Research ArticleCardiology

Cyclin A2 Induces Cardiac Regeneration After Myocardial Infarction Through Cytokinesis of Adult Cardiomyocytes

See allHide authors and affiliations

Science Translational Medicine  19 Feb 2014:
Vol. 6, Issue 224, pp. 224ra27
DOI: 10.1126/scitranslmed.3007668


Cyclin A2 (Ccna2), normally silenced after birth in the mammalian heart, can induce cardiac repair in small-animal models of myocardial infarction. We report that delivery of the Ccna2 gene to infarcted porcine hearts invokes a regenerative response. We used a catheter-based approach to occlude the left anterior descending artery in swine, which resulted in substantial myocardial infarction. A week later, we performed left lateral thoracotomy and injected adenovirus carrying complementary DNA encoding CCNA2 or null adenovirus into peri-infarct myocardium. Six weeks after treatment, we assessed cardiac contractile function using multimodality imaging including magnetic resonance imaging, which demonstrated ~18% increase in ejection fraction of Ccna2-treated pigs and ~4% decrease in control pigs. Histologic studies demonstrate in vivo evidence of increased cardiomyocyte mitoses, increased cardiomyocyte number, and decreased fibrosis in the experimental pigs. Using time-lapse microscopic imaging of cultured adult porcine cardiomyocytes, we also show that Ccna2 elicits cytokinesis of adult porcine cardiomyocytes with preservation of sarcomeric structure. These data provide a compelling framework for the design and development of cardiac regenerative therapies based on cardiomyocyte cell cycle regulation.


Heart disease remains the leading cause of death in the industrialized world. This can be partially attributed to the inability of cardiomyocytes to divide in a clinically relevant manner (1). The heart therefore responds to injury primarily through scar formation. Molecular and cellular strategies to induce cardiac regeneration have been the focus of intense studies over the past decade (2). These efforts have largely focused on the use of stem cells, whether extrinsic or intrinsic to the heart, in an attempt to generate new cardiomyocytes. However, most of the clinical trials based on stem cell therapeutics have not resulted in adequate and sustained improvement in cardiac contractile function. Notably, there has not been any evidence of actual cardiomyocyte differentiation arising from such cells (3, 4).

Despite recent evidence of cardiomyocyte turnover in the healthy adult heart, with widely ranging rates of turnover reported (5, 6), cardiac regeneration in response to injury, such as myocardial infarction (MI), is limited and its mechanisms are not clearly defined. Although adult mammals have lost their cardiac regenerative capacity due to unclear reasons, certain metazoan species teach us important lessons. Namely, urodele amphibians, such as the newt, have retained an extraordinary capacity to replace lost anatomical structures through epimorphic regeneration (7). Newt regeneration hinges on the local plasticity of differentiated cells that remain after injury or tissue removal. Mechanistically, this involves reentry to the cell cycle with loss of differentiated characteristics to generate a “local progenitor cell” of restricted potentiality (7).

The adult zebrafish heart has been reported to have the capacity to regenerate up to 20% of its volume (8). Several studies support a mechanism whereby regeneration results from proliferation of cardiomyocytes adjacent to the area of injury (9, 10). Zebrafish harboring a temperature-sensitive mutation in mps1, a gene encoding a mitotic checkpoint kinase (8), fail to regenerate the heart with extensive fibrosis noted in the excised areas. These results are highly relevant to our current study, given that a single gene mutation blocking cardiomyocyte mitosis can impede the regenerative process and thus create a permissive environment for fibrosis to occur.

We and others have shown that cyclin A2 (Ccna2), a key cell cycle regulatory gene that complexes with its cyclin-dependent kinase partners to mediate both the G1-S and G2-M transitions of the cell cycle (1113), is silenced shortly after birth in mammalian cardiomyocytes (14, 15). We subsequently demonstrated that Ccna2 mediates cardiac repair by inducing cardiomyocyte mitoses after MI in two small-animal models of MI (16, 17). As a prerequisite to the potential for clinical application, we present a therapeutic efficacy study of Ccna2-mediated cardiac repair in a large-animal model that closely mimics human anatomy and physiology, which is validated by others (18, 19). Large-animal models of cardiovascular disease are an absolute necessity on the road to translation of successful clinical therapies (20, 21). Furthermore, mechanisms of cardiac repair may be widely divergent across species. This is illustrated in the case of Sca1+ stem cells found in mouse hearts (22) with a complete lack of Sca1 in the human genome (23). To this end, we demonstrate mechanistic evidence of cardiomyocyte cytokinesis induced in adult porcine cardiomyocytes captured by time-lapse microscopy.


Cyclin A2 markedly enhances cardiac function in infarcted porcine hearts

Animals were subjected to MI via occlusion of the left anterior descending artery (LAD) with a platinum coil with a 67% survival rate at 48 hours after MI. We used four independent but complementary in vivo technologies to assess cardiac function. These included left ventricular (LV) angiography, two-dimensional (2D) and 3D echocardiography, and magnetic resonance imaging (MRI), which were performed at several time points before and after MI (Fig. 1, A to E; see movies S1 to S3 for representative movies of 2D and 3D echocardiography and MRI). Groups of animals surviving MI (n = 18) were blindly randomized to receive either adenovirus with Ccna2 (Adeno-Ccna2) or adenovirus lacking Ccna2 (Adeno-Null). One week after MI, animals underwent left lateral thoracotomy and received injections of 1012 virus particles at 10 equidistant sites surrounding the infarct area, in a manner we have previously described (17) (Fig. 1F). Of 18 animals, 16 survived this surgery. CCNA2 expression was confirmed with immunoblotting performed 7 days after injection in two separate animals that had not undergone infarction (Fig. 1, C and D). There was about sevenfold higher levels of protein expression in the inferior wall (P = 0.008) in Adeno-Ccna2–injected versus Adeno-Null–injected animals and about sevenfold higher levels of protein expression in the anterior wall (P = 0.178) of Adeno-Ccna2–injected versus Adeno-Null–injected animals. CCNA2 expression was also confirmed at 6 weeks after injection of viruses and was still expressed, although at lower levels than detected at 7 days after injection (fig. S1).

