c-kit⁺ precursors support postinfarction myogenesis in the neonatal, but not adult, heart

Results and Discussion

To determine whether neonatal hearts are capable of regenerative cardiac myogenesis, cryoinfarctions were performed in postnatal day (PN) 1–3 mice harboring a transcription marker for c-kit (c-kit^BAC-EGFP; ref. 14), a gene product linked to CPC status in the heart (10, 14, 15, 17–19). Infarction resulted in the localization of (c-kit)EGFP⁺ cells and a marked induction of c-kit and EGFP mRNA within the area of injury, far greater than observed in identically infarcted adult hearts (14, 19, 20)(Fig. 1 A–C and Fig. S1A). Individual (c-kit)EGFP⁺ cells of varying morphology were observed throughout the neonatal infarct, but not in areas remote from the injury. The early cardiac transcription factor Nkx2.5 was identified in numerous Nkx2.5⁺ and (c-kit)EGFP⁺/Nkx2.5⁺ immature cells, and Nkx2.5 mRNA was induced within the neonatal infarct, but neither c-kit⁺/Nkx2.5⁺ cells nor Nkx2.5 message induction were observed in time matched adult infarcts (Fig. 1D and Fig. S1 B and C), excluding its expression as part of the revascularization process (10). Consistent with the (c-kit)EGFP marker identifying cells that include multipotent cardiovascular precursors (9–11), (c-kit)EGFP⁺ cells were observed at various stages of endothelial cell (EC, Flk-1⁺; PECAM-1⁺) or cardiomyocyte (α-actinin⁺; Nkx2.5⁺) differentiation (Fig. 1 D–G). Mast cells (tryptase⁺ or toluidine blue⁺) expressing (c-kit)EGFP were only observed at the epicardial border and were not consistent with the marked expansion of c-kit(EGFP)⁺ cells at the neonatal infarct border (Fig. S1D). Quantitative morphometry indicated that, at day 3 after infarction (D3), 67.2 ± 2.2% of (c-kit)EGFP⁺ cells within the infarct and border zone were of cardiac phenotype, increasing to 87.7 ± 4.9% at D7; the great majority of the remaining (c-kit)EGFP⁺ cells were identified as endothelial cells (EC) (Fig. 1G).

Most importantly, neonatal cryoinfarction lesions underwent marked neomyogenesis and regeneration following injury (Fig. 2 A and B); by D5, many striated cardiomyocytes were intermixed within areas of fibrosis. By the first week following injury, cardiac myocytes were observed within clusters of (c-kit)EGFP⁺ cells, and by D21, the infarcted region was infiltrated with numerous α-actinin positive myocytes, some of which were (c-kit)EGFP⁺, and fibrotic areas were progressively remodeled by 3 mo (Fig. 2 B–D). Quantitative analysis indicated an increase in cellular content within the scar area to 20.4 ± 4.1% at 3 wk and 43.2 ± 3.0% at 3 mo (P < 0.001; Fig. S2A). Thus, following an ablative infarction, c-kit transcriptionally active CPC migrate to the site of injury and progressively differentiate into heart cells, resulting in a regenerative response to the injury.

In vivo pulse-chase labeling with BrdU indicated that cardiac myogenesis was accompanied by a marked expansion of the (c-kit)EGFP⁺ pool that was localized to the infarct (Fig. 2 E and F). At day 3 postinfarction (PI), the number of (c-kit)EGFP⁺ cardiomyocytes within the infarct was 6.7-fold greater than in remote areas; overall, 20.5 ± 4.7 and 32.1 ± 2.9% of (c-kit)EGFP⁺ myocytes incorporated BrdU at D3 and D5, respectively. Moreover, (c-kit)EGFP⁺ cells were a major component of all myocytes showing evidence of DNA replication within the infarct, accounting for 25.4 ± 8.5 and 42.0 ± 6.9% of all BrdU⁺ cardiomyocytes at D3 and D5, respectively, compared with 2.2 ± 1.3 and 7.0 ± 4.1% in remote regions (Fig. 2F). By contrast, determination of histone H3 phosphorylation (pHH3) in myocytes, which reflects the pool of currently expanding myocytes, indicated a similar stimulation of mitosis in existing myocytes in remote and infarcted regions. pHH3⁺ cardiomyocytes increased from a control of 0.07 ± 0.01 to 0.22 ± 0.03% and 0.23 ± 0.08%, in remote and infarcted regions, respectively (Fig. S2 B and C). Taken together, these data are consistent with a generalized expansion of existing myocytes throughout the heart, but a localized expansion of (c-kit)EGFP⁺ cells and cardiac differentiation within the infarct. Quantitatively, by D5, ∼32% of (c-kit)EGFP⁺ cardiomyocytes localized to the infarct have passed through S phase, and more than 75% of (c-kit)EGFP⁺ have adopted a cardiac fate (Fig. 1G), indicating that postinfarct cardiac myogenesis arises at least partially from a local expansion of the (c-kit)EGFP⁺ CPC pool. The pancardiac expansion of myocytes following neonatal infarction is not associated with c-kit expression as evidenced by minimal c-kit⁺/BrdU⁺ cells outside of the infarct zone (Fig. 2F).

