OUP user menu

Epicardial adipose stem cell sheets results in greater post-infarction survival than intramyocardial injections

Hadhami Hamdi, Valérie Planat-Benard, Alain Bel, Etienne Puymirat, Raghed Geha, Laetitia Pidial, Hany Nematalla, Valérie Bellamy, Philippe Bouaziz, Séverine Peyrard, Louis Casteilla, Patrick Bruneval, Albert A. Hagège, Onnik Agbulut, Philippe Menasché
DOI: http://dx.doi.org/10.1093/cvr/cvr099 483-491 First published online: 12 May 2011


Aims Intramyocardial injections of cells can damage tissue and enhance dissociation-induced cell death. We assessed whether epicardial delivery of cell sheets could overcome these issues in a rat model of chronic myocardial infarction.

Methods and results Eighty-two rats that had undergone coronary ligation and simultaneous harvest of fat tissue to yield the adipose-derived stromal cell (ADSC) fraction were randomized 1 month after infarction to receive injections of either control medium (n= 24) or 10 × 106 autologous ADSC (n= 37) or the epicardial deposit, onto the infarcted area, of a trilayered ADSC sheet (10 × 106, n= 21) prepared by culturing cells on temperature-sensitive dishes. Some treated rats received green fluorescent protein labelled ADSC. Survival, function, and cell engraftment were blindly assessed after 2 months. Prior to implantation, cell sheets and suspended cells were assessed for the expression of extracellular matrix constituents and molecules involved in angiogenesis and cardiac remodelling. The survival rate of rats receiving the cell sheets was significantly higher than after cell injections (73 vs. 41%, P = 0.01). This correlated with the absence of left ventricular (LV) remodelling in the cell sheet group, as end-diastolic volume only increased by 2.8% compared with baseline [95% confidence interval (CI): −18.7%; +30.0%, P = 0.81] vs. increases of 25.9% (−0.4%; +59.2%, P = 0.05) and 51.2% (+18.6%; +92.8, P = 0.001) in the cell and medium injection groups, respectively. Sheets also resulted in a greater cell engraftment possibly related to the greater expression of extracellular matrix constituents.

Conclusion The better preservation of LV geometry afforded by ADSC sheets is associated with increased survival and engraftment, which supports the concept of an epicardial delivery of cell-seeded biomaterials.

  • Adipose-derived stromal cells
  • Transplantation
  • Chronic ischaemic heart failure
  • Cell sheet
  • Extracellular matrix
  • Temperature-responsive surface

1. Introduction

The efficacy of cardiac cell therapy is still hampered by the high rate of cell death occurring over the first days after transplantation. In a previous head-to-head comparative study,1 we have shown that the epicardial delivery of a cellularized biomaterial (either a myoblast-seeded collagen patch or a sheet of myoblasts generated by culturing cells onto temperature-responsive dishes) resulted in better outcomes than conventional hand-held transepicardial injections. This result, which was in keeping with those previously reported by the Japanese group which has pioneered this technology,24 was attributed to the generation, during the culture period, of an extracellular matrix preserving cell cohesiveness and thereby reducing the component of apoptotic death related to their pre-injection proteolytic dissociation.5,6 So far, however, this mechanistic assumption has been supported by few data. The present two-step study was therefore designed (i) to characterize, in vitro, the composition of cell sheets with regard to the presence of extracellular matrix constituents, and then (ii) to compare in an in vivo rat model of myocardial infarction, cells sheets, and intramyocardial injections, with survival as the primary endpoint. The reason for this choice is that most, if not all, experimental studies have reported that cells, regardless of their phenotype, improved left ventricular (LV) function, a finding in sharp contrast with the still rather marginal benefits of this therapy in patients. It was thus reasoned that survival would be a more sensitive, more robust and equally relevant marker of treatment effects. Adipose-derived stromal cells (ADSC) were selected as the candidate cells because (i) they can be harvested and grown in a straightforward fashion, and (ii) they secrete a wide array of angiogenic and anti-apoptotic factors7,8 that likely account for their beneficial effects on perfusion and function in both small9,10 and large11 animal models of myocardial infarction.

2. Methods

2.1 Animals

Eighty-two female Wistar rats (Janvier) weighing an average of 200–250 g were used in this study. All procedures were approved by our institutional Ethics Committee and complied with the European legislation (European Commission Directive 86/609/EEC) on animal care.

