Cardiac Bmi1+ cells contribute to myocardial renewal in the murine adult heart

Introduction The mammalian adult heart maintains a continuous, low cardiomyocyte turnover rate throughout life. Although many cardiac stem cell populations have been studied, the natural source for homeostatic repair has not yet been defined. The Polycomb protein BMI1 is the most representative marker of mouse adult stem cell systems. We have evaluated the relevance and role of cardiac Bmi1+ cells in cardiac physiological homeostasis. Methods Bmi1CreER/+;Rosa26YFP/+ (Bmi1-YFP) mice were used for lineage tracing strategy. After tamoxifen (TM) induction, yellow fluorescent protein (YFP) is expressed under the control of Rosa26 regulatory sequences in Bmi1+ cells. These cells and their progeny were tracked by FACS, immunofluorescence and RT-qPCR techniques from 5 days to 1 year. Results FACS analysis of non-cardiomyocyte compartment from TM-induced Bmi1-YFP mice showed a Bmi1+-expressing cardiac progenitor cell (Bmi1-CPC: B-CPC) population, SCA-1 antigen-positive (95.9 ± 0.4 %) that expresses some stemness-associated genes. B-CPC were also able to differentiate in vitro to the three main cardiac lineages. Pulse-chase analysis showed that B-CPC remained quite stable for extended periods (up to 1 year), which suggests that this Bmi1+ population contains cardiac progenitors with substantial self-maintenance potential. Specific immunostaining of Bmi1-YFP hearts serial sections 5 days post-TM induction indicated broad distribution of B-CPC, which were detected in variably sized clusters, although no YFP+ cardiomyocytes (CM) were detected at this time. Between 2 to 12 months after TM induction, YFP+ CM were clearly identified (3 ± 0.6 % to 6.7 ± 1.3 %) by immunohistochemistry of serial sections and by flow cytometry of total freshly isolated CM. B-CPC also contributed to endothelial and smooth muscle (SM) lineages in vivo. Conclusions High Bmi1 expression identifies a non-cardiomyocyte resident cardiac population (B-CPC) that contributes to the main lineages of the heart in vitro and in vivo. Electronic supplementary material The online version of this article (doi:10.1186/s13287-015-0196-9) contains supplementary material, which is available to authorized users.