Fig. 1. The porcine infarct model and multimodality imaging were used in the study design.

Yorkshire female swine underwent baseline imaging studies to assess EF, were subjected to MI followed by treatment with either Adeno-Ccna2 or Adeno-Null, and studied at two time points after treatment. (A) Overview of study design. (B) Still image of LV angiography used to assess EF. (C) Still image of 2D echocardiography used to assess EF (movie S1). (D) Still image of 3D echocardiography used to measure EF (movie S2). (E) Representative image of porcine chest imaged by MRI to obtain EF (movie S3). (F) Virus was delivered 1 week after MI after performing left lateral thoracotomy to expose the heart. Virus was injected at 10 equidistant sites surrounding the infarct zone, identified by white area (example: dotted region indicated by an arrow). (G) Immunoblotting was performed on porcine heart tissue 1 week after injection of virus to confirm expression of CCNA2 in Adeno-Ccna2–injected animal and exclude expression in Adeno-Null–injected animal. (H) CCNA2 expression normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was higher in the Adeno-Ccna2–injected animal versus the Adeno-Null–injected animal.

We wanted to determine whether treatment with Adeno-Ccna2 was inducing recovery of cardiac contractile function. We assessed ejection fraction (EF), a measure of cardiac contractile function, using four distinct modalities to measure LV end-diastolic volume (EDV) and LV end-systolic volume (ESV) before MI, at 1 week after MI before virus injection, and at 7 weeks after MI (6 weeks after virus injection). Baseline studies had not shown any significant differences between animals (table S1). With respect to MRI, two time points were examined (1 week after MI pretreatment and 7 weeks after MI). For each time point and each modality, EF was calculated by using the following formula: EF = (EDV − ESV)/EDV × 100% (24) (table S2 and Fig. 2). LV angiography (Fig. 2A) demonstrated ~31 and ~11% enhancement of EF in treated animals and control animals, respectively (P = 0.031). With 2D echocardiography analysis (Fig. 2B), we noted ~33 and ~10% enhancement of EF in treated versus control animals (P = 0.041). 3D echocardiography imaging (Fig. 2C) resulted in ~27 and ~8% enhancement of EF-treated and control animals, respectively (P = 0.032). With MRI, which is widely considered the “gold standard” for measurement of cardiac function (25) (Fig. 2D), there was ~18% enhancement of EF in treated animals and a ~4% decrease in control animals (P = 0.035).

Fig. 2. Cyclin A2–treated animals exhibit marked enhancements of LVEF after MI using four independent imaging modalities.

All black bars represent measurements made at 1 week after MI before virus injection, and all gray bars represent measurements made at 7 weeks after MI (6 weeks after injection). (A) LV angiography demonstrates significantly greater LVEF enhancement in experimental animals versus controls (P = 0.0311, n = 8 each group). (B) 2D echocardiography demonstrates significantly greater LVEF enhancement in experimental animals versus controls (P = 0.0408, n = 8 each group). (C) 3D echocardiography also exhibits significantly greater LVEF enhancement in experimental animals versus controls (P = 0.0316, n = 8 each group). (D) With MRI, there was a significant enhancement of LVEF in experimental animals and a decline in LVEF in control animals (P = 0.0350, n = 5 each group).

To assess whether changes in scar size could be detected, we performed MRI delayed enhancement imaging at time points 1 and 7 weeks. We noted a trend toward a reduction in scar size at 7 weeks in the Adeno-Ccna2–treated animals (fig. S2).

Cyclin A2 elicits cell cycle activation in vivo but may not have increased endogenous progenitor cell homing

All animals were sacrificed at 7 weeks after MI after cardiac function measurements were made, and tissue specimens were prepared for histologic studies. To assess the presence of mitotic cardiomyocyte nuclei in vivo, mitoses were detected using anti–phospho–histone H3 antibody (H3P) (16) and colocalized to cardiomyocytes with anti–α-actinin antibody using epifluorescence imaging. Three animals each from the Adeno-Null and Adeno-Ccna2 group were used. An average of 1102 cardiomyocytes was analyzed per animal by investigators blinded to the treatment arm (Fig. 3A). There was a more than threefold increase in the percentage of mitotic cardiomyocytes in the anterior walls in treated animals versus controls (P = 0.002) (Fig. 3B). Z-stack analysis was used to confirm that mitotic nuclei were embedded within cardiomyocytes (Fig. 3C and movie S4).

Fig. 3. Cyclin A2 elicits cardiomyocyte cell cycle activation in infarcted porcine tissues but does not appear to increase endogenous stem/progenitor cell homing.

The scale bar for all images is equivalent to 20 μm. In all images, tissues were co-immunostained with antibody to α-actinin (green) and either H3P, Ki-67, ABCG2, or c-Kit (red) with 4′,6-diamidino-2-phenylindole (DAPI) (blue) used to identify nuclei. (A) Ventricular porcine tissues from experimental versus control animals immunostained with antibody to H3P (red). (B) A significantly greater percentage of H3P+ nuclei could be identified within cardiomyocytes in sections from experimental animals compared with controls (P = 0.0019, n = 3306 cardiomyocytes examined per group from a total of three animals per group). (C) Z-stack analysis of cardiomyocyte from (A) illustrates that the mitotic nucleus is embedded within the cardiomyocyte. (D) Ventricular porcine tissues from experimental versus control animals were immunostained with antibody to Ki-67 (red). (E) There was a significantly higher Ki-67+ cardiomyocyte index in sections from experimental animals compared with controls (P = 0.0014, n = 3306 cardiomyocytes in each group from a total of three animals per group). (F) Z-stack analysis of cardiomyocyte from (D) illustrates that the Ki-67+ nucleus is embedded within a cardiomyocyte. (G) Ventricular porcine tissues from experimental versus control animals immunostained with antibody to ABCG2 (red). (H) Quantitative real-time polymerase chain reaction (PCR) did not reveal significant differences in expression of ABCG2 between experimental and control ventricular tissues. (I) Ventricular porcine tissues from experimental versus control animals immunostained with antibody to c-Kit (red). (J) Quantitative real-time PCR did not reveal significant differences in expression of c-Kit between experimental and control ventricular tissues. NS, not significant.

