Developmental origin of postnatal cardiomyogenic progenitor cells

Aim: To trace the cell origin of the cells involved in postnatal cardiomyogenesis. Materials & methods: Nkx2.5 enhancer-eGFP (Nkx2.5 enh-eGFP) mice were used to test the cardiomyogenic potential of Nkx2.5 enhancer-expressing cells. By analyzing Cre excision of activated Nkx2.5-eGFP+ cells from different lineage-Cre/Nkx2.5 enh-eGFP/ROSA26 reporter mice, we traced the developmental origin of Nkx2.5 enhancer-expressing cells. Results: Nkx2.5 enhancer-expressing cells could differentiate into striated cardiomyocytes both in vitro and in vivo. Nkx2.5-eGFP+ cells increased remarkably after experimental myocardial infarction (MI). The post-MI Nkx2.5-eGFP+ cells originated from the embryonic epicardial cells, not from the pre-existing cardiomyocytes, endothelial cells, cardiac neural crest cells or perinatal/postnatal epicardial cells. Conclusion: Postnatal Nkx2.5 enhancer-expressing cells are cardiomyogenic progenitor cells and originate from embryonic epicardium-derived cells.

tamoxifen labeled 80% of the pre-existing cardiomyocytes, suggesting a higher percentage of the new-forming cardiomyocytes originated from stem/progenitor cells. Hsieh et al. previously reported that stem cells refresh mammalian cardiomyocytes in mice post-injury based on indirect pulse-chase evidence [4]. Some other evidence also shows that cardiomyocytes may arise from progenitor or stem cells [9][10][11][12][13].
Cardiac progenitor cells with different surface markers, such as c-Kit + cells [9], Sca-1 + cells [10], and side population cells (ATP-binding cassette transporter expressing cells) [11], have been identified in postnatal mammalian hearts, and all of those cells are able to differentiate into cardiomyocytes in vitro. Cardiospherederived cells, which have been isolated from the adult mammalian hearts and are a heterogeneous collection of cells, also reportedly differentiate into cardiomyocytes [14,15]. Myocardial progenitors (Wt1 progenitors) in the epicardial layer of mouse hearts are reported to form cardiomyocytes within the damaged heart [12]. Chong et al. also found a population of adult cardiacresident mesenchymal stem cells with multilineage differentiation potential [13]. However, the markers used to identify those populations of cells were neither tissue-nor cardiac lineage-specific and the origin of those progenitors remains unclear.
Nkx2.5, a homeodomain-type transcription factor, is one of the earliest transcription factors expressed during embryonic cardiogenesis. Nkx2.5 is required for terminal differentiation and morphogenesis of the early developing heart [16,17]. Using Nkx2.5 enh-eGFP mice that express eGFP only in Nkx2.5 progenitor cells but not in cardiomyocytes, Wu et al. documented the ability of Nkx2.5-enh-eGFP cells in the developing hearts to undergo bipotential differentiation into cardiomyocytes and smooth muscle cells [18]. The finding that perinatal loss of Nkx2.5 leads to conduction and contraction defects indicates the importance of Nkx2.5 in the postnatal cardiac development [19].
The aim of this study was to determine if the postnatal Nkx2.5 enhancer expressing cells are cardiomyogenic progenitor cells and to trace the developmental origin of these progenitor cells.

Surgery
MI was created by permanent ligation of the left anterior descending coronary artery approximately 2 mm beneath the left atrial appendage after the mice were anesthetized via intraperitoneal injection of a combination of ketamine (100 mg/kg) and xylazine (10 mg/kg), intubated with ventilator support, and underwent left thoracotomy.

