BMP and Notch Signaling Pathways differentially regulate Cardiomyocyte Proliferation during Ventricle Regeneration

Adult mammalian hearts show limited capacity to proliferate after injury, while zebrafish are capable to completely regenerate injured hearts through the proliferation of spared cardiomyocytes. BMP and Notch signaling pathways have been implicated in cardiomyocyte proliferation during zebrafish heart regeneration. However, the molecular mechanism underneath this process as well as the interaction between these two pathways remains to be further explored. In this study we showed BMP signaling was activated after ventricle ablation and acted epistatic downstream of Notch signaling. Inhibition of both signaling pathways differentially influenced ventricle regeneration and cardiomyocyte proliferation, as revealed by time-lapse analysis using a cardiomyocyte-specific FUCCI (fluorescent ubiquitylation-based cell cycle indicator) system. Further experiments revealed that inhibition of BMP and Notch signaling led to cell-cycle arrest at different phases. Overall, our results shed light on the interaction between BMP and Notch signaling pathways and their functions in cardiomyocyte proliferation during cardiac regeneration.


Introduction
Myocardial infarction (MI) is a leading cause of death worldwide [1,2]. In adult mammals, the heart, as one of the least regenerative organs, replaces the infarcted myocardium with noncontractile scar tissue [3]. On the contrast, zebrafish are capable to completely regenerate the injured hearts [4][5][6]. New cardiomyocytes have been shown to arise from proliferation of pre-existing cardiomyocytes through genetic fate-mapping experiments [7,8]. Though natural cardiomyocyte turnover rate is low in adult humans [9,10], accelerating such process after acute insult is therapeutically valuable. Thus, utilizing cellular and genetic factors that stimulate cardiomyocyte proliferation may provide viable solutions for promoting human cardiac regeneration.
Zebrafish cardiac regeneration involves multiple signaling pathways [11][12][13]. BMP signaling is vital for vertebrate cardiovascular development [14,15], and its function in cardiac injury and repair has become more appreciated. Since inhibitory effects of BMP2 on fibroblast function have been reported [16], exogenous BMP2 reduced the size of infarcted tissue by diminishing apoptotic cardiomyocytes in the border zone in a mouse permanent coronary occlusion model [17]. While BMP4 enhances apoptosis and hypertrophy of cultured mammalian cardiomyocytes [18], BMP7 signaling attenuates myocardial fibrosis by inhibiting TGF-β responses in a rat MI model [19]. BMP10 has been demonstrated to enhance inflammation activity after MI and Ivyspring International Publisher exogenous BMP10 results in cardiomyocyte cell-cycle re-entry [20]. Thus, BMP signaling has been suggested as a good candidate pathway for modulating cardiac regeneration [21], but its role in cardiomyocyte proliferation requires further investigation.
Notch signaling pathway also plays important roles in cardiac regeneration [22,23]. Notch signaling is activated in the endocardium of injured zebrafish hearts and its inhibition impedes cardiomyocyte proliferation and heart regeneration [6,[24][25][26]. Endocardial Notch signaling has been shown to promote myocardial proliferative events through inducing BMP10 expression in adjacent cardiomyocytes [27][28][29]. However, the interaction between these two pathways and their differential regulation of cardiomyocyte proliferation remains elusive. In this study we utilized the zebrafish larval ventricle ablation model and a cardiomyocyte-specific FUCCI (fluorescent ubiquitylation-based cell cycle indicator) system to tackle these questions. We demonstrated the epistatic relationship between BMP and Notch signaling pathways during ventricle regeneration and further revealed their differential functions in cardiomyocyte proliferation, blocking of these pathways lead to cell-cycle arrest at different phases. Overall, our results shed light on molecular mechanism of BMP and Notch signaling pathways in cardiomyocyte proliferation during cardiac regeneration, which could lay a foundation for future development of therapeutic interventions.

