treated hearts compared to all the other groups examined (Fig. 5D). Moreover, the proportion of cardiac progenitor-like cells present in the peri-infarcted myocardium and evaluated by counting nuclei expressing the early cardiac commitment GATA-4 marker was significantly higher in mice treated with cocultured cells than in those treated with HBSS, MSC, or CM alone (Fig. 5E). To determine whether these cells originated from BM, we quantified the proportion of GATA-4+/GFP+ in chimeric mice reconstituted with GFP-BM cells at day 3 postinfarction. This time point was chosen as optimal for discriminating homing to self-renewal of circulating progenitor cells at the infarction site. We found that GATA-4+/GFP+ cells were more abundant in coculture-treated mice (Fig. 5F) indicating that the increased concentration of GATA-4+ cells at the myocardial injury site of these animals was due at least in part to a greater mobilization of BM derived cells having a cardiac progenitor-like phenotype. Finally, the rate of apoptotic CM after caspase-3 immunostaining was found similar in mice peri-infarct area treated with hMADS and cocultures at days 3 and 7 postinjection (Fig. 5G).
Consistent with our in vitro data, our in vivo results indicate that improved cardiac cell therapeutic efficacy of hMADS through coculture is associated with an enhancement of the angiogenesis and homing of BM-derived cells that exhibit a cardiac progenitor-like phenotype in the myocardium. The fact that neither endothelial or cardiac trandifferentiation nor permanent cell fusion were observed in infarcted hearts treated with naïve or cocultivated hMADS (Supporting Information Fig. 6) suggests that the repairing properties of engrafted cocultured cells are likely to be mediated by a paracrine process.
Coculture-Mediated MSC Paracrine Changes Are Reinducible by New Exposure to CM
To gain further insights on whether the paracrine regenerative effects of hMADS was stimulated by the coculture, secretome of hMADS was examined at day 4 corresponding to the stage at which cells were injected. Luminex and real-time PCR assays showed that at this time point, secretome differences between hMADS grown alone or in coculture were flattened for most of the soluble factors (Fig. 6A, 6B) suggesting that the higher regenerative capacity of cocultivated hMADS was unlikely due to a higher production of human cardioprotective factors at the time of cell injection. Nevertheless, following second exposure to CM to mimic what happens after intramyocardial delivery of cocultivated cells, the first primed hMADS increased their transcriptional expression and release of human derived-VEGF, MCP-3, HGF, IL-6, and GROα (Fig. 6C and not shown) revealing that the paracrine stimulation of hMADS through coculture is transient by de novo contact with CM.
In the attempt to correlate our in vitro observations with the in vivo regenerative potential of cocultivated cells, we examined if the transplanted cocultivated stem cells could directly communicate with resident CM by examining heterocellular exchanges of human stem cell mitochondria at the site of infarction. Immunohistochemistry analysis with anti-human mitochondria and cardiac troponin T antibodies revealed the presence of some CM containing human mitochondria at the peri-infarct borders 3 days after treatment with either hMADS alone or in coculture (Fig. 6D). These observations strongly suggest that stem cells and resident CM physically interacted with each other in vivo and that such interaction might boost the paracrine activities of cocultivated hMADS.
