Bmi‐1 high‐expressing cells enrich cardiac stem/progenitor cells and respond to heart injury

Abstract Bmi‐1 gene is well recognized as an oncogene, but has been recently demonstrated to play a role in the self‐renewal of tissue‐specific stem cells. By using Bmi‐1GFP /+ mice, we investigated the role of Bmi‐1 in cardiac stem/progenitor cells and myocardial repair. RT‐PCR and flow cytometry analysis indicated that the expression of Bmi‐1 was significantly higher in cardiac side population than the main population from CD45−Ter119− CD31− heart cells. More Sca‐1+ cardiac stem/progenitor cells were found in Bmi‐1 GFP hi subpopulation, and these Bmi‐1 GFP hi heart cells showed the potential of differentiation into SMM + smooth muscle‐like cells and TnT+ cardiomyocyte‐like cells in vitro. The silencing of Bmi‐1 significantly inhibited the proliferation and differentiation of heart cells. Otherwise, myocardial infarction induced a significantly increase (2.7‐folds) of Bmi‐1 GFP hi population, mainly within the infarction and border zones. These preliminary data suggest that Bmi‐1hi heart cells are enriched in cardiac stem/progenitor cells and may play a role in myocardial repair.


| INTRODUCTION
Polycomb complex protein Bmi-1 is encoded by the BMI1 gene. A number of previous studies have demonstrated the roles of Bmi-1 in the development and progression of various types of malignant tumours, 1 such as leukaemia, 2,3 colorectal cancer, 4 and medulloblastomas. 5 These studies have found that down-regulation of Bmi-1 in cancer stem cells suppresses tumour growth. 3,6,7 Beyond its role as an oncogene, up-regulation of Bmi-1 in various tissuespecific stem cells, [8][9][10] such as hematopoietic stem cells (HSC), 2,3,8 intestinal stem cells, 11 and epithelial stem cells in the pancreatic, prostate, lung, and others, [12][13][14][15][16][17][18] has been demonstrated to play essential roles in the self-renewal and the maintenance of stemness. Reduced expression of Bmi-1 has also recently been found to enhance the beating of cardiomyocytes (CM) induced from neonatal and adult mouse fibroblasts by directly reprogramming. 19 However, little has been known about Bmi-1 expression in cardiac stem/progenitor cells.
Actually, the identity, origin and physiological role of endogenous cardiac stem/progenitor cells in adult mammals are still debated. For a long time, adult mammalian heart was thought to be a terminally differentiated organ. However, considerable evidence has shown the low turnover rate of CM. 20,21 There are at least two possible resources for the new born CM: preexisting CM 22,23 or cardiac stem/ progenitor cells. [24][25][26][27] By now, different markers and methods have been applied for the identification and expansion of resident cardiac stem/progenitor cells, such as the c-kit-positive cells, 26 Sca-1-positive cells, 27 cardiac side population (SP), 24 and cardiosphere derived cells. 28 Using an inducible genetic labelling approach, we have recently defined cardioblasts, the small non-myocyte cells express cardiac transcription factors and sarcomeric proteins and form mature CM in vivo after transplantation. 25 Endogenous cardioblasts are rarely evident in the normal adult mouse heart, but will be significantly activated after myocardial infarction. The cardioblasts do not arise from haematogenous seeding, CM dedifferentiation, or mere expansion of a preformed progenitor pool. 25 In this study, we investigated the potential role of Bmi-1 on cardiac stem/progenitor cells by using Bmi-1-GFP-knock-in mice, in which GFP was expressed under the endogenous transcriptional regulatory elements of the Bmi-1 gene, and the levels of Bmi-1 expression in cells could be quantified by GFP fluorescence. 3 We found that the subpopulations of cells with high expression of Bmi-1 in heart tissue enriched in SP and Sca-1-positive cardiac stem/progenitor cells, and showed a significantly increase in number in response to myocardial infarction.

| Animals and genotyping
The procedures for all animal experiments were approved by the Animal Care and Use Committee of the Shanghai Ruijin Hospital, Shanghai Jiaotong University School of Medicine, China and the Cedars-Sinai Medical Center, Los Angeles, CA, USA. All methods were performed in accordance with the relevant guidelines and regulations. Bmi-1 GFP/+ mice from JAX Lab, originally generated by Dr.
Weissman group in Stanford University were inbred in the animal centre of Shanghai Ruijin Hospital, Shanghai, China. Eight-to 12-week-old mice were used for experiments. Mice genotyping was verified by PCR of tail genomic DNA. 3

