Parallels between oncogene-driven cardiac hyperplasia and heart regeneration in zebrafish Free

Epimorphic organ regeneration and tumor formation are somewhat related processes because they depend on enhanced cell proliferation in a functional body part. Both phenomena are triggered by a disturbance of tissue homeostasis through the disruption of organ integrity or genetic aberrations, respectively. The key difference is the opposite outcome for the organism: the epimorphic regeneration reconstructs the damaged organ, whereas a tumor ruins the organ architecture. Whether the hyperplastic nature of regeneration-competent and oncogene-transformed cells shares other common cellular and molecular mechanisms is still being disputed (Charni et al., 2017; Corradetti et al., 2021; Milanovic et al., 2018; Oviedo and Beane, 2009; Pomerantz and Blau, 2013; Sarig and Tzahor, 2017; Stiehl and Marciniak-Czochra, 2017; Wong and Whited, 2020).

The susceptibility for oncogenic diseases can substantially vary in different cell types. In the adult mammalian heart, the risk of hyperplasia is low, probably owing to a lack of active stem cells and the post-mitotic nature of cardiomyocytes (Cai and Molkentin, 2017; Maleszewski et al., 2017; Maroli and Braun, 2021). Consistently, human myocardial tumors, called rhabdomyomas, are extremely rare, reported mostly in newborn infants (Freedom et al., 2000; Uzun et al., 2007). This neonatal pathology is thought to arise from fetal cardiomyocytes that are capable of cell divisions (Haubner et al., 2016; Mollova et al., 2013). The underlying molecular triggers of these diseases are poorly understood. In conditional transgenic mouse models, overexpression of potent proliferative stimulants can force the cell-cycle entry of mammalian cardiac cells (Lin et al., 2015; Monroe et al., 2019; Wei et al., 2011). Short and transient activation of cardiomyocytes can be beneficial for healing myocardial infarction. However, a prolonged proliferative stimulation frequently causes a pathogenic enlargement of the heart. Thus, the mechanisms governing cell-cycle regulation in cardiomyocytes are fundamental for developing medically relevant approaches.

In contrast to mammals, adult zebrafish can increase the size of their heart through proliferation of functional cardiomyocytes (González-Rosa et al., 2018; Jaźwińska and Blanchoud, 2020; Pronobis and Poss, 2020). Manipulations of physiological conditions and overexpression of proliferative inducers can lead to the development of cardiomegaly due to hyperplasia and hypertrophy (Narumanchi et al., 2021). Under normal conditions, the growth of the heart is endogenously controlled during homeostasis and regeneration. After injury, the zebrafish myocardium dedifferentiates and re-establishes the damaged tissue through cell proliferation (González-Rosa et al., 2017; Han et al., 2019; Jopling et al., 2010; Kikuchi, 2015; Sanz-Morejón and Mercader, 2020; Tzahor and Poss, 2017; Uygur and Lee, 2016). The regenerative capacity is preserved even after multiple cycles of cryoinjury and regeneration in the same animal (Bise et al., 2020). In the cryoinjury model, the peri-injury myocardium, which is located in a zone ∼100 µm from the lesion site, activates the regenerative program and contributes to the new myocardium (Pfefferli and Jaźwińska, 2017; Wu et al., 2016). Within 1-2 months, most of the injured myocardium is restored (Chablais et al., 2011; Gonzalez-Rosa et al., 2011; Schnabel et al., 2011). Various mitotic signaling pathways regulate cardiomyocyte proliferation, including NRG1/ErbB, MAPK and Akt signaling, and their deregulation may lead to cardiomegaly or hypertrophy (Honkoop et al., 2019; Missinato et al., 2018; Tahara et al., 2021; Zheng et al., 2021). An intriguing question is how does zebrafish myocardium react to conditional oncogene expression during homeostasis and regeneration?

Approximately 20-30% of all human cancers are attributed to RAS alterations (Gimple and Wang, 2019). Various tumors have been linked to a missense gain-of-function mutation in the HRAS protein that substitutes glycine at the position 12 with valine (G12V) (Keeton et al., 2017; Li et al., 2018). In zebrafish, tissue-specific overexpression of GFP-HRAS^G12V results in melanoma, leukemia, glioblastoma and chondroma (Lieschke and Currie, 2007; MacRae and Peterson, 2015; Mayrhofer et al., 2017; Mayrhofer and Mione, 2016; Santoriello and Zon, 2012). These findings demonstrate that HRAS^G12V acts as an oncogene in zebrafish. Another related oncogene, KRAS^G12D, causes rhabdomyosarcoma during development (Chen and Langenau, 2011; Storer et al., 2013). The effects of the activated RAS have not yet been characterized in differentiated cardiomyocytes in zebrafish.

RAS proteins are GTPases that regulate several pathways, such as ERK/MAPK and PI3K/AKT/mTOR signaling (Gysin et al., 2011; Keeton et al., 2017; Shaw and Cantley, 2006). The zebrafish RAS-driven melanoma and rhabdomyosarcoma models showed that a combined suppression of ERK/MAPK and PI3K/mTOR signaling is necessary to synergistically impair tumor growth (Fernandez del Ama et al., 2016; Le et al., 2013). Whether inhibiting only one of these pathways is sufficient to suppress HRAS-dependent tumorigenesis is controversial.

In this study, we developed a cardiac-specific tamoxifen-dependent Gal4-ERT2/UAS model to achieve uniform but conditionally regulated expression of HRAS^G12V in zebrafish cardiomyocytes. We investigated whether the oncogenic HRAS-driven phenotype and normal heart restoration are dependent on TOR signaling. Bioinformatic analyses were applied to determine common and distinctive molecular signatures of oncogenic HRAS-expressing myocardium versus regenerating heart. Finally, we assessed whether a pulse of HRAS^G12V expression before cryoinjury can influence the regenerative outcome. The strength of our approach relies on comparison of adult zebrafish cardiomyocytes that were stimulated by conditional oncogenic factors or intrinsic regenerative responses.

