Cardiac injury modulates critical components of prostaglandin E2 signaling during zebrafish heart regeneration

The inability to effectively stimulate cardiomyocyte proliferation remains a principle barrier to regeneration in the adult human heart. A tightly regulated, acute inflammatory response mediated by a range of cell types is required to initiate regenerative processes. Prostaglandin E2 (PGE2), a potent lipid signaling molecule induced by inflammation, has been shown to promote regeneration and cell proliferation; however, the dynamics of PGE2 signaling in the context of heart regeneration remain underexplored. Here, we employ the regeneration-competent zebrafish to characterize components of the PGE2 signaling circuit following cardiac injury. In the regenerating adult heart, we documented an increase in PGE2 levels, concurrent with upregulation of cox2a and ptges, two genes critical for PGE2 synthesis. Furthermore, we identified the epicardium as the most prominent site for cox2a expression, thereby suggesting a role for this tissue as an inflammatory mediator. Injury also drove the opposing expression of PGE2 receptors, upregulating pro-restorative ptger2a and downregulating the opposing receptor ptger3. Importantly, treatment with pharmacological inhibitors of Cox2 activity suppressed both production of PGE2, and the proliferation of cardiomyocytes. These results suggest that injury-induced PGE2 signaling is key to stimulating cardiomyocyte proliferation during regeneration.

In situ hybridizations. Hearts were processed as described above, and serial sections were subjected to in situ hybridization studies with DIG-labelled RNA probes directed against cox2a, cox2b and ptger2a using previously described methods 5,41 . cDNA fragments corresponding to the first 900-bp of each gene were synthesized (www.IDTDNA.com) and cloned into the pMiniT vector (www.NEB.com). Antisense probes were synthesized with either SP6 or T7 Polymerase using the Roche DIG Labelling Kit (SP6/T7) in accordance to the manufacturer's suggested protocol (www.Roche.com). Negative controls included cox2a riboprobe but no secondary antibody, and secondary antibody only. Images were captured using settings described under the Immunohistochemistry methods section.
Gene expression studies of whole heart tissues. Ventricles were isolated in ice-cold PBS, placed immediately in Ambion TRIzol Reagent (Invitrogen -Thermo Fisher Scientific, Waltham, MA #2302700), and homogenized using an electric homogenizer. RNA
Liquid chromatography and mass spectrometry. An ACQUITY UPLC BEH C18 column, 130 Å, 1.7 μm, 2.1 mm × 100 mm with in-line filter (Waters, Milford, MA) was equilibrated at 35 °C at a flow rate of 300 μl/ min with LC solvent A on an ACQUITY UPLC H-Class system (Waters, Milford, MA). A small portion of each sample, 30 μl, was loaded onto the column at a flow rate of 300 μl/min of 100% LC solvent A followed by a linear gradient to 20% LC solvent B (acetonitrile-isopropanol(50:50, v/v)) over six min, then increased to 55% LC solvent B over 0.5 min and 100% LC solvent B over 5.5 min and held for four min. Column was primed with 100% LC solvent A for three min between samples. Metabolites were directed to an ACQUITY TQ detector mass spectrometer equipped with an electrospray ionization source set to negative ion mode (Waters, Milford, MA). Specific analytes were detected via multiple reaction monitoring and quantified using a standard curve of known concentration using MassLynx software (Waters, Milford, MA). Each standard (Cayman Chemical, Ann Arbor, MI) and internal standard was resolved at the follow retention time (RT) with the following mass transitions: arachidonic acid, RT

Cardiac injury triggers an elevation in PGE 2 .
Prostaglandins are powerful lipid signals synthesized at the site of injury that regulate the inflammatory response. To profile cardiac prostaglandins and quantify changes associated with injury, we used Liquid Chromatography Tandem Mass Spectrometry (LC-MS/MS). In uninjured hearts, PGE 2 concentrations were significantly higher than those detected for other prostanoids. PGD 2 and 6-keto PGF 1α , a stable metabolite of prostacyclin, barely registered above background levels, while TXB 2 , a Thromboxane A metabolite, and PGF 2α , were undetected ( Fig. 1A). To define prostaglandin levels during regeneration, we amputated ~20% of the adult ventricle, allowed regeneration to proceed for 3 days, and extracted ventricles for analyses. At 3 days post-amputation (dpa), concentrations of PGE 2 were again, significantly higher than all other prostanoids examined (Fig. 1B). Furthermore, we found that at 3 dpa, concentrations of PGE 2 increased by more than a 60% relative to uninjured hearts (Fig. 1C). Together, these experiments identify PGE 2 as the most abundant prostanoid in the zebrafish heart, and establish an injury-induced increase in PGE 2 concentrations.
