Reactivation of Notch signaling is required for cardiac valve regeneration

Cardiac Valve Disease is one of the most common heart disorders with an emerging epidemic of cardiac valve degeneration due to aging. Zebrafish can regenerate most of their organs, including their heart. We aimed to explore the regenerative potential of cardiac valves and the underlying molecular mechanisms involved. We used an inducible, tissue-specific system of chemogenetic ablation and showed that zebrafish can also regenerate their cardiac valves. Upon valvular damage at larval stages, the intracardiac flow pattern becomes reminiscent of the early embryonic stages, exhibiting an increase in the retrograde flow fraction through the atrioventricular canal. As a result of the altered hemodynamics, notch1b and klf2a expression are ectopically upregulated, adopting the expression pattern of earlier developmental stages. We find that Notch signaling is re-activated upon valvular damage both at larval and adult stages and that it is required during the initial regeneration phase of cardiac valves. Our results introduce an animal model of cardiac valve specific ablation and regeneration.

in adult stages are not available to surgery, and there are no available valve specific promoters, we developed an inducible two component system using specific GAL4 driver lines to express the Nitroreductase gene in valve cells and induce their damage by adding Metronidazole in the fish water. We found that reactivation of the Notch signaling pathway, following valvular damage is necessary for the initiation of valve regeneration in both larval and adult stages.

Results
An inducible valve ablation system reveals the regenerative potential of zebrafish cardiac valves. There is currently a lack of well-characterized promoters for valve specific expression pattern. We have conducted a large-scale screen for the GAL4 transgenic fish using the gene trap and enhancer trap methods at the National Institute of Genetics, Mishima, Japan. In the course of the screen, we identified two cardiac valve driver lines with different pattern and intensity of expression within the cardiac valves Tg(hspGFFDMC73A) and Tg(gSAIzGFFD703A). Tg(hspGFFDMC73A) drives, at 72hpf, robust expression in both Valve Endocardial Cells (VECs) (cuboidal cells at the AV boundary, arrowheads in Supplementary Fig. 1B) and future Valve Interstitial Cells (VICs) which at these stages largely overlap with the TCF positive, mesenchymal looking cells (arrows in Supplementary Fig. 1C) and this expression pattern remains on up to adulthood. Tg(gSAIzGFFD703A) shows weaker expression and is restricted to VECs at 72hpf (Supplementary Fig. 1D-F). We crossed both driver lines to the Tg(UAS-E1b:NfsB-mCherry) to test the regenerative potential of cardiac valves initially at larval stages. We optimized a metronidazole-induced ablation of valve cells at 96 hours post fertilization (hpf) (Fig. 1A), which resulted in the reproducible ablation of >80% NTR positive cells ( Fig. 1 and Supplementary Fig. 2A). In order to test if cardiac valve cell ablation actually had the predicted effect on cardiac function, we quantified the intracardiac blood flow dynamics and confirmed an increase (4,91 folds) in the retrograde flow fraction of the hemodynamic pattern (Supplementary video 1 untreated, 2 ablated quantified in Fig. 1E,I and Supplementary Fig. 3A). The flow pattern of the embryos with valvular ablation is reminiscent of earlier stages (72 hpf) during valve  Supplementary Fig. 1F). The ablation was confirmed to be mediated via apoptosis, as detected in the MTZ treated embryos with TUNEL assay (Supplementary Fig. 4).We washed off metronidazole and followed larvae for the following eight days. We identified that GFP and mCherry positive cells started reappearing already at 2 days post ablation ( Fig. 2A,B compare with 2D,E) and they were comparable to the untreated larvae by eight days post ablation ( Fig. 2C compare with 2F and quantified in Supplementary Fig. 2B). Concerning the flow pattern at this developmental stage, no differences were observed between retrograde blood flows for both untreated (Supplementary video 3) and recovering from treatment embryos (Supplementary video 4 and quantified in Supplementary Fig. 3D).

