V-ATPase proton pumping activity is required for adult zebrafish appendage regeneration

The activity of ion channels and transporters generates ion-specific fluxes that encode electrical and/or chemical signals with biological significance. Even though it is long known that some of those signals are crucial for regeneration, only in recent years the corresponding molecular sources started to be identified using mainly invertebrate or larval vertebrate models. We used adult zebrafish caudal fin as a model to investigate which and how ion transporters affect regeneration in an adult vertebrate model. Through the combined use of biophysical and molecular approaches, we show that V-ATPase activity contributes to a regeneration-specific H+ ef`flux. The onset and intensity of both V-ATPase expression and H+ efflux correlate with the different regeneration rate along the proximal-distal axis. Moreover, we show that V-ATPase inhibition impairs regeneration in adult vertebrate. Notably, the activity of this H+ pump is necessary for aldh1a2 and mkp3 expression, blastema cell proliferation and fin innervation. To the best of our knowledge, this is the first report on the role of V-ATPase during adult vertebrate regeneration.

Although humans are unable to regenerate after severe organ loss or amputation of body parts, other metazoans have such a capacity. The teleost Danio rerio (zebrafish) is able to regenerate several internal organs and the fins. The latter constitutes a great model to study adult vertebrate regeneration due to its easy access and non-vital function [1].

The caudal fin is composed of segmented bony rays or lepidothrichia that encircle the intra-ray mesenchyme. They are separated by inter-ray connective tissue and covered with epithelium. Blood vessels, nerves and pigment cells complete the fin [2]. Upon amputation, a new fin is produced roughly within two weeks through a process called epimorphic regeneration, including three main stages: wound healing (0–12 hours post amputation - hpa), blastema formation (12–48 hpa), and regenerative outgrowth (48 hpa to 2 weeks) [3]–[5]. Importantly, the blastema is the crucial structure for epimorphic regeneration. This heterogeneous cell population arises from dedifferentiation of mature cells [6]–[8], possibly in combination with cellular transdifferentiation and/or a resident stem cell pool, and contains the morphogenetic information required to give rise and re-pattern all the missing tissues [9]. Regeneration is regulated by the orchestrated action of several signalling pathways activated after injury, including Wnt (canonical and non-canonical), Fgf, Shh, Bmp, Activin-βA, Notch and Retinoic acid [1] [10].

Alongside classical signalling pathways, the relevance of ion channels and transporters for regeneration is becoming increasingly evident. Their coordinated activity results in the differential accumulation of ions, thus electric charge, across cells membranes. The electrical properties of all organisms arise from this charge segregation [11]. Although the generation of an endogenous wound electric current (EC) is a universal and essential response to wounding [12], the maintenance of endogenous ECs after wound closure is restricted to regenerating structures [13] [14]. In fact, it has been long known that ECs are essential to regeneration [15] [16]. In the last decade, successful efforts have begun to unveil the ionic nature of these electric cues and the molecular players that generate them. For instance, the ionic composition of the ECs at rat corneal wounds is now described and is actively regulated by specific ion transporters [17]. Moreover, cellular hyperpolarization caused by the H+ pump V-ATPase has proven essential for regeneration of Xenopus larval tail, by promoting cell proliferation and neural patterning [18]. Also, during planarian regeneration, the proton,-potassium transporter H+,K+-ATPase ensures the membrane depolarization required to specify anterior polarity in regenerating tissues and for tissue remodelling via apoptosis [19]. Other studies have identified ion transporters that generate electric signals involved in cell migration, proliferation, differentiation and apoptosis (reviewed in [20] [21]). Interestingly, all these cell behaviours are required for regeneration.

Ion transporters also generate chemical gradients that contribute to instruct specific cell behaviours [21]. For example, the increase in intracellular sodium, mediated by the voltage-gated sodium channel NaV1.2, is required for cell proliferation and tissue innervation during Xenopus tadpole tail regeneration [22]. In the zebrafish eye, the V-ATPase regulates retinoblast proliferation and survival, possibly through the acidification resulting from H+ accumulation [23]. The same H+ pump is essential for activation of Wnt, JNK and Notch signalling, by regulating endosomal pH [24]–[26].