Introduction
The adult mammalian heart was long considered a terminally differentiated organ with no capacity to replace aged or damaged cardiomyocytes (CM) [1]. This view was challenged by considerable evidence of low but intrinsic CM turnover in the adult mouse heart [2], although the contribution of adult CM turnover to heart homeostasis and the origin of the new cells remain unclear. There is compelling evidence that this low turnover rate throughout heart life mainly reflects the activity of a reservoir of cardiac stem cells (CSC) [3] that might reside in physiological niches [4,5]. Dedifferentiation and division of preexisting adult CM populations were recently proposed to contribute notably to heart turnover after myocardial infarction [6,7].
Adult stem cells maintain and repair host tissues in adult organisms, and self-renewal, differentiation, and prevention of senescence of these cells are thus critical for tissue homeostasis. Adult resident cardiac stem/progenitors are defined primarily by the expression of cell surface markers such as c-KIT [8], SCA-1 [9,10], ATP-binding cassette ABCG2 [11], PDGFRα [12] or combinations of these. The diversity of these findings has hindered a broad, unambiguous consensus for identification and molecular definition of endogenous CSC [3]. Several authors have emphasized the need for careful in vivo lineage tracing of CSC progeny to ascertain how this cell type contributes to CM replenishment during homeostasis or after myocardial injury [13][14][15]. Recent lineage tracing studies yielded interesting results. SCA-1 was reported to label a noncardiomyocyte population in adult heart that clearly contributes to CM generation during homeostasis and normal aging (4.55 ± 0.87 %) [16]. In a c-kit lineage tracing study, c-KIT + CSC appeared to make a small contribution to the generation of new CM (0.03 ± 0.008 %) in adult mouse heart [17]. Two additional lineage tracing studies, although not directly related to CSC regulation, should be mentioned. Malliaras et al. identified small nonmyocyte cells termed cardioblasts (CdBs) that express sarcomeric α-actinin (38 %), α-MHC (39 %) and SCA-1 (55 %) but not c-KIT [18]. Activated CdBs apparently did not originate from hematogenous seeding, cardiomyocyte dedifferentiation, or mere expansion of a preformed progenitor pool; they, therefore, appear to arise by activation/differentiation of the endogenous CSC pool. Finally, a cardiac progenitor population defined as TIE-1 + CPC, a classical endothelial marker, is able to generate 70 % of the SCA-1 + intermediate perivascular progenitors that contribute (up to 3 %) to CM homeostatic turnover origin [19]. These findings reinforce the hypothesis that cardiac progenitor cells support the low CM turnover in the adult heart. Further genetic lineage tracing studies will help to elucidate the complex relationship between these partially different populations.
Bmi1, a member of the Polycomb repressive complex 1 (PRC1), is a transcription factor involved in many biological processes including embryonic development, organ formation, tumorigenesis, and stem cell stabilization and differentiation [20]. Bmi1 has a crucial role during selfrenewal and maintenance of hematopoietic, neural, intestinal, bronchioalveolar, pancreatic, prostate, lung and epithelial stem cells, as well as in the tongue and in rodent incisors [21][22][23][24][25][26][27][28][29]. There is little information on the role of Bmi1 in the adult heart. Upregulation of Bmi1 expression is cardioprotective against doxorubicin-induced damage [30]. A recent study demonstrated that, by controlling senescence, Bmi1 expression in adult mouse CM is limiting dilated cardiomyopathy and heart failure [31]. Although a significant proportion of c-KIT + human and porcine CSC expressed low BMI1 levels [32,33], no specific study has addressed the functional relevance of this factor.
We hypothesized that adult cardiac progenitor cells may be characterized by high Bmi1 expression, as in other adult stem cell compartments [21][22][23][24][25][26][27][28][29]. Using a validated lineage tracing strategy to track activity of the Bmi1 locus, we show that the adult heart contains a resident noncardiomyocyte population of Bmi1-expressing progenitor cells that constitute a fraction of the SCA-1 + population. Bmi1-CPC (B-CPC) show enriched expression of several multipotency and stemness markers, and their numbers increase throughout the lifetime of the mouse. B-CPC contribute significantly to the generation of de novo CM, endothelial and SM cells throughout life.

Immunodetection analysis
Heart immunohistochemistry was performed as previously described [7]. Specific yellow fluorescent protein (YFP) detection with anti-GFP antibody was confirmed by control immunofluorescence analysis of heart sections of TM-injected Rosa26 YFP/+ mice and of non-induced Bmi1-YFP mice (Bmi1-YFP NI ); no signal was observed for either ( Fig. 2a). For immunodetection, sections were fixed in 2 % paraformaldehyde (PFA) and rinsed in PBS or PHEM buffer (25 mM Hepes, 10 mM EGTA, 60 mM PIPES, 2 mM MgCl 2 ; all from Sigma). Slides were rinsed in blocking buffer (BB; 0.5 % porcine skin gelatin, 0.1 % bovine serum albumin; BSA; Sigma), incubated in 150 mM glycine (Merck, Madrid, Spain) (10 min, room temperature (RT)), followed by sodium borohydride (Sigma; 10 min) and finally in PBS with 0.1 % Triton X-100 (Sigma). Preparations were incubated with primary antibodies (see Additional file 1: Table S1) (1-3 h, RT), washed and incubated with the appropriate secondary antibody (1 h). Slides were incubated with Sytox Green and mounted in ProLong antifade reagent (both from Invitrogen, Madrid, Spain). Images were captured with a Leica SP5, Zeiss LSM 700 or LSM 780 coupled to a two-photon Spectra-Physics Mai Tai laser scanning confocal microscope and were assembled with ImageJ software (NIH). Processing, including assignment of pseudo-colors and changes in brightness, was applied uniformly to the entire image exclusively to equalize the appearance of multiple panels in a single figure. Immunocytochemistry was performed as above.