Notch1 is reactivated ectopically upon valvular damage and is necessary for valve regeneration.
In order to understand the underlying mechanisms of cardiac valve regeneration, we used several transgenic lines and in situ hybridization experiments. We identified that the Notch reporter line showed ectopic upregulation of expression following valvular damage. Both the shear-stress sensitive kruppel-like factor 2a (klf2a) and notch1b are ectopically upregulated 24 hours following valvular ablation ( Supplementary Fig. 5,B, D compare with 5,A,C). In the destabilized version of the notch reporter line Tg(TP1:VenusPEST), Notch signaling was activated in endocardial cells adjacent to the damaged area ( Fig. 3B), while in the Tg(TP1:h2bmCherry) the expression domain of mCherry positive cells is expanded throughout the endocardium ( Fig. 3B compare with 3A), since it represents the accumulating activation of ectopic Notch signaling over time. When DAPT was added in the water during the regeneration phase, no ectopic Notch activation was observed as expected. In addition, the regeneration process also halted, as monitored by the lack of reappearing UAS-E1b:NfsB-mCherry positive cells at the valve region ( Fig. 3E compare with 3D, 3C and 3H, compare with 3G, 3F and quantified in Supplementary Fig. 6).  Fig. 7). We also dissected Tg(hspGFFDMC73A/UAS-E1b:NfsB-mCherry) hearts that carry the Tg(TP1:VenusPEST) transgene and showed that Notch is upregulated at the valve region (Supplementary Movies 7,8 and Supplementary Fig. 8,C,D compare with S8A,B, quantified in S8E). We allowed the fish to recover in system water or system water with 5 μM DAPT. Adult animals that were allowed to recover

Discussion
In this study, we describe the ability of cardiac valves to regenerate after chemogenetic ablation in larval and adult zebrafish. Moreover, we introduce 2 new transgenic zebrafish lines with valve expression promoters. These lines could further contribute to valve specific expression or silencing of genes of interest. Recent advances on imaging and the identification of novel cardiac valve mutants and gene networks, have helped deciphering the effect of intracardiac blood flow dynamics and shear stress on the endothelial cells that are destined to become valvular. Klf2a is the best-characterized flow sensitive transcription factor to date 6,21,22 . Some of its key downstream targets include the Cerebral Cavernous Malformations proteins (CCM) 23,24 and Notch that are very important for proper valve morphogenesis 1,11,12,[25][26][27] . Endocardial notch1b activation requires functional primary cilia 28 and klf2a 6,21,22 . Klf2a has been recently shown to be also required for myocardial trabeculation integrity during development via the modulation of Fgf signaling 29 as well as for myocardial reprogramming via its well-established endocardial hemodynamic response and Notch mediated function 30 . The CCM proteins appear to function antagonistically to the activation of β1 integrin by shear stress 31 . Knockdown of β1 integrin suppresses the cardiovascular defects www.nature.com/scientificreports www.nature.com/scientificreports/ of ccm mutant embryos 23 . In addition to the intracardiac flow dynamics and endocardial/myocardial interactions at the valve-forming region, it is worth noticing that there is extensive extracellular matrix components (ECM) also known as cardiac jelly. One of its major components is hyaluronic acid, produced by the Has2 enzyme that is tightly regulated during AV development to restrict the AV region via the BMP signaling pathway 32 as well as via mir23 33 . Wnt signaling is also required for valve development 34,35 . Fibronectin synthesis has been shown to be flow dependent and downstream of Klf2a, coupling the mechanosensory system to ECM composition 36 .
It is becoming clear that valve regeneration would require several steps, including the proliferation of endocardial cells, their transformation to interstitial cells and the tightly regulated production of several ECM components. Therefore, cardiac valve tissue engineering (CVTE) would require the combination of optimized biomaterials with different cell types and conditioning protocols making it a very challenging process. Despite the challenges, CVTE is expected to be a promising therapeutic alternative to mechanical and bioprosthetic valves. Both of these types require life-long anticoagulative therapies and accumulate damages due to the stressful hemodynamic microenvironment of cardiac valves, since they do not contain any living cells with adaptive or regenerative potential. Such potential is particularly vital in pediatric patients whose cardiac valves need to adapt to the growing size of their hearts. The optimization of recellularization protocols to accommodate the need for replacement valves that could grow/adapt with somatic outgrowth requires the best understanding of valve development and regeneration mechanisms 20,37 . Here we addressed the initial step of valve regeneration that results from the functional consequence of a dysfunctional valve: the increase of retrograde blood flow. We propose a system where the immature flow patterns are sensed by the endocardial cells, and the increase of the retrograde intracardiac flow pattern, due to a dysfunctional valve, could be the stimulus for its regeneration. Following injury, the mechanosensitive transcription factor klf2a is upregulated, and a Notch dependent developmental program is activated for valve regeneration. This is reminiscent of the embryonic expression pattern of these signaling pathways (Fig. 5). Recapitulating development is a recurrent scenario during the regeneration process of different organs. Studies of in vitro cell systems that incorporate biomechanical stress and conditioning www.nature.com/scientificreports www.nature.com/scientificreports/ as well as in silico computational modeling of hemodynamics for tissue-engineered heart valve shape optimization 37 would greatly benefit from in vivo experimental systems, such as the zebrafish, that show endogenous regenerative potential of their cardiac valves.

