Abstract
Sexual differences are frequently observed in the onset and prognosis of human cardiovascular diseases, but the underlying mechanisms are not clearly known. Here, we report that zebrafish heart regeneration is faster in females, can be accelerated by estrogen and suppressed by estrogen-antagonist tamoxifen. Transcriptomic analyses suggested heart injuries triggered more pronounced immune and inflammatory responses, previously shown to enhance heart regeneration, in females, leading to increased STAT1 expression and leukocyte filtration. These responses could be enhanced by estrogen treatment in males. We also showed that injuries to the heart, but not other tissues, increased plasma estrogen level and expression of estrogen receptors, especially esr2a, in zebrafish hearts. Although this effect was stronger in female fish, the resulting endocrine disruption was sufficient to induce the expression of female-specific protein vitellogenin in male zebrafish. Altogether, this study reveals that heart regeneration is modulated by an estrogen-inducible inflammatory response to heart injury, which in turn stimulates estrogen signalling predominately in females. These findings elucidate a previously unknown layer of control in zebrafish heart regeneration and provides a new model system for the study of sexual differences in human cardiac repair.
Introduction
Cardiovascular diseases (CVD) are the primary cause of death worldwide: causing the death of 17.9 million people in 2015, more than 30% of the global mortality (Roth et al., 2017). Gender differences have been reported on risk factors, clinical manifestation and recovery of CVD (reviewed by Maas and Appelman, 2010; Ostadal and Ostadal 2014; EUGenMed et al., 2015). In various mammalian models of cardiac defects, females consistently demonstrate a lower mortality, less severe disease phenotypes and better functional recovery than the male counterparts (Du 2004, Czubryt et al., 2006; Du et al., 2006). An understanding of the molecular mechanism underlying these gender differences will provide an important gateway towards better CVD prevention and treatment. While the protective effects of estrogen on the cardiovascular system have been widely reported, most of these studies focused on how estrogen ameliorates CVD risk factors, by the maintenance of vasodilation and in the prevention of atherosclerosis (Pare et al., 2002; Mendelsohn 2002; Babiker et al., 2002; Fliegner et al., 2010). But estrogen may play more direct roles in cardiac development, functions and pathology, as estrogen can protect cardiomyocytes from apoptosis caused by ischaemia–reperfusion in vitro (Liu et al., 2011). Gene expression profiles of mammalian cardiomyocytes (CM) are sexually dimorphic (Isensee et al, 2008, Witt et al, 2008, Tsuji et al, 2017), and estrogen receptor expression is deregulated in some cardiomyopathies (Mahmoodzadeh et al., 2006). Despite the apparent involvement of estrogen in CVD, clinical translation of these findings has been lagging, partly due to inconclusive results from large-scale studies on the benefits of post-menopausal hormone replacement treatment on CVD prevention (Low et al., 2002). This is attributed, at least in part, to the lack of a mechanistic understanding of the roles of estrogen in cardiomyocyte biology, which prevents optimal study design and subject selection in clinical studies (Whayne & Mukherjee, 2015).
While adult human cardiomyocytes are virtually unable to re-enter the cell cycle, many vertebrates, including zebrafish, newts and axolotls, demonstrate a remarkable ability to regenerate their hearts (reviewed by Garcia-Gonzalez and Morrison 2014; Vivien et al., 2016). Among these organisms, zebrafish has emerged as one of the most well-established model organisms for the study of heart regeneration. Even after the resection of up to 20% of its ventricle, a zebrafish can completely regenerate its heart within 2 months with no scarring (Poss et al., 2002). Apart from physical amputation, cryoinjury (Chablais et al., 2011; González-Rosa et al., 2011) and genetic ablation (Wang et al., 2011) of cardiomyocytes can also trigger proliferation, usually measured by the expression of proliferating cell nuclear antigen (PCNA) or phosphor-histone H3. Heart regeneration is not localized to the proximity of the wound, but involves the precise coordination of cardiomyocytes proliferation, migration (Itou et al., 2012) and extracellular matrix remodelling (Wang et al., 2013). Fate mapping experiments have established that the proliferating cardiomyocytes originated from mature cardiomyocytes (Jopling et al., 2010; Senyo et al., 2013; Zhang et al., 2013). The rapid disappearance of the sarcomeric structures and re-expression of embryonic genes, such as embryonic cardiac myosin heavy chain gene (embCMHC), suggest that regeneration largely involves the expansion of dedifferentiated cardiomyocytes (Jopling et al., 2010; Sallin et al., 2015). Biochemical and gene expression analyses have identified the involvement of growth factors such as tgfb1, pdgf, shh and igf2b, and signalling pathways such as Notch, Jak1/Stat3 and NFkB in zebrafish heart regeneration (reviewed by Kikuchi et al., 2015; González□Rosa et al., 2018). A recent study has attributed the remarkable regenerative capacity of zebrafish heart to this organism’s enhanced inflammatory and immune responses, in comparison with other fish species (Lai et al., 2017). Hypoxia, as a result of cardiac damage, also plays a positive role in cardiomyocyte proliferation (Jopling et al., 2012), whereas hyperoxia and miR-133 inhibit regeneration (Yin et al., 2012).