Although H3P is a mitosis-specific marker, we additionally investigated cardiomyocyte cell cycle activation in vivo by assaying for expression of Ki-67 (26), which is present in all active phases of the cell cycle (G1, S, G2, and mitosis) but is absent from resting cells (G0) (Fig. 3D). There was a 3.5-fold increase in the percentage of cardiomyocytes expressing Ki-67 in the anterior walls of treated animals versus controls (P = 0.001) (Fig. 3E). As above, Z-stack analysis confirmed that Ki-67+ nuclei were embedded within cardiomyocytes (Fig. 3F).

We sought to determine whether differences in endogenous stem or progenitor cell homing could be playing a role in cardiac recovery of the Adeno-Ccna2 animals. Because putative stem cells were only scarcely identified on immunofluorescence analysis, we used quantitative PCR for ABCG2 and c-Kit (Fig. 3, G to J). There was no significant difference in fold change for each gene analyzed between the treated animals and controls (27). This suggested that increased numbers of endogenous stem cells that homed to the damaged myocardium may not have been the most important mechanism underlying the regenerative effect of Ccna2. However, this does not exclude the possibility that earlier activation of endogenous stem cells during the time course of recovery contributed to the repair process.

Structural evidence from in vivo studies supports a regenerative role for cyclin A2 after MI

Tissue sections from the anterior wall peri-infarct regions were analyzed from three treated animals and three control animals to determine the extent of fibrosis, the size and number of cardiomyocytes, and any changes in vascularity. Gomori’s trichrome staining was applied to about 20 sections of Adeno-Ccna2–treated and 20 sections of Adeno-Null–treated animals, and the collagen–to–muscle density ratios for each section were computed with ImageJ software (Fig. 4, A and B). There was a significantly enhanced collagen–to–muscle density ratio, indicative of greater fibrosis, in the Adeno-Null–treated animals (Fig. 4C).

Fig. 4. Structural changes in peri-infarct tissues of Adeno-Ccna2–treated animals indicate that Ccna2 is mediating regeneration of cardiomyocytes without a significant enhancement of vascularization.

The scale bar for all images is 100 μm. In images (D) to (K), tissues were immunostained with antibody to α-actinin (green) and either labeled with WGA or co-immunostained with SMA (red) with DAPI (blue) used to identify nuclei. (A and B) Peri-infarct tissue sections were stained with Gomori’s trichrome stain to delineate blue (collagen) versus red (muscle) in Adeno-Null–treated (A) and Adeno-Ccna2–treated (B) animals. (C) There was a significant decrease in collagen/muscle density in Adeno-Ccna2–treated animals (P = 0.0034, n = 20 sections per group). (D) Tissues from Adeno-Null–treated animals were stained with flourophore-labeled WGA. (E) Tissues from Adeno-Ccna2–treated animals were stained with flourophore-labeled WGA. (F) Cardiomyocyte cell lengths were reduced in a significant manner in Adeno-Ccna2–treated animals (P < 0.0001, n = 140 cells per group). (G) Cardiomyocyte cell widths were reduced in a significant manner in Adeno-Ccna2–treated animals (P < 0.0001, n = 140 cells per group). (H) The number of cardiomyocytes per unit area was significantly increased in the peri-infarct tissues of Adeno-Ccna2–treated animals compared with Adeno-Null–treated animals (P = 0.0027, n = 20 sections per group). (I) Vascular density differences were not found to be statistically significant in the peri-infarct tissues of Adeno-Ccna2–treated versus Adeno-Null–treated animals (P = 0.2632, n = 20 sections per group from a total of three animals per group). (J) Tissues from Adeno-Null–treated animals were immunostained with antibody to SMA. (K) Tissues from Adeno-Ccna2–treated animals were immunostained with antibody to SMA.

To determine whether treatment with Adeno-Ccna2 elicited hypertrophy of cardiomyocytes, tissue sections from peri-infarct regions were stained with fluorophore-labeled wheat germ agglutinin (WGA) to delineate cell membranes (28) and α-actinin to colocalize the WGA to cardiomyocytes (Fig. 4, D and E). We then measured lengths and widths of cardiomyocytes for which the cell borders and nuclei could clearly be seen using Zeiss AxioVision software (fig. S3). The lengths and widths of cardiomyocytes in the Adeno-Ccna2–treated animals were smaller than those of control animals, thus excluding a hypertrophic effect of Adeno-Ccna2 treatment (Fig. 4, F and G). Although the size reduction was not of great magnitude, the effect was significant. We used the delineation of cardiomyocyte cell borders to count the number of cardiomyocytes per unit area, and found that the numbers of cardiomyocytes were significantly greater in peri-infarct regions of Adeno-Ccna2–treated animals compared with control animals (P = 0.003) (Fig. 4H). To ascertain whether vasculogenesis was a contributing factor to the enhanced cardiac function noted in the Adeno-Ccna2–treated animals, tissue sections from peri-infarct regions were immunostained with antibody to smooth muscle actin (SMA), and total area of SMA fluorescence as a fraction of section area delineated by DAPI-stained nuclei was computed (Fig. 4, I to K). A trend toward enhanced vascularity was noted in the Adeno-Ccna2–treated animals compared to controls, but there was no significant difference.