2.2 Myocardial infarction model and transplantation

Permanent ligation of the left anterior descending coronary artery was performed as described previously.1 Briefly, rats were anaesthetized with isoflurane (Baxter, Maurepas, France), 3% at induction and 2% for maintenance and tracheally ventilated at a rate of 70/min and with an 0.2 mL average insufflate volume (Alphalab, Minerve, Esternay, France). Analgesia was performed for 2 days after surgery with a 10 mg/kg subcutaneous injection of ketoprofen (Merial, Lyon, France). The heart was exposed through a left thoracotomy, and the left coronary artery was permanently ligated between the pulmonary artery trunk and the left atrial appendage. One month after myocardial infarction, rats underwent a baseline echocardiographic assessment of LV function. Following a median sternotomy, animals were then randomly allocated to receive three in-scar injections of phosphate-buffered saline (PBS) (controls, n= 24) or autologous ADSC (10 × 106, n= 37) or underwent the epicardial deposition of a triple-layer autologous ADSC sheet (10 × 106 cells, n= 21) overlaying the infarcted area. Several volumes were used in the ADSC injection group: 80, 100, and 150 μL. Intramyocardial injections were performed with a 29-gauge needle, while the autologous ADSC sheets spontaneously adhered to the surface of the heart. Two months after treatment, all animals underwent a repeat echocardiographic assessment of LV function and were euthanized with an overdose of isoflurane. All rats were treated with cyclosporin A (10 mg/kg/d, ip, Novartis) 1 day before transplantation and daily until sacrifice. Although cells were of autologous origin, this immunosuppression regimen was deemed necessary because some cells were transduced with enhanced green fluorescent protein (eGFP) (see below), which is known for triggering immune reactions involving the bystander activation of autoreactive T cells by viral-induced inflammatory cytokines.12

2.3 Assessment of LV function

Pre- and post-transplantation cardiac function was evaluated before transplantation and 2 months thereafter by transthoracic echocardiography (Sequoia 516, equipped with a 13 MHz phased-array linear 15L8 probe, Siemens®) in animals sedated with 2% isoflurane. Parasternal two-dimensional long-axis views allowed to measure LV end-diastolic surface, LV end-systolic surface, LV end-diastolic length, and LV end-systolic length and LV volumes were calculated as (8/3π) × (area²/length). Ejection fraction (%) was calculated as (LVEDV − LVESV) × 100/LVEDV). All measurements were made on digital loops in triplicate and averaged by an investigator blinded to the treatment group.

2.4 Isolation of ADSC and preparation of cell sheets

Autologous cells from the stromal vascular fraction were isolated from inguinal subcutaneous adipose tissue at the time of myocardial infarction in each rat.11 After enzymatic digestion and filtration, mature adipocytes were separated from the stromal fraction by centrifugation. The pellet was re-suspended, and cells were seeded in DMEM-F12 medium supplemented with 10% foetal bovine serum (FBS) and maintained in 5% CO2 till the sub-confluence. Sub-confluent ADSC were obtained after 6 days.13 ADSC were transduced with a third-generation lentivirus vector expressing eGFP cDNA (BiVIC Vectorology Platform, IFR150, Toulouse) using a MOI of 50 in the presence of protamine sulphate (4 µg/mL, Sigma). More than 30% of ADSC were eGFP-positive, as determined by flow cytometry.

Cell sheets were formed by plating of 3.6 × 105 cells/cm² ADSC onto thermo-responsive dishes (UpCell™) and kept in a 37°C incubator for 24 h (ADSC-Sheet-24 h) for transplantation studies and for 24 and 48 h (ADSC-Sheet-24 h, ADSC-Sheet-48 h) for in vitro characterization studies. Cell sheets detached spontaneously at room temperature within 30 min, and were then washed once with PBS, thereby yielding a scaffold-free monolayered ADSC graft. Three cell sheets were stacked one above the other to yield the triple-layer construct, which was used in transplantation experiments.

2.5 Characterization of ADSC sheets

2.5.1 Histological examination

ADSC sheets were fixed in 4% formalin overnight and embedded in paraffin. Sections with 5µm thickness were stained with haematoxylin and eosin for visualization of general morphology. Images were taken with an inverted microscope (Leica DMIL) equipped with a digital camera (Qicam, Qimaging).