While the zebrafish heart is different from the human counterpart on cellular and anatomical levels (Hu et al., 2000; Jensen et al., 2013), the conservation of the genetic circuitry between zebrafish and mammals (Howe et al., 2013), and the technical and genetic resources offered by the zebrafish research community, make this organism an ideal model for the study of the gender bias associated with human CVD. Some observations have suggested a role of sex hormones in zebrafish cardiac development and functions. For instance, inhibitors of estrogen synthesis induced phenotypes similar to congestive heart failure and tamponade in zebrafish embryos (Allgood et al., 2013). These conditions could be reversed by estradiol (E2) treatment, which has also been shown to affect heart rates during zebrafish embryonic development (Romano et al., 2017). Despite these findings, the entire literature on zebrafish heart regeneration is based on studies on a single sex (usually male) and sexual differences in regeneration have never been investigated. In this study, we examine, for the first time, the sexual dimorphism of zebrafish in heart regeneration and in response to factors that influence the rate of cardiac regeneration.
Results
Zebrafish heart regeneration is sexually dimorphic
Female and male zebrafish, matched for age and weight, were subjected to cardiac damage by cryoinjury. On day 7 post cryoinjury (7 dpc), female hearts contained a significantly higher number of PCNA-positive cells (Fig 1A) and less vimentin immunoreactivity (Fig 1C) than in the male counterpart (Fig 1B and 1D), indicating more cell proliferation and fewer scar-forming fibroblasts in the regenerating female heart. After one month of regeneration, female hearts contained significantly smaller scar tissues compared to males (Fig 1E, F). These observations prompted us to investigate the effect of estrogen on zebrafish heart regeneration. Male zebrafish exposed to E2 after cryoinjury displayed a ∼5-fold increase in cardiomyocyte proliferation in the vicinity of the injured area (Fig 1G, H). Male cardiomyocytes were highly sensitive to E2: the number of PCNA-positive cells in the wound increased by ∼4-folds when exposed to E2 at a concentration as low as 0.01nM (Fig S1A, B). E2 treatment of male fish also increased cardiomyocyte dedifferentiation, as judged by the expression of embCMHC (Fig 1I, J, Fig S1E), and scar reduction (Fig 1K, L). Although E2 treatment could also accelerate heart regeneration in females (Fig 1K, L), the effect was less pronounced (∼2-fold reduction of scar size) than in males (∼3-fold reduction of scar size, Fig 1L). Conversely, treatment of female fish with tamoxifen, an antagonist of estrogen receptors (Jordan, 2006; Xia et al., 2016), resulted in a ∼4-fold decrease in cardiomyocyte proliferation (Fig 1G, H), ∼2-fold decrease in embCMHC expression (Fig 1I, J, Fig S1E), and ∼2-fold increase in scar size (Fig 1K, L). Moreover, echocardiographic measurements, that allowed the non-invasive measurement of cardiac performance during regeneration (Wang et al., 2017), indicated that E2 treatment of male fish after cryoinjury accelerated the restoration of Fractional Shortening (FS) time to the pre-injury level (Fig S1F). FS is a parameter that indicates the cardiac contractile force and has previously been shown to correlate with zebrafish cardiac recovery during regeneration (Hein et al., 2015). The increase in FS recovery by E2 treatment confirms the promoting effect of this hormone on the recovery of physiological functions after cardiac damage in male zebrafish.
To determine whether female and male zebrafish responded differently to environmental conditions that alter heart regeneration rates, heart-injured fish were exposed to hyperoxia, a condition known to suppress the induction of HIF1a after cardiac injury and decelerate regeneration (Jopling et al., 2012). As expected, hyperoxia reduced the occurrence of PCNA-positive cells in the regenerating hearts of both sexes (Fig S2). However, the effect of hyperoxia on the frequency of PCNA-positive cells was stronger in females (∼40% in females, less than 30% in males). As a result, the sexual difference in cardiomyocyte proliferation was significantly reduced in hyperoxic condition. As hyperoxia also eliminated the effect of E2 on cardiomyocyte proliferation in male zebrafish (Fig S2), estrogen is likely to regulate mechanisms that are upstream of HIF1a in the heart regeneration regime.
Enhanced immune and inflammatory responses to heart injury in female zebrafish
To elucidate the mechanism underlying the sexual difference in zebrafish heart regeneration, we performed comparative transcriptomic analyses on RNA extracted from female and male hearts at 7 dpc. A total of 1050 genes were found to be differentially expressed (by more than two-fold) between two sexes with statistical significance across 3 biological replicates (Table S1). Gene ontology (GO) analyses revealed that a majority of biological processes enriched for the female-biased genes are related to immunological processes, such as immune response, inflammatory response and chemotaxis for immune cells. On the other hand, male-biased genes are more diverse in function, ranging from protein homeostasis, stress response to muscle contraction (Fig 2A). Gene set enrichment analysis (GSEA, Subramanian et al., 2005) confirmed that immune-related pathways are among the most sexually dimorphic in post-injured zebrafish hearts (Fig S3). As female heart regeneration is faster, the over-representation of immune-related genes in heart-injured females might be a mere reflection of the fact that at 7 dpc the female hearts were at more advanced stage of regeneration than the males. A comparison of the published transcriptomic profiles of zebrafish hearts collected at 2 and 5 dpc (Lai et al., 2017), however, shows that immune-related genes were not among the most differentially expressed genes between these two days (Fig S4), indicating that the observed female-biased expression of immune-related genes after heart injury was far more pronounced than can be explained by the difference in heart regeneration rate between sexes.