To investigate, whether the adult zebrafish myocardium is susceptible to oncogene-driven activation, we assessed the effects caused by conditional EGFP-HRAS^G12V overexpression (HRAS-OE) in cardiomyocytes. To this aim, we generated a transgenic fish line containing a cardiac-specific promoter, cmlc2, upstream of Gal4 fused to a tamoxifen-binding ERT2 domain (Akerberg et al., 2014). To facilitate screening of transgenic fish, we linked the cmlc2:Gal4-ERT2 cassette to a lens marker with a crystallin alpha-a promoter and Kusabira Orange 2 protein: cryaa:KO2. These transgenic fish were crossed with UAS:GFP-HRAS^G12V, and the double transgenic fish were named cmlc2/GFP-HRAS. Control fish were cmlc2:Gal4-ERT2; UAS:mRFP, abbreviated as cmlc2/RFP (Fig. 1A).

In the absence of 4-hydroxytamoxifen (4-OHT), Gal4-ERT2 is retained in the cytoplasm, preventing its function as a transcriptional activator (Akerberg et al., 2014; Gerety et al., 2013). To validate the robustness of this mechanism in our double transgenic lines, we analyzed cmlc2/GFP-HRAS and cmlc2/RFP juvenile fish without inducing treatment. Immunofluorescence analysis of heart sections showed very rare GFP-HRAS-positive cardiomyocytes (3.56%±0.77; n=3 hearts, three or four sections per heart), whereas no expression was observed for RFP (Fig. S1A). We concluded that our newly generated line, cmlc2:Gal4-ERT2, does not markedly activate the UAS-responders in absence of 4-OHT, as predicted.

To assess the inducibility of the transgenic system in adult zebrafish, we applied one overnight pulse of 4-OHT, and collected the hearts at 4 days post-treatment (4 dpt) for immunostaining of transverse sections (Fig. 1B). Quantification of the ventricular area at the level of the atrio-ventricular valve revealed a similar organ size between both fish groups (Fig. 1C,E). Importantly, ∼80% of Tropomyosin-labeled ventricle expressed fluorescent proteins (Fig. 1C,F). In cmlc2/RFP hearts, RAS immunoreactivity was observed only in the valve, demonstrating that endogenous RAS is not detectable in the myocardium (Fig. 1D). In cmlc2/GFP-HRAS hearts, ∼80% of the Tropomyosin-positive area was immunolabeled with the RAS antibody, suggesting a detection of overexpressed transgenic protein (Fig. 1D,G). The high proportions of the labeled myocardium demonstrate that the Gal4-ERT2/UAS system is highly responsive to 4-OHT-mediated induction (Fig. 1G).

To examine the reversibility of the transcriptional activation in the transgenic system, we collected hearts at 14 dpt (Fig. S1B). After this recovery period, no GFP/RFP or RAS immunoreactivity was detected in the myocardium (Fig. S1C-G). This result suggests that these fluorescent proteins were degraded and that the transgene expression was switched off at this time. We concluded that the conditional induction of the Gal4-ERT2/UAS system can be de-activated after discontinued 4-OHT treatment.

To investigate the effects of prolonged HRAS^G12V overexpression on heart growth, we applied four overnight pulses of 4-OHT interspaced by a few days and collected hearts 16 days after the first treatment (Fig. 1H). We selected this time point for the termination of the experiment, because after a longer recovery period, some cmlc2/GFP-HRAS fish started to display behavioral abnormalities, suggesting impaired well-being. In these fish, histological analysis of heart sections revealed gigantic ventricles with denser trabeculation at the expense of luminal cavities, suggesting cardiomegaly (Fig. 1I; Fig. S2). Consistent with the previous experiment, ∼80% of the ventricular myocardium expressed fluorescent reporters and RAS (Fig. 1J,L,M). The ventricular area of sections was 60% larger in cmlc2/GFP-HRAS fish than in cmlc2/RFP control (Fig. 1K). Taken together, the EGFP-HRAS^G12V-overexpressing (HRAS-OE) myocardium rapidly enlarged within 16 days of 4-OHT treatment.

In zebrafish, oncogenic RAS-induced melanoma and rhabdomyosarcoma models showed that only the combined suppression of MAPK/ERK and PI3K/mTOR signaling can synergistically impair tumor growth (Fernandez del Ama et al., 2016; Le et al., 2013). To determine whether inhibition of only one of these pathways can prevent HRAS^G12V-mediated overgrowth of the adult heart, we targeted rapamycin-sensitive kinase TOR, which is known to promote phosphorylation of its downstream target: the ribosomal protein S6 (pS6) (Fig. 2A) (Saxton and Sabatini, 2017; Simanshu et al., 2017). As a proof-of-concept, we first performed an experiment with one pulse of tamoxifen followed by 3 days of 0.1% DMSO (control) or 0.5 µM rapamycin treatment (Fig. 2B). In the DMSO control condition, pS6 immunoreactivity was very rare in cmlc2/RFP hearts, whereas in cmlc2/GFP-HRAS fish, it labeled ∼25% of the myocardium (Fig. 2C,D). This suggests that oncogenic HRAS enhanced the TOR signaling readout in the heart.