2.5.2 Colony-forming unit formation and differentiation assays

ADSC were cultured on either classic TPP® flasks (ADSC) or seeded for 24 h onto temperature-responsive dishes (ADSC-Sheet-24 h). All cell cultures were then trypsinized and cells collected for further use. For colony-forming unit (CFU)-f assays, the cells were plated at 400 cells per 25 cm2 flask in DMEM-F12 supplemented with 10% of newborn calf serum. After 14 days of culture, the cells were fixed with methanol and stained with May–Grünwald–Giemsa (Sigma). The number of colonies were counted and recorded. For differentiation assays, the cells were plated at a density of 1.5 × 104 cells/cm2. Adipogenic differentiation was induced at confluence in DMEM-F12 supplemented with 10% FBS, 1 μM dexamethasone, 0.5 mM isobutyl-methylxanthine, and 60 μM indomethacin for 3 days, and isobutyl-methylxanthine was then removed for the next 18 days. Medium was changed every 2–3 days. The acquisition of the adipogenic phenotype is characterized by cells that contain lipid vesicles and was determined by staining the monolayers with 0.5% Oil Red-O solution. Osteogenic differentiation was induced 1 day after plating in DMEM-F12 supplemented with 10% FBS, 0.1 μM dexamethasone, 3 mM NaH2PO4, and 50 μg/mL ascorbic acid for 21 days. Medium was changed every 2–3 days. Osteogenic mineralization was assessed by staining with 40 mM Alizarin red (pH 4.1, Sigma).

2.5.3 RT² Profiler PCR array and qRT–PCR

This array is a pathway-focused gene expression profiling using real-time PCR. The system allows one to identify genes involved in the extracellular matrix constitution, cell adhesion, and remodelling. RNA was isolated using RNeasy Mini kit (Qiagen). The single strand cDNA from 1 μg total RNA was synthesized using RT² First Strand Kit (SABioscience). Real-time PCR was performed according to the user manual of RT² Profiler PCR array system (SABioscience) using SYBR Green PCR Master Mix in a Light Cycler 480 system (Roche Diagnostics). Rat extracellular matrix and adhesion molecules PCR array (SABioscience) were repeated three times for each condition and data were analysed using Excel-based PCR Array Data Analysis Templates (SABioscience). In addition, qRT–PCR was used to quantify transcripts for rat-specific hypoxia-inductible factor 1 alpha (HIF1α), vascular endothelial growth factor-A (VEGF-A), vascular endothelial growth factor receptor-1 (VEGFR-1), insulin-like growth factor 1 (IGF-1), transforming growth factor beta 1 (TGF-β1), and cyclin D1. The messenger RNA expression levels of target genes were normalized to hypoxanthine-guanine phosphoribosyltransferase signals as housekeeping gene, and all experiments were performed in triplicate.

2.5.4 Western blot analysis

ADSC and ADSC-Sheet-24 h lysates (40 µg/lane) were separated by sodium dodecyl sulphate–polyacrylamide gel electrophoresis before electrophoretic transfer onto a nitrocellulose membrane (Bio-Rad). Western blot analysis was carried out using anti-metalloprotease tissue inhibitor type 1 (Timp1, 1:500, rabbit polyclonal, Santa Cruz Technology), anti-integrin alpha 2 (Itag2, 1:500, mouse monoclonal, Santa Cruz Technology), anti-elastin microfibril interfacer 1 (Emilin1, 1:500, rabbit polyclonal, Santa Cruz Technology), anti-phospho-Akt 1/2/3 (Ser 473) (1:1000, rabbit polyclonal, Santa Cruz Technology), and anti-pan actin (Clone C4, 1:20000, mouse monoclonal, Millipore) antibodies. Antibody-reacting bands were visualized after development with peroxidase-conjugated secondary antibodies (Pierce Biotechnology) and a chemiluminescent detection system (ECL-Plus; GE Healthcare). Bands were quantified by densitometric software (Scion Image, NIH).