Cell isolation, culture and flow cytometry
Hearts were collected from Bmi1-YFP mice five days after TM induction, perfused with PBS to remove blood cells, and processed by enzymatic digestion using 0.1 % collagenase IV (Sigma) and 10 μg/ml DNAse (Roche, Madrid, Spain) (40 min, 37°C). The resulting single cell suspension was passed through a 40 μm filter to remove debris. YFP + cells were separated from total heart mass with a BD Fac-sAria II Special Order System cell sorter fitted with a 488 nm laser to excite YFP (collected in the 525/50 channel). To discriminate YFP + from autoflorescent cells, a 488 nm laser was used to excite cells, followed by collection in the 585 channel (phycoerythrin). For flow cytometry analysis, cardiac cells from hearts of TM-induced Bmi1-YFP mice were incubated with the following primary and secondary antibodies as indicated: APC (allophycocyanin)-conjugated rat anti-c-KIT, APC-rat anti-SCA-1/Ly6a, biotin-rat anti-SCA-1/Ly6a, biotin-rat anti-CD45, biotin-rat anti-CD31 (all at 1:100; all from BD Pharmingen, Madrid, Spain), and streptavidin-Alexa Fluor 405 conjugate (1:500; Invitrogen). Labeled cells were examined with a BD Facs Canto II flow cytometer and data analyzed using Facs DIVA Software.
For Fluo-4 AM labeling, we prepared a stock solution of 1 mM Fluo-4 AM (Invitrogen) in DMSO with an equal volume of 20 % Pluronic F-127 DMSO (1:1 ratio); the working concentration was 1 μM. Fluo-4 AM was added to DMEM supplemented with 100 U/ml penicillin, 100 mg/ml streptomycin and 2 mM L-glutamine; cells were incubated in the dark (20-30 min). We washed the cells and added fresh DMEM without phenol red (Sigma) and images were acquired by confocal microscopy as for Ca 2+ fluorescence. Fluo-4 was excited with the 488 nm line of an argon laser and 505 nm signal emissions were collected. Images were captured in a time series (xyt, pixel dwell 1.58 μs) and 2D images (512 × 512 lines) were obtained and stored for offline analysis.

B-CPC differentiation potential
To evaluate spontaneous endothelial/smooth muscle differentiation potential of Bmi1-CPC, cells were obtained from eight-week-old TM-induced Bmi1-YFP mouse hearts and seeded on gelatin-coated plates as above. After seven to ten days, plates were fixed with 2 % PFA in PHEM buffer (15 min, RT) and processed for immunocytochemistry. For cardiomyocyte differentiation, B-CPC cardiospheres were plated on a monolayer of NBRC or adult transgenic mouse GFP + CM derived from the beta-actin GFP mouse strain (Jackson Laboratory). For B-CPC/adult mouse CM co-culture, we used M-199 medium, 0.05 % FBS, 0.001 % ITS and 25 μM blebbistatin. After four to five days co-culture, cells were fixed with 2 % PFA in PHEM buffer (15 min, RT) and processed for immunocytochemistry.

Bone marrow transplant
To generate bone marrow (BM) chimeras, eight to ten week-old C57BL/6 Bmi1-YFP mice were TM-induced, lethally irradiated (one dose each of 4.75 and 4.5 Gy, separated by 24 h) and then transplanted (i.v.) with 10 7 whole BM cells isolated age-matched C57BL/6 Act-RFP mice [35]. At two months post-TM induction (54 days after BM transplant), after confirmation of full chimerism, transplanted Bmi1-YFP mouse hearts were digested and analyzed as above.