The (c-kit)EGFP marker reports current c-kit transcriptional activity and thus cannot definitively identify the developmental origin of myocytes observed within the infarction. To unequivocally establish the ability of undifferentiated (c-kit)EGFP⁺ cells to adopt cardiac fates, we infarcted neonatal c-kit^BAC-EGFP/αMHC-mCherry mice and FACS purified cells for in vitro differentiation experiments within 6 h of infarction, to prevent dedifferentiation of myocytes. Pure populations of EGFP⁺/mCherry⁻ (noncardiac) cells began to express mCherry after incubation in cardiac differentiation medium (14) (Fig. 2G). By 12–13 d, 12.0 ± 1.4% of cells expressed mCherry (Fig. S2 F and G), whereas no EGFP⁻/mCherry⁺ cells were observed to express EGFP in parallel experiments. Moreover, purified EGFP⁺/mCherry⁻ cells did not simply express mCherry, but formed a contractile cardiac phenotype with spontaneous contractions (Movies S1 and S2). FACS-purified neonate (c-kit)EGFP⁺ cells also adopted cardiac fates when injected into adult infarcts (SI Methods and Fig. S2H), whereas FACS-purified adult (c-kit)EGFP⁺ cells do not undergo cardiac differentiation in vitro or when transplanted into adult infarcts (14, 15). Thus, although our data do not exclude expansion of existing myocytes (16), they provide evidence of a pool of precursor cells that are capable of supporting myogenesis postinfarction in neonatal mice.

We next sought to further discriminate between progenitor and cardiomyocyte contributions to the expansion of (c-kit)EGFP⁺ cells, and attendant postinfarction myogenesis. We infarcted triallelic (c-kit^BAC-EGFP/ αMHC-CreER/R26-floxed STOP-βGal) neonatal mice, in which myocytes were marked with β-gal by tamoxifen-induced recombination before birth, resulting in a minimum of 3 d between tamoxifen exposure and infarction. Infarcted regions contained a markedly lower percentage of recombined cells than remote areas at D5, the time point of maximum (c-kit)EGFP⁺ cells within the infarct, and differing distinctly from the localized expansion of (c-kit)EGFP⁺ cells (Fig. 2 H and I and Fig. S2D). To exclude (c-kit)EGFP⁺ cells that adopt an EC fate, we compared the percentage of recombined (c-kit)EGFP⁺ myocytes to the overall cardiomyocyte recombination rate. In three experiments, the recombination rate of (c-kit)EGFP⁺ myocytes was 68.8 ± 4.2% that of EGFP⁻ cardiomyocytes (P < 0.01; Fig. S2E), consistent with a nonmyocyte source for at least one-third of these cells, particularly when viewed in light of the induction of c-kit in differentiated myocytes following injury (14).

Our data indicate a major expansion of the (c-kit)EGFP⁺ cells in response to cardiac injury; however, myogenesis associated with these cells could reflect an influx of bone-marrow-derived c-kit⁺ cells following injury, as ∼50% of c-kit⁺ cells have been reported to be copositive for the hematopoietic lineage marker CD45 in adult mice (20). We observed that less than 3% of (c-kit)EGFP⁺ cells were CD45⁺ during the first week following infarction of the neonatal mouse (Fig. 1G). CD45⁺/(c-kit)EGFP⁻ cells migrated to the infarct in large numbers, but did not participate in the formation of new myocytes or vascular structures. Rather, these cells surrounded vessels and myocytes, persisting for weeks following the injury (Fig. 3 A–D). Unlike (c-kit)EGFP⁺ cells, CD45⁺ cells also did not coexpress cardiac α-actinin (Fig. 3B) or spatially overlap with PECAM-1⁺ cells (Fig. 3 C and D), further establishing (c-kit)EGFP⁺/CD45⁻ cells as the precursor pool.

Although controversial in the adult heart (21, 22), neonatal c-kit⁺/CD45⁻ bone-marrow-derived stem cells could contribute to the formation of new heart cells, as neonatal cells might possess enhanced differentiation potential compared with adult hematopoietic stem cells. To test this possibility, we transplanted pooled marrow from 5-d-old c-kit^BAC-EGFP/pCAGGS-dsRed (23) mice into irradiated adult B6 recipients, and infarcted these mice following reconstitution. Virtually no (c-kit)EGFP⁺ cells were observed in infarcts in these experiments, although dsRed cells did home to the site of injury as expected (Fig. 3E), and microscopic evaluation failed to reveal (c-kit)EGFP⁺ or dsRed⁺ myocytes or vascular endothelial cells, excluding hematopoietic stem cells as the myogenic/vascular precursors. This experiment is limited by the potential loss of transdifferentiation capacity of neonatal bone marrow during reconstitution (14–21 d), but reconstitution with PN2-3 neonatal marrow for 2–4 d also failed to result in the migration of any (c-kit)EGFP⁺ cells to the heart (SI Methods). To further confirm the cardiac origin of (c-kit)EGFP⁺ cells, we transplanted infarcted c-kit^BAC-EGFP neonatal ventricular tissue into SCID mice. Three to five days after transplantation, the infarcted tissue showed induction of (c-kit)EGFP transcriptional activity in cells within the transplant, many of which adopted cardiac fates (Fig. 3F). By contrast, striated myocytes were not observed in adult transplanted tissue (see below), excluding the induction of EGFP in differentiated myocytes. Thus, cardiac CPC migrate to the region of cardiac injury, undergo cell division, and initiate a wave of cardiovascular differentiation.