2.6 Tissue processing, histological, and immunohistochemical assessment

After the last echocardiographic assessment and sacrifice, hearts were removed and separated in two halves by a short-axis section through the mid-portion of the infarcted area. The blocks were immediately fixed in Tissue-Tek (Sakura) and frozen in liquid-nitrogen-cooled isopentane until they were sliced into 7 µm thick cryosections using an ultramicrotome (Leica). Haematoxylin and eosin staining was used to delineate the area of myocardial infarction. The presence of grafted cells was detected by immunofluorescence using an antibody directed against GFP (1:800, rabbit polyclonal, Invitrogen). Engraftment was then assessed by a semi-quantitative score ranging from 0 (no cells), 1 (minimal amount of cells), 2 (moderate amount of cells) to 3 (large amount of cells). ADSC cells were also identified using antibodies directed against CD90 (1:200, mouse monoclonal, BD Biosciences) and CD73 (1:200, mouse monoclonal, BD Biosciences). In addition, immunolabelling was performed with antibodies directed against alpha-actinin (1:50, mouse monoclonal, Sigma-Aldrich), a rat-specific endothelial cell antibody (RECA, clone HIS52, 1:30, mouse monoclonal, Serotec), CD68 (Clone ED-1, 1:100, mouse monoclonal, Serotec), and c-kit (1:50, mouse monoclonal, Dako). Binding of primary antibodies was detected by incubating the sections with FITC- or Texas-red-conjugated anti-mouse IgG (1:40, Vector) or anti-rabbit IgG (1:40, Vector) antibodies. Fibrosis was assessed by Sirius red staining. For each heart, 18 fields spanning the entire infarct area were analysed with a microscope (Leica DMIL) equipped with a digital camera (Qicam, Qimaging).

2.7 Statistical analysis

The changes in LV function parameters [LV ejection fraction (LVEF), LV end-diastolic volume (LVEDV), and LV end-diastolic volume (LVESV)] were compared between treatment groups using a parametric analysis of variance with time (pre- and post-transplantation), treatment group, and their interaction, and are expressed as ratios of geometric means with their 95% confidence intervals. Comparison of survival in the different groups of animals was analysed by a log-rank test with the Kaplan–Meier method. Statistical significance was set at the 5% threshold, and all statistical analyses were performed with the SAS statistical software version 9.1 (Cary, NC 27513, USA).

The post-transplantation percentages of fibrosis were analysed using a parametric analysis of covariance with the treatment group and the pre-transplantation LVEF as covariates. The numbers of vessels per mm² were compared between groups using a mixed model for clustered continuous data, taking into account the intra-heart correlation (as multiple slices were used for a single heart), and including the treatment group and the pre-transplantation LVEF as covariates. A square root transformation of the numbers of vessels was done to satisfy hypotheses required for the parametric model and confidence intervals were estimated by the delta method.

3. Results

3.1 Characterization of the ADSC sheet

ADSC sheets were detached spontaneously at room temperature, thereby yielding a scaffold-free monolayered ADSC graft (Figure 1A). These sheets feature a ‘carpet’ of cells well connected with each other, and cell morphology was not modified following 24 h on the temperature-responsive surface (Figure 1B and C). ADSC properties were not affected either, as cells cultured for 24 h on temperature-responsive surfaces maintained their capacity to form CFU-f and to differentiate along the adipogenic and osteogenic lineages (Figure 1D and E).

Figure 1

In vitro characterization of ADSC sheets. ADSC sheets detached spontaneously at room temperature after a 24 h culture onto temperature-responsive surfaces, thereby yielding a scaffold-free monolayered ADSC graft (A). Haematoxylin–eosin staining shows that the cell sheet represents a carpet of cells well interconnected with each other (B and C). Differentiation assays and CFU-f in ADSC (D) and ADSC-Sheet-24 h (E). For these assays, ADSC were cultured onto classic or temperature-responsive surfaces during 24 h, and assays were performed after trypsinization of these cells. Adipogenic (Ad) and osteogenic differentiation (Os) of ADSC is characterized by Oil-Red-O and Alizarin red staining, respectively. Bar: 10 mm (A), 30 µm (B), 12 µ (C), and 50 µm (DE).