To test how much the observed female-specific gene expression pattern in heart regeneration was shaped by estrogen, the respective effects of E2 treatment of male fish, and tamoxifen treatment of female fish, on the transcriptome of injured hearts were investigated. Fig 2B shows that tamoxifen treatment reversed almost all female-specific gene expression signatures in the injured heart, whereas genes involved in immune response and neutrophil chemotaxis were upregulated in estrogen-treated males. Among the genes whose expression level in injured hearts are sexually dimorphic, and/or reciprocally regulated by E2 and tamoxifen, are interferon-gamma (IFNg), interferon regulatory factor 1b (IRF1b) and various isoforms of CXCL11 (Fig 2C).). In mammals, both IRF1 and CXCL11 are IFNg-inducible (Yang et al., 2007; Flodstro□m & Eizirik, 1997), suggesting the interferon-gamma pathway is instrumental to the sexual dimorphism in heart regeneration. In consistent with the transcriptomic data, Q-PCR analysis showed that the expression level of IFNg in the female heart increased by ∼20-folds rapidly (1 dpc) after injury, while remained unchanged in males (Fig 2D). IFNg expression after heart injury was significantly stimulated by E2 treatment of male fish and suppressed by tamoxifen treatment of females (Fig 2E, F), in consistent with the role of estrogen in IFNg expression in other tissues (Fox et al., 1991; Hao et al., 2013). There was a significant increase of STAT1, a target of IFNG (Horvath, 2004), at 7 dpc in female hearts but not in males (Fig 2G), consistent with the female-specific induction of IFNg induction (Fig 2D). The upregulation of STAT1 in female hearts was reversed by tamoxifen. In males, E2 treatment resulted in a small but significant increase in STAT1 (Fig 2G, H). STAT3 was also induced in female hearts after heart injury, but this induction was not tamoxifen-sensitive (Fig 2G), suggesting that, although STAT3 is essential to injury-induced cardiomyocyte proliferation in general (Fang et al., 2013), STAT1 may be more important for orchestrating sexual difference in regeneration rates. Interestingly, STAT1 has recently been identified as an estrogen-responsive gene in mammals (Young et al., 2017). As IFNg and STAT1 are important for the activation (Ohmori & Hamilton, 2001) and retention (Hu et al., 2008) of macrophages and neutrophils in the inflamed area, we asked how the sexually dimorphic activation of STAT1 affected the occurrence of leukocytes in cardiac wound. Figure 2I shows that, at 1 dpc, the number of L-Plastin positive leukocytes (Shinomiya, 2012) detected in the injured area was significantly higher in females than in males (Fig 2I, J).
Estrogen receptor esr2a is induced by heart injury
In our transcriptomic analyses, we noticed that the expression level of an estrogen receptor (ER) was significantly upregulated in male hearts at 7 dpc (data not shown). To investigate this phenomenon in more detail, the expression of all three nuclear ERs (Lu et al., 2017) was examined in zebrafish hearts after cryoinjury or sham operation, as well as in the heart of fish that had not undergone any surgery (uninjured). Sham operation was performed by the identical procedures that preceded cryoinjury, including the opening of chest skin and tissue to expose the heart. Fig 3A shows that at 7 days post-surgery, the expression of esr1 and esr2a was induced (∼3-fold compared to uninjured) in male hearts. The increase in expression of these two ERs in female hearts was even more pronounced (up to 4-folds for esr1 and 9-folds for esr2a, Fig 3B). Sham operation also induced in expression of esr2a (in female) and esr1 (in both sexes), but to a lesser extent compared to the effect of cryoinjury. Esr2b was suppressed by both cryoinjury and sham operation in either sex. Fig 3C shows that E2 treatment led to a dramatic increase in esr2a expression in injured male hearts (∼6-folds higher than injured males without E2 treatment), whereas the effect on esr1 and esr2b was comparatively minor. The expression of esr2a, but not esr1 and esr2b, in injured female hearts was suppressed by tamoxifen (Fig 3B). Cryoinjury appeared to sensitise esr2a towards E2, as the induction of esr2a expression by E2 treatment was much less pronounced in uninjured male hearts (∼2-folds) than at 7 dpc (∼6-folds) (Fig 3E).