Cyclin A2 mediates cytokinesis in cultured adult porcine cardiomyocytes

In addition to noting significantly increased cardiomyocyte mitoses in treated animals in vivo, definitive evidence of actual cytokinesis necessitated in vitro studies with time-lapse imaging. Isolated adult porcine cardiomyocytes were plated at equal densities and transfected with either Adeno-Ccna2 or Adeno-Null viruses (the same viruses used for the in vivo studies). For delineation of sarcomeric structure in cardiomyocytes, all cells were transfected with adenovirus containing α-actinin–mCherry, which we constructed to allow for proper folding of the virally delivered α-actinin into the living cardiomyocyte sarcomere (Adeno-act-mCherry) (29). This enables confirmation of cardiomyocyte identity before and after cytokinesis and tracking of sarcomere dynamics, whereas antibodies can result in artifact and can only be used at one time point after cell fixation. We observed coexpression of α-actinin and CCNA2 in cultured adult cardiomyocytes (fig. S4). We performed time-lapse microscopic imaging of live cells to capture cardiomyocyte cytokinesis (Fig. 5A; still images from movie S5). We calculated the cytokinetic index of adult porcine cardiomyocytes by counting cytokinetic events observed in 34 regions of interest (ROIs) (table S3). The cytokinetic index was ~15-fold higher in Adeno-Ccna2–transfected cardiomyocytes (P = 0.001) (Fig. 5B). Most remarkably, sarcomere structure was preserved before and after cytokinesis (Fig. 5, C and D, for magnified image at lower light intensity of cell from Fig. 5A before it begins to divide). The sarcomeric pattern is clearly visible before the cell underwent division, and the daughter cells were noted to be mononuclear after they had been fixed and stained with DAPI (Fig. 5, A, final panel, and G). Upon magnification of a daughter cell, the presence of sarcomeric structure is easily noted (Fig. 5, E and F). Additionally, these cells were fixed and immunostained with antibody to troponin Tc (Fig. 5, H and I) as further confirmation of their cardiomyocyte identity. Clusters of other cardiomyocytes that had not taken up the Adeno-act-mCherry could be seen adjacent to the daughter cells. The daughter cells were co-immunostained with antibodies to α-actinin (as the mCherry fluorescence had faded by this step) and CCNA2 to ensure that they had been transfected with Adeno-Ccna2 and that they express both proteins (fig. S5B).

Fig. 5. Cyclin A2 induces cytokinesis in adult porcine ventricular cardiomyocytes.

Time-lapse microscopy imaging of live cells was performed after cells were transfected with Adeno-Ccna2 versus Adeno-Null. All cells were transfected with Adeno-act-mCherry for identification of the sarcomere before and after cytokinesis. The scale bar in all images is equivalent to 20 μm. (A) Still images from movie S4 at various time points, except the last panel, where the cells were fixed and stained with DAPI. The daughter cells appear mononuclear. (B) Cytokinetic index was calculated by counting cytokinetic events observed over 72 hours of time-lapse imaging in 34 ROIs (table S3) and dividing by total numbers of cardiomyocytes expressing Adeno-act-mCherry in ROIs. The cytokinetic index was ~15-fold higher in experimental versus control cells (P = 0.0001, n = 16 ROIs CCNA2, n = 18 ROIs Null). (C) The first panel of Fig. 4A has been magnified ×4 and brightness decreased 90% to visualize Adeno-act-mCherry fluorescence as it has been incorporated into the sarcomere. (D) Grayscale image of Fig. 4C. (E) The last panel of Fig. 4A has been magnified ×7 after fixation and staining with DAPI to visualize Adeno-act-mCherry fluorescence; sarcomere is visible after cytokinesis. (F) Grayscale image of Fig. 4E. (G) Magnification of the second daughter cell from Fig. 4A. (H) Adeno-act-mCherry (red) fluorescent cells from Fig. 4A were immunostained with antibody to troponin Tc only (cyan), and nuclei were immunostained with DAPI to further confirm cardiomyocyte identity of the cells that underwent cytokinesis.


We have previously demonstrated that Ccna2, normally silenced in postnatal mammalian myocardium (14, 15), displays regenerative capacity in a transgenic mouse model of constitutive cardiac Ccna2 expression (16). It also mediates cardiomyocyte regeneration when exogenously delivered to infarcted hearts in genetically naïve adult rats (17). We now find that the therapeutic delivery of Ccna2 1 week after MI in the porcine heart induces cardiomyocyte cell cycle activation in vivo with marked enhancement of cardiac function as noted on LV angiography, 2D and 3D echocardiography, and MRI. We also observed significantly decreased fibrosis, lack of a hypertrophic response, significantly increased numbers of cardiomyocytes per unit area, and no significant differences in vascularity in the peri-infarct tissues of animals treated with Ccna2. These data provide proof of concept that cell cycle manipulation results in potent therapeutic cardiac regeneration. Using the same viral vector encoding Ccna2 complementary DNA (cDNA) that was delivered to the porcine heart in vivo, our in vitro studies confirm that it elicits cytokinesis in isolated adult porcine cardiomyocytes. The preservation of sarcomeric structure in the daughter cells resulting from cytokinesis is notable and would not have been possible to study without dynamically labeling the sarcomeric actinin within the cardiomyocytes.

To our knowledge, most clinical trials of stem cell therapy in the heart have provided inconsistent evidence of improvements in LVEF. In contrast, we have used a comprehensive multimodality approach to demonstrate improvements in LVEF. Previous studies of cell types used in clinical trials have also not clearly demonstrated mechanistic proof of either stem cell differentiation to cardiomyocytes or significant proliferation of native cardiomyocytes. Thus, delineating intrinsic molecular pathways for cardiac regeneration is critically important. In any translational pathway of using cell cycle regulators for actual clinical use, precautions must be in place to prevent the expression of cell cycle activators in extracardiac tissues due to the potential for oncogenic transformation (30). These safety concerns would necessitate tissue-specific activation of cyclin A2, and such studies are currently under way by our laboratory, using a variety of approaches. A limitation of this report is being addressed in a next-generation gene therapy study using a cardiomyocyte-specific promoter driving cyclin A2 expression. This will ensure that the effects noted in this report are specifically due to cardiomyocyte cell cycle activation, and not to activation of other cell types such as endothelial cell–induced angiogenesis with concurrent improvement of myocardial perfusion. Such a conclusion would also be strengthened by in vivo myocardial perfusion assessment to exclude significant contribution by angiogenesis.