In order to identify extracellular matrix proteins and/or cell adhesion molecules implicated into the formation of cell sheets, we used the quantitative RT2 Profiler™ PCR array, which revealed that there was no significant difference in the gene expression profile, regardless of whether cells (at the same density of 3.6 × 105 cells/cm2) were cultured onto classical or temperature-sensitive surfaces, for 24 or 48 h (Figure 2A). However, comparison of PCR results between ADSC-Sheet-24 h and suspended ADSC showed that five genes were significantly up-regulated (fold changes): Itag2 (3.02, P = 0.013), Emilin1 (2.54, P = 0.010), Timp1 (1.79, P = 0.019), Cspg2 (1.67, P = 0,044), and extracellular matrix protein 1 (1.66, P = 0.012) (Figure 2B). To confirm the validity of these genomic data, we selected the significantly up-regulated genes to further study their changes in expression by real-time PCR (data not shown) and western blot (Figure 2C). Our results confirmed that Emilin1, Itag2, and Timp1 proteins were up-regulated in ADSC-Sheet-24 h compared with suspended ADSC. Prior to implantation, cell sheets and suspended cells were also assessed for the expression of a cell survival signalling mediator (Akt phosphorylation) (Figure 2C) and for some of the key molecules involved in angiogenesis (HIF1α, VEGF-A, and VEGFR-1) and in adaptative cardiac remodelling (TGF-β1, IGF-1). Our results demonstrated that all these factors were up-regulated in ADSC-Sheet-24 h compared with suspended ADSC (data not shown).

Figure 2

Extracellular matrix composition of ADSC sheets. Changes in extracellular matrix and adhesion molecule gene profiles of ADSC cells cultured during 24 h (A, upper panel) or 48 h (A, lower panel) at the same density (3.6 × 105 cells/cm2) onto classic (ADSC-24 h, ADSC-48 h) or temperature-responsive surfaces (ADSC-Sheet-24 h, ADSC-Sheet-48 h). Note that the culture surface does not influence the extracellular matrix and adhesion molecule gene profile expression. Comparison of gene expression profiles between suspended ADSC and ADSC sheets (ADSC-Sheet-24 h) showing up- and down-regulation of five genes (B). Corresponding changes in protein expression level in ADSC and ADSC-Sheet-24 h (C). Note the up-regulation of Emilin1, Itag2, TIMP1, and phospho-Akt (Ser 473). Data are expressed as mean ± SD, n= 4.

3.2 Survival

There was a significant difference (P< 0.02 by the log-rank test) in survival between the three groups, the highest rate being yielded by the cell sheet group (Figure 3A). Of note, whereas pair-wise comparisons failed to detect a significant difference in the survival rate between control and cell-injected rats (P = 0.09), they showed that the cell sheet group had a significantly higher survival compared with the cell-injected rats (P = 0.01). Most of the deaths occurred over the first 3 weeks after the procedure, after which survival curves tended to level off and became parallel between groups. Reducing the volume of injection (from 150 to 100 µL and further down to 80 µL) failed to significantly alter the mortality rate in the cell-treated group (P = 0.68 by the log-rank test). Likewise, mortality adjusted for delivered volume was still not significantly different between medium- and cell-injected hearts (Figure 3B).

Figure 3

Kaplan–Meier survival curves after 2-month follow-up. (A) Percentages of surviving animals, in control (n= 14 out of 24), cell-injected (n= 15 out of 37), and cell-sheet-treated (n= 15 out of 21) groups. Note that rats receiving the cell sheet had a significantly higher survival compared with the cell-injected rats (P = 0.01). (B) Percentages of surviving animals in cell-injected rats treated with 80 µL (n= 5 out of 11), 100 µL (n= 6 out of 13), and 150 µl (n= 4 out of 13). No significant difference was observed in the survival rate between the animals treated with these different volumes of injection (P = 0.68). Significance was analysed with the log-rank test.