Cardiac injury leads to endocrine disruption in zebrafish
As ERs are estrogen inducible genes, the upregulation of esr2a in heart-injured fish prompted us to investigate whether the level of E2 in those fish have also changed. Fig 3F shows that the plasma E2 level in male cryoinjured and sham-operated fish was 20% and 10% higher than that of the control respectively. Both cryoinjury and sham operation led to an even larger increase in E2 level in females (∼5-fold) (Fig 3G). The induction of ER expression did not generally occur in all types of injured tissue, as the plasma E2 levels in male and female fish on day 7 after caudal fin amputation were virtually unchanged as compared to untreated controls (Fig 3H), and the expression of esr1 and esr2a in the fin was not significantly altered after fin amputation (Fig 3I, J). Hence, heart injury and sham operation increased estrogen level and ER expression in zebrafish hearts, especially in females, and this endocrine disruption appeared to be specifically linked to injuries of the heart and its proximity, but not other tissue regeneration/repair systems. We hypothesised that hypoxia, due to heart injury and/or blood loss, may be the stimulus of estrogen induction. However, hyperoxia did not significantly affect the increase of E2 after heart damage in either sex (Fig 3K).
To examine the biochemical consequence of this hormonal change, we utilized the sexual dimorphism in zebrafish plasma proteins (Babaei et al., 2013; Li et al., 2016). Quantitative proteomic analysis was used to study the relative abundances of 18 known sexually dimorphic plasma proteins in the plasma collected from untreated, sham-operated and ventricular amputated male zebrafish (Fig S5 and Table S2). Most female-biased plasma proteins increased after cardiac damage, as compared to sham operation, except Nucleoside Diphosphate Kinase (NDPK), a protein known to be marginally female-biased (Li et al., 2016). Conversely, all but two male-biased plasma proteins were downregulated after ventricular amputation. As the two outliers, Myoglobin and Phosphoglycerate Kinase 1, are expressed predominately in muscle, the increase of these proteins in plasma might be a result of the release of CM cellular content into circulation. These data indicate that although the impact of heart injury on estrogen level in males was relatively minor, it was sufficient to shift the gender characteristic of the male zebrafish plasma towards a more feminised composition.
Vitellogenin accumulate in the regenerating heart of adult male zebrafish
Our plasma proteomic analysis indicates the presence of vitellogenin isoforms in the plasma of male zebrafish after heart injury (Fig S5). Western blotting detected vitellogenin in the plasma of male zebrafish as early as day 1 after both cryoinjury (Fig 4A) and heart amputation (data not shown), as compared to sham or untreated fish. Vitellogenin is specifically expressed in females (reviewed by Hara et al., 2016) and are routinely used as biomarkers for endocrine disruption in males (García-Reyero et al., 2004; Scott and Robinson, 2008). As vitellogenin has been reported as an acute phrase response protein (Tong et al., 2010; Zhang et al., 2011), it is possible that vitellogenin was induced in cardiac-damaged males as a result of infection. However, the abundance of vitellogenin in the plasma was much higher in heart-damaged than in sham-operated fish, even though the latter also suffered from extensive tissue damages. Moreover, plasma proteomics showed that most of the known acute phase response proteins (Roy et al., 2017), such as the complement and coagulating factors, were down-regulated in the heart regenerating fish as compared to the sham (Fig S6 and Table S3). This suggests that the presence of vitellogenin in male plasma after heart injury is not a consequence of injury-related infection or inflammation; but is specifically associated with heart regeneration.
We examined the tissue distribution of vitellogenin in male zebrafish during heart regeneration. On day 5 after cardiac injury, vitellogenin was detected in the heart, but not in the gill, kidney and liver (Fig 4B). Vitellogenin was observed in the entire regenerating heart, not restricted to the wound. Whole mount immunohistochemistry confirmed that vitellogenin accumulated in the injured heart of male fish (Fig 4C) and was not observed in the proximity of the chest wound of the sham-operated fish. This indicates that the vitellogenin accumulation in male fish is specifically associated with cardiac damages and is not related to general wound healing. Moreover, vitellogenin was not undetectable on the male caudal fin on day 7 after amputation (Fig 4D), confirming that vitellogenin accumulation is not a general consequence of tissue regeneration or repair. Furthermore, E2 treatment of male fish increased the level of vitellogenin in the regenerating hearts, while tamoxifen treatment of female fish reduced this accumulation (Fig 5).
Discussion
In this study, we revealed the roles of estrogen in several aspects of zebrafish heart regeneration. First, we demonstrated by cellular (Fig 1, 2), anatomical (Fig 1, 4), physiological (Fig 3) and biochemical (Fig 3) parameters that zebrafish heart regeneration is sexually dimorphic. As far as we know, this is the first evidence of sexual dimorphism in heart regeneration. Sexual differences in the regeneration of other tissues has been observed in mammals but has not been documented for the heart (Harada et al., 2003; Blankenhorn et al., 2003; Deasy et al., 2007). In zebrafish, the pectoral fin is the only tissue that has been shown to demonstrate a sexually dimorphic regenerative capacity (Nachtrab et al., 2011). Like their hearts, males regenerate the pectoral fin more slowly and often incompletely. However, this phenomenon was not observed in other fins (Nachtrab et al., 2011), suggesting that this sexual dimorphism is related to the unique role of pectoral fins in reproduction (Kang et al., 2013; McMillan et al., 2013).