Given that the RAS pathway involves MEK and PI3K upstream of TOR (Fig. 2A), we aimed to test the role of these intermediate transducers in the pS6 response in our transgenic model. Using the same experimental design as for rapamycin, we applied specific inhibitors of MEK and PI3K that were validated in zebrafish (Fig. S3A,B). We found that, in HRAS-OE hearts, blocking of either MEK or PI3K displayed a tendency for reduction of pS6 immunofluorescence (Fig. S3C-E). Thus, both signaling transducers are involved in the molecular cascade from HRAS to TOR. Taken together, GFP-HRAS^G12V overexpression upregulates the TOR pathway in the myocardium, and pS6 immunoreactivity serves as a readout of this mechanism in our transgenic system.

After this validation, we designed a 16-day experiment with four pulses of 4-OHT interspaced by DMSO or rapamycin treatment (Fig. 2E). To test whether rapamycin affected transgene expression, we imaged GFP and RFP fluorescence of whole hearts. We found that these fluorescent proteins were present in both groups, suggesting that rapamycin did not interfere with conditional induction of transgene expression (Fig. 2F). Next, we measured the ventricular areas of the dissected whole hearts. In cmlc2/RFP, no size difference was detected between rapamycin- and DMSO-treated groups (Fig. 2F,G). By contrast, in HRAS-OE hearts, the size of rapamycin-treated ventricles was ∼40% smaller than in DMSO-treated control (Fig. 2F,G). We conclude that the inhibition of TOR is sufficient to markedly counteract cardiac overgrowth caused by oncogenic HRAS.

Rapid growth can be induced at the juvenile stage by lower fish density in the aquarium (Wills et al., 2008). To determine whether TOR signaling is upregulated in this context, we assessed 12-week-old fish transferred from standard (five fish per liter) to low-density condition (three fish per 10 l) for 10 days, which resulted in an increased body mass (Fig. S4). We found that this stimulation was not associated with enhanced pS6. Thus, different mechanisms underlie low density-stimulated growth and HRAS^G12V-induced heart enlargement.

To identify how rapamycin prevents a cardiac phenotype, we performed transcriptomic analysis. We tested four experimental groups, namely 4-OHT-pulsed cmlc2/RFP and cmlc2/GFP-HRAS in combination with either rapamycin or DMSO treatment (Fig. 3A). We first examined cmlc2/RFP fish in both treatments. In this control strain, rapamycin-treated hearts were transcriptionally nearly undistinguishable from the DMSO group (Fig. 3B, Fig. S5A,B). This demonstrates few side-effects of 0.5% rapamycin on ventricles.

We then compared both 4-OHT-pulsed transgenic strains after DMSO treatment. We found 8011 differentially expressed genes (DEGs) in HRAS-OE hearts, compared with RFP-OE control (padj<0.05 and absolute Fold change >1.5) (Fig. S5C and Table S1). This shows that oncogenic HRAS induced large transcriptomic changes in the ventricle.

Finally, we compared HRAS-OE hearts between rapamycin and DMSO conditions. The results displayed 2907 DEGs (Fig. S5D and Table S2). Interestingly, most of these genes (2261 DEGs) were inversely deregulated by rapamycin towards the normal expression levels observed in RFP-OE hearts (Fig. 3B-D). Accordingly, we called them ‘rapamycin-reverted genes’ in HRAS-OE hearts, which likely represent candidate factors associated with the rescue from cardiac overgrowth.

Among these rapamycin-reverted genes, 1326 were upregulated and 935 were downregulated in DMSO-treated HRAS-OE hearts (Fig. 3B-D). Among the upregulated subset, we identified an enrichment of immune response, cell cycle regulation and cell adhesion (Fig. 3D and Figs S5E, S6). We focused first on cell proliferation, which is the most evident mechanism promoting tissue growth. We performed PCNA immunostaining and detected cardiac nuclei by myosin light chain 7 (Myl7) and DAPI colocalization. Although almost no PCNA-positive cardiomyocytes were present in the cmlc2/RFP control, cmlc2/GFP-HRAS fish comprised nearly 40% proliferating cardiac nuclei; importantly, this number was decreased to 25% in the rapamycin-treated group (Fig. 3E,F). This finding demonstrates that enhanced cell proliferation contributed to cardiomegaly; this effect was partially reverted by the inhibition of TOR.

We then focused on the downregulated genes in DMSO-treated HRAS-OE hearts (Fig. 3B,D). Here, we identified factors involved in translation, heart contraction, metabolic processes and myocyte differentiation (Fig. 3B,C and Fig. S6). Given that cardiomyocyte dedifferentiation might enhance the proliferative capacity, we assessed expression of an embryonic isoform of cardiac myosin heavy chain (embCMHC) using N2.261 antibody, which immunolabels the developing zebrafish ventricle until 14 days post-fertilization (Sallin et al., 2015). Adult control cmlc2/RFP ventricles lacked this immunoreactivity, consistent with the normal differentiation status of cardiomyocytes (Fig. 3G). By contrast, cmlc2:GFP-HRAS hearts expressed embCMHC in ∼20% of the myocardium, suggesting extensive dedifferentiation of adult cardiomyocytes upon HRAS^G12V overexpression (Fig. 3G). Importantly, this effect was fivefold reduced after rapamycin treatment (Fig. 3G,H). We conclude that HRAS-OE cardiomyocytes were less differentiated and that this state was reverted by TOR inhibition.

The immune response was one of the main biological processes enriched in our transcriptomic analysis (Fig. 3D). To address the involvement of leucocytes in our model, we visualized phagocytes using a L-Plastin antibody (Bise et al., 2019; Morley, 2012). We observed enhanced infiltration of these cells in the HRAS-OE myocardium, and this phenotype was reverted to control levels by rapamycin (Fig. S7).