3.3 LV function

Baseline functional parameters did not differ significantly between the three groups (Table 1). After exclusion of deaths that occurred during the follow-up period, 14, 15, and 15 animals in the control, ADSC injection group, and ADSC cell sheet group, respectively, remained available for the 2-month functional outcome analysis. Prevention of LV remodelling was observed in the cell sheet group where LVEDV was virtually unchanged [+2.8% (95% confidence interval (CI): −18.7%; +30.0%) compared with baseline; P = 0.81], whereas it increased by 25.9% (95% CI: −0.4%; 59.2%) in the cell injection group (P = 0.05) and by 51.2% (95% CI: 18.6%; 92.8%) in the placebo medium-injected control group (P = 0.001). These conclusions were further supported by between-group pair-wise comparisons of changes in LVEDV over time (i.e. post- vs. pre-transplantation values). Thus, comparison of changes between the cell-sheet-treated and control rats yielded a negative value [mean: −32% (95% CI: −51.5%; −4.7%, P = 0.02)] confirming the lesser degree of remodelling in the former group. A similar conclusion was drawn from the comparison of changes in LVEDV between the cell sheet and cell injection groups [−18.3% (95% CI: −41.4%; +13.8%)]. Analysis of changes in LVESV showed similar patterns with a mean of 0.7% (95% CI: −26.7%; +34.6%) decrease compared with baseline values in the cell sheet group (P = 0.96) vs. increases of 34.0% (95% CI: −1.2%; 81.6%, P = 0.05) and 110.4% (95% CI: 53.6%; 188.2%, P< 0.0001) in the cell injection and control groups, respectively. These differences in LV volumes translated into a trend for a better, albeit not significant, preservation of EF in the cell sheet group [+1.63% (95% CI: −5.0%; +8.3%)] compared with the cell injection group [−3.3% (95% CI: −9.9%; +3.3%)] and in the control group [−20.5% (95% CI: −27.4%;−13.6%)].

View this table:
Table 1

Summary of functional results

GroupnLVEDV (mL)LVESV (mL)LVEF (%)
Control140.33 ± 0.090.52 ± 0.22*0.16 ± 0.050.37 ± 0.21**53.0 ± 6.132.5 ± 13.3**
ADSC150.33 ± 0.090.41 ± 0.100.18 ± 0.070.24 ± 0.0845.6 ± 9.442.3 ± 9.2
ADSC-Sheet150.39 ± 0.120.40 ± 0.130.22 ± 0.100.22 ± 0.0943.9 ± 7.645.5 ± 9.2
  • Values are given as mean ± SD.

  • LV, left ventricular; LVEDV, LV end-diastolic volume; LVESV, LV end-systolic volume; LVEF, LV ejection fraction; Post-Tx, post-transplantation.

  • *P= 0.001 vs. baseline.

  • **P<0.0001 vs. baseline.

  • P=0.05 vs. baseline.

3.4 Cell engraftment

Two months after transplantation, the sheet-scar interface harboured a mixed population of endothelial cells, macrophages, and myofibroblasts. The grafted cells were detected by CD90 and eGFP immunostaining (Figure 4). In the ADSC sheet hearts, many cells were identified within the epicardial construct (Figure 4A and B) and also in infarcted myocardium (Figure 4C), suggesting that some of these cells had migrated into the myocardium and specifically located around the vessels. Conversely, in the ADSC-injected group, only rare cells were sparsely identified between the residual cardiomyocytes (Figure 4D). The number of engrafted cells was greater in the cell sheet than in the cell injection group. In the epicardial layers, four of the four eGFP-cell-injected hearts were in the 0–1 (no or minimal amount of cells) range. In contrast, among the six eGFP-cell-sheet-treated hearts which were analysed, two fell in this 0–1 category, while four were in the 2–3 (moderate or large number of cells) range (P = 0.07). Interestingly, quantification of intramyocardial eGFP+ cells showed that their number was still higher in the cell sheet (three hearts in the 0–1 range and three in the 2–3 range) than in the cell injection (four of four hearts in the 0–1 range) groups. Because eGFP-positive cells could mistakenly be interpreted as donor cells while they represent macrophages that have engulfed eGFP released by dead cells, sections were also stained for CD68, a marker for macrophages but there was no overlay between the two stainings (Figure 5A and C). Two months after implantation, some of these eGFP+ cells co-expressed endothelial-cell-specific markers but none co-expressed cardiomyocyte-specific alpha-actinin (Figure 5DI). Only few c-kit+ cells were identified in each cell-treated group. There was no significant difference in the extent of fibrosis and angiogenesis among the three groups (data not shown).

Figure 4

Detection of grafted cells. Two months after implantation, grafted cells were identified by CD90 (A) and eGFP (BD) immunostaining in ADSC sheet-grafted (AC) and ADSC-injected (D) hearts. More cells were detected within the epicardial construct (AB) and in the infarcted myocardium (C, arrow) after the sheet application compared with direct intramyocardial injections (D, arrows). Bar: 50 µm (A); 25 µm (BD).

