De novo cardiomyogenesis is limited to ≈1% per year in the adult mammalian heart.1 Whether newly formed cardiomyocytes are derived from division of pre-existing myocytes or from differentiation of resident cardiac progenitor cells is a topic of debate. Cardiac progenitor cells have been posited as a source of endogenous cardiomyocyte renewal and as cells that can be harvested, expanded in vitro, and delivered therapeutically after infarction. Stem cell antigen 1 (Sca-1), which was initially described as a surface marker of murine hematopoietic stem cells, has been reported to mark resident cardiac progenitor cells,2 and a previous study reported frequent contribution of Sca-1–expressing cells to cardiomyocytes.3 However, the transgenic approach used in this study resulted in more widespread expression than just Sca-1–expressing cells, complicating interpretation of the results.4 Because more recent fate mapping studies questioned the existence of resident cardiac progenitor cells, we aimed to test the hypothesis that endogenous Sca-1–expressing cells are progenitors for cardiomyocytes in vivo under physiological and pathophysiological conditions.
We generated a tamoxifen-inducible genetic lineage-tracing Sca-1 mER-Cre-mER knock-in mouse model (Sca-1mCm/+)5and cross-bred to a Rosa26 tdTomato reporter line (Sca-1mCmR26tdTomato) to assess the fate of cellular descendants of Sca-1–expressing cells in vivo. Adult Sca-1mCmR26tdTomato animals were treated with tamoxifen daily for 7 days, and hearts were harvested 7 days after the last tamoxifen injection to assess recombination. Recombination in the heart averaged 33% in Sca-1+/CD31+ populations and 13% in the Sca-1+/CD31– population (Figure [A] and [B]). No tdTomato+ cells were identified by flow cytometry in Sca-1mCmR26dTomato mice that did not receive tamoxifen. Histology (Figure [C]) demonstrated a mostly endothelial expression pattern in Sca-1mCmR26dTomato hearts that colocalized with CD31 expression. Additionally, there was overlap between Sca-1 and tdTomato staining as expected, as well as some NG2-expressing perivascular cells that overlapped with tdTomato. In tissue sections, TdTomato was not identified in cardiomyocytes. Additionally, tamoxifen-dependent tdTomato expression in tissues known to express Sca-1 (kidney, lung, liver, ileum) confirmed the presence of tdTomato+ cells, predominately in an endothelial pattern; this was absent in tissue from animals that received no tamoxifen (data not shown).
Next, we set out to test whether new cardiomyocytes were derived from descendants of Sca-1–expressing cells after myocardial infarction (MI). Adult Sca-1mCmR26dTomato mice were treated with tamoxifen daily for 7 days, followed by a 7-day chase period. The mice then underwent MI or were allowed to age normally. Mice were euthanized 6 months later for examination of the heart. We enzymatically dissociated cardiomyocytes and quantified either fraction of tdTomato+cardiomyocytes from live cells (uninjured), or after immunostaining for cardiac troponin T and tdTomato (post-MI). This allowed unambiguous distinction of cardiomyocytes and nonmyocytes (Figure [D]). We first assessed the numbers of tdTomato+ cardiomyocytes at baseline before injury and detected 6 tdTomato+ cardiomyocytes from 5 hearts (0.0002%; Figure [E]). We then assessed labeling of cardiomyocytes after MI and detected a total of only 19 tdTomato+cardiomyocytes from 7 hearts (Figure [F]). This accounts for 0.007% tdTomato+ cardiomyocytes after infarction on average. Importantly, not a single cardiomyocyte expressed tdTomato in animals that were not exposed to tamoxifen, demonstrating that this is not a leaky tracing system.
The results of this study do not support the hypothesis that Sca-1–expressing cells have the ability to significantly contribute new cardiomyocytes to the heart after MI in the mouse. Despite using the best available techniques for lineage tracing, including knocking a tamoxifen-inducible Cre recombinase into the endogenous Sca-1 gene, using a sensitive Cre-reporter strain, and enzymatically dispersing cardiomyocytes for single-cell screening, we were unable to detect significant contributions of the Sca-1 lineage to cardiomyocytes.
Our study contrasts the study by Uchida et al,3 which showed a significant contribution of Sca-1+ cells to cardiomyocytes. The main difference between these 2 studies relates to the lineage-tracing approach (ie, a transgenic promoter fragment [Uchida et al] versus the endogenous Sca-1 locus dependent Cre expression [this study]). The transgenic system used by Uchida et al has been reported to generate a large number of false-positives, meaning the promoter fragment is active in cells that normally do not express Sca-1.4 Therefore, we believe the transgenic promoter fragment may have overestimated the true cardiogenic potential of endogenous Sca-1–expressing cells in this model. In our model, the recombination rates in the heart were reasonable, but still only resulted in 0.007% of cardiomyocytes to become labeled after MI. Thus, although this number might be an underestimation, it is 2 orders of magnitude too low to explain a 1% per year cardiomyocyte turnover rate. When coupled with data demonstrating that most cardiomyocyte renewal can be accounted for by division of pre-existing cardiomyocytes, our findings reduce the likelihood that there is a cardiac progenitor cell population in mammalian species.