The recruitment of immune cells could be caused by cellular damage (Di Micco et al., 2006; Primo and Teixeira, 2020). To assess genomic instability, we performed immunofluorescence analysis of phospho-Histon H2AX (γ-H2AX), a biomarker of DNA double-stranded breaks (Martin and Bonner, 2006). We found no enhanced γ-H2AX immunoreactivity in HRAS-OE hearts, compared with control hearts (Fig. S8A-C). Furthermore, this marker was not induced after one pulse of 4-OHT treatment at the onset of HRAS^G12V expression in the myocardium (Fig. S9A-C). Thus, no evidence of genomic instability in cardiomyocytes was identified in our model.

To test whether broader cytotoxic mechanisms are triggered by HRAS^G12V overexpression, we also tested immunoreactivity against activated-Caspase 3 (act-Casp3), a biomarker of cell apoptosis. A few act-Casp3-positive cells were detected only in HRAS-OE hearts, predominantly at the outer layer of the compact myocardium (Fig. S8D,E). No enhanced act-Casp3 immunoreactivity was found after one pulse of 4-OHT treatment (Fig. S9D,E). Thus, tissue damage probably does not represent the main mechanism for immune cell recruitment and cardiomyocyte dedifferentiation in HRAS-OE hearts.

To compare the effect of rapamycin on cardiomyocyte dedifferentiation between the HRAS-OE myocardium and regenerating hearts, we used the cryoinjury model. Although TOR signaling has been investigated in heart regeneration after ventricular resection or genetic cell ablation (Chávez et al., 2020; Miklas et al., 2022), its role in cardiomyocyte dedifferentiation in the peri-injury zone after cryoinjury is still unclear. We focused on the early and late regeneration phases at 7 and 30 days post-cryoinjury (dpci) (Fig. 4A). At 7 dpci, we found that pS6 was induced in cardiomyocytes within 100 µm of the wound margin, and rapamycin treatment abolished this pattern (Fig. 4B). At 14 and 30 dpci, pS6 immunoreactivity declined, suggesting a transient role for TOR signaling during cardiomyocytes dedifferentiation (Fig. S10). Rapamycin treatment reduced both restorative responses, as assessed by PCNA and embCMHC (Fig. 4C-G). Consistently, at 30 dpci, histological staining detected larger fibrotic tissue after rapamycin treatment, compared with control (Fig. 4E,H). We concluded that the rapamycin treatment suppresses the activity of TOR in the peri-injury zone, resulting in impaired regeneration after cryoinjury.

Our experiments suggest that TOR signaling is a common mechanism required for cardiomyocyte activation in response to oncogenic HRAS and cryoinjury. To compare the molecular signature between HRAS-OE and regenerating heart conditions, we added transcriptomic analysis of untreated cmlc2/GFP-HRAS hearts at 7 dpci and the uninjured state (Fig. 5A,B). First, we identified ∼3600 DEGs between uninjured and 7 dpci samples (Fig. S11A,B). The transcriptomes of HRAS-OE and regenerating ventricles were grouped together with 2250 similarly deregulated genes (Fig. 5C-E; Table S3, Fig. S11C). These data suggest that cardiac regeneration and HRAS-driven growth involve unexpectedly many common molecular mechanisms.

Among upregulated genes, we identified numerous factors of the cell cycle and several previously characterized regeneration markers, such as aldh1a2, anln, col12a1a, col12a1b, cxcr4b, itgb3b, junba, junbb, mdka, meis1a, pdgfrb, tagln, tgfb1b, tncb and wnt1b (Fig. S11D-G). Among the metabolic factors, two enzymes of the serine synthesis pathway (SSP) were upregulated, consistent with a recent study on heart regeneration (Ogawa et al., 2021). Furthermore, dozens of genes linked to extracellular matrix and immune response were similarly upregulated in oncogenic HRAS-OE hearts and regenerating hearts (Fig. S12). For further analysis, we selected a non-fibrillar collagen, ColXII, that is known to be upregulated during spinal cord and heart regeneration (Marro et al., 2016; Wehner et al., 2017). Similarly, the HRAS-OE myocardium was abundantly infiltrated with ColXII fibers (Fig. 5F,G). This finding suggests a similar modulation of the matrix composition to create the appropriate microenvironment for myocardial growth of the adult heart.

Among common downregulated genes, we found many sarcomeric contraction family genes, such as myosin light chain kinases, troponin, alpha-actinin and myomesin genes (Fig. 5E; Fig. S11F). Indeed, immunofluorescence analysis validated that Myomesin was downregulated in HRAS-OE hearts and the peri-injury myocardium (Fig. 5H,I). Thus, dedifferentiated cardiomyocytes share similar structural characteristics in the regeneration and oncogene-driven conditions.

To understand the molecular differences between HRAS overexpression versus regeneration, we analyzed genes that were deregulated in the former but not the latter process (Fig. 6A). Using this approach, we found that HRAS-OE hearts distinctively upregulated 1040 genes, many of them are involved in catabolic processes, including proteasomes and cathepsins (Fig. 6B and Table S4). The group of 937 downregulated genes included factors of translation, histone modifications and tumor suppression (Fig. 6C and Table S4). We concluded that beside the mechanistic parallels between both conditions, each of them was also associated with its unique genetic profile.