Figure 5

Phenotypic characterization of eGFP+-grafted cells. Two months after implantation, eGFP+ cells (green fluorescence, A, D, G) were co-immunostained with anti-CD68 (B), anti-rat-specific endothelial cell (RECA) (E), and anti-alpha-actinin (H) antibodies (red fluorescence). Merged images are shown in panels C, F, and I. Nuclei were stained in blue by DAPI. Bar: 50 µm (AC, GI); 25 µm (DF).

4. Discussion

The major finding of the present study is that, compared with commonly used intramyocardial injections, the epicardial delivery of trilayered ADSC sheets is associated with greater numbers of retained cells, a better preservation of LV geometry, and ultimately, higher rates of post-infarction survival.

There is now compelling clinical evidence that intramyocardial injections of cells are overall safe, regardless of whether injections are made trans-epicardially or endocardially14,15 even though electromechanically guided intramyocardial injections of stem cells or genes have been reported to cause a significant release of cardiac biomarkers.16 However, our results draw attention to the fact that injection-induced myocardial tissue disruption can be more damaging than it has usually been considered so far, possibly because most experimental studies commonly report the numbers of animals dying after the index infarct but often fail to detail mortality rates occurring during the follow-up period. Regardless of the cell type, this tissue damage is likely contributed by puncture-induced inflammation along the needle tracks,17 and the formation of multiple intramyocardial clusters, which can act as electrical barriers that impair direct wave propagation and set the stage for malignant arrhythmias. This could explain why intramyocardial injections of either skeletal myoblasts or bone marrow-derived cells cause more arrhythmic events than when the same cells are delivered through the coronary venous system.18 The potentially detrimental effects of direct intramyocardial cell injections on electrical stability are further highlighted by the recent findings of Song et al.,19 showing that engraftment sites can behave as electroanatomic substrates that initiate re-entries and are associated with a high (31.6%) rate of post-infarction sudden deaths in their occlusion–reperfusion rat model. In the present study, the poorer outcomes after cell injections compared with those of the placebo medium could be explained, in the latter case, by the expected lack of intramyocardial aggregates and the fast clearance of the injected fluid. Although injection-induced tissue damage has been suggested to be volume-dependent,16 this view is not supported by our finding that reducing this volume by almost one half failed to alter the mortality rate. Cell size is also unlikely to be involved in the poor outcomes of the injection group as ADSC are not larger than myoblasts or mesenchymal stem cells, which have already been largely investigated. However, because cell survival after sheet implantation has been shown to be dose-dependent,20 we used tri-layered constructs and consequently injected a higher number of cells than in previous mouse21 or rat studies9,10 for matching the number of cells harboured in the sheets. While this increased intramyocardial cell concentration may have contributed to reduce the survival rate of injected rats and account for the difference in outcomes compared with previous studies, it is noteworthy that it failed at preventing adverse LV remodelling compared with the epicardial sheet-based delivery approach. Similar conclusions have actually been made in a hamster model of dilated cardiomyopathy22 and in a rat model of chronic myocardial infarction3 where myoblast sheet application was compared with intramyocardial injections of myoblasts entailing the same dosing (10 million) as in the present study.

The increased survival rate seen in the sheet group was paralleled by a greater degree of epicardial cell engraftment, with the caveat that the identification of donor cells on the basis of their positive staining for CD90 and eGFP may have been partly confounded by two issues: (i) some CD90-positive cells could be mesenchymal-type host cells that populated the sheet and (ii) some eGFP-positive cells may be host macrophages that had uptaken the fluorescent dye released by transplanted cells upon their death. However, the lack of overlay between eGFP and CD68, a marker for macrophages (Figure 5), suggests that this was not at least a consistent event that could invalidate our conclusions on the comparative engraftment rates based on eGFP positivity. Paradoxically, direct injections even resulted in smaller numbers of intramyocardial cells than delivery of cell sheets, thereby confirming a high death rate in the former case and the possibility for some sheet-derived cells to migrate in the myocardium, as previously shown in other studies entailing the use of epicardially delivered cell constructs.2325 Although it is tempting to relate the greater number of retained cells in the sheet group with the better preservation of function, the mechanism of action of the cells remains elusive. As ADSC failed to convert into new cardiomyocytes, it is likely that grafted cells acted paracrinally on host-associated tissue-protective signalling pathways. Among others, one attractive hypothesis is that factors released by sheet-bound cells could reactivate the embryonic developmental programme in the underlying epicardial cells, thereby driving them towards an epithelial to mesenchymal (EMT) transition that, in turn, could generate new cell constituents of the heart.26 Our failure to identify a greater expression of c-kit+, a hallmark of the EMT, in the sheet group does not exclude this hypothesis as EMT would be expected to occur transiently early after transplantation, whereas our assessment was made 2 months later. For that same reason, we did not assess apoptosis whose histologic patterns have usually waned after 2 months.27 Finally, a paracrine action of the grafted cells on extracellular matrix remodelling is not in contradiction with our finding of a similar extent of fibrosis across the three groups. Namely, scar size may have been overestimated in the sheet group because remnants of the sheet containing extracellular matrix and staining positively for Sirius red could not be distinguished from the infarct-related fibrosis and were thus integrated in the overall calculations.