The observed female superiority in heart regeneration is likely due to the effect of estrogen: E2 accelerated heart regeneration in males (Fig 1), and tamoxifen retarded it in females (Fig 1). Our gene expression analyses suggested that female hearts demonstrated a stronger immune and inflammatory responses to cryoinjury. In particular, the expression of IFNg, as well as IFNg-inducible factors, in regenerating hearts was highly sexually dimorphic and E2 sensitive (Fig 2). As inflammation is essential for tissue regeneration (Eming et al., 2017), including cardiac repair (Frangogiannis, 2015), our observation is consistent with the higher heart regeneration rate in female zebrafish. Importantly, a comparison between fish species with different heart regeneration rates attributes the high heart regeneration capacity of zebrafish to this organism’s enhanced inflammatory response to heart injury (Lai et al., 2017). These authors also discovered that treatment with poly I:C, a viral mimic known to stimulate IFNg-responsive genes (Farina, et al., 2010), can enable heart regeneration in medaka, a species that is normally unable to repair heart injuries. The immune-promoting effects of estrogen have been well documented in fish and mammals (Burgos-Aceves et al., 2016 Taneja, 2018). This study reveals, for the first time, that the sexual differences in immune functions can modulate tissue regeneration.
In this study, we have also discovered a previously unknown aspect of tissue regeneration in zebrafish: the spontaneous endocrine disruption after cardiac damage towards feminisation. Our data reveal that cardiac damage triggers the secretion of E2 (Fig 3F, G) and expression of estrogen receptors, notably esr2a (Fig 3A, B), in both sexes. Interestingly, a genetic variant of esr2 has been identified as a risk factor of myocardial infraction (Domingues-Montanari et al., 2008). In addition, ERb, a mammalian homologue of esr2a, has recently been found to locate in mitochondria and play bioenergetic roles (Liao et al., 2015). We observed a ∼9-fold increase in esr2a expression in female hearts at 7 dpc. Such dramatic increase in ER expression, compounded by the ∼5-fold increase of plasma E2 level, will likely sensitise estrogen-responsive genes and amplify the estrogen-dependent inflammatory response of the female heart towards injury. We postulate that this can further enhance the female prominence in heart regeneration, and firmly establish the sexual dimorphism of this process (see Fig 6 for a model that summarises the data presented in this study).
The stimulation of estrogen level and receptor by heart injury also operates in male fish, hitherto to a lesser extent. Interestingly, the level of serum estrogen among acute myocardial infarction and unstable angina patients were found significantly higher than in the normal group and intensive care patients (Aksut et al., 1986), suggesting that cardiac lesions, but not other traumas, can lead to an increase in estrogen level in humans. The clinical significance of this observation is undermined by the lack of understanding of the underlying mechanism. We now show that this phenomenon is evolutionarily conserved.
Sham operation can precondition zebrafish for heart regeneration (de Preux Charles et al., 2016a, b). Lai et al (2017) postulated that the preconditioning was mediated by the upregulation of immune-related genes in the heart of sham-operated zebrafish, but could not identify the cause of this change in gene expression. Here, we discovered that sham operation could induce plasma E2 level and the expression of ER in zebrafish hearts, suggesting that an estrogen-dependent inflammatory mechanism might be the cause of preconditioning. It is not known how cardiac injury and sham operation induces E2 level in zebrafish. It is not caused by hypoxia, a physiological consequence of heart damage, as hyperoxic treatment did not prevent E2 increase (Fig 3K). Moreover, environmental hypoxia is known to decrease, rather than increase, estrogen levels in zebrafish (Shang et al., 2006). Zebrafish offers an excellent animal model for delineating the molecular mechanism of this process, as biochemical pathways for sex hormone synthesis have been studied in detail in this organism (Segner, 2009).
It is interesting to examine how the systemic increase of estrogen as a result of heart injury may impact the functions of other organs during heart regeneration. One of the consequences of the endocrine disruption is the detection of vitellogenin in males, an occurrence generally regarded as the hallmark of endocrine disruption by environmental agents (García-Reyero et al., 2004; Scott & Robinson, 2008). The accumulation of vitellogenin in regenerating hearts, but not in tissues (Fig 4), suggests a functional role in cardiac regeneration, rather than as a collateral consequence of estrogen secretion. The canonical function of vitellogenin is to transport lipids, calcium and phosphate to the developing oocytes (Arukwe & Goksøyr, 2003, Hara et al., 2016). We hypothesise that vitellogenin might play similar roles in the regenerating heart. Lipids are instrumental to cardiac remodelling in reptiles (Riquelme et al., 2011), and our previous micro-CT imaging of cardiac-damaged male zebrafish has shown the accumulation of lipids around the regenerating heart (Babaei et al., 2016). Future investigations will involve the characterisation of the vitellogenin cargoes isolated from the plasma and heart of fish after cardiac damage.
Competing Interests
The authors declare that they have no conflicts of interest with the contents of this article.