RT-qPCR and genomic PCR analysis
RNA was extracted from hearts of eight-week-old TMinduced Bmi1-YFP mice, or from the indicated subpopulations (Bmi1-CPC and SCA-1 CPC) purified using the sorting strategy described above, with a Cells-to-CT kit (Ambion, Thermo, Madrid, Spain). RNA from ES cells was prepared as previously described [36]. Complementary DNA was obtained by reverse transcription with the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Madrid, Spain). cDNAs were analyzed by real time PCR using the Power SYBR Green PCR Master Mix (Applied Biosystems). Amplification, detection and data analysis were carried out with an ABI PRISM 7900HT Sequence Detection System. The crossing threshold values for individual mRNAs were normalized to GusB expression for mRNA. Changes in mRNA expression were denoted as the x-fold change relative to the control. (See Additional file 2: Table S2 for primers used).

Statistical analysis
Statistical analysis was performed with Prism 5.0 (Graph-Pad Software). Significance between groups was evaluated in all experiments as detailed in the figures. A value of P < 0.05 was considered significant. All replicates considered are biological replicates.

Results and Discussion
Murine SCA-1 + CPC population expresses Bmi1 at similar levels to embryonic stem cells We hypothesized that adult cardiac progenitors would show high Bmi1 expression, as it has been previously defined in other adult stem cell systems [21][22][23][24][25][26][27][28][29]38]. The cardiac SCA-1 + population (SCA-1 + CPC) contains cardiac progenitor cells [10]. We initially compared expression of Bmi1 and other stemness-related genes in SCA-1 + CPC and embryonic stem cells (ES), and found low but comparable Bmi1 levels in both populations (see Additional file 3: Figure S1A). Bmi1 expression in these cells was also very sensitive to oxygen culture conditions and passage number (see Additional file 3: Figure S1B). These preliminary studies confirmed Bmi1 expression in SCA-1 + CPC at levels comparable to ES, and established culture conditions appropriate for maintenance of Bmi1 expression. We therefore studied whether Bmi1 levels corresponded to the entire SCA-1 + CPC population or whether defined subpopulations were associated with its expression.
We used flow cytometry to characterize non-CM YFP + cells derived from Bmi1-YFP hearts at 5d-postTM (see Additional file 4: Figure S2A); most were SCA-1 + (95.9 ± 0.4 %, n = 16), although they made up only 5.4 ± 0.4 % of the total SCA-1 + population (Fig. 1d). These non-CM YFP + cells were mainly CD45 − (<1 %) and c-KIT − (<0.5 %) but CD31 + (94 %) (see Additional file 4: Figure S2A). We used RT-qPCR to compare the gene expression profile of the sorted YFP + population with that of the closely related SCA-1 + YFP − cardiac population (S-YFP − ). The YFP + population showed higher expression of stemnessassociated genes and transcripts related to muscle contractility (Fig. 1e). These results suggest that Bmi1 + cells are probably a heterogeneous population, we cannot currently distinguish whether this heterogeneity is due to non-progenitor cells or to different progenitor populations.
Analysis of the YFP + compartment (the progeny of Bmi1 + cells) at one-year post-TM induction showed that this population increased during this period (2 x 10 4 ± 4 x 10 3 cells/heart; Fig. 1f, g), with an expression profile similar to that of 5d-postTM cells (see Additional file 4: Figure S2B). Analysis after conventional TM induction (5d-postTM) of aged (one-year-old) mice showed the presence of the YFP + population (Fig. 1f, g).
These results suggest that the Bmi1 + non-myocyte population (B-CPC) is maintained throughout the mouse lifespan and is apparently not in equilibrium with more primitive precursors that would dilute initial labeling.