Gene expression determination by RT-qPCR
The hearts from the mice were dissected and digested with collagenase solution (collagenase A, 10 mg/ml and collagenase B, 10 mg/ml [both from Roche Diagnostics] in 10 mM HEPES (Sigma-Aldrich) buffered solution in 20% fetal calf serum) at 37°C. The external (epicardial/subepicardial) and internal (endocardial/subendocardial) parts of the heart were obtained by digesting the whole heart with collagenase for 1 h and the myocardial part was obtained from trituration and digestion of the remaining heart tissue. Cells from digested hearts were lysed with Trizol (Invitrogen, CA, USA). Total RNA was purified and stored at -80°C. cDNA was generated using a SuperScript III (Invitrogen) synthesis kit. Reverse transcription-quantitative polymerase chain reaction (RT-qPCR) was performed using Roche Light-Cycler 480II System (Roche Diagnostics, IN, USA) for 40 cycles. The primers were: GATA4, forward: 5′-TCT CAC TAT GGG CAC AGC AG-3′, reverse: 5′-CGA GCA GGA ATT TGA AGA GG-3′; Nkx2.5, forward: 5′-GCT ACA AGT GCA AGC GAC AG-3′, reverse: 5′-GGG TAG GCG TTG TAG CCA TA -3′.
Labeling of the pre-existing cardiomyocytes 4-OH tamoxifen (Sigma-Aldrich) in DMF (1 mg/10 μl) was dissolved in sunflower oil at a concentration of 10 mg/ml then injected intraperitoneally into αMHC-MerCreMer/Nkx2.5 enh-eGFP/R26R-Tomato or R26R-LacZ mice at a dose of 2 mg three-times a week for 2 weeks (i.e., a total of 12 mg) to label the pre-existing cardiomyocytes with either red fl uorescence or as β-galactosidase+.
Labeling variant stages of epicardial Wt1 cells 4-OH tamoxifen in DMF (1 mg/10 μl) was dissolved in sunflower oil at a concentration of 10 mg/ml then injected intraperitoneally into Wt1 CreERT2 /Nkx2.5 enh-eGFP/R26R-LacZ mice. The labeling of peri-MI Wt1 cells was by administering 4-OH tamoxifen at a dose of 2 mg injected intraperitoneally three-times a week for 1 week prior to MI and three-times a week for 1 week after MI (i.e., a total of 12 mg) to label the adult epicardial Wt1 cells β-galactosidase+. The labeling of perinatal Wt1 cells was achieved by administering 2 mg of 4-OH tamoxifen intraperitoneally into the dams on embryonic day (ED) 18.5. The embryonic Wt1 cells were labeled with 1 mg 4-OH tamoxifen on ED 10.5 in combination with 1 mg/kg/day progesterone on ED 10.5-11.5.

Excision PCR
The eGFP positive cells from the digested hearts of variant lineage-Cre/Nkx2.5 enh-eGFP/R26R-LacZ: ±/+/± mice were sorted and cultured in differentiation medium. Cellular DNA was isolated using the Univers All Extraction kit (Yeastern Biotech, Taipei, Taiwan) and analyzed by PCR for the presence of excision. The primers were: forward: 5′-TGG CTT ATC CAA CCC CTA GA-3′, and reverse: 5′-GTT TTC CCA GTC ACG ACG TT-3′. GAPDH was used as the control.

Statistical analysis
Numerical data are presented as mean ± standard deviation in the text and as mean ± standard error of the mean in the figures. Statistical analysis was performed using two-tailed t-test to compare the mean of two groups, and using ANOVA with Bonferroni's post hoc to make multiple comparisons. Statistical analysis was performed using SPSS 13.0 for Windows software (SPSS Inc., IL, USA). Probability values of p < 0.05 were considered statistically significant.

Myocardial injury triggers cardiogenesis gene expression
To determine if myocardial injury triggers cardiogenesis, we did coronary artery ligation and sham operation on 6-to 8-week-old C57BL/6J mice. The hearts were harvested from the mice 0, 1, 3, 5, 7, 9, 11, 14 and 21 days after MI (n = 4 to 5 in each group) ( Figure 1A). Because studies in zebrafish have shown that cardiac injury activates the epicardial cell layer and initiates cardiac regeneration at the subepicardial layer [27,28], differential gene expression analysis was performed. The external (epicardial/subepicardial) and internal (endocardial/subendo-cardial) parts of the heart were obtained by digesting the whole heart with collagenase for 1 h and the myocardial part was obtained from trituration and digestion of the remaining heart tissue. At present, there are no good and standard methods to separate epicardial layers from the myocardium. We therefore used collagenase digestion method to separate epicardium from myocardium. In theory, cells in the outer layers (e.g., epicardium/subepicardium) and near chambers (e.g., endocardium) would be separated first, followed by cells in the myocardium. Parts of cells in the myocardium might also be separated. The percentage of the epicardial cells using the 1 h whole-heart collagenase digestion protocol should be higher than that in the entire heart. At the external and internal parts of the hearts, the expression of cardiogenesis genes GATA4 and Nkx2.5 significantly increased following MI. GATA4 expression peaked on day 11 (15.00 ± 13.24 on day 11 vs 1.00 ± 0.35 pre MI; p = 0.029), and Nkx2.5 expression peaked on day 21 after MI (9.74 ± 4.73 on day 21 vs 1.00 ± 0.91 pre MI; p = 0.002) ( Figure 1B). At the myocardial part of the hearts, alternations in gene expression were vague ( Figure 1B, lower panels) (unit: expression fold over no MI).