Among identified genes, two members of the high mobility group (HMG) protein family, hmga1b and hmgb3a, have been uniquely deregulated in HRAS-OE hearts. The roles of HMG genes have been widely investigated in human carcinogenesis (Niu et al., 2020; Wang et al., 2022). Mammalian HMGB1, HMGB2 and HMGB3 share similar amino acid sequences and identical functional regions (Niu et al., 2020). The subcellular distribution of HMGB1 has been characterized, showing that this nonhistone chromatin protein can be transported from the nucleus into the cytoplasm and is released extracellularly to act as a chaperone (Chen et al., 2022; Kwak et al., 2020). Given the involvement of HMGB1, HMGB2 and HMGB3 genes in human cancer, the biochemical similarity of these homologues, their localization-dependent protein functions, and the evolutionary conservation between the human and zebrafish HMGB1 orthologues, we aimed to assess the HMGB1 distribution in oncogenic HRAS-OE zebrafish hearts. We found that in cmlc2/RFP and regenerating hearts, HMGB1 was expressed at a low level in nuclei of some cardiomyocytes and non-myocytes (Fig. 6D,E and Fig. S13). In HRAS-OE hearts, HMGB1 immunoreactivity was enhanced with a dotty pattern, suggesting vesicular distribution (Fig. 6E). We concluded that the non-nuclear localization of HMGB1 is a distinctive feature of the HRAS-OE myocardium, contrasting the regenerative program.

Taking advantage of the conditional regulation of the GAL4-ERT2/UAS system, we aimed to investigate the effects of transient HRAS expression on heart regeneration. To determine whether oncogene-driven growth can impact the regenerative program, we designed an experiment with one 4-OHT treatment before cryoinjury and analyzed hearts at 7 and 14 dpci (Fig. 7A). At 7 dpci, AFOG staining revealed a similar wound tissue in cmlc2/GFP-HRAS and cmlc2/RFP hearts (Fig. 7B). By contrast, at 14 dpci, cmlc2/GFP-HRAS hearts comprised markedly less collagenous tissue, compared with control (Fig. 7C). In four hearts out of eight, a ‘myocardial bridge’ was observed along the wound margin spanning the continuity of the ventricular wall, which is a feature of more advance stages, e.g. 21 dpci (Chablais and Jaźwińska, 2012b; Chablais et al., 2011). This analysis suggests that myocardial replacement was more efficient after transient preconditioning with HRAS^G12V overexpression.

To determine the mechanism of such advanced regeneration, we assessed cardiomyocyte dedifferentiation using the embCMHC antibody. At 7 dpci, cmlc2/GFP-HRAS ventricles displayed a 2.5-fold increase in embCMHC staining with atypical expression along the outer myocardial wall (Fig. 7D,F). At 14 dpci, embCMHC staining was similar between cmlc2/GFP-HRAS and cmlc2/RFP (Fig. 7E,F). We conclude that short-term HRAS expression transiently boosts cardiomyocyte dedifferentiation at the beginning of regeneration.

To investigate cell proliferation, we performed PCNA immunostaining at 7 dpci. We found a twofold increase in PCNA-positive cells, suggesting more efficient regeneration (Fig. 7I,J). Next, we assessed the injury size relative to the ventricular tissue, which was detected by F-actin. At 7 dpci, in both cmlc2/GFP-HRAS and cmlc2/RFP hearts, the injury size was ∼20% of the ventricular section (Fig. 7G). Importantly, at 14 dpci, the proportion of the wounded area was three times smaller in cmlc2/GFP-HRAS hearts than in control (Fig. 7G). The total size of the ventricle was similar between groups (Fig. 7H). This demonstrates that regeneration was not accompanied by extensive growth, consistent with our previous experiments with one pulse of 4-OHT (Fig. S1C-E). We concluded that short-term HRAS^G12V expression preconditioned cardiac restoration through increasing cardiomyocyte dedifferentiation and proliferation at the early phase, without causing extraordinary enlargement of the organ.

We examined the longstanding concept of a link between oncogene-driven hyperplasia and post-traumatic regeneration. This topic has previously been addressed by comparison between non-identical tissue types (Milanovic et al., 2018; Oviedo and Beane, 2009; Sarig and Tzahor, 2017; Wong and Whited, 2020). The power of our approach relies on experiments with the same organ, namely the ventricle of adult zebrafish, which was challenged by either oncogenic or traumatic injury. Another strength of our comparison is the similar temporal scale of experiments in adult organism. To achieve the oncogenic hyperplasia model, we established a cardiac specific, 4-OHT-inducible and reversible transgenic system based on the Gal4-ERT2/UAS transgenic lines. To induce regeneration, we applied the cryoinjury method (Chablais et al., 2011; Gonzalez-Rosa et al., 2011; Schnabel et al., 2011). These two experimental setups allowed the identification of common and distinct mechanisms underlying the response of the zebrafish heart to oncogenic versus regenerative stimuli.

As our oncogene-driven model is novel, we first characterized its suitability (Fig. 8A). Overexpression of GFP-HRAS induced massive heart growth within 16 days, a time-point until which the fish did not display any behavioral abnormality. Transcriptomic analysis identified more than 8000 deregulated genes (Fig. 8B). HRAS-OE cardiomyocytes showed a reactivation of the embryonic cardiac program, as monitored by embCMHC immunofluorescence and enhanced proliferation. Although the HRAS oncogene is predicted to directly account for these effects, we cannot exclude the possibility that secondary processes additionally stimulate the pro-regenerative program. Overall, the aberrant structure of the myocardium with dedifferentiated cardiomyocytes suggests that oncogene-driven cardiac overgrowth represents a hyperplasia model.

We found that this phenotype was efficiently suppressed by rapamycin-mediated inhibition of TOR (Fig. 8C). This result is interesting because previous studies with the same transgene in other tumor models have suggested the requirement of the combined inhibition of TOR and ERK/AKT signaling (Fernandez del Ama et al., 2016; Le et al., 2013). Thus, the downstream effectors of the oncogenic HRAS might be tissue dependent. Interestingly, transcriptomic analysis demonstrated that approximately one-third of HRAS-deregulated genes were reverted towards their normal expression after rapamycin treatment. In particular, we validated that cardiomyocyte dedifferentiation and proliferation processes were reduced. In addition, the recruitment of L-plastin-expressing leucocytes was suppressed by rapamycin treatment, a finding consistent with the mammalian systems (Janes and Fruman, 2009). Altogether, inhibition of just one pathway, TOR, was sufficient to rescue the hyperplastic phenotype, providing a basis for further analysis of the identified candidate genes.