Despite the increasing amount of data, a comprehensive description of the role of ion channels and transporters in regeneration is far from complete. In this study, we coupled biophysical and molecular approaches to address the ion nature and respective ion transporters involved in regeneration in an adult vertebrate (zebrafish). We show that a H+ outward current (efflux) is specifically set during caudal fin regeneration and that the V-ATPase, which is the main H+ pump in animal cells, contributes to that efflux. The onset and intensity of both V-ATPase expression and H+ efflux vary with the amputation plane along the proximal-distal (PD) axis in a way that correlates with the regeneration rate. Particularly, we demonstrate that inhibition of V-ATPase activity impairs regeneration and that proximal stumps have a stronger dependence on V-ATPase activity compared to distally amputated fins. We then investigated how the activity of this H+ pump articulates with molecular signalling pathways to affect cell behaviour and give rise to the missing tissues. We show that V-ATPase is required for aldh1a2 and mkp3 expression, blastema cell proliferation and normal fin innervation.

All procedures involving animal use were approved by the Ethics Committee for Animal Welfare at Instituto Gulbenkian de Ciência (IGC), according with directives from Direção Geral de Veterinária (PORT 1005/92). All surgeries were performed under Tricaine anesthesia, and all efforts were made to minimize animal suffering.

For most experiments, we used AB wild type (WT) zebrafish (Danio rerio) from the IGC fish facility. The mutant line atp6v1e1hi577tg (AB) was acquired from ZIRC. All fish were raised and maintained under standard procedures [27]. Adult fish (6–9 month old) were kept at 30°C in isolated tanks with re-circulating water

Prior to manipulation, fish were anesthetized in Tricaine (Sigma-Aldrich #A5040), 1 mM for embryos/larvae and 0.6 mM for adults. Larval fin folds were amputated 2 days post-fertilization (dpf) as described [28]. Adult caudal fins were amputated with surgical razor blades along the dorsoventral axis, 1–2 ray segments before the first ray bifurcation (distal amputation) or 2 segments distal to the most posterior scale covering the fin base (proximal amputation). For amputation at both planes (proximal-distal amputation, PD), the dorsal and ventral halves of the fin were amputated at either planes; a third cut along the proximodistal axis, halfway through the dorsoventral axis, completed tissue removal. Upon manipulation, animals returned to their tanks to regenerate. For tissue collection, fins were re-amputated at several regeneration time points, 1-2 bone ray segments proximal to the first amputation.

SIET (Applicable Electronics LLC, USA) is a non-invasive technique that measures stable and low magnitude extracellular ion-specific fluxes in aqueous media. Ion-selective electrodes (ISE) were built, connected to the SIET system and calibrated as described in [29]–[31], with adaptations to adult zebrafish (Table S1). Fish were anesthetized in recording medium (Table S2) (adapted from [32]) and accommodated in a custom-made plate (Fig. 1A). The ISE was positioned in front of the distal end of the fin, aligned with the fin in the “X” and “Z” axes and moved between one position close to the rays (10–20 μm) and another 70 μm farther away to measure direct current (DC) voltage potential. A reference measurement was taken away from the fish. The electrical signal was recorded in ASET software (Science Wares) and transformed into ion fluxes using the Nernst equation and Fick's law, as described [33]. Fluxes were classified according to the direction: influx and efflux. We measured potassium-, sodium-, calcium-, protons- and chloride- specific fluxes. Measurements were performed at the 3rd, 4th and 5th dorsal and ventral rays. One-way ANOVA and post-hoc Tukey HSD test were used to compare H+ flux at all regeneration stages screened, and paired T-test for comparisons within each time point.