One could argue that the higher mortality seen in the injection groups simply reflected a natural selection of the animals which, just by chance, had the initially larger myocardial infarcts. This hypothesis is not supported by our pre-transplantation echocardiographic measurements showing similar alterations in LV function in all groups. Rather, it is conceivable that for a given level of baseline functional impairment, the greater survival rate of cells in the sheet group25 allowed them to release a greater amount of the cardioprotective factors that have been identified by the characterization of conditioned media.2 Conceivably, the donor-cell-derived paracrine effects could have facilitated the stabilization of LV volumes seen in this study, thereby contributing to reduce heart failure-related mortality. This increased survival of cells when they are patterned in a sheet form is likely the result of their better cohesiveness and anchoring to the self-secreted extracellular matrix, as suggested by our in vitro genomic and proteomic findings of the up-regulation of compounds such as Emilin-1 and integrin α-2, both involved in cell-cell and cell-matrix component adhesion2831 and tissue inhibitor of metalloproteinase-1, which is rather involved in the modulation of extracellular matrix turnover but also in the regulation of cell growth and apoptosis.32,33 The two and a half greater expression of phosphorylated Akt in the sheet group compared with suspended cells is also consistent with the benefits of cell cohesiveness, as afforded by the sheet, on survival signals and fits with our in vivo findings of a greater engraftment in the sheet group. A such, these data sharply contrast with the high rate of apoptotic death inherent in the proteolytic disruption of the cellular microenvironment, which occurs when cells are enzymatically detached for injection.34 This assumption is supported by the assessment of cell survival kinetics using bioluminescence imaging and showing that adipose stomal cells rapidly die off after injection in the acutely infarcted mouse heart.35 The higher engraftment rate seen in the sheet group could also reflect, in addition to the enhanced survival intrinsic to the maintenance of intercellular connexions, the avoidance of cell death induced by the physical stress that cells withstand during poorly controlled manually operated injections with regard to duration and pressure. Of note, the better survival of sheet-treated animals compared with those receiving injections documented in our ischaemic scenario has also been reported in a hamster model of dilated cardiomyopathy.22

Thus, our data suggest that in the setting of cardiac surgery, which offers the unique opportunity to directly manipulate the heart, the epicardial delivery of a biomaterial might be less traumatic and more effective than transmyocardial injections. While cell sheets prepared onto temperature-responsive dishes are useful for providing the proof-of-concept and have the advantage of avoiding the use of foreign scaffolds, their delicate handling characteristics may preclude realistic wide-scale clinical applications. Cell seeding onto bioresorbable matrices providing a stronger structural support can then represent an acceptable alternative1 and there is increasing evidence that the benefits of these constructs are further optimized by populating them with mixed cell populations (i.e. cells with a cardiomyogenic differentiation potential and cells featuring trophic properties) whose cross-talks may synergize their cardioprotective effects.2325,36

Conflict of interest: none declared.


This work was supported by INSERM, by the Fondation de l'Avenir (grant number ET9–547) and by the Fondation Desmarets. H.H. was supported by fellowships from the Agence Universitaire de la Francophonie and the Fondation LeDucq. P.B. was supported by fellowships from the Fondation pour la Recherche Médicale.


  • These authors contributed equally to this work.


View Abstract