Enzymes critical to PGE 2 synthesis are upregulated in regenerating adult hearts. Having shown that PGE 2 is elevated in the heart after injury, we next asked how enzymes critical to PGE 2 synthesis are modulated during regeneration. COX enzymes catalyze the first rate-limiting step in prostanoid synthesis. Mammals have two COX isozymes, constitutively expressed COX1, and inducible COX2. Three Cox isozymes have been identified in the zebrafish, a single ortholog of mammalian COX1, and two orthologs of mammalian COX2: Cox2a and -2b 43 . Downstream of the Cox enzymes, prostaglandin E synthase (Ptges), the zebrafish ortholog of mammalian PTGES, catalyzes the terminal step of PGE 2 biosynthesis 44 .
To characterize relative expression of the Cox enzymes, we conducted qPCR studies in uninjured and 3 dpa regenerating hearts. These assays showed that in uninjured ventricles, cox1 expression was significantly elevated relative to both cox2a (>log 2 6.5-fold) and cox2b (>log 2 2.5-fold) ( Fig. 2A). At 3 dpa however, cox2a levels had increased more than log 2 2.5-fold relative to uninjured hearts. No significant changes were observed in cox2b or cox1 transcripts (Fig. 2B). Mirroring the response of cox2a, expression of the terminal prostaglandin synthase ptges was also significantly upregulated (>log 2 0.5-fold) after amputation injury when compared to uninjured levels (Fig. 2C).
To define the spatial distribution of cox2a and cox2b during heart regeneration, we performed in situ hybridizations on uninjured, 1, 2 and 3 dpa regenerating hearts using DIG labelled RNA probes. While cox2a expression was observed in CMs at all time points, expression became enriched within the resection injury zone at 2 and 3 dpa (Fig. 2D). This localized signal suggests potential expression in immune, epicardial and/or endocardial cells. Expression of cox2b displayed a more uniform pattern and intensity across the early stages of heart regeneration www.nature.com/scientificreports www.nature.com/scientificreports/ (Fig. 2E). These signals, however, were absent when heart sections were hybridized with cox2a riboprobe in the absence of an anti-DIG antibody (Fig. 2D) or with anti-DIG independently (Fig. 2E).
Together, these results demonstrate that the zebrafish dynamically responds to cardiac injury by upregulating both cox2a and ptges, two enzymes critical to PGE 2 synthesis. Importantly, increased expression of these genes correlates directly with increased levels of PGE 2 observed in ventricles subsequent to injury.
Cox2a expression is highest in epicardial cells at 3 dpa. Multiple cell types contribute to the injury-stimulated production of prostaglandins, the end products of Cox activity. To localize expression of the Cox enzymes in the injured heart, and identify potential sites of PGE 2 synthesis, we utilized Fluorescence-Activated Cell Sorting (FACS) to isolate distinct cell populations for qPCR studies. CMs, epicardial cells, endocardial/ endothelial cells, and macrophages were sorted from the dissociated ventricles of Tg(cmlc2:EGFP); (tcf21:DsRed), Tg(fli1a:EGFP), and Tg(mpeg1.YFP) strains, respectively (Fig. 3A-C). Purity of the sorts was validated by qPCR, which revealed a log 2 6 to 7-fold enrichment of cell-specific markers in fluorescent(+) vs. fluorescent(−) cells (Fig. 3D).
qPCR analysis showed that at 3 dpa, cox2a expression was lowest in cmlc2(+) CMs and fli1a(+) endocardial/endothelial cells. Relative to cmlc2(+) cells, we observed a trend towards increased levels of cox2a levels in mpeg1(+) macrophages, which have classically been associated with cox2 expression 45,46 . Unexpectedly, we found cox2a expression was ~log 2 2-fold higher in tcf21(+) epicardial cells relative to macrophages (Fig. 3E). Expression of cox2b, which showed a muted response to injury, was highest in fli1a(+) cells (Fig. 3F). There was no significant difference in cox1 expression among the resident cardiac cells assayed; however, Cox1 expression in mpeg1(+) cells was significantly lower than that observed in fli1a(+) and tcf21(+) cells (Fig. 3G). From these studies, we identified tcf21(+) epicardial cells as the primary site of inducible cox2a expression during the early stages of zebrafish heart regeneration. Injury activates a differential shift of PGE 2 receptor expression in the heart. Having shown that the zebrafish heart responds to injury by upregulating the expression of cox2a and downstream production of PGE 2 , we next examined the expression of PGE 2 receptors. Mammals express four G protein-coupled PGE 2 receptors, EP1, EP2, EP3, and EP4. Zebrafish orthologs of the mammalian EP1-4 receptors are Ptger1a, Ptger2a, www.nature.com/scientificreports www.nature.com/scientificreports/ Ptger3, and Ptger4a, respectively. Examination of receptor expression in the uninjured heart demonstrated that the level of ptger3 was higher by a log 2 factor of ~2 relative to both ptger2a and ptger4a (Fig. 4A). Ptger1a was undetected in the ventricles (data not shown).