Methods
Key Resources Table
Experimental Model
Zebrafish used for heart cryoinjury, ventricle amputation, fin amputation, immunofluorescence, western blot, echocardiography, Mass spectrometry, Quantitative PCR (q-PCR) was acquired from the Zebrafish International Resource Center (ZIRC; University of Oregon, Eugene, OR, USA). Fish were maintained in a recirculating system at 28°C with a photoperiod of 14 h of light/10 h of darkness, and fed with dry meal (TetraMin) three times a day, supplemented with live brine shrimp once a day (Westerfield, 2000). All the animal procedures used in this study were approved by the Department of Health, Hong Kong, SAR, China (refs (17-18) in DH/HA&P/8/2/5 Pt.1), and the experiments were conducted in accordance with the relevant guidelines and regulations in Hong Kong, SAR, China.
Method Details
Animal surgery
Adult zebrafish (12 to 18-month-old, the weight is 0.35-0.37g, with cardiac weight accounting for 0.5% of body weight) were anesthetised by immersion in system water containing 0.04% MS-222 (ethyl-3-aminobenzoate methanesulfonate salt; E10521; Sigma-Aldrich) for 3−5 min and then immobilized on a Petri dish. After the removal of the ventral scales by forceps, a small incision was made through the body wall and the pericardium by using forceps and micro-dissection scissors, tearing the tissue rather than making a clean cut in order to facilitate healing. Once the pericardial sac was opened, the heart ventricle was exposed by gently squeezing the abdomen. For sham operation, the heart was gently pushed back into the body without any further treatment. For ventricular amputation, a portion of the ventricle was excised by using surgical fine scissors (Poss et al., 2002). For cryoinjury, the ventricle was touched for 10-12 seconds by metal probe pre-chilled in liquid nitrogen (Chablais et al., 2011). For fin amputation, the caudal fin of zebrafish was cut using a blade after anesthetization (Nachtrab et al 2011). After the operation, fish were placed in a tank of fresh water, and reanimation was enhanced by pipetting water onto the gills for a couple of minutes. At time points indicated in the Results, the zebrafish were sacrificed by immersion in overdose concentration of MS-222. Hearts were dissected from the fish and their plasma were collected as previously described (Babaei et al., 2013).
Chemical exposure
Adult male and female zebrafish, untreated or after surgery, were incubated in water containing 17β-estradiol (E2; E1132, Sigma-Aldrich) at various concentrations (0, 0.01, 1 and 100 nM), tamoxifen (T9262, Sigma-Aldrich) at various concentrations (0 nM, 10 nM, 100 nM, 1000 nM), or DMSO (D5879, Sigma-Aldrich), which was used to prepare the E2 and tamoxifen, at matching concentrations. The water was changed every day, and the fish were continuously exposed to the vehicle or drugs for durations indicated in Fig 1-5.
Hyperoxic treatment
Zebrafish were exposed to hyperoxia by using the method described by Jopling et al (2012). Briefly, zebrafish were placed in an autoclave bag (EURTUBO) with 2L fish water. 100% oxygen was used to fill the bag. The bag was tightly sealed with a rubber band and placed in an incubator at 28°C. The average O2 concentration in the water was 20 mg/L. The control zebrafish was maintained in the bag with normoxic condition, and left bag open. The average O2 concentration in the water was 6.5 mg/L.
Western blotting
Proteins from zebrafish hearts were extracted by using RIPA Lysis and Extraction Buffer (89901, Thermo scientific) according to manufacturer’s instruction. BCA Protein Assay (23225; Thermo Scientific) was used to determine the total protein concentration of zebrafish plasma and heart protein extract, according to the manufacturer’s instruction. 20µg of protein sample per lane was separated in a 10% SDS polyacrylamide gel, transferred to polyvinylidene difluoride (PVDF) membrane (10600023; GE Healthcare life science) by using Mini-Trans Blot Electrophoretic Transfer Cell system (1703930; Bio-Rad Laboratories) according to the manufacturer’s instructions. The blot was blocked in 5% no fat milk in PBST (0.05% Tween 20 in 1X PBS) for 1 hour at room temperature, following by incubation in primary antibodies diluted in PBST overnight at 4□. The following primary antibodies were used: mouse anti-GAPDH (60004-1; Proteintech Group, Inc.) at 1:10000, mouse anti-zebrafish vitellogenin, JE-2A6 at 1:2000, mouse anti-embCMHC at 1:500, rabbit anti-SATA1(R1408-2, Huabio Inc.) and rabbit anti-STAT3(ET1607-38, Huabio Inc) at 1:2000. For secondary antibodies, HRP-conjugated rabbit anti-mouse IgG (AP160P; Millipore) at 1:5000, and HRP-conjugated goat anti-rabbit IgG (AP307P; Millipore) at 1:5000 were used. The proteins were detected with the EMD Millipore Luminata Western HRP chemiluminescence substrate (WBLUF0500; Millipore) and the signals were visualized with the Western blotting system (C600; Azure Biosystems, Inc.). The band densities were quantified using Image J.