Studies in rodents have demonstrated that oncogenic signaling mediators, such as constitutively active ERBB2 or mutations of Hippo pathway effectors, partially reprogram mammalian cardiomyocytes to a more fetal and proliferative state, which might lead to pathological hypercellularity (D'Uva et al., 2015; Lin et al., 2015; Monroe et al., 2019). In pigs, microRNA therapy can stimulate cardiac repair; however, the dose needs to be tightly controlled to prevent a persistent myoblastic phenotype, leading to detrimental arrhythmia (Gabisonia et al., 2019). A tetracycline transactivator-controlled model has been developed in mice to transiently overexpress HRAS^G12V in the mammalian heart (Wei et al., 2011). Using this approach, overexpression of the HRAS oncogene resulted in pathogenic hypertrophy. Our study expands this topic by examination of TOR signaling, as the downstream target of HRAS, and by transcriptomic comparison between HRAS-OE hearts and regenerating hearts (Fig. 8D). Besides the broad similarities at the level of cardiomyocyte dedifferentiation, metabolic switch and microenvironmental changes, we also identified genes that are differently regulated between both systems, such as hmga1b and hmgb3a. In control and regenerating hearts, HMGB1 was localized at low level in nuclei, whereas in HRAS-OE hearts, this protein was predominantly non-nuclear. In human cells, secretion-ready cytoplasmic HMGB1 has been linked to the regulation of authophagy and stress responses (Kwak et al., 2020; Chen et al., 2022). Thus, it will be interesting to determine the functional significance of this factor in the relevant contexts.

Finally, we addressed a question about the effects of a short-term oncogene expression on heart regeneration (Fig. 8E). A pulse of HRAS expression accelerated regeneration, as evident at 14 dpci. This suggests that preconditioning with transient and reversible overexpression of the HRAS oncogene is beneficial for zebrafish heart regeneration. Our finding is line with a recent study in mice, demonstrating that short-term expression of pluripotency genes (OSKM) can induce cardiomyocyte dedifferentiation and regenerative proliferation (Chen et al., 2021). The effects depend on the level and the duration of transgene overexpression, ranging from pro-regenerative stimulation to heart tumor formation in mice. Thus, the concept of exogenous stimulation with potent factors to reawaken the proliferative program might still represent a valuable approach for regenerative biology and medicine.

Our database offers a platform to further compare the identified factors with other zebrafish and mammalian HRAS-related disease models. These comparative studies might also be relevant for medicine, e.g. in cases of the rare human tumors known as cardiac rhabdomyoma, which are reported in infants (Freedom et al., 2000; Uzun et al., 2007). Some pediatric cardiopathologies, such as Costello syndrome and cardiofaciocutaneous syndrome, are associated with mutations in RAS pathway genes (Kratz et al., 2011). Interestingly, a medical case report found a regression in newborn cardiac neoplasia after treatment with mTOR inhibitors (Sugalska et al., 2021). The zebrafish model could contribute to understanding the biological mechanisms that are relevant for the treatment of symptomatic cardiac rhabdomyomas in children.

Animals used for this study were adult zebrafish, if not specified differently for specific experiments. We used mixed females and males. Wild-type fish were the AB strain (Oregon). A Tg(cryaa:KO2;cmlc2:Gal4-ERT2) transgenic line was generated in this study, as described below. Other previously published lines are: Tg(cmlc2::DsRed2–nuc) (Rottbauer et al., 2002), Tg(UAS:mRFP1) (Asakawa et al., 2008), Tg(UAS:eGFP-H-RAS^G12V) (Santoriello et al., 2009), Tg(cmlc2:nucDsRed) (Rottbauer et al., 2002) and Tg(careg:eGFP) (Pfefferli and Jaźwińska, 2017). Identification of both UAS strains was performed by PCR of the caudal fin tissue: Tg(UAS:mRFP1), primers forward 5′-cgtcatcaaggagttcatgc-3′ and reverse 5′-tggtgtagtcctcgttgtgg-3′; Tg(UAS:eGFP-H-RASG12V), primers forward 5′-AGCTGACCCTGAAGTTCATCT-3′ and reverse 5′-GTACTGGTGGATGTCCTCAAAAG-3′. Tg(cryaa:KO2;cmlc2:Gal4-ERT2) fish were identified by detection of orange fluorescent protein in the lens using a Leica AF M205 FA stereomicroscope. For identification of transgenic fish (tissue biopsy or fluorescence stereomicroscopy), animals were anesthetized with buffered solution of 0.6 mM tricaine (MS-222 ethyl-m-aminobenzoate, Sigma-Aldrich) in system water.

Before heart collection, fish were euthanized. All assays were performed using different animals that were randomly assigned to experimental groups. The exact sample size (n) is described for each experiment in the figure legends and was chosen to ensure the reproducibility of the results.

The animal housing and all experimental procedures were approved by the cantonal veterinary office of Fribourg, Switzerland. All experiments were performed in accordance with relevant guidelines and regulations.