BMP signaling is activated during larval ventricle regeneration
Recent research suggested that BMP signaling activation was an injury response in adult zebrafish hearts [21]. To investigate the role of BMP signaling in the larval ventricle regeneration model, we treated the ventricle-specific genetic ablation line Tg(vmhc:mCherry-NTR) with metronidazole (MTZ) at 3 days post fertilization (dpf) [6]. Whole-mount in situ hybridization (WISH) staining revealed that gene expressions of several BMP signaling components were dramatically increased in ablated hearts compared to that in control hearts at 5 dpf/ 2 days post MTZ-treatment (dpt) (Fig. 1A-D'). BMP ligand bmp10 was weakly expressed in the ventricle of control hearts, probably due to the trabeculae formation at this stage. After ablation, bmp10 expression was strongly induced, especially in the atrium (Fig. 1A, A'). Expression of BMP receptor bmpr1aa and signal transducer smad1 was also up-regulated in ablated hearts, while expression of BMP signaling effector id2b showed a much dramatic increment after ablation (Fig. 1B-D'). To visualize the transient activation of BMP signaling in regenerating hearts, we used a reporter line Tg(Bre:dGFP) which expressed destabilized GFP in BMP-activated cells [30,31]. Control Tg(vmhc:mCherry-NTR; Bre:dGFP) larvae displayed weak GFP signal in the ventricle at 4 dpf, besides other extracardiac signals. After ventricle injury, BMP signaling was activated in the ablated heart, mainly in the atrium, at 1 dpt. The signal intensity was gradually enhanced in the atrium and extended to the ventricle at 2 dpt ( Fig. 1E-G). The activation of BMP signaling during ventricle regeneration was also confirmed by immunofluorescence staining of another marker, phosphorylation of Smad1/5/9 (pSmad1/5/9), which showed a similar pattern as Bre:dGFP ( Fig. 1H-J). To further validate these results, we also performed immunostaining of pSmad1/5/9 in Tg(vmhc:mCherry-NTR; Bre:dGFP) larvae. The results suggested that BMP signaling was strongly activated in myocardium during ventricle regeneration, while a weaker epicardial BMP signal could be detected occasionally ( Fig. 1K-N'). Overall, our results confirmed that BMP signaling was activated during larval ventricle regeneration.

BMP signaling acts downstream of Notch signaling during ventricle regeneration
Previous study showed post-injury upregulation of bmp10 expression was blocked after Notch signaling inhibition [29], which suggested that BMP signaling was regulated by Notch signaling during regeneration. Whether BMP signaling regulated Notch signaling in a reciprocal manner remained unclear. We used pharmacological approach to study the epistatic relationship between these two signaling pathways. Inhibition of Notch signaling by DAPT treatment significantly reduced bmp10 expression in ablated hearts at 2 dpt as revealed by WISH staining ( Fig. 2A, B). BMP signaling activation was also blocked as revealed by diminished Bre:dGFP signals in DAPT-treated ablated hearts (Fig. 2D, E). Treatment of BMP signaling inhibitor Dorsomorphin (DM) blocked downstream transduction of BMP pathway, but had no effect on ligand expression (Fig.  2C, F). On the other hand, WISH analysis of notch1b expression in ablated hearts at 1 dpt showed no difference between DM-treated group and control group, while notch1b upregulation was attenuated after DAPT treatment ( Fig. 2G-I). We observed a similar pattern in the Notch signaling reporter line Tg(tp1:dGFP) that inhibition of BMP signaling by DM treatment did not affect the activation of Notch signaling in the endocardium around atrioventricular canal ( Fig. 2J-L, asterisk). Thus, our results suggested that BMP signaling acted epistatic downstream of Notch signaling during ventricle regeneration.
We then examined the effect of Notch or BMP signaling blockage on re-activation of early cardiogenic transcription factors. Expression of nkx2.5, gata4 and tbx20 was dramatically increased in ablated hearts compared to control hearts at 2 dpt as revealed by WISH experiments (Fig. 3C, D, I, J, O, P). After DAPT treatment, cardiogenic factor re-activation was blocked in both chambers (Fig. 3E, F, K, L, Q, R) similar as previously reported [6,29]. However, DM treatment showed heterogeneous effect on cardiogenic factor re-activation in the two chambers, with an apparent stronger reduction in the ventricle (Fig. 3G, H, M, N, S, T). These results suggested BMP and Notch signaling pathways might have differential functions in cardiogenic factor re-activation.