B-CPC spatial distribution and differentiation potential
To locate Bmi1 + cells in mouse heart, we used mainly a Bmi1-YFP transgenic model and complemented the study with Bmi1-tomato mice for additional information. GFP immunostaining from 5d-postTM Bmi1-YFP hearts showed broad YFP distribution not observed in negative controls (Fig. 2a). B-CPC cells were located in variablesized clusters of compact cells with cramped nuclei (Fig. 2b) scattered throughout heart sections, with preferential perivascular (Fig. 2c) and inter-sarcomeric localization (Fig. 2d). We found broad distribution of B-CPC cells both in atria and ventricles (see Additional file 5: Figure S3). GFP immunostaining of cardiac explants [39] from 5d-postTM Bmi1-YFP mice showed that a large proportion of the bright, rounded cells that migrated on the fibroblast-like layer were YFP + . In orthogonal projections, the rounded BMI1 + cells selectively occupied the upper layer (see Additional file 6: Figure  S4A). Like their freshly sorted counterparts 5d-postTM, explanted YFP + cells expressed BMI1 (see Additional file 6: Figure S4B). This result suggests that Bmi1 + cells are related to previously reported populations now being evaluated in clinical trials [40,41].
Cardiac progenitor cells are able to differentiate to the main heart lineages. Although several methods have been tested for in vitro differentiation, results for terminal differentiation with sarcomeric structure and beating cardiomyocytes are poor [17,42,43]. We examined the differentiation potential of B-CPC in vitro, based on our previous work with SCA-1 + CPC [44]. Only a small percentage of sorted YFP + B-CPC developed sporadic microvascular networks composed of VE (vascular endothelial)-cadherin + and SMA + (smooth muscle actin) cells ( Fig. 2e), indicating that although B-CPC initially express CD31, they are not committed to the vascular lineage. It has been reported that hematopoietic stem cells, in different stages of development, or cardiac progenitor cells are characterized by the expression of endothelial markers, such as CD31 and Tie2 [19,40,45]. Moreover, CD34 expression has been found in various stem cell systems [46]. These findings indicate that these proteins have various cellular functions and are not restricted to mature endothelium [47].
Co-culture of neonatal rat CM with B-CPC (Bmi1-YFP mice) promoted their cardiac differentiation, with coexpression of sarcomeric α-actinin (SαA) and YFP (Fig. 2f ). Although the frequency of mononucleated YFP + cells that co-stained for SαA was low, orthogonal projection of these cells confirmed co-expression (Fig. 2f,  right). Results were similar when we co-cultured B-CPC (Bmi1-tomato mice) with adult mouse GFP + CM (βactin GFP mice) to discard fusion events. Although we did not detect marked mature differentiation, some tomato + cells initiated a SαA expression and CM transition program without fusion, detectable by double GFP + tomato + staining (Fig. 2g). These data suggest that the B-CPC population contains progenitors able to contribute all three main heart lineages.