| Evaluation of SP cells in heart cells and bone marrow cells
Heart SP and main population (MP) were prepared as previously described with modification. 24 Briefly, heart tissue of Bmi-1 GFP/+ mice was minced into about 1 mm 3 pieces and digested with 0. Bone marrow cells were obtained from the same Bmi-1 GFP/+ mice as previously described. 29 Single cell suspensions were incubated with

| Myocardial infarction model
To further investigate how the Bmi-1 expression will be changed in response to heart injury, myocardial infarction was made in Bmi-1 GFP/+ mice (8-12 weeks old) by permanent ligation of the left anterior descending coronary artery (LAD). Mice were randomly allocated into either LAD ligation or sham operation group by random number table, 9 mice/group in total. Mice were performed with tracheal intubation, tidal volume 0.7 mL, respiratory rate 120 breaths per minute. A left thoracotomy was performed through the fourth intercostal space.
After removing the pericardium, LAD was ligated with 7-0 silk suture under direct vision of surgical microscopy. Mice received a left thoracotomy alone were used for control. All mice survived after the successful surgical procedures, and no mouse died during the follow-up period.
Six mice from each group were euthanized 1 week after surgery, and the heart for evaluating the subpopulation of cells with high expression of Bmi-1 by flow cytometry as described above. Another three mice from each group were euthanized 2 weeks after surgery, and excised hearts were snap-frozen for histological analysis, and cryosections (10 μm) were stained following standard procedures. Images of histological staining were taken using Olympus microscope. All analyses were conducted by individuals blind to treatment allocation.

| Statistical analysis
Data and statistical analysis were done by GraphPad Prism 6.0.
Results are presented as mean ± SEM unless specified otherwise.
Comparisons between any two groups were performed with twotailed unpaired Student's t-test. Each experiment was performed at least three times and the differences were considered statistically significant when P < 0.05. Representative flow cytometry plots (E) and quantitative data (F) showed more Sca-1 + cells in Bmi-1 GFP hi than GFP int populations of heart cells. Sca-1 + cells were gated from CD45 − Ter119 − CD31 − heart cells. *P < 0.05 high (hi), intermediate (int), and negative (−) subpopulations (Figure 1). As congruent with previous report, 3 HSC (Lin − c-kit + Sca-1 + HSCs) were dramatically enriched in the GFP hi cells when compared to the GFP int and GFP − cells ( Figure 1F-H). Similarly, higher expression of GFP was also observed in the Lin − c-kit + Sca-1 + HSCs than the Lin − c-kit + progenitors or Lin + matured cells from bone marrow of Bmi-1 GFP/+ mice ( Figure S1).
We next examined whether Bmi-1 will be also highly expressed in cardiac stem/progenitor cells. Because almost all GFP + cells were Bmi-1 + and 96.69% ± 2.61% of Bmi-1 + heart cells expressed GFP through immunofluorescent images (Figure 2A), we used GFP + cells to represent Bmi-1 + cells in flow cytometry assay. After negative deletion of matured hematopoietic cells and endothelial cells by using antibodies against CD45, Ter119, and CD31, we divided these CD45 − Ter119 − CD31 − heart cells into cardiac SP and MP as the previously stated 24 ( Figure 2B). We found that the Bmi-1 expression was significantly higher in SP cells than in MP cells ( Figure 2C). We also tried to gate CD45 − Ter119 − CD31 − SP cells by flow cytometry, and then measured the Bmi-1 expression by fluorescence intensity.
Sca-1 was known as one of the common marker for stem/progenitor cells. Our results also showed that Sca-1 + cardiac stem/progenitor cells were enriched in Bmi-1 GFP hi population from CD45 − Ter119 − CD31 − heart cells ( Figure 2E). The percentage of Sca-1 + cells gated from Bmi-1 GFP hi heart cells was almost 2.5-folds higher than that of from Bmi-1 GFP int population (19.98 ± 4.88% vs. 7.82 ± 2.77%, P < 0.05; Figure 2F). So, our data from BM and heart cells indicated that Bmi-1 GFP hi subpopulation enriched not only in HSC but also in cardiac stem/progenitor cells.

| Bmi-1 GFP hi heart cells could differentiate into SMM + smooth muscle-like cells and TnT + CM-like cells in vitro
We also purified CD45 − Ter119 − D31 − Bmi-1 GFP hi and GFP − subpopulations from Bmi-1 GFP/+ mice heart, and then evaluated their potency of myocardial differentiation in vitro (Figure 3). These Bmi-1 GFP hi cells grew well after 7 and 14 days of culture, but cells were rarely grown from the Bmi-1 GFP − cells ( Figure 3A). Immunostaining showed that some cells grown from Bmi-1 GFP hi cells were positively expressed with SMM heavy chain (12.96% ± 2.70%) and cardiac Troponin T (TnT) (26.03% ± 3.58%), suggesting the differentiation into smooth muscle-like cells and CM-like cells (Figure 3B,C).