To generate the Tg(cryaa:KO2;cmlc2:Gal4-ERT) line, the pDestTol2crya:KO2 construct was first produced by replacing the Venus cassette of pDestTol2crya:Venus plasmid (kindly provided by Roehl lab) with a PCR fragment of the KO2 reporter [primers (F) 5′-TTGGCGCGCCATGGTGAGCGTGATCAAGCC-3′ and (R) 5′-GGAATTCCATATGTTAGGAGTGGGCCACGGCG-3′] using the restrictions sites AscI and NdeI. The p5E-cmlc2 plasmid was generated by subcloning a PCR fragment of the cmlc2 promoter [primers (F) 5′- GGGGTACCGTGACCAAAGCTTAAATCAGTTGT-3′ and (R) 5′-CGGGATCCGGAGAAGACATTGGAAGAGCC-3′] in the p5E-MSC plasmid (kindly provided by Dr H. Roehl, University of Sheffield, UK) using the KpnI and BamHI restriction sites. The final pDest-cryaa:KO2-cmlc2:Gal4-ERT construct was generated using multisite Gateway assembly of p5E-cmlc2, pME-Gal4-ERT2-VP16 (kindly provided by Dr S. Stewart, University of Oregon, USA) (Akerberg et al., 2014), p3E-SV40polyA (kindly provided by Dr H. Roehl) and pDestTol2cryaa:KO2. Each plasmid was co-injected with the pCS2FA-transposase mRNA into one-cell-stage wild-type embryos (Felker and Mosimann, 2016). Founder fish (F0) were identified based on red fluorescent eyes in Tg(cryaa:KO2;cmlc2:Gal4-ERT2).

For the induction of the Gal4-ERT2/UAS system, adults were incubated in 2.5 μM 4-hydroxytamoxifen (4-OHT, Sigma-Aldrich). This solution was made from a 10 mM stock solution dissolved in DMSO. The duration of treatments was 18 h in the dark at the indicated time-points. Control animals were kept in water with 0.1% DMSO. The mTOR inhibitor rapamycin (Selleckchem) was dissolved in DMSO at a stock concentration of 10 mM and used at a final concentration of 0.5 μM for experiments in Figs 2 and 3 (neoplasia model) and 1 μM for experiments of Fig. 4 (regeneration model). This difference in concentration was justified by the fact that oncogene-overexpressing hearts were also subjected to regular pulse treatments with another chemical, 4-OHT. To avoid any pharmacological stress, we used the minimal concentration of Rapamycin with a functional effect in our assays. The inhibitor of MEK was 0.1 µM PD184352, also called CI-1040 (HY-50295, MedChem Express), the inhibitor of PI3K was 10 µM LY294002 (HY-10108, MedChem Express). Zebrafish were treated with drugs at a density of three adults per 200 ml of water.

For heart cryoinjury, the fish were immersed in analgesic solution of 5 mg/L lidocaine for 1 h before procedure. Ventricular cryoinjuries were performed according to our video protocol (Chablais and Jaźwińska, 2012a). Briefly, anesthetized fish were placed ventral side up on a damp sponge under a stereomicroscope. After chest skin incision, a stainless steel cryoprobe precooled in liquid nitrogen was applied on the ventricle for 23-25 s. To stop the procedure, water was dropped on the tip of the cryoprobe and fish were immediately returned into water. The recovery of fish after the procedure was monitored and assisted. To collect the heart, fish were euthanized in 0.6 mM tricaine solution and on wet ice. The heart was removed from the body, as shown in our video protocol (Bise and Jaźwińska, 2019). For the assessment of the organ size, the dissected hearts were imaged before fixation using a Leica AF M205 FA stereomicroscope.

Hearts were fixed in 4% paraformaldehyde (PFA) overnight at 4°C, followed by washes in PBS (3×10 min each). After fixation, hearts were equilibrated in 30% sucrose at 4°C, embedded in tissue-freezing media (Tissue-Tek O.C.T.; Sakura) and oriented in blocks for sectioning along the transverse planes of the organ. Hearts were sectioned at 16 μm with a cryostat. Sections were collected on Superfrost Plus slides (Thermo Fisher Scientific) and allowed to air dry for ∼1 h at room temperature. The material was stored in airtight boxes at −20°C.

Before use, slides were brought to room temperature for 10 min, the area with sections was encircled with PAP pen (Vector) to keep liquid on the slides and left for another 10 min at room temperature to dry. Slides were then transferred to coplin jars containing 0.3% Triton-X in PBS (PBST) for 10 min at room temperature. The slides were transferred to a humid chamber. Blocking solution (5% goat serum in PBST) was applied on the sections for 1 h at room temperature. Subsequently, sections were covered with ∼200 µl of primary antibody diluted in blocking solution and incubated overnight at 4°C in a humid chamber. They were washed in PBST in coplin jars for 1 h at room temperature and again transferred to the humid chamber for incubation with secondary antibodies diluted in blocking solution. The slides were washed in PBST for 1 h at room temperature and mounted in 90% glycerol in 20 mM Tris (pH 8) with 0.5% N-propyl gallate.