Cardiomyogenic progenitor cells exist in the postnatal mammalian heart
The increased expression of cardiogenesis genes (i.e., GATA4, Nkx2.5) after myocardial injury hinted that cardiac lineage-specific progenitor cells existed in the postnatal heart. Nkx2.5 is one of the earliest transcription factors expressed during embryonic cardiogenesis. A 2.1 kilobase enhancer located 9.5 kilobase upstream of the translation start of murine Nkx2.5 along with a 500 base-pair Nkx2.5 base promoter was used to generate cardiac-specific Nkx2.5 enh-eGFP mice, in which eGFP is expressed specifically in Nkx2.5 cardiac progenitor cells in the developing hearts [18,29]. The mice were used to examine if cardiomyogenic progenitor cells exist in the postnatal mammalian heart (Supplementary Figure 1). Nkx2.5 enh-eGFP positive cells were isolated (Figure 2A, left panel) from the hearts of postnatal Nkx2.5 enh-eGFP mice, and cultured in differentiation medium [18] to test the cardiomyogenic potential of postnatal Nkx2.5 enh-eGFP positive cells. The isolated Nkx2.5 enh-eGFP positive cells did not express cTnI and striation in the first 2 days after sorting (Figure 2A (Figure 2B, a) and cardiac precursor marker (Nkx2.5) ( Figure 2B, b), suggesting they are cardiac progenitor cells. The Nkx2.5 enh-eGFP positive cells presented with small round cells and were primarily located at the outer layer of compact myocardium just beneath the epicardium (Supplementary Figure 2).