Caudal fins were amputated at the distal plane and collected at 3, 24, 48 and 96 hpa. Intact fins (0 hpa) were also collected and used as control. Two replicates were used per time point, each consisting of three fins. Total RNA was isolated using RNeasy Mini Kit (Qiagen). Scanned arrays were analyzed first with GCOS 1.4 software to obtain Absent/Present calls and for subsequent analysis with dChip 2010 (http://www.dchip.org). The arrays were normalized to a baseline array with median CEL intensity by applying an Invariant Set Normalization Method [34]. Normalized CEL intensities of the ten arrays were used to obtain model-based gene expression indices based on a PM (Perfect Match)-only model. All genes compared were considered to be differentially expressed if the 90% lower confidence bound of the fold change between experiment and baseline was above 1.2 in both replicate datasets and if the transcript had at least one Present Call per replicate.

Gene fragments for atp6v1e1b, atp6v1a, wnt10b, mkp3 and aldh1a2 were cloned by PCR amplification from zebrafish cDNA, according to [35], using sequence-specific primers (Table S3). PCR products were ligated into pGEM-T Easy (Promega #A1360). For atp6v1e1b, we used pBluescript II KS+ (Stratagene #212207). Digoxigenin-labeled RNA probes were then synthesized. After collection, caudal fins were processed and used for in situ hybridization, as described [8]. In fins amputated at the PD plane the final precipitation reaction was stopped when the proximal regenerate was full stained.

Caudal fins were fixated overnight in 0.2% PFA (for ATP6V1A) or 4% PFA (for the remaining antigens) and processed according to [8]. All methanol and acetone incubations were excluded when staining for rhodamin-phalloidin (Invitrogen R415). The following primary antibodies were used: anti-ATP6V1A (Genscript A00938), anti-Pan-cadherin (Abcam ab6528-100), anti-phospho-Histone-3 (H3P) (Millipore 06-570), anti-active-Caspase-3 (Abcam ab13847), anti-acetylated α-tubulin (Sigma T7451).

Proliferating cells were detected by whole mount immunohistochemistry against H3P. The number of H3P-positive cells was counted in the regenerating mesenchyme and in the mesenchyme immediately below the amputation plane. We used three-dimensional projections of confocal images through the mid 20 μm of the mesenchymal depth, and pan-cadherin to exclude epithelial cells. Statistical analysis was performed using paired T-test.

Fin folds of 15 AB WT and 15 atp6v1e1bhi577tg/- (AB) mutant larvae were amputated at 2 dpf. At 5 dpf the regenerated area was measured and compared to the area of non-amputated larvae of the same genotype using independent T-test. The two fish lines were compared separately due to genotype- specific morphometric differences. Phenotypic defects were also investigated by visual inspection.

Different morpholinos (Gene Tools, LLC) were used to block translation of V-ATPase subunit v1e1b. Fluorescein-tagged morpholinos (fluo-MO) and corresponding mismatch controls (cfluo-MO) were used at 1 mM. Vivo-morpholino (vivo-MO) and control vivo-MO (cVivo-MO) were delivered at 0.5 mM. Their sequences were as follows (5′ to 3′): fluo-MO-1 TCGGCATCGCTGAGCGCCA, cfluo-MO-1 TCcGCATgGCTGAcCGCgA, fluo-MO-2 TGCAGATCCTGCTCCTGCTGCTTTA, cfluo-MO-2 TGgAcATCgTGCTgCTcCTGCTTTA, vivo-MO TCGGCATCGCTGAGCGCCA, cVivo-MO CCTCTTACCTCAGTTACAATTTTATA. Fins were amputated at either the proximal or distal level. Each atp6v1e1b-specific morpholino was injected into one half of the caudal fin at 2 or 16 hpa, and the other half received the corresponding control. The same number of fish received the inhibitory molecule at the dorsal and ventral part. Fluo-MO experiments were performed on AB WT fish, and vivo-MO was delivered to atp6v1e1bhi577aTg/+ (AB) fish (heterozygous fish carrying a recessive mutation for atp6v1e1b). For fluo-MOs, injection was immediately followed by electroporation of the whole fin. Both procedures were performed as described [36]. The areas of regenerating tissue after MO and control MO delivery were compared with a paired T-test. Results were plotted as the difference (%) in treated regenerate area compared to the control, calculated using the formula: (MO area x 100/control MO area) -100 = % MO area relative to the control.