As c-kit⁺ cells participate in the formation of new vessels within adult infarcts, which are devoid of CPCs (14, 15), our results indicate that at least two distinct pools of (c-kit)EGFP⁺ cells participate in cardiac repair. Furthermore, because marrow-derived cells appeared to play at most a very minor role in revascularization of the neonatal infarct (Fig. 3 A–D), our results might reflect a previously unappreciated endogenous source for such cells in the adult heart. To further define the pool of c-kit–expressing cells participating in adult cardiac repair, we performed adoptive transfer experiments using adult dsRed marrow transplanted to c-kit^BAC-EGFP recipients, as well as c-kit^BAC-EGFP donor marrow transplanted to dsRed recipients, and infarcted the mice following reconstitution. As shown in Fig. 4A, dsRed-labeled marrow cells prominently occupy the infarct, but these cells play little or no role in the formation of vascular structures, whereas host (c-kit)EGFP⁺ cells markedly participate in new vessel formation. Conversely, despite complete reconstitution, (c-kit)EGFP⁺ marrow cells were rarely detected in the infarcted hearts of cardiac mCherry (αMHC-mCherry) mice and never adopted cardiac or vascular lineages, although numerous CD45⁺ cells were observed (Fig. 4 B and C). Thus, as in the neonate, there is little evidence of traffic of c-kit⁺ hematopoietic stem cells to the infarcted adult heart. As (c-kit)EGFP⁺ cells are not found in adult mice (14), these results indicate activation of the reporter in resident cells that migrate to the infarct and adopt strictly vascular fates, a process similar to recent reports of perivascular precursors that adopt EC and smooth muscle fates in vitro, and underlie atherosclerotic injury responses (24–26). To definitively confirm that adult c-kit⁺ cells derive from the surrounding, nonablated cardiac tissue, we transplanted the left ventricular freewall of adult c-kit^BAC-EGFP/(pCAGGS)dsRed mice into SCID mice s.c. immediately following cryoinfarction. As in neonatal mice, (c-kit)EGFP⁺ cells were observed 5 d following infarction/transplantation of adult heart tissue; however, (c-kit)EGFP⁺ cells contributed only to the formation of vascular structures (Fig. 4D). Highlighting the distinct molecular features of the neonatal and adult processes, the expression of sca-1, which has been implicated in angiogenic responses mediated by adventitial vascular precursors (24, 26), was significantly up-regulated in adult infarcts, but is not a feature of the neonatal response (Fig. 4E). Thus, in adult infarcts, resident vascular precursors, rather than authentic CPC, migrate from the surrounding heart tissue and differentiate to form vascular structures.

Taken together, our results suggest that the failure of myogenesis following adult infarction is not a result of a context-dependent limit to differentiation associated with the infarct per se, but depends critically on the developmental state of the heart, consistent with data indicating that the early human heart possesses significant regenerative potential (8). Recent data indicate that early cardiac regeneration involves activation of myocyte cell cycle reentry in the entire neonatal heart, resulting in widespread dedifferentiation and proliferation, without formation of a fibrotic scar (16). Our data do not exclude the participation of non-terminally differentiated cardiomyocytes in neonatal infarct repair, but the local expansion of undifferentiated, (c-kit)EGFP⁺ cells at the area of infarction, the induction of the early cardiac transcription factor Nkx2.5 in these cells, and their adoption of multiple developmental fates, coupled with a limited expansion of previously differentiated heart cells at the site of injury, clearly establishes the role of CPC in this process. Strikingly, our explant studies demonstrate the intrinsic capacity of the neonatal heart for myogenesis, a capacity that temporally correlates with the existence of authentic CPC (14, 15, 19). We further demonstrate that the neonatal and adult heart walls contain a reservoir of angiogenic precursors, and that activation of these cells results in c-kit expression. Finally, our data highlight the critical distinction in developmental potential of adult heart cells that express the stem cell factor receptor, c-kit. Mobilization of c-kit⁺ cells in the infarcted adult heart from surrounding heart tissue results in extensive angiogenesis/vasculogenesis. Although incapable of cardiac myogenesis, these c-kit⁺ vascular progenitors may be attractive targets for directed cardiac differentiation in vivo.

Methods

Supplementary Material