Mass spectrometry
Proteins from plasma collected from individual fish (typically about 20ug) were precipitated in 1 ml of acetone for 60 min at -20°C and pelleted at 1000g for 10 min at 4°C. The pellet was air-dried (30 min) and then dissolved in 8M urea in 10mM Tris-HCl (pH 8.0). After clarification (10000g, 10 min, RT), the supernatant was reduced by 10mM dithiothreitol (R0861, Thermo Scientific) in 50mM ammonium bicarbonate (30 min, RT), and alkylated in 50mM iodoacetamide (90034, Thermo Scientific) in 50mM ammonium bicarbonate (20 min, RT). LysC (Roche) was then added to the protein mixtures and incubated for 120 min at RT. The samples were then diluted with ammonium bicarbonate (Sigma-Aldrich; 50 mM, pH 8.0) so that the final concentration of urea was 1 M, followed by the addition of 1 ug trypsin (Roche). The mixtures were incubated overnight at 37°C and dried in a vacuum centrifuge. Peptides were dissolved in a small amount of 0.1% trifluoroacetic acid (Sigma-Aldrich; TFA), followed by purification with the use of ZipTip (Millipore), following manufacturer’s instruction.
Mass spectrometry was performed on the Thermo Scientific Q-Exactive Plus Mass Spectrometer (Thermo Fisher Scientific, Bremen, Germany) coupled to a Thermo Ultimate 3000 RSLCnano HPLC system. About the 0.5 ug of the peptide mixture was loaded onto a 250nl OPTI-PAK trap (Optimize Technologies, Oregon City, OR) custom packed with Michrom Magic C18 5um solid phase (Michrom Bioresources, Auburn, CA). Chromatography was performed a Thermo EasySpray 25cm × 75um C18 2um column, using 0.2 % formic acid in both Solution A (98%water/2% acetonitrile) and Solution B (80% acetonitrile/10% isopropanol/10% water), in a gradient from 2%B to 35%B over 140 min at a flowrate of 325 nl/min. The Q-Exactive Plus mass spectrometer was set up with a FT survey scan from 340-1500 m/z at resolution 70,000 (at 200m/z), followed by HCD MS/MS scans on the top 15 ions at resolution 17,500. The MS1 AGC target was set to 1e6 and the MS2 target was set to 2e5 with max ion inject times of 50ms and 75ms respectively. Dynamic exclusion placed selected ions on an exclusion list for 30 seconds. Charge exclusion was used to perform MS/MS only on +2, +3, and +4 ions.
MS/MS peak lists were exported as an .mgf file and proteins were identified via automated Mascot database searching (Matrix Science 2.2.04) of all tandem mass spectra against of the Danio rerio Protein Index protein sequence database that contained all zebrafish protein entries (143,725 sequences) from NCBI RefSeq (version 51). The instrument setting for the Mascot search was specified as “ESI-Trap.” Parameters used for the database search were as follows: a maximum of two missed cleavages; carbamidomethylation of cysteine as a fixed modification and oxidation of methionine, acetylation of protein N-term and Gln→pyro-Glu conversion as variable modifications; trypsin as the enzyme; a peptide mass tolerance of 10 ppm; a fragment mass tolerance of 0.6 Da; and an ion score of 35 as the cut-off, using a significance threshold of p < 0.05. After peptide identification, any peptides that had conflicting assignments were resolved, either one of identical proteins or by assignment to proteins with the largest number of peptides already present (by following Occam’s Razor principle).
The acquired MS data (in Thermo “.raw” format) from 3 biological replicates in the SO and Untreated groups (i.e., three separate plasma samples each collected from one individual fish) and 4 biological replicates in the VA group were quantitated by using Progenesis LC-MS (version 2.5, Nonlinear) as previously described (Babaei et al., 2013).
Histology
Organs dissected from zebrafish were fixed with 4% paraformaldehyde at 4°C overnight, dehydrated and embedded in paraffin as previously described (Chablais et al., 2011). To measure scar size, the entire heart was serially sectioned at the thickness of 5µm. The sections were deparaffined, rehydrated and stained with picrosirus red (ab150681, Abcam). On each section, the labeled area and the area of the whole ventricle were measured by using ImageJ (National Institutes of Health, Bethesda, MA, USA), and the percentage of the scar volume to the entire ventricle was calculated by the summation of the data from all the sections (Xu et al., 2018). For the whole mount paraffin sections, the fish were fixed with 2% PFA and 0.05% Glutaraldehyde in 80% HistoChoice (H120, Amresco Inc.) with 1% sucrose and 1% CaCl2 at 4□ for 24h, and the paraffin sections were prepared according to Kong et al. (Kong et al., 2008).