B-CPC contribute to adult cardiomyocyte turnover
To assess the role of B-CPC in homeostatic conditions, we traced YFP + cells in the adult heart throughout the TM-induced Bmi1-YFP mouse lifetime (Fig. 3a). Immunohistochemical analysis of Bmi1-YFP heart sections from 2 to 12 months post-TM administration showed YFP + CM, which co-localized with SαA staining (Fig. 3b). To confirm these results and further evaluate the appearance of YFP + CM, we isolated and plated the enriched CM fraction for dual analysis by flow cytometry and immunostaining.
No YFP expression was detected in the CM-enriched population shortly after induction (5d-postTM), either by cytometry (<1 YFP + CM in 1.35 x 10 5 total cardiomyocytes or < 0.0022 ± 0.0016 %) (Fig. 3d) or in plated preparations (Fig. 4). To confirm the specific Bmi1 expression in the YFP + sorted fraction 5d-postTM and not in the CM compartment, we performed RT-qPCR analysis, which showed preferential Bmi1 expression mainly in the YFP + non-CM fraction (>15-fold; see Additional file 7: Figure  S5A, top). No BMI1 protein expression was detected by western blot in the sorted YFP − non-CM compartment or in the enriched CM fraction (see Additional file 7: Figure  S5A, bottom). To further confirm the lack of Rosa26 locus recombination in CM after TM induction, we isolated the CM-enriched fraction from Bmi1-YFP mice 5d-postTM, and analyzed the structure of the transgenic Rosa26 locus by PCR. There was no detectable recombination in the CM-enriched fraction from Bmi1-YFP mice compared with positive and negative controls (see Additional file 7: Figure S5B).
In addition, we analyzed Cre expression by immunocytochemistry in plated CM; no Cre expression was detected in the enriched CM fraction of Bmi1-YFP mice compared to the CM fractions of Myh6-YFP or WT mice (see Additional file 7: Figure S5C). This result strongly suggests that the cardiac Bmi1-expressing fraction does not include a minor subpopulation of adult CM with specific dedifferentiation/proliferation potential [6]. Given the wide variety of Bmi1 functions, we cannot currently rule out Bmi1 expression in intermediate progenitors or trace levels in differentiated cells. The Bmi1-CreERT construct is an inducible transgenic model in which a specific Cre expression threshold is necessary to trigger Rosa26 locus recombination [48]. Although we cannot completely exclude basal Bmi1 expression in CM, we confirmed that the lineage-tracing results were not due to ectopic Cre expression by CM.
Analysis of CM-enriched fractions of Bmi1-YFP chimera heart showed YFP + CM, with no mRFP + CM at two months post-transplant ( Fig. 6f-h) compared to CM preparations from Act-mRFP control mice (Fig. 6f ). These results were confirmed by immunocytochemistry in plated CM (Fig. 6h). We observed no fused cells (orange) in CM from chimeric Bmi1-YFP hearts (Fig. 6h). The results suggest that, as for other putative cardiac stem cells [18,50], BM-derived cells do not contribute notably to the B-CPC population, either directly or after fusion with resident non-CM cells.