Treatment of ischemic cardiac tissue with stem cell therapy is one of the most promising strategies to prevent progression toward heart failure. In this active research area, intensive efforts are currently devoted to characterize the mechanisms by which MSC can repair broken hearts. Such basic knowledge is of critical importance for the rational design of novel MSC-based cell therapies and optimization of existing cell therapies. Several studies, including the ones based on clinical trials, tend to attribute cardiac benefit of transplanted MSC to paracrine effects . However, it remains so far unclear if diffusible mediators involved in cardiovascular regeneration are produced by intramyocardially engrafted MSC in a constitutive fashion or in response to their contact with the microenvironment of injured hearts. This study enlightens this fundamental but largely unexplored question through coculture assays of hMADS with mouse adult CM in a suffering state to mimic the in vivo conditions of heart-resident CM following a myocardial infarction [14, 28]. This study confirms that, like other types of MSC, also hMADS constitutively secrete chemical factors that limit myocardial loss and scarring and stimulate angiogenesis [29-31]. In addition, data presented here provide new insights and clarifies how the “innate” humoral regenerative function of MSC is enhanced after transplantation through cell interaction and communication with distressed CM. More specifically, we provide in vitro evidence that cell-to-cell communication between stem and cardiac cells triggers changes in the hMADS secretome expression and consequently enhance the hMADS effectiveness in promoting angiogenesis and chemoattraction of BM-derived mesenchymal progenitors, these two latter processes being of key importance for cardiac repair [4-7]. Our report also provides partial identification of the hMADS-released factors responsible of the above stated effects. In particular, we report that upregulated secretion of VEGF and HGF by cocultivated hMADS account for the enhanced proangiogenic activity while over secretion of VEGF, HGF, SDF-1α, and MCP-3 by the means of coculture participates to the heightened BM-derived progenitor recruitment potential of hMADS. .Additionally, although not formally demonstrated in our study, stimulated release of other factors such as IL-6 and GRO-α by cocultivated hMADS may indirectly enhance their proangiogenic effects by, respectively, activating the secretion of VEGF and/or the myocardial homing of BM-derived endothelial progenitors [32, 33]. Concomitantly with the increased production of soluble factors considered, the enhanced salutary effects of cocultivated hMADS may also be due to downregulation of “deleterious” diffusible molecules that contribute to adverse left ventricular postacute MI. These include MCP-1, previously reported to exacerbate myocardial inflammation , and MMP-3, incriminated in extracellular matrix degradation . Although we cannot exclude the involvement of other important factors, it is curious to note that in this study early cell-to-cell communication events between MSC and CM does not seem to affect the paracrine action exerted on heart-derived progenitor cells, cardiac fibroblasts, and apoptotic CM. However, such mechanisms could take place later in the coculture and, alternatively, hMADS could act on these cell types by a paracrine-independent pathway.
By characterizing the cell-to-cell communication routes involved in the hMADS secretome alteration, we discovered a novel functional facet of nanotubular connections between stem and cardiac cells consisting of a direct regulation of the regenerative paracrine properties of MSC. This way of cell-to-cell communication has been considered so far as a common salvage process by which MSC could directly rescue various kinds of somatic cells such as CM [14, 36], astrocytes , and endothelial cells  exposed to stress signals. Whether this process allowing bidirectional intercellular exchanges has an impact on the fate of MSC is still controversial. In fact, while some studies showed that transfer of cardiac cellular components through MSC by tunneling conduits favor MSC transdifferentiation into cardiac lineage, works by our group and others failed to confirm this mechanism. This apparent discrepancy is likely due to differences in the developmental stage of CM used for coculture. Indeed, studies involving cardiac MSC transdifferentiation are generally conducted with healthy neonatal cardiac cells capable of division [37-39]. In contrast, such a phenomenon is not observed in nonreplicative adult CM characterized by a suffering state [14, 40]. To our knowledge, we provide here the first evidence that membrane tunneling bridges, enriched of f-actin and microtubules and already found to connect MSC and CM [14, 41], are clearly involved in the hMADS paracrine switch that leads to heart repair. The key mediators mainly responsible of the increased angiogenic and chemotactic effects of hMADS cells following coculture include VEGF, HGF, SDF-1α, and MCP-3. We postulate that the paracrine activation of hMADS might be triggered by stress signals sent by CM along TNT. Although the nature of these signals remains to be elucidated, miRNA, proteins, or organelles such as lysosomes or mitochondria might contribute to different extents to the phenomenon observed.
Our study shows that part of hMADS paracrine changes is mediated by other pathways of cell-to-cell communications, as illustrated by the “human-specific” secretion of IL-6, GRO-α, MMP-3, and MCP-1. For instance, uptake by hMADS of large microparticles (up to 0.4 µm)  or apoptotic bodies of cardiac origin [10, 43] might stimulate the release of IL6 and GRO-α, while soluble factors or exosomes/small microvesicles  might decrease the production of MMP-3 and MCP-1. However, our data suggest that these communication pathways are not responsible of the functional changes observed in hMADS during coculture.
In relation to clinical considerations, we show that in vitro preconditioning of stem cells with CM through coculture improves stem cell therapeutic efficacy. This is an exciting finding since a higher functional recovery of hearts engrafted with cocultured hMADS is accompanied by more efficient paracrine mechanisms that promote angiogenesis and BM cell myocardial homing. The increased beneficial effects of cocultivated versus naïve hMADS are unlikely due to the presence of mouse CM in the cocultures because these cells were dead at the time of intramyocardial cell injection or to differences in their secretome expression. Although our study does not formally demonstrate that the improved regenerative properties of cocultured cells in vivo is due to TNT interactions with dying CM, this hypothesis is supported by the fact that (a) TNT communication between stem and cardiac cells probably occur in vivo, as suggested by mitochondria exchanges from stem to myocardial cells preferentially ensured by these structures and (b) in vitro, MSC subjected to a second CM exposure activated their secretome more efficiently than during the first one and as a function of the CM ratio. Based on these observations, the activation of hMADS secretome following engraftment might be expected to be greater than in vitro since MSC in vivo should encounter a broader range of CM.