In vivo evidence that Nkx2.5 enhancerexpressing cells differentiate into cardiomyocytes
To confirm if Nkx2.5 enhancer-expressing cells differentiate into cardiomyocytes in vivo, we created an experimental MI by ligating the left anterior descending coronary artery of 6-8 week-old inducible Nkx2.5 enh-Cre/R26R-LacZ mice that express Cre under the control of both Nkx2.5 cardiac enhancer and the tetracycline transactivator using tet-off system [20] with R26R-LacZ as a reporter. Doxycycline was administered from conception to MI. There was no LacZ+ staining cells in the unoperated group, indicating no significant leakage of the tet-off system ( Figure 3A, a). Six weeks following MI, the descendent cells of post-MI Nkx2.5 enhancerexpressing cells were identified by LacZ staining (Figure 3A, b-d). These galactosidase+ cells differentiated into striated cardiomyocytes (labeled by cTnT) (Figure 3B), accounting for 3% of all cardiomyocytes.  Although it has been proposed that a stem cell pool might contribute to postnatal cardiomyocyte renewal in mammals [4], the origin of such cells remains uncertain.
An experimental MI was then created on different lineage-specific Cre±/Nkx2.5 enh-eGFP/R26R-LacZ± mice, the hearts were harvested, digested and sorted for the Nkx2.5 enh-eGFP positive cells 1 week post-MI. Cellular DNA of eGFP positive cells and unsorted cells were isolated. PCR was performed to assess Cremediated excision to trace the origin of activated postnatal Nkx2.5 enh-eGFP positive cells ( Figure 5A). The negative excision bands of the sorted Nkx2.5 enh-eGFP positive cells from the Tie2-Cre and Pax3-Cre series confirmed that the activated Nkx2.5 enh-eGFP positive cells following MI do not originate from endothelial cells and c ardiac neural crest cells ( Figure 5B & C).
To further confirm that the activated Nkx2.5 enh-eGFP positive cells do not arise from the pre-existing cardiomyocytes, the pre-existing cardiomyocytes were labeled with β-galactosidase+ with 4-OH tamoxifen intraperitoneal injection into the αMHC-MerCreMer/Nkx2.5 enh-eGFP/R26R-LacZ mice at a dose of 2 mg threetimes a week for 2 weeks (i.e., a total of 12 mg) prior to MI. That study confirmed that the activated Nkx2.5 enh-eGFP positive cells did not originate from the pre-existing mature cardiomyocytes because the isolated Nkx2.5 enh-eGFP positive cells did not show excision (n = 7) ( Figure 5D). The result suggested that the cardiomyocytes do not undergo Nkx2.5 fetal gene re-expression, nor do they d edifferentiate into Nkx2. These assays revealed that the postnatal cardiac progenitor cells arose from epicardial cells, not from endothelial cells, cardiac neural crest cells, or pre-existing cardiomyocytes.
Embryonic epicardium: developmental origin of postnatal cardiac progenitor cells Gata5-Cre lineage analysis suggested an epicardial origin of postnatal Nkx2.5 cardiac-lineage progenitor cells; however, the origin could be adult, perinatal or embry- The peri-MI Wt1 epicardial cells were labeled with 4-OH tamoxifen intraperitoneal injection at a dose of 2 mg three-times a week for 1 week prior to MI and three-times a week for 1 week after MI (i.e., a total of 12 mg). The PCR results showed that the activated Nkx2.5 enh-eGFP positive cells did not derive from the peri-MI Wt1 epicardial cells as the sorted GFP+ cells showed negative results ( Figure 6B). The unsorted cells showed negative PCR result ( Figure 6B). LacZ staining of the heart sections from the Wt1 CreERT2 /R26R-LacZ mice, which had been labeled peri-MI Wt1 cells following 4-OH tamoxifen injection before and after MI, also revealed no β-galactosidase positive cells ( Figure 7A). Together, the PCR and LacZ staining results confirmed that no or very few adult Wt1 epicardial cells existed even following MI.
The perinatal Wt1 epicardial cells were labeled with 4-OH tamoxifen injection on ED 18.5 ( Figure 6C). The negative PCR results revealed that the activated Nkx2.5 progenitor cells did not come from perinatal Wt1 epicardial cells ( Figure 6C). The embryonic Wt1 cells were subsequently labeled with 4-OH tamoxifen on ED 10.5, and the positive excision PCR confirmed the embryonic epicardial origin of postnatal Nkx2.5 cardiac progenitor cells ( Figure 6D). LacZ staining of the heart sections from Wt1 CreERT2 /R26R-LacZ mice, in which the embryonic Wt1 cells were labeled with 4-OH tamoxifen on ED 10.5, showed that embryonic epicardium-derived cells were present in both the epicardium and myocardium of the adult heart ( Figure 7B).