The following primary antibodies were used: mouse anti-tropomyosin at 1:100 (developed by J. Jung-Chin Lin and obtained from Developmental Studies Hybridoma Bank, CH1), rabbit anti-Ras at 1:500 (Abcam, ab52939), mouse anti-PCNA Clone PC10 (Dako, M0879) at 1:500 following antigen retrieval, mouse anti-embCMHC (N2.261) at 1:50 (developed by H. M. Blau, obtained from Developmental Studies Hybridoma Bank), guinea pig anti-ColXIIa (kindly provided by F. Ruggiero, Lyon, France), rat anti-RFP at 1:200 (5F8-10, Chromotek), rabbit Myl7 at 1:200 (GTX128346, GeneTex), mouse anti-A4.1025 at 1:100 (developed by H. M. Blau, obtained from Developmental Studies Hybridoma Bank), rabbit anti-HMGB1 at 1:200 (GTX101277, GeneTex; the immunogen sequence used to generate the HMGB1 antibody is 86% and 88% identical to the corresponding sequence of the zebrafish HMGB1a and HMGB1b proteins, according to the information provided by the company), mouse anti-Myomesin at 1:50 (mMaC myomesin B4; developed by J.-C. Perriard, obtained from Developmental Studies Hybridoma Bank), rabbit anti-pS6 ribosomal protein (phospho-Ser240/244; D68F8) at 1:2000 (5364, Cell Signaling Technology), chicken anti-L-plastin at 1:1000 (kindly provided by Prof. P. Martin, Bristol, UK), rabbit anti-Histone H2A.XS139ph (phospho Ser139) at 1:200 (GTX127342, GeneTex) and rabbit anti-active-Caspase 3 at 1:10000 (ab13847, Abcam). The secondary antibodies (at 1:500) were Alexa conjugated (Jackson ImmunoResearch Laboratories). Phalloidin-CruzFluor-565 (Sigma-Aldrich) was used at 1:500 to label actin filaments. DAPI (Sigma-Aldrich) was applied to detect nuclei.

Aniline Blue, Acid Fuchsin and Orange-G (AFOG) triple staining was performed as previously described (Chablais et al., 2011). The imaging of heart sections was performed using a Zeiss Axioplan2 microscope.

Fluorescent images of sections were captured using a Leica confocal microscope (TCS SP5) and the image J 1.49c software was used for subsequent measurements. Each biological replicate (n) corresponds to one fish. For each fish, three or four non-adjacent ventricular sections were analyzed that were interspaced from each other by ∼100 μm within the organ. For regenerating hearts, the largest sections of the cryoinjured part have been selected for AFOG and immunofluorescence analysis. Error bars correspond to the standard error of the mean (s.e.m.). The significance of differences was calculated using an unpaired two-tailed t-test or a two-way ANOVA with Šidák multiple comparisons test. Statistical analyses were performed with the GraphPad Prism software.

Specific features were measured according to the following parameters: ventricular section area, the total area outlined by the margin of the ventricle; identification of the ventricular myocardium/cardiomyocytes, the area with fluorescent staining of a cardiac marker, i.e. Tropomyosin, A4.1025, Myl7 or F-actin within the ventricular section; fluorescent myocardium, the overlap of the area with GFP or RFP immunoreactivity, and the area labeled with a cardiac marker in the ventricle; Ras-positive myocardium, the overlap of the area with Ras immunoreactivity and the area labeled with a cardiac marker in the ventricle; PCNA quantification, the number of PCNA-positive dots of a nuclear size that overlap with DAPI and a cardiac marker; embCMHC quantification, the overlap of N2.261 immunoreactivity with the area labeled with a cardiac marker in the ventricle; pS6 quantification, the overlap of pS6 immunoreactivity with the area labeled with a cardiac marker in the ventricle; L-plastin quantification, the area of L-plastin immunoreactivity within the area of ventricular section.

RNA was isolated from three pooled ventricles dissected in Hanks Buffered Saline Solution supplemented with 5 mg/ml heparin (Sigma-Aldrich). Three biological replicates were prepared for each experimental condition. Total RNA was extracted using QIAzol Lysis Reagent (QIAGEN) and purified using the RNeasy Micro Kit (QIAGEN). A DNase digestion (RNase-free DNase set, QIAGEN) was performed to eliminate genomic DNA. The concentration and quality of RNAs were measured using a TapeStation system (Agilent). cDNA was synthesized and amplified using the SMART-seq Low Input RNA kit for Sequencing. RNAseq libraries were prepared using Nextera XT DNA Library Preparation Kit (Illumina) and were sequenced using the Illumina NovaSeq 6000 system.

The quality of the sequencing data was assessed using fastqc v. 0.11.5 (Andrews. S. 2010; Babraham Bioinformatics; https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) and RSeQC v. 2.6.4 (Wang et al., 2012). Reads were mapped to the reference genome GRCz11.94 using HiSat2 v. 2.1.0 (Kim et al., 2015). PCR duplicates were removed using UMI-Tools v1.1.1 (Smith et al., 2017). The number of reads aligning to genes was counted using FeatureCounts v. 1.6.0 (Liao et al., 2014) and differential expressed genes were identified using DESeq2 (Love et al., 2014). Significant differentially expressed genes were defined as those with absolute fold change greater than 1.5 and a Benjamini-Hochberg adjusted P-value of less than 0.05. To identify common gene expression in regeneration and in cardiac neoplasia, a multi-factor differential gene expression analysis was carried out using DESeq2. In this analysis the treated samples (7 dpci and HRAS, DMSO) were compared with the controls (uninjured and RFP, DMSO) with the system (regeneration and neoplasia) as co-factor. Principal component analysis, volcano plots and heatmaps were generated using the ggplot2 v 3.3.5 package of R. Gene Set Enrichment Analysis and Gene Ontology enrichment analysis were performed using the Bioconductor ClusterProfiler package v 3.18.1 (Yu et al., 2012). GO terms with an adjusted P-value <0.05 were considered as significantly enriched pathways.

We thank V. Zimmermann for excellent technical assistance and for fish care, and Dr P. Nicholson (NGS Platform, University of Bern) for RNA-sequencing experiments. Confocal microscopy was performed at the Bioimage Core Facility (University of Fribourg). We are grateful to Prof. C. Lengerke (University of Tübingen) and Prof. M. Mione (University of Trento) for sharing transgenic fish lines and for the initiation of this study; to Prof. F. Ruggiero (Institut de Génomique Fonctionnelle de Lyon) for providing ColXII antibody; and to Prof P. Martin (University of Bristol) for L-plastin antibody.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.201412.reviewer-comments.pdf