| Bmi-1 knockdown inhibited the proliferation and myocardial differentiation of non-CM cells
To confirm the role of Bmi-1 on the proliferation and differentiation, we tried to silence the expression of Bmi-1 in non-CM cells by siRNA (Figure 4). The efficiency of knockdown was showed in

| Bmi-1 GFP hi population increased in response to myocardial infarction
To further evaluate the role of Bmi-1 expression in myocardial repair, we analysed the change of Bmi-1 GFP hi heart cells in Bmi-1 GFP/+ mice after LAD ligation. Interestingly, CD45 − Ter119 − CD31 − Bmi-1 GFP hi population was significantly increased in the infarcted heart when compared to the control heart received sham operation (4.04 ± 1.55% vs 1.47 ± 0.12%, P < 0.05). However, myocardial infarction did not induce significant changes in the GFP int and GFP − populations (Figure 5A,B). Immunofluorescence staining also clearly showed some clusters of Bmi-1 hi cells within the infarction and border zones (Figure 5C). Moreover, many CM were observed around the clusters of Bmi-1 hi cells. Although the absence of direct evidence, these results indirectly suggested that the heart cells expressed with high level of Bmi-1 might be involved in myocardial repair after injury.

| DISCUSSION
As congruent with a previous report, 3 our data indicated the percentage of lin − c-kit + Sca-1 + HSC were much higher in Bmi-1 GFP hi population compared to GFP int and GFP − cells in Bmi-1 GFP/+ mice ( Figure 1). On the other hand, Bmi-1 GFP was highly expressed in lin − c-kit + Sca-1 + HSC when compared with Lin − c-kit + progenitors or Lin + matured cells from bone marrow ( Figure S1). Actually, the high level of Bmi-1 expression has been demonstrated to be critical on the self-renewal of HSC and the maintenance of hematopoietic function. 8,31 Although cardiac-specific Bmi-1 deletion during embryogenesis does not affect cardiogenesis, 32 it has been reported that quercetin could minimize doxorubicin-induced cardiotoxicity by modulating Bmi-1 expression. 33 There is very limited information on the role of Bmi-1 expression in cardiac stem/progenitor cells. By using Bmi-1-GFP-knock-in mice, 3 34 and myocardial repair following acute injury. 35 Furthermore, Bmi-1 expression is associated with reactive oxygen species levels. 36 But we made a more detailed distinction. In this study, we The proportion of cTnT + cardiomyocytelike cells was at 5 days after TSA treatments. NC: negative control. *P < 0.05; **P < 0.005; ***P < 0.001 identification of cardiac stem/progenitor cells. Second, it is impossible for us to exactly quantify the Bmi-1 hi cells in the whole heart because we collected these small size heart cells for analysis by removing the CM. Third, we do not know whether the subpopulations of small size heart cells with the expression of Bmi-1 at intermediate and low levels are actually generated from these Bmi-1 hi cells followed by a gradually reduction of Bmi-1 expression during the differentiation process. 3 Finally, it is hard for us to collect enough number of Bmi-1 hi heart cells for in vivo implantation into a damaged heart. The role of Bmi-1 is also required to investigate in vivo by using Bmi-1 knockout animals. So, the role of Bmi-1 hi heart cells for functional myocardial repair is still questionable.
In summary, Bmi-1 was expressed higher in cardiac SP than MP from CD45 − Ter119 − CD31 − heart cells. More Sca-1 + cells were found in Bmi-1 GFP hi population, and more Bmi-1 GFP hi cells in Sca-1 + population. The CD45 − Ter119 − CD31 − Bmi-1 GFP hi cells from Bmi-1 GFP/ + mice could differentiate into SMM + smooth muscle-like cells and TnT + CM-like cells in vitro. The silencing of Bmi-1 significantly inhibited the proliferation and differentiation of heart cells. Bmi-1 GFP hi population increased in heart of mice 1 week after infarction and some clusters of Bmi-1 hi cells were observed within the infarction and border zones. Based on our data, heart cells with high Bmi-1 expression seem to be enriched in cardiac stem/progenitor cells and possibly play a role during myocardial repair.

ACKNOWLEDG EMENTS
We thank Dr. Eduardo Marbán of the Cedars-Sinai Heart Institute for generously providing access to laboratory resources required to obtain the data in Figure 5. We thank Dr. Linheng Li in Stower Institute for his project direction and data sharing, and core facility in Stower Institute for flow cytometry analysis. We thank Dr Naoki Hosen and Dr Tong Yin http://orcid.org/0000-0003-3037-7629