Injury induced a differential shift in receptor expression, and at 3 dpa, ptger2a was upregulated by more than log 2 0.6-fold while ptger3 was downregulated by more than log 2 1-fold relative to uninjured hearts. We observed no significant change in expression for ptger4a (Fig. 4B). In situ hybridization studies revealed ptger2a expression was similar to the spatial dynamics of cox2a levels. Ptger2a was expressed throughout the heart and enriched within the wounded apex in 1, 2 and 3 dpa hearts (Fig. 4C). These results demonstrate that in the zebrafish, cardiac injury activates a dynamic shift in PGE 2 receptor expression, upregulating the proliferation-associated EP2 ortholog ptger2a, while downregulating the functionally opposing receptor, ptger3.

Activation of the Cox2-PGE 2 circuit stimulates cardiomyocyte proliferation. PGE 2 has been
shown to promote cell proliferation in multiple contexts. To better understand the impact of Cox2 activity on PGE 2 synthesis and CM proliferation, we subjected animals to ventricular amputation and treated them with either NS-398, a small molecule shown to selectively inhibit Cox2 activity in the zebrafish 47 , or DMSO control. At 3 dpa, ELISA assays demonstrated that PGE 2 concentrations were reduced by more than 47% in NS-398 treated hearts relative to controls (Fig. 5A). To determine effects of suppressed Cox2 activity upon heart regeneration, we quantified CM proliferation indices at 3 dpa, a time during regeneration previously demonstrated to exhibit high levels of proliferative activity 48 . Immunostaining for Myocyte enhancer factor −2 (Mef2-green), a nuclear CM marker, and Proliferating cell nuclear antigen (Pcna-red), a nuclear marker for proliferating cells, showed that NS-398 treatment suppressed CM proliferation by approximately 70% relative to controls at 3 dpa (Fig. 5B-D). biological replicate for each cell type. Student's t-test *P < 0.05) (E) qPCR studies in 3 dpa FACS sorted cells showed cox2a expression is significantly higher in tcf21(+) cells, relative to all other cell types examined. (F) qPCR of FACS sorted cells showed that at 3 dpa, cox2b expression was significantly higher in fli1a(+) cells relative to all other cell types examined. (G) There was no significant difference in cox1 expression among the resident cardiac cells assayed. cox1 expression in mpeg1(+) cells was significantly lower than that observed in fli1a(+) and tcf21(+) cells. Gene expression was calculated relative to cmlc2(+) cells. (mean = ±s.e.m. n = 2-5 biological replicates; 12-45 pooled ventricles per replicate. One-way ANOVA followed by Tukey's multiple comparisons test. *P < 0.05, **P < 0.01, ***P < 0.001). (2020) 10:3095 | https://doi.org/10.1038/s41598-020-59868-6 www.nature.com/scientificreports www.nature.com/scientificreports/ To confirm these findings, we examined the effects of another selective Cox2 inhibitor, Celecoxib. At 3 dpa, Celecoxib treatment also had a significant effect, reducing CM proliferation indices by almost 50% relative to controls (Fig. 5E). Collectively, these studies demonstrate that, in the injured zebrafish heart, Cox2 activity drives both PGE 2 synthesis and CM proliferation during the early stages of heart regeneration.

Discussion
Prostaglandin E 2 (PGE 2 ) is a potent inflammatory mediator, with pleiotropic effects. In the present study, we have identified a pro-regenerative role for PGE 2 during cardiac regeneration in adult zebrafish. After ventricular amputation, the injured zebrafish heart revealed increased levels of PGE 2 , as well as upregulated expression of cox2a and ptges, two enzymes critical to PGE 2 synthesis (Figs. 1, 2). Importantly, pharmacologic inhibition of Cox2 activity suppressed both PGE 2 and CM proliferation, supporting a crucial role for the Cox2-PGE 2 circuit in initiating the regenerative response (Fig. 5).
In addition to the wound environment, we unexpectedly identified the epicardium as a potential source of inflammation-associated prostaglandin signaling in the regenerating heart. During the inflammatory response, inducible Cox2 has canonically been associated with immune cells recruited to the site of injury, notably macrophages 19 . However, during conditions of homeostasis, Cox2 is constitutively maintained in some tissues including the vascular endothelium, where it supports the synthesis of prostanoid vasodilators 49,50 . The zebrafish genome encodes two Cox2 genes, Cox2a and −2b. During regeneration, FACS studies revealed expression of the most inducible Cox enzyme, cox2a, was highest in the epicardium, while more constitutively expressed cox2b was most prominent in endocardial/endothelial cells (Fig. 3). These findings raise the intriguing possibility that Cox2a and Cox2b have divergently evolved in specific tissues, such that Cox2b is expressed in endothelial cells to maintain vascular tone, while in the epicardium, Cox2a facilitates a rapid inflammatory response.