BMP and Notch signaling pathways differentially influence cardiomyocyte cell-cycle progression
Previous studies reported that both BMP and Notch signaling regulated cardiomyocyte proliferation [29]. To resolve the discrepancy in the zGem+ signals observed above, we examined other proliferation markers, such as immunostaining of phospho-histone H3 (referred as pH3) and EdU pulsed labelling, in Tg(vmhc:mCherry-NTR; myl7:mAG-zGeminin) fish. The overlapped signals of pH3 and EdU with zGem+ could help to eliminate the interference caused by proliferative cells in adjacent tissues, such as endocardium, epicardium and blood cells. Proliferating cardiomyocytes were in different cell-cycle stages (Fig. 6A-F''), an EdU+/zGem+ cell represented a cardiomyocyte at S phase (white arrowheads), a pH3+/zGem+ cell represented a cardiomyocyte at M phase (orange arrowheads), while a zGem+ only cell without pH3 or EdU signals was considered in G2 phase. We then assessed the marker distribution in fish treated with DAPT or DM at 48 hpt (Fig. 6G-L) and quantified the numbers of proliferating cardiomyocytes in different cell-cycle phases (Fig. 6M-O). In terms of BMP signaling inhibition, DM treatment reduced the numbers of EdU+/zGem+ cells in control hearts (2.8 ± 1.5 vs. 0.8 ± 0.8, N=12, 14, respectively) and ablated hearts (3.8 ± 2.6 vs. 1.8 ± 1.7, N=16, 13, respectively). Similarly, the numbers of pH3+/zGem+ cells decreased in control hearts (1.2 ± 1.1 vs. 0.4 ± 0.6, N=13, 14, respectively) and ablated hearts (2.2 ± 1.5 vs. 0.8 ± 0.8, N=17, 17, respectively) upon DM treatment. However, the numbers of zGem+ only cells remained unchanged between groups with or without DM treatment. On the contrast, Notch signaling inhibition with DAPT treatment significantly reduced the number of zGem+ only cells in the ablated heart (29.5 ± 7.5 vs. 15.5 ± 8.6, N=18, 17, respectively), reflecting a significant reduction in the number of G2 phase cardiomyocytes. DAPT treatment also reduced the numbers of EdU+/zGem+ cells in control hearts (2.8 ± 1.5 vs. 0.7 ± 0.8, N=12, 15, respectively) and ablated hearts (3.8 ± 2.6 vs. 0.7 ± 1.7, N=16, 16, respectively). The reduction in the numbers of pH3+/zGem+ cells upon DAPT treatment was not statistically significant. We also plotted the data based on the proportions of marked cardiomyocytes in different cell-cycle phases (Fig. 6P). Reduced percentages of cardiomyocytes in S and M phases and increased percentage of G2 phase cardiomyocytes implied that DM treatment may cause cell-cycle arrest at G2 phase. Blocking Notch signaling pathway may possibly led to G0/G1 phase arrest because of the universally dwindled number of proliferative cardiomyocytes. Taken together, our results suggested that BMP and Notch signaling pathways regulated myocardium proliferation through differentially influence cardiomyocyte cell-cycle progression.