Albeit a direct implication TNT effects is extremely difficult to assess in vivo, this issue could be addressed indirectly by comparing the efficacy of cell therapy as a function of increasing doses of administered cells and/or by using different routes of cell delivery. In this context, as cells may be unable to interact with CM because they may clump when delivered at high concentrations or be retained in interstitial spaces when intravenously injected, the repairing processes involving TNT should be minimized
Besides a role for TNT, we cannot exclude that other mechanisms could account for the heart repair properties of cocultivated MSC. These might involve such as those promoted by direct cell-to-cell contact with CM or cell-to-cell communications with neighboring endothelial, fibroblast, and/or resident cardiac stem cells. Furthermore, our study raises questions about the in vivo biological significance of “stem” mitochondrial transfer into CM. Despite the fact that such a process was found to protect CM against apoptosis in vitro (Supporting Information Fig. 7), this role was not confirmed by our in vivo experiments since stem cell administration into infarcted myocardium did not significantly decrease cardiac apoptosis. Although the implications of this phenomenon in vivo remains to be confirmed, we speculate that a transfer of human stem cell mitochondria should dramatically improve the bioenergetic status of the cardiac cells and perhaps also the overall metabolism to combat injury.
Our study, addressing for the first time the cardiac regenerative potential of hMADS in vivo, argues against the commonly accepted concept that MSC infusion affords functional benefit in the infarcted myocardium [1-3, 45, 46], since hMADS fail to promote significant functional recovery of injured heart function. Nevertheless, similar negative findings about the therapeutic effectiveness of MSC have been also previously documented by others [47-52]. This apparent discrepancy might be explained by several factors including the source and species of MSC, number and mode of cell delivery, animal model, and extent of left ventricular dysfunction . Nevertheless, of important note is that our model significantly differs from previously described studies regarding the immunological context in which our experiments were done. Indeed, most of the positive studies on human MSC have used immunodeficient animals [53-57] while, in sharp contrast, our experiments were carried out on immunocompetent mice. In this xenogeneic context, hMADS should induce an immune rejection earlier after their transplantation, thus compromising heart repair, as reported for human engrafted MSC in rat infarcted myocardium [42, 49, 58]. Importantly, the improved regenerative properties of cocultured cells are unlikely to result from an immune survival advantage over their naive counterparts as both human cells were similarly cleared in immunocompetent mouse hearts. Nevertheless, we cannot exclude that the enhanced benefit of cocultured cells may be partially explained by a differential response of cocultured and naive hMADS toward the inflammation associated to myocardial infarction. Although the role of inflammation requires further investigations, our in vitro observations support the idea that TNT cell-to-cell communication can be improved when hMADS are primed with inflammatory cytokines prior to coculture. To conclude, our study provides new insights on the mechanisms underlying the paracrine beneficial activities of MSC and supports the intriguing possibility that reparative capacities of stem cells can be enhanced by exogenously manipulating these properties.
Our study provides the first evidence that cell-to-cell communication through TNT with distressed CM improves the paracrine regenerative function of human MSC and their ability to repair infarcted heart.
We are grateful to A. Henry and X. Decrouy for helpful flow cytometry and confocal microscopy technical assistance, respectively, as well as B. Chazaud (INSERM 1016, Paris), R. Motterlini and R Foresti, (INSERM U955, Creteil) for critical manuscript reading. This work was supported by funding from French National Institute of Health and Medical Research (INSERM) and Association pour la Recherche et l'Etude des Maladies Cardiovasculaires (AREMCAR).
F.F.: conception and design, collection and assembly of data, data analysis and interpretation, and manuscript writing; P-F.L., O.L.C., T.D., R.S., C.T., A.S., and J.R.: collection and assembly of data and data analysis and interpretation; R.M., A.G., C.M., and M.S.: collection and assembly of data; L.H.: administrative support and final approval of manuscript; J-L.D-R.: conception and design, financial supports, and final approval of manuscript; A-M.R.: conception and design, collection and assembly of data, data analysis and interpretation, and manuscript writing. F.F. and P-F.L. contributed equally to this article.