An emerging role for the epicardium as a source of inflammatory signaling has been highlighted in both mammalian and zebrafish models. In mice, epicardial cells were found to be a source of YAP-mediated IFNγ production following myocardial infarction, orchestrating the recruitment of immune-suppressive regulatory T cells (T regs ) 11 . PGE 2 has been implicated in both YAP activation 36 , and T reg recruitment 17 , suggesting that in the epicardium, crosstalk between these pathways could mediate the inflammatory response. Our current work supports this study, identifying the epicardium as a potential locus of reparative, inflammatory signaling.
In comparison to our qPCR studies (Figs. 2B and 4B), in situ hybridizations revealed more uniform expression for cox2a, −2b and ptger2a in response to injury (Figs. 2D,E and 4C). One potential explanation for the expression differences observed between these two methodologies is their relative sensitivity. Highly sensitive qPCR assays are more likely to amplify subtle differences in transcript levels, compared to the semi-quantitative nature of in situ hybridizations. Interestingly, we also noted that while in situ studies identified cox2 transcripts in both injury and remote zones of the adult heart, CM proliferation appeared to be limited to the wound area (Fig. 5B). It is possible this localized proliferation may be attributable to additional factors, either related, or unrelated to Cox2 activity, that are spatially restricted during regeneration. . Injury activates a differential shift in PGE 2 receptor expression in the heart. (A) qPCR determination of PGE 2 receptor expression showed that in uninjured hearts, ptger3 levels were significantly higher than ptger2a or ptger4a. Gene expression was calculated relative to ptger2a. (mean ± s.e.m. n = 4-5 biological replicates; 3-5 pooled ventricles per replicate. One-way ANOVA followed by Tukey's multiple comparisons test. ***P < 0.001). (B) Relative to uninjured hearts, expression of ptger2a was significantly upregulated, while ptger3 was significantly downregulated at 3 dpa. (mean ± s.e.m. n = 5 biological replicates; 3-5 pooled ventricles per replicate. Student's t-test. *P < 0.05). (C) Representative in situ hybridization images of ptger2a expression in uninjured, 1, 2 and 3 dpa hearts. (n = 4 biological replicates, brackets mark approximate amputation zone; arrowheads demarcate signal within the injury zone). (2020) 10:3095 | https://doi.org/10.1038/s41598-020-59868-6 www.nature.com/scientificreports www.nature.com/scientificreports/ In the downstream PGE 2 signaling circuit, we found that PGE 2 receptor expression was dynamically modulated in the regenerating heart. PGE 2 signals through the autocrine and paracrine activation of four G protein-coupled receptors to initiate diverse downstream pathways. Activation of EP2, the ortholog of zebrafish Ptger2a, directly promotes cell proliferation in the context of regeneration and cancer 23,51 , polarizing neutrophils and macrophages towards a reparative phenotype 18,19 . By contrast, EP3, the ortholog of zebrafish Ptger3, appears to activate pathways in opposition to EP2, inhibiting cell proliferation 25,26 and exacerbating pathologic inflammation 13,52 . Our profiling studies of PGE 2 receptor expression revealed that injury drives the upregulation of ptger2a and reciprocal downregulation of ptger3 in cardiac tissues (Fig. 4), suggesting that in the zebrafish heart, PGE 2 signaling is directed towards a restorative response.
Whether modulation of the PGE 2 circuit after injury has an enduring impact on cardiac regeneration remains an open question. Our work showed that in the regenerative zebrafish heart, injury triggers the upregulation of genes critical to PGE 2 synthesis (Fig. 2). We furthermore demonstrated that activation of the PGE 2 signal is essential to stimulate CM proliferation, a major driver of heart regeneration (Fig. 5).
In other model systems, PGE 2 has been shown to directly promote the proliferation of multiple cell types, including skeletal muscle stem cells 28 and primary human CMs 15 . However, a large body of work also supports a role for PGE 2 in governing the recruitment, retention, and pro-regenerative polarization of neutrophils, macrophages, and T cells 17,18,53 . Therefore, it is likely that PGE 2 stimulates CM proliferation directly, as well as indirectly, through orchestrating restorative immune cell activity. This study, documenting the injury-induced modulation of PGE 2 signaling during cardiac regeneration, seeds the field for future research to define the mechanisms through which PGE 2 promotes the early, regenerative response. (E) Reduction of CM proliferation was greater than 54% in animals treated daily with Celecoxib, relative to controls. (mean ± s.e.m. n = 7-9 hearts from clutchmates; three sections were quantified per heart and results averaged. Student's t-test. *P < 0.05).