Discussion
In recent years, evidence supporting the existence of cardiomyocyte renewal in postnatal mammalian hearts has been mounting [3,5,6]. However, the origin of the cells contributing to postnatal cardiomyogenesis remained uncertain. We confirmed that Nkx2.5 enhancer-expressing cells are present in postnatal hearts and expand remarkably following myocardial injury. Using inducible Nkx2.5 enh-Cre/R26R-LacZ mice [20] to lineage trace postnatal Nkx2.5 enhancer-expressing cells, we confirmed that Nkx2.5 enhancer-expressing cells contribute directly to postnatal cardiomyogenesis after myo-cardial injury. These studies determined the cardiomyogenic differentiation potential of those Nkx2.5 enhancer-expressing cells both in vitro and in vivo. Lineage tracing studies further confirmed those cardiomyogenic progenitor cells were indeed embryonic epicardium-derived cells.
In the current study, the number of Nkx2.5 enhancer-expressing cells was low in the postnatal heart, but was activated following cardiac injury. The reactivation of the Nkx2.5 progenitor cells suggested that they might help to repair or regenerate the injured myocardium.
Postnatal cardiomyocytes reportedly dedifferentiate following cardiac injury and re-express markers of embryonic cardiomyocytes [30][31][32]. Although the increased expression of the cardiogenesis genes GATA4 and Nkx2.5 after MI could be due to the  result of either dedifferentiation or fetal gene reexpression of the cardiomyocytes, we believe that is unlikely to be the case because the pre-existing cardiomyocytes did not undergo Nkx2.5 re-expression after MI ( Figure 5D).
Embryonic epicardial progenitor cells, marked by Tbx18 or Wt1, reportedly contribute to embryonic cardiomyogenesis [25,33]. A recent study indicated that Wt1+ cells, via priming by thymosin-β4, might transdifferentiate into cardiomyocytes following cardiac injury in a mouse model [12]. Those authors found that the adult heart can respond to injury with modest increases in Wt1 progenitors but without initiating a cardiogenic program [12]. The data generated in this study showed that postnatal Nkx2.5 progenitor cells arise from embryonic Wt1 epicardial cells, but not from adult Wt1 epicardial cells. The adult Wt1 epicardial progenitor cells described by Smart et al. seemed different from the postnatal Nkx2.5 progenitor cells. Further, Chong et al. described a population of adult cardiac-resident mesenchymal stem cell-like stem cells (cardiac colony-forming-unit fibroblasts; cCFU-Fs) with the expression of PDGF-α exhibited multipotency including cardiomyocytes, endothelial cells, smooth muscle cells, bone, cartilage, adipose tissue, etc. [13]. Those authors also confirmed the epicardial origin of the cCFU-Fs. The cCFU-Fs also seemed different from Nkx2.5 cardiac progenitor cells because they did not express Nkx2.5 [13]. Further research will be required to determine whether cCFU-Fs and postnatal Nkx2.5 progenitor cells represent hierarchically related progenitors.
Using GATA5-Cre and Wt1 CreERT2 lineage mice, the lineage tracing performed in this study confirmed the epicardial origin of postnatal Nkx2.5 progenitor cells. However, GATA5-Cre recombinase activity is not only expressed in epicardium. Instead, GATA5-Cre recombinase is also expressed in the cavities and their linings, the hepatobiliary system, mesenchyme, renal and urinary system (based on information on the Mouse Genome Informatics Website). Tamoxifen-inducible Wt1 CreERT2 mouse line was reported inefficiently to recombine the epicardium and its cellular derivatives [34]. Our initial tests revealed that injection of high-dose 4-OH tamoxifen (e.g., 1.5 mg or more) led to embryonic death. We thus labeled the embryonic Wt1 cells with low dose (e.g., 1 mg) 4-OH tamoxifen on ED 10.5, which might resulted in low recombination efficiency obtained with the tamoxifen-inducible Wt1 CreERT2 line in epicardial cells. Even so Wt1 CreERT2 recombinase activity is still detected in the postnatal Nkx2.5 progenitor cells, strongly suggests the embryonic epicardial origin of postnatal Nkx2.5 progenitor cells.
The discovery of cardiogenesis gene expression in the external and internal part of the heart ( Figure 1B) and additional lineage tracing using GATA5-Cre and Wt1 CreERT2 line (Figure 7) strongly suggest an epicardial origin for postnatal Nkx2.5 cardiac p rogenitor cells.
Our study identifies Nkx2.5 enhancer-expressing cells as a source for postnatal cardiomyogenesis. However, the results cannot exclude the possibility that postnatal cardiac regeneration occurs through c ardiomyocyte proliferation.

Conclusion
The presented study demonstrated that Nkx2.5 cardiomyogenic progenitor cells existed in the postnatal mammalian heart and originated from the embryonic epicardium.

Future perspective
The major challenge in cardiovascular medicine is the inability to replace the large number of cardiomyocytes lost after cardiac injury. This study demonstrates the cell type and the origin of the cells involved in postnatal cardiomyogenesis. These results will facilitate cell therapy for cardiac regeneration, the pharmacological targeting of the regenerating cells, enhancing endogenous cardiac regeneration and further understanding the mechanisms of cardiovascular diseases.