Methods
Zebrafish transgenic lines. gSAIzGFFD703A and hspGFFDMC73A zebrafish lines were generated during a large-scale screen for zebrafish transgenic Gal4 lines by the gene trap and enhancer trap methods at the Kawakami lab 38,39 . More information on the insertion site and expression pattern can be found at http:// kawakami.lab.nig.ac.jp/ztrap/. The transgenic lines Tg(UAS:eGFP1A) 38 and Tg(UAS-E1b:NfsB-mCherry) c26418 , which we will refer to as NfsB-mCherry for short in figures, were crossed with the above lines to create double or triple transgenic lines. hspGFFDMC73A/Tg(UAS-E1b:NfsB-mCherry) c264 double transgenics were crossed with Tg(Tp1:venus-PEST) and/or Tg(Tp1:h2B:mCherry) 40 in order to create appropriate Green/Red combinations of the transgenics to distinguish Notch signaling reporter expression following valve cell ablation. Tg(7xTCF-Xla. Siam:nlsmCherry) ia5 line 35 was also used to image expression pattern and the overlap of mesenchymal-like TCF positive cells with gSAIzGFFD703A line.

Metronidazole (Mtz) treatment of zebrafish embryos and adults. Mtz (Sigma) was diluted in
embryo water/0,2%DMSO in a final concentration of 4 mM for hspGFFDMC73A/UAS-E1b:NfsB-mCherry 3 dpf embryos and at 10 mM for gSAIzGFFD703A/UAS-E1b:NfsB-mCherry 3 dpf embryos. 0,2% DMSO in embryo water was used as a control. Mtz treatment (concentration and exposure time) was optimized for the two lines and the different developmental stages to ensure reproducibility between larvae and adults as well as >80% of successful elimination of NTR+ cells. Embryos were treated for 12 and 20 hours for the 73 A and 703 A lines, respectively. Embryos were then washed-off Mtz and left for recovery for 8days. For adult transgenic fish hspGFFDMC73A/ UAS-E1b:NfsB-mCherry Mtz treatments, fish were treated with 5 mM in system water for 12 hours and left for the mentioned interval to recover before proceeding to heart extractions. Euthanasia was carried out by prolonged immersion in water with an overdose of tricaine methane sulfonate (MS222, 300 mg/l).
DAPT treatments. 4 dpf fish were treated in 70 μΜ DAPT diluted in EW/1%DMSO until 6 dpf, where they were fixed and immunostained. Adult DAPT treatments took place as previously described with 5 μΜ DAPT final concentration in system water. www.nature.com/scientificreports www.nature.com/scientificreports/ Immunofluorescence. Zebrafish embryos were fixed in 4% PFA and stained using the Zn5 (Dm:grasp, Alcama) Ab with Alexa anti-mouse 633 as a secondary antibody. Embryos were then embedded in 4% agarose and sectioned in a vibratome at 180 μΜ sections. Adult hearts were extracted, fixed in 4%PFA, embedded, sectioned and then stained on 180 μΜ vibratome sections. Tropoelastin Ab was previously described 41 with Alexa anti-rabbit 633 used as the secondary Ab.
Zebrafish embryos' live imaging and retrograde flow fraction quantification. 5 dpf embryos were anesthetized in tricaine, embedded in 1% low melting agarose and imaged under a Leica confocal microscope. For brightfield videos of 4 dpf embryos ORCA Hamamatsu camera was used and quantification of flow fractions during independent heart beats was measured as previously described 7 . Bitplane Imaris was used to get 3D videos of stacks of adult zebrafish cardiac valve confocal images.