Although we had observed significantly greater cardiomyocyte mitoses and increased numbers of Ki-67+ nuclei in anterior wall tissue sections from treated pigs compared with control pigs, the enhancement of cell cycle activation at the 6-week post-treatment time point may not entirely account for the remarkable increases in cardiac function. We had previously observed a “time window” of cardiac repair in Ccna2 transgenic mice that had undergone MI and found that most cardiomyocyte mitoses were observed in the first 3 weeks after injury (16). In our current model, a greater degree of cell cycle activation likely occurred at an earlier time point also, especially in light of the observation that expression of Adeno-Ccna2 was significantly reduced at the 6-week time point as compared to expression levels at 1 week. Additionally, one cannot exclude nonproliferative effects of cyclin A2 on cardiomyocytes, such as an increase in electromechanical coupling, that would need to be evaluated in future studies. We noted higher numbers of cardiomyocyte mitoses in the Adeno-Null control animals than we expected. This may be due to the higher percentage of mononuclear cardiomyocytes present in large animals compared with rodents (31). This is also consistent with observations that mononuclear cardiomyocytes can more easily be induced to divide than multinucleated cells (31).

Endogenous cardiac stem cells could be activated in our porcine infarction model to partially contribute to the enhancement of LVEF. In our previous study of MI induced in Ccna2 transgenic mice, we had not noted any changes in levels of expression of endogenous stem cell markers c-Kit and ABCG2 (16). We surmised that mitoses noted in small cells expressing cardiac markers were likely due to the proliferative stimulus provided by Ccna2 once the transgene was activated in cardiac progenitors. Here, we examined expression of stem cell markers c-Kit and ABCG2 in porcine cardiac tissue by immunohistochemistry and quantitative PCR. We found no differences in levels of expression in either of these markers in pigs treated with Ccna2 adenovirus compared with controls treated with null adenovirus 6 weeks after treatment. These results do not exclude the potential for greater proliferative ability of progenitors that have arisen from either c-Kit– or ABCG2-positive stem cells, especially at an earlier time point than was examined in our study.

Our structural studies in vivo lend further support to Ccna2’s regenerative effect on cardiomyocytes. We note a significant decrease in fibrosis in the peri-infarct tissues of the treated animals and a greater number of cardiomyocytes compared with controls. The size of the cardiomyocytes in the treated group was smaller, thus excluding hypertrophy as a mechanism to explain the enhanced contractile function. These results are reminiscent of our earliest studies of the effect of constitutive expression of Ccna2 in murine myocardium (15) in that we noted hyperplasia, but not hypertrophy, as the outcome of sustained Ccna2 expression. Surprisingly, we did not note a significant increase in vascularity with respect to smooth muscle cell proliferation in the animals treated with Ccna2, perhaps because vascular cells maintain a fully active cell cycle repertoire unlike mammalian cardiomyocytes and may thus be less amenable to cell cycle perturbation.

In summary, these data provide compelling evidence that cell cycle regulation of cardiomyocytes can be successfully exploited to effect significant cardiac regeneration in a large-animal model of MI. Moreover, our data not only support a regenerative role for Ccna2 at the whole-organ level and cell cycle reentry of cardiomyocytes in vivo but also highlight Ccna2-mediated cytokinesis of adult porcine cardiomyocytes captured by time-lapse imaging with preservation of sarcomeric structure after cell division. An alternative to cell therapy for cardiac regeneration can now be considered as a translational strategy that may recapitulate the developmental pathways mediating cardiomyocyte division in lower vertebrates and embryonic mammalian hearts.


Study design

Study design. Before MI creation at time 0 (Fig. 1), cardiac function was assessed with LV angiography and 2D and 3D echocardiography to rule out heart disease at baseline. In our previously published rat MI model of cyclin A2–mediated cardiac repair, cyclin A2 gene therapy demonstrated a 56% increase in EF (32) relative to control animals at 6 weeks after delivery of therapy. Therefore, in our porcine model for this study, we estimated that a sample size of eight animals would provide at least 90% power to detect a 10% increase in EF in animals receiving the cyclin A2 virus compared to animals receiving null vector after 6 weeks at a two-sided α level of 0.05. One week after MI, animals underwent cardiac function testing with LV angiography, 2D and 3D echocardiography, and MRI. Then, left lateral thoracotomy for direct myocardial injection of either Adeno-CCNA2 or Adeno-Null was performed. Six weeks after virus administration, cardiac function was again tested with LV angiography, 2D and 3D echocardiography, and MRI. The animals were then sacrificed for histological studies.

Randomization. One week after MI, animals were randomized to receive either Adeno-CCNA2 or Adeno-Null (Fig. 1).

Blinding. Cardiac function assessments (LV angiography, 2D and 3D echocardiography, and MRI) were performed and interpreted by cardiologists blinded to the treatment arms. All histology sections were immunostained and analyzed in a blinded manner.

Replication. Imaging parameters for echocardiography and MRI were analyzed in triplicate. Immunoblots and quantitative PCRs were performed in triplicate. For all other in vivo and in vitro studies, replication parameters are indicated in the figure legends.

Guidelines. Results were reported in accordance to the ARRIVE (Animal Research: Reporting of In vivo Experiments) guidelines listed through the EQUATOR Network library.

Animal care and biosafety

Juvenile female Yorkshire Landrace pigs weighing 17 to 22 kg were used in this study in accordance with the guidelines from Guide for the Care and Use of Laboratory Animals and approved by the Institutional Animal Care and Use Committee (IACUC) at the Mount Sinai School of Medicine. Housing was in accordance with the IACUC Committee at Mount Sinai School of Medicine.

Creation of MI

For MI creation, animals underwent cardiac catheterization via femoral access with an 8-French sheath. A 7-French hockey stick guide catheter (Cordis INFINITI, Johnson & Johnson) engaged the left coronary artery, and a baseline coronary angiogram was performed. An 8-mm-long, 4.0-mm VOYAGER over-the-wire balloon (Abbott) was inflated after the first diagonal branch of the LAD to 3 atm for 1 hour. Subsequently, a platinum embolic coil (0.035 inch, 40 mm in length, 5 × 3 mm in diameter, Cook Medical Inc.) was placed with a 4-French AR catheter (Cordis Infiniti, Johnson & Johnson) into the LAD. The 48-hour survival rate after MI was 67% (18 survived MI, 3 died during MI creation, and 6 died within 48 hours after MI).