For immunohistochemistry, antigen retrieval was performed on dewaxed sections in sodium citrate buffer (10 mM sodium citrate, 0.05% Tween 20, pH 6.0) at 95°C for 15 min. Primary antibodies used were: mouse anti-vimentin (ab8978; Abcam) at 1:200; mouse anti-PCNA (sc-56; Santa Cruz) at 1:200; mouse anti-zebrafish vitellogenin, JE-2A6 (V01408102; Biosense) at 1:200, mouse anti-embCMHC (N2.261;Developmental Studies Hybridoma Bank) at 1:50, rabbit polyclonal anti-GFP (ab13970; Abcam) at 1:200, mouse anti-L-plastin (sc-133218; Santa Cruz Biotechnology Inc.) at a dilution of 1:100. The secondary antibodies used were: Cy3-conjugated goat anti-mouse (A10521; Invitrogen) or Alexa Fluor 488-conjugated goat anti-rabbit (A11034; Invitrogen) antibodies at 1:500. The sections were amounted with cover slide in 50% Glycerol in PBS and images were acquired using Olympus BX61 microscope.
Plasma E2 concentrations measurement
Plasma from three individual females or males were pooled, and three biological replicates were performed. The concentration of E2 and testosterone in plasma were measured using Estradiol ELISA Kit (582251, Cayman).
Quantitative PCR (q-PCR) and RNA sequencing
Total RNA was extracted from zebrafish heart, fin or liver using NucleoZOL reagent (740404; MACHEREY-NAGEL GmbH & Co. KG). Three individuals were pooled together as one biological replicate, and three biological replicates were performed. 1µg total RNA was decontaminated using RQ1 RNA-free DNase (M6101; Promega Co.) and then cDNA was synthesized using the PrimeScript reverse transcription (RT) reagent kit (6210B; Takara Bio. Inc.) according to the manufacturer’s instructions. The expression of each gene was determined by q-PCR using the SYBR Premix Ex Taq (RR402A; Takara Bio. Inc.), and β-actin was used as the reference gene. q-PCR was performed in triplicate for each gene, and the results were analyzed using the ΔΔCT method. All levels of expression of genes were normalized to β-actin and fold change was calculated by setting the normal female to 1.
For RNA sequencing, total RNA was extracted and decontaminated, the sequencing was performed using BGISEQ-500 platform, averagely generating about 21.83 M reads per sample. The reference genome can be accessed at: http://www.ncbi.nlm.nih.gov/genome/50?genome_assembly_id=210873. The average mapping ratio with reference genome is 90.81%, the average mapping ratio with gene is 71.22%; A total of 25,900 genes were detected. The sequencing and the primary analysis were performed by BGI (Shenzhen, China), three bio-replicates of each samples were performed. We did comparative GSEA (Gene set enrichment analysis) and drew the union enrichment maps using an R package, HTSanalyzeR2 (https://github.com/CityUHK-CompBio/HTSanalyzeR2). Here, we focus on Biological Process Gene Ontology. By permutating 1000 times, we get the statistical significance and choose adjusted P value < 0.05 as significant results.
Echocardiography
Zebrafish was anaesthetized with 0.02% MS-222 and fixed in a sponge upside-down. Echocardiographic studies were performed using Vevo LAZR Multi-modality Imaging Platform (FUJIFILM VisualSonics) under B-mode (50 MHz, 77 fps) and PW Doppler mode (40 MHz, 25 kHz PRF) at 20□ as previously described (González-Rosa et al., 2014; Hein et al., 2015; Wang et al., 2017). Ejection Fraction (EF) and Fractional Shortening (FS) were acquired using a plug-in of Vevo LAB (Vevo Strain) under B-mode images. The window was set to be 300 frames long, starting with diastalsis. PW Doppler images collected were analyzed using Vevo LAB (FUJIFILM VisualSonics) for E/A ratio calculation. 5 respective sets of E/A values were calculated for each sample.
Quantification and Statistical Analysis
Three different images were taken for each heart. The percentages of proliferating cells in ventricle or injured area were the ratio of PCNA positive cells/DAPI. For untreated fish, all nuclei in the whole ventricle were counted; for cryoinjured fish, only the nuclei in the injured area (marked by white dash lines) in Fig 1, Fig S1, and Fig S2 were counted. The area of vimentin, embCMHC, and L-plastin expressions were quantified in the injured area using Image J and the percentage of the expression with respect to the area of the injured area was calculated (de Preux Charles et al., 2016a). For quantification of proliferating cardiomyocyte, the PNCA+/GPF+/ double positive cardiomyocyte within 200 µm of the vicinity of the injured area were quantified, and normalized with injured area in Fig 1G, H. The data were expressed as the mean±S.D.(standard deviation). Statistical analysis was performed using Student’s two-tailed t-test and one way-ANOVA.
Supplementary fiels
Acknowledgements
SHC and YWL conceived and coordinated the study; SSX, SF and FB designed, performed and analyzed the experiments; FJX, KFW, and LS analyzed the echocardiographic data; TML and RR conducted the proteomic analyses; LNZ and XW do GSEA analysis; SSX and YWL wrote the paper; All authors reviewed the results and approved the final version of the manuscript. This work was supported by a general research fund grant [Project No. CityU 160213] from the Research Grants Council (RGC) of the Hong Kong Special Administrative Region, China, to SHC, and a Strategic Research Grant [Project No. CityU 7004661] from City University of Hong Kong to YWL. We thank the University Research Facility in Life Sciences of Hong Kong Polytechnic University for providing Vevo 2100 for this research. We thank all members of the SHC and YWL labs for fruitful discussions.