Concanamycin A (concA, Sigma C9705) was dissolved in 100% DMSO and diluted in standard Danieau medium to working concentrations of 100 and 500 nM; the corresponding control solutions were 0.1% and 0.5% DMSO respectively, both prepared in Danieau medium. ConcA and control were injected every 12 h between 6–42 hpa. The experimental design and procedure was as described for vivo-MO.

For in situ hybridization and in vivo protocols, images were obtained with a Leica Z6APO stereomicroscope equipped with a Leica DFC320 color camera. Immunostainings were imaged on a Zeiss LSM 710 confocal microscope. The ImageJ software was used to analyze the Z stacks and to measure regenerate area and blastema length. All statistical analyses were done on SPSS 16.0 software.

The individual contribution of K+, Na+, H+, Ca2+ and Cl− - specific fluxes to the ECs during adult zebrafish fin regeneration was investigated using SIET (Fig. 1B, File S1). From the five ion-species tested, H+ was the only one with a dynamic pattern in stages specific to regeneration events (later than wound healing). Before amputation (0 hpa) and during wound healing (6 hpa) fins maintained a small H+ efflux close to the background noise (p>0.05, independent T-test). However, by 24 hpa, when the wound had healed and a blastema was forming, an outward current was established. This efflux was 14-fold higher than the efflux detected in intact fins (p<0.05, one-way ANOVA), and remained at high intensity until the end of blastema formation (48 hpa). From 72 hpa on, it decreased towards levels closer to the uninjured tissue (Fig. 1B). SIET measurements show for the first time that H+ current is specifically set during adult vertebrate appendage regeneration, suggesting that some mechanism of H+ extrusion is activated in cells during regeneration.

To find candidate H+ transporters that could be associated to regeneration and to the detected H+ efflux, we took advantage of an Affymetrix microarray previously done in our laboratory to assess gene expression during regeneration after distal amputation. Upon a specific analysis for proton transporters' expression, we focused on the V-ATPase, one of the main H+ transporters in animal cells. In fact, all V-ATPase subunits present in the microarray were upregulated between 24 and 96 hpa (Fig. 1C). This expression pattern matches that of H+ efflux, suggesting that the regeneration specific H+ extrusion could be associated to the increased V-ATPase expression.

The regeneration-associated expression of this H+ pump was confirmed by in situ hybridization for different V-ATPase subunits. Neither atp6v1a nor atp6v1e1b were detected in intact fins (Figures A and B in File S2). In contrast, 24 h after distal amputation, both atp6v1a and atp6v1e1b were expressed in the blastema-forming region and in scattered cells below the amputation plane (Fig. 1D, G and Figure C in File S2). By 48 hpa, their expression domain was expanded, accompanying the growth of the blastema (Fig. 1E, H). At 72 hpa both transcripts were faint but could still be observed in the distal regenerating region (Fig. 1F, I).

For V-ATPase detection at the protein level, we targeted the cytoplasmic subunit Atp6v1a. In the intact fin, Atp6v1a was mainly restricted to the epidermis, in a scattered pattern (Figures D-F in File S2). On the contrary, by 24 hpa, this subunit was strongly upregulated in the blastema (Fig. 1J, K) and was present all over the blastema until48 hpa (Fig. 1L). At that time, Atp6v1a also became evident in the regions where ray segment joints were forming. At 72 hpa it was also present in the areas of lepidotrichia regenerative outgrowth and in the intra-ray mesenchyme (Fig. 1M). Atp6v1a did not co-localize with rhodamin- phalloidin, which labels cortical F