Discussion
In this study we affirmed the upregulated expression of multiple BMP signaling components by WISH and the activation of BMP signaling in the myocardium and epicardium by pSmad1/5/9 immunostaining and Bre:dGFP reporter signals during larval ventricle regeneration. In zebrafish larvae, bmp10 has been reported to express in the endocardium, while bmp10l is expressed in the myocardium [38,39]. The expression of bmp6 and id2b is endocardial specific during heart regeneration [40]. Wu et al. reported that bmp2b and bmp7 was expressed in the endocardium and epicardium of cryo-injured hearts, while pSmad1/5/9 signals were up-regulated in the myocardium [21]. Thus, we speculate that BMP signaling activation in myocardium may be promoted by enhanced ligand secretion from the endocardium after ventricle ablation, probably through interaction with other endocardial signaling pathways [26,29,41].
It is suggested that multiple signaling pathways eventually converge together and control cardiomyocyte cell cycle progression through certain function axes [42], implying the importance to understand the complex crosstalk among different signaling pathways. Cardiomyocyte proliferation in injured hearts with defective endocardial Notch signaling can be partially restored by WNT antagonist treatment, demonstrating the crosstalk between Notch and WNT/β-catenin signaling [43]. Moreover, endocardial Notch signaling promotes cardiomyocyte reprogramming and cardiac regeneration through activating myocardial Erbb2 and BMP signaling [29]. Our results proved that inhibition of Notch signaling by DAPT abolished the upregulated expression of ligand bmp10 and blocked BMP signaling activation. BMP pathway blockage by DM treatment did not reciprocally affect expression of notch1b nor signals of Notch signaling reporter, suggesting BMP signaling acts epistatic downstream of Notch signaling. Whether this regulation is cell autonomous or not requires further investigation. Our results showed a broader expression of bmp10 by WISH than the more confined Notch signaling revealed by tp1:dGFP fluorescence, suggesting at least certain cell non-autonomous mechanism is involved in this process. Interestingly, we also observed differential influence of Notch and BMP signaling on the re-activation of cardiogenic factors during regeneration. Bmp signaling exerts opposite effects on chamber differentiation during early heart development [44,45], so one possible explanation is that BMP pathway regulates this process in a chamber-specific way by which the ventricle, instead of the atrium, is more susceptible. Another possibility is that the BMP pathway may not be directly involved in the regulation of cardiogenic factor re-activation, but rather that its inhibition on cardiomyocyte proliferation leads to a decrease in the number of ventricular cardiomyocytes as well as the expression level of cardiogenic factors. This is consistent with a previous report that BMP10 is responsible for cardiomyocyte proliferation while NRG1 regulates cardiomyocyte differentiation downstream of Notch signaling during mice ventricle trabeculation [27].
Many studies on the regulation of cell proliferation focus on the change in cell cycle progression [42,46,47]. Blocking Notch signaling pathway has been reported in several models to cause G0/G1 phase arrest by influencing p27 Kip1 and other elements, thereby affecting proliferation [48][49][50]. On the contrary, few studies have been conducted on the mechanism of BMP pathway in regulating proliferative events. Overexpression of Id1 increases Cdk4 levels and reduces p21 Clip1 , thus promoting cell cycle progression in mouse cardiomyocytes [51]. BMP signaling responses to laminar shear stress by activation of p21 which leads to G2/M phase arrest of human bladder transitional carcinoma cells [52]. In our larval ventricle regeneration model, inhibition of BMP signaling reduced the numbers of EdU+/zGem+ and pH3+/zGem+ cardiomyocytes but had no effect on the number of zGem+ only cardiomyocytes, implying cell-cycle arrest at G2 phase. Blocking Notch signaling pathway may possibly led to G0/G1 phase arrest because of the universally dwindled number of proliferative cardiomyocytes. Our study provides novel insights into and potential direction for the understanding of cardiomyocyte proliferation in zebrafish ventricle regeneration process. The molecular mechanisms how these signaling pathways regulate proliferation, and whether it is conserved in different species and under different situations warrants further investigation.

EdU labeling
The larvae were incubated with 500 μM EdU for 1 h in E3 water with 2% DMSO to facilitate EdU solubilization. After pulse labeling, larvae were rinsed with E3 water, anaesthetized with 0.2% tricaine and fixed in 4% PFA. The CLICK-IT reaction for EdU labeling was performed according to the manufacturer's instruction (Thermo Fisher Scientific).

Imaging
Live imaging and immunostaining images were obtained using a Zeiss LSM880 confocal microscope or an Olympus IX83 inverted microscope. The numbers of differentially marked cardiomyocytes were counted with ZEN software or cellSens software.

Statistical analysis
Values were presented as mean ± SD. Statistical significance was defined as a threshold of P < 0.05 determined by Student's t-test between two groups, ANOVA analysis between more than two groups or Binomial test in quantification of the percentage of recovered hearts.