Intramyocardial injection of adenovirus

Replication-deficient adenoviral vectors containing murine cyclin A2 or null content (control) driven by the cytomegalovirus (CMV) promoter were made previously by our laboratory (17) and packaged at the University of Iowa Gene Transfer Vector Core. One week after MI creation, anaesthetized animals were placed in the lateral recumbent position and a left lateral thoracotomy was performed in the fourth intercostal space. After dissection of the pericardium, 1012 adenovirus particles containing either a cyclin A2 or a null vector were injected around the circumference of the peri-infarct zone in 0.1-ml aliquots with a 27-gauge needle. Of 18 animals, 16 (89%) survived the surgery.


Nonsurgical procedures were performed under propofol (8 to 15 mg/hour), and surgery under isoflurane anesthesia (0.8 to 1.2% in 100% oxygen). Pigs were euthanized using intravenous injection of Euthasol (pentobarbital, phenytoin, 1 ml/4.5 kg). After the hearts were removed and sectioned, viable myocardium was identified by staining the myocardium with tetraphenyl tetrazolium chloride.

Assessment of myocardial function and structure

We assessed myocardial function and structure at baseline before MI creation, at 1 week after MI (before adenovirus injection), and at 7 weeks after MI (6 weeks after adenovirus injection). Blinded echocardiography with an iE33 ultrasound machine (Philips Medical Systems) was performed during end-expiratory breath hold. 2D and 3D images were obtained in the standard LV apical, parasternal long axis, and short axis views. QLAB software (Philips) was used for echocardiographic analysis, which included EF calculation (using modified Simpson’s formula) and speckled tracking strain quantification. We performed LV angiography at baseline, 1 week after MI creation, and 6 weeks after virus delivery with an Integris H5000 single-plane fluoroscopy system (Philips Medical Systems). Cardiac MRI was done 1 week after MI creation before adenovirus injection and 6 weeks after virus delivery before sacrifice. Cine, perfusion, and delayed enhancement images were acquired with a 3-T magnet (Philips Achieva) with electrocardiogram gating during end-expiratory breath hold. To determine scar size, delayed enhancement images were acquired 15 min after the administration of gadopentate dimeglumine (0.2 mM/kg) (Magnevist, Bayer Medical Solutions) in an inversion-recovery fast gradient echocardiography sequence. LV function analysis was calculated with Argus software (Siemens Medical Solutions). With all imaging modalities, images were acquired and analyzed by an investigator blinded to the study arm.

Immunostaining of porcine ventricular tissues

Porcine heart ventricular 6-μm-thick frozen sections were air-dried and then fixed for 5 min in 2% paraformaldehyde. Sections were blocked with 10% donkey serum (Jackson ImmunoResearch) for 1 hour at room temperature. Each section was incubated with the primary antibody for 1 hour at room temperature, followed by a secondary antibody for an additional 1 hour at room temperature, and counterstained with DAPI. Slides were imaged with a Zeiss LSM-510 Meta confocal microscope (Carl Zeiss). The following primary antibodies were used for staining: mouse anti–α-actinin (Santa Cruz Biotechnology #15335), rabbit anti-H3P (Upstate Biotechnology #06570), rabbit anti-BCRP/ABCG2 (Abcam #ab63907), and rabbit anti–c-Kit (Abcam #ab5506). Alexa Fluor 488 and Alexa Fluor 568 secondary antibodies were purchased from Molecular Probes (Invitrogen). All immunofluorescence sections were analyzed by an investigator blinded to the study arm.

Mitoses in vivo

Porcine heart ventricular 5-μm-thick sections were immunostained as described in Materials and Methods with antibodies against α-actinin (cardiomyocyte marker) and H3P (mitosis marker) (Millipore #06-570). Sections were analyzed with a Zeiss LSM-510 Meta confocal microscope (Carl Zeiss). All nuclei contained within α-actinin–positive cells were counted. All nuclei positive for H3P that colocalized to α-actinin–positive cells were counted, and mitotic index was calculated as the fraction of H3P+ nuclei in cardiomyocytes/total cardiomyocyte nuclei. We also analyzed the cell proliferation marker Ki-67. Porcine heart ventricular sections were immunostained with anti–Ki-67 (BD Pharmingen #556003, BD Biosciences). Images were analyzed as above, and Ki-67+ index was calculated. To confirm the cardiomyocyte-specific H3P and Ki-67 immunopositivity, Z-stack analysis was carried out at ×63 magnification with the Z-stack tool of AxioVision microscopy software (AxioVision Release 4.7, Carl Zeiss). After Z-stack analysis, the “3D cut view” (see Fig. 3, C and F) and “3D movie” (fig. S4) were constructed with the same software.

Real-time quantitative PCR

Quantitative PCRs were performed with TaqMan gene expression probes on the ABI Prism 7900HT sequence detection system (Applied Biosystems). The PCR protocol consisted of 40 cycles at 95°C (15 s) and 60°C (1 min). Fold changes in gene expression were determined using the comparative CT method (ΔΔCT method) (27) with normalization to the endogenous control GAPDH. TaqMan probes used for reverse transcription PCR (RT-PCR) experiments are as follows: ABCG2 (#Ss03393456, Applied Biosystems) and c-Kit (#Ss03380145, Applied Biosystems).

Comparative CT method (ΔΔCT method)

The threshold cycle number (CT) was obtained as the first cycle at which a statistically significant increase in fluorescence signal was detected. The data were normalized by subtracting the CT value of GAPDH from that of ABCG2 or c-Kit. Each reaction was done in triplicate, and the CT values were averaged. The ΔΔCT was calculated as the difference of the normalized CT values (ΔCT) of the cyclin A2– and null-treated control samples: ΔΔCT = ΔCT treated − ΔCT control. ΔΔCT was converted to fold of change by the following formula: fold of change = 2−ΔΔCT (27). The fold differences in gene expression are represented as means ± SD. RT-PCR was performed on RNA isolated from 6 samples each from anterior wall and inferior wall sections for Adeno-Ccna2 (n = 3) and Adeno-Null (n = 3); thus, 12 samples per animal were analyzed. The fold differences calculated with the ΔΔCT method are usually expressed as a range, which is a result of incorporating the error of the ΔΔCT value into the fold difference calculation. The error bars represent the top and bottom range of the fold difference. P values were determined by a two-tailed paired Student’s t test from the ΔCT values.