B-CPC in the general context of cardiac repair and cardiomyocyte turnover
Cardiac progenitors are routinely isolated based on expression of the stem cell hematopoietic marker SCA-1 [9], although SCA-1 appears to label a heterogeneous population with predominantly endothelial potential [16,50]. Genetic deletion of Sca-1 showed that resident CPC do not respond efficiently to pathological damage in vivo, consistent with the impaired growth and survival of cardiac progenitor cells in vitro [50]. Sca-1 labels a non-CM population in heart that contributes to CM generation during homeostasis (4.55 ± 0.87 %) [16]. The SCA-1 + CPC contribution to the repair process is only found after ischemic damage and pressure overload, but not after acute myocardial infarction. Finally, using the Rosa26-Confetti reporter mouse strain, the authors propose limited expansion of Sca1-derived clones and limited pluripotential capacity, which suggests that only a small Sca-1 + cell subset differentiates into the CM lineage [16]. It is tempting to consider that Bmi1-CPC, which make up 5.4 ± 0.4 % of SCA-1 + CPC, are related to this proposed subset.
Malliaras et al. described a transient amplifying cell population, termed cardioblasts (CdB). These cells are defined by expression of SCA-1 and mature cardiac markers such as Myh6 or SαA but c-KIT − , and have multipotent properties. CdB are activated and increase in an experimental model of cardiac injury (AMI). The fact that CdB are a SCA-1 + subset, combined with the similarity to our results, strongly suggests partial overlap between these subpopulations. Further studies are needed to fully understand this question.
The classical endothelial marker Tie-1 was recently reported to define a cardiac progenitor population [19]. These cells (c-KIT low ) generate 70 % of SCA-1 + intermediate perivascular progenitors and contribute to homeostatic CM turnover. Our Bmi1-CPC RNA expression profile (unpublished results) showed moderate expression of Tie-1 and other endothelial driver constructs used by the authors for confirmation [19], as well as high CD31 levels (see Additional file 4: Figure S2A). These data suggest a functional relationship between the two populations. Specific research will be needed to explore the interplay between both populations and their role in the adult endothelium.
All of these cardiac progenitor populations, linked by SCA-1 expression, show partially overlapping similarities with Bmi1-CPC. The apparent discrepancies in cardiac progenitor identity and responses nonetheless remain to be resolved. Some of the cardiac stem-like populations identified might represent distinct developmental or physiological stages of a single resident stem cell population [51].
Our lineage tracing analysis shows that high Bmi1 expression in the adult heart labels a resident population that contributes to CM turnover, suggesting a role for Bmi1 in heart turnover. Pulsed Bmi1 + cells remained in (See figure on previous page.) Fig. 6 Evaluation of extracardiac Bmi1 + cells for de novo YFP + CM generation. a Scheme for BM (RFP + ) transplant into a lethally irradiated, TM-induced Bmi1-YFP mouse. b-e FACS analysis of the non-CM fraction of Bmi1-YFP mice after BM transplant (n = 3). b No YFP + or RFP + heart cells were detected in non-induced Bmi1-YFP mice. c FACS analysis of induced chimeric Bmi1-YFP hearts two months post-BM transplant confirmed the presence of YFP + (3.53 ± 0.87 %) and RFP + cells (45.5 ± 15 %). d Negative control for SCA-1 and CD45 staining of the non-CM fraction of Bmi1-YFP hearts (left) and percentage of SCA-1 + CD45 − cells after the staining (20 ± 3.74 %; right). e Percentage of SCA-1 + CD45 − in RFP + cells in the non-CM fraction of chimeric Bmi1-YFP hearts two months post-transplant (0.26 ± 0.06 %). (f-h) FACS and ICC of the CM compartment of chimeric Bmi1-YFP mice two months post-transplant (n = 3). f No RFP or YFP expression was detected in Bmi1-YFP CM analyzed by FACS 5d-postTM (left). RFP + CM from Act-RFP mice (right). g No RFP + CM were detected in chimeric Bmi1-YFP mice two months post-transplant. h ICC of chimeric Bmi1-YFP CM two months post-transplant. Bars, 100 μm. Data shown as mean ± SD. YFP yellow fluorescent protein, CM cardiomyocytes, FACS fluorescence-activated cell sorting adult mouse heart for over one year (currently observed to two years of age; not shown), although they appear to decline in aged mice (two-years-old), and probably become more senescent. Gene expression, combined with the non-drastic dilution of the label over this period, implies intrinsic self-maintenance with no apparent contribution of more primitive or extracardiac progenitors to de novo CM formation [52,53] and our findings here.
A number of studies have attempted to identify a cardiac non-stem cell source for adult cardiomyocyte replacement, suggesting that adult CM dedifferentiate and proliferate to give rise to de novo CM [6,7]. In our experimental model using the αMHC promoter, we confirmed that at 5d-postTM, a large proportion of CM are labeled (see Additional file 7: Figure S5B,C). This temporal window is also sufficient to ensure clearance of TM, given its estimated 6 h half-life in the mouse, which rules out long-term TM effects [54]. In contrast, in our Bmi1-YFP mouse model, we found no indication of Cre expression or activity in the CM population at 5d-postTM. Although we cannot currently exclude additional sources of heart turnover from other stem cell-like populations or a possible functional connection between putative differentiation of rare Bmi1 + CM, our results support the idea that Bmi1 + non-myocyte cells (B-CPC) contribute to homeostatic cardiomyocyte turnover in the mammalian adult heart.

Conclusions
Bmi1 + non-CM cells (B-CPC) form a small fraction of the SCA-1 population; they express high levels of stemness and cardiac lineage specification markers, contribute to the main heart lineages in vitro, and are capable of in vivo self-maintenance. B-CPC is likely to be a heterogeneous population that contains progenitor cells; further studies are needed for a better understanding of the population and its specific subpopulations. Improved characterization of the biology of these cells will help to define in vivo hallmarks as well as potential ontogenic relationships with other adult cardiac progenitor-like populations and to identify the factors needed to harness their potential for effective cardiac cell therapy.