Fibrosis assessment

We used Gomori’s Trichrome Stain Kit (Blue Collagen) (Thermo Scientific) to evaluate fibrosis in porcine heart tissue sections (33). The kit stains cytoplasm or muscle as red, collagen as blue, and nuclei as black. To stain tissue sections on glass slides, the manufacturer’s instructions were followed with slight modifications. Briefly, cryosectioned tissues were fixed in 2% paraformaldehyde solution prepared in 1× perfusion buffer containing calcium chloride (1 mg/ml). Fixation was carried out at 37°C for 30 min. Slides were washed once in 1× phosphate-buffered saline (PBS) and then hydrated in deionized water for 5 min. Slides were then placed in Bouin’s fluid (provided in the kit) and incubated at 56°C for 1 hour. Further steps of staining were carried out as mentioned in the instruction manual. Slides were mounted with glycerol after drying the samples at 40°C for 15 min. Light microscopy at ×20 magnification was carried out with Axioplan 2 microscope (Carl Zeiss). Images of randomly selected areas from each group (n = 3) were analyzed with ImageJ software. Signals of red, green, and blue were split (RGB), and integrated densities of red signal (representing muscle) and blue signal (representing collagen or fibrosis) were calculated. The values of integrated density of collagen (blue) of each group were normalized with that of muscle (red) and were plotted to analyze the change in fibrosis.

Cardiac cell size, density, and vascularity analysis

WGA is known to bind the sarcolemma of cardiac cells (28). We used WGA tagged with Alexa Fluor 647 (Invitrogen #W32466) to delineate the cardiomyocyte boundaries and z-discs on histological sections of adult porcine heart. Cardiac tissue sections were first immunostained with anti–sarcomeric α-actinin antibody (Thermo Scientific #MA1-22863) as described in the “Immunostaining of porcine ventricular tissues” section. Before mounting, immunostained sections were washed once with 1× HBSS (Hanks’ balanced salt solution) and incubated in WGA solution (5 μg/ml) (prepared in 1× HBSS) at room temperature for 15 min. Sections were washed twice with 1× PBS and were mounted for fluorescence microscopy. Images were taken at ×20 magnification with Zeiss AxioVision Observer Z1 inverted microscope. For length and width measurements, the “line measurement” tool of AxioVision microscopy software (AxioVision Release 4.7, Carl Zeiss) was used as depicted in fig. S3. Longitudinal sections were selected for length and width measurements of the cardiomyocytes. Cardiomyocytes immunopositive for sarcomeric α-actinin and with DAPI staining (nuclei) were used for the analysis. The measurements from z-disc to z-disc of cardiomyocytes yielded length of the cardiomyocytes, whereas measurements across the cell yielded width. About 140 cells in randomly selected sections in each group were measured. Only those cells in which cell borders and z-discs were clearly visible were used in this analysis. Individual cells stained with sarcomeric α-actinin and WGA were also counted per unit imaged area and plotted to analyze cellular density in each group. To analyze vascular density in each group, histological sections were immunostained with anti–α-SMA (Sigma #A2547) and DAPI. Multiple images (each of area 12.46 mm2) of tissue sections were acquired with the “mosaic X” tool of AxioVision microscopy software (AxioVision Release 4.7, Carl Zeiss) at ×10 magnification. Images were analyzed by ImageJ software, and percent vascular (SMA) area was calculated for each group of pigs (n = 3).

Adenovirus construction

The cardiac α-actinin coding region was amplified by PCR from a mouse cDNA library (OriGene, cDNA clone MGC:62771 IMAGE:6308412). The coding sequence contained the ATG start codon but not the TAA stop codon (29). This fragment was subcloned to the vector pShuttle-CMV-mCherryN1 (provided by T. Weber). This construct was generated by subcloning the mCherry fragment to the pEGFPN1 plasmid (Clontech). The mCherryN1 fragment was subsequently subcloned to the pShuttle-CMV vector (provided by B. Vogelstein, Howard Hughes Medical Institute). The pShuttle-CMV-mCherryN1 construct can be used to subclone gene fragments that lack the stop codon to the multiple cloning sequences, thus producing fusion proteins with the N terminus of the mCherry protein. The resulting fusion protein α-actinin–mCherry is similar to the α-actinin–GFP (green fluorescent protein) fusion protein that has been demonstrated to become specifically localized within the sarcomeres of cardiomyocytes (29). Adenovirus carrying this vector was then packaged at the University of Iowa Vector Core.

Isolation of cardiomyocytes from porcine hearts

Ventricular samples (~1 cm2) from euthanized adult pigs were collected in 10-ml perfusion buffer (34) at room temperature. Samples were washed twice with fresh perfusion buffer after 20 min of incubation at room temperature. They were minced with sterile scissors and then incubated in 5 ml of collagenase type 2 (Worthington) solution (2 mg/ml) prepared in sterile-filtered 1× perfusion buffer. Enzymatic digestion was carried out at 37°C for 30 min with intermittent shaking. They were poured in a 100-mm petri plate and triturated with 10-ml pipette gently. Equal volume of stop solution [10% fetal bovine serum (FBS) in perfusion buffer] was added and mixed. Cells were filtered through a sieve of pore size 100 μm and pelleted by centrifuging at 40g for 2 min. Cell pellet was washed twice and resuspended in cardiomyocyte culture medium (CMC) formulated by adding 13% FBS, 5% horse serum, 1× nonessential amino acid, 2 mM sodium pyruvate, penicillin (100 U/ml), streptomycin (100 μg/ml), and fungizone (0.5 μg/ml) to Dulbecco’s modified Eagle’s medium (DMEM)/F12 (50:50). Cells at a density of ~2 × 105 per plate were seeded in 100-mm untreated polystyrene plates (Fisher Scientific). Nonadherent cells were collected every 24 hours and centrifuged at 20g for 2 min at room temperature. Cell pellet was washed with serum-free DMEM and seeded on new polystyrene plates with CMC. On day 4, cells were washed and plated on 1% bovine serum albumin (BSA)–coated glass bottom plates at a cell seeding density of 350 to 500 cells/cm2 (to avoid cell clumping). Medium was carefully removed the next day without disturbing the cells on the glass surface, and fresh CMC was replenished. Adenoviruses (1:500 dilution in perfusion buffer and stored in 4°C) were added and mixed slowly to achieve multiplicity of infection of 60. After 48 hours of incubation, transfection was confirmed by observing desired fluorescence in live cell imaging with Zeiss AxioVision Observer Z1 inverted microscope (Carl Zeiss). Medium was changed with fresh CMC, and time-lapse microscopy was performed.

Time-lapse microscopy

To capture cell division events in cardiomyocytes in vitro, we carried out time-lapse microscopy using the Zeiss AxioVision Observer Z1 inverted microscope in a humidified chamber in the presence of 5% CO2 at 37°C. Multiple random points with cells expressing mCherry (red) were selected in wells cotransfected with Adeno-act-mCherry and either Adeno-Ccna2 or Adeno-Null adenoviruses. The positions were marked with the “position-list” tool in the AxioVision microscopy software (AxioVision Release 4.7, Carl Zeiss). Only the channel for Texas red was used (for detection of mCherry) to acquire images for 72 hours. The fluorescein isothiocyanate (green) channel of the microscope was kept closed during the time-lapse imaging to avoid cell death from exposure to ultraviolet rays in this channel. Images were taken at intervals of 30 min. The objective lens of 10× was used for all time-lapse imaging. Time-lapse movies were generated after the end of each experiment and exported as .MOV files.

Immunostaining of cultured cardiomyocytes

After time-lapse microscopy, cells were fixed with 4% paraformaldehyde at room temperature for 20 min and washed with 1× PBS. Cells were permeabilized by incubating them in 0.5% Triton X-100 solution for 20 min at room temperature. Blocking was carried out with 5% BSA solution for 20 min at room temperature. Primary antibody (1:100 diluted) was added and incubated at room temperature in a moist chamber for 1 hour. Cells were washed three times with 1× PBS. Secondary antibody tagged with fluorophores was diluted (1:300) in 0.05% BSA in 1× PBS. Cells were incubated in secondary antibody at room temperature for 45 min in the dark. Cells were washed three times with 1× PBS. Nuclear staining was carried out with incubation of cells in DAPI solution (0.5 μg/ml) for 5 min. Cells were washed with 1× PBS and covered with fluorescent mounting medium (Dako). In each immunostaining experiment, suitable isotype control antibodies were used to confirm the specificity of test antibodies. After time-lapse microscopy, cells were fixed and immunostained with anti–troponin Tc (Santa Cruz Biotechnology) and images were acquired. Troponin Tc antibody was stripped off with stripping buffer (35) (composition: 60 mM tris buffer at pH 8.0 with 20 mM dithiothreitol and 2% SDS), and cells were re-immunostained with anti–cyclin A2 (Abcam) and anti–sarcomeric α-actinin (Thermo Scientific). Images were acquired with Zeiss AxioVision Observer Z1 inverted microscope.

Statistical analysis

All data generated in this study were analyzed with an unpaired t test. Data are presented as means ± SEM unless otherwise indicated. Two-sided α level of 0.05 was used, and normal distribution was confirmed with frequency histograms. Significance was considered to be P < 0.05.


Fig. S1. Cyclin A2 protein expression was determined at 6 weeks after injection of adenoviruses.

Fig. S2. MRI delayed enhancement studies exhibit a trend toward scar reduction in cyclin A2–treated animals.

Fig. S3. Measurement of cell size and number was performed by delineating cardiomyocyte cell borders.

Fig. S4. Cyclin A2 expression in cardiomyocyte nuclei was confirmed with immunofluorescence analysis in fixed cells.

Fig. S5. Immunofluorescence control experiments verify cardiomyocyte identity.

Table S1. Baseline studies of EDV and ESV as measured with 2D and 3D echocardiography exhibit no differences in baseline EF.

Table S2. EDV and ESV were measured with LV angiography, 2D and 3D echocardiography, and MRI.

Table S3. Cytokinetic indices of adult porcine cardiomyocytes were measured over 34 ROIs.

Movie S1. 2D echocardiography was used for EF assessment of porcine hearts.

Movie S2. 3D echocardiography was performed to measure EF of porcine hearts.

Movie S3. MRI was also used for EF determination of porcine hearts.

Movie S4. Rotational analysis of Z-stack image of mitotic nucleus in Fig. 3C.

Movie S5. Cytokinesis of adult porcine cardiomyocyte is captured by live cell imaging using time-lapse microscopy.


  1. Acknowledgments: We thank M. Zaidi and D. Wolgemuth for critical review of the data and manuscript. We thank T. Weber for the vector pShuttle-CMV-mCherryN1 used to construct our Adeno-act-mCherry virus. We thank M. Scheel at the University of Iowa Vector Core for guidance in vector construction. We thank R. Huq and the Mount Sinai Microscopy Core Facility for support in Axioplan 2 microscopy imaging. We thank J. Tripodi and B. Rawal for their technical assistance. Funding: H.W.C. is funded by the National Heart, Lung, and Blood Institute (R01-HL088255 and R41HL088867) and a sponsored research agreement from Broadview Ventures. Author contributions: S.D.S. and Y.K. performed the porcine imaging studies, induced MIs, and performed surgical procedures with injection of viruses. A.K.R. performed all in vitro cardiomyocyte studies and in vivo tissue analysis. S.D.S., A.K.R., and H.W.C. designed the study, analyzed the data, and co-wrote the manuscript. R.K.C. and H.W.C. constructed all adenoviruses. R.J.K. and R.B. performed in vivo tissue analyses. G.G.-M., J.S., and M.J.G. performed and analyzed in vivo imaging studies. Competing interests: H.W.C. is a founder and equity holder in VentriNova Inc. and is co-inventor on issued patents (10/557,873; 11/267,431; 11/535,444) and patent applications (EPO 4752668.6; Canada 2526490; Hong Kong 6110012.7; United States 12/642,055; EPO 6815444.2; Canada 2,660,661; Hong Kong 9102068.4) associated with this work and owned by Columbia University. The other authors declare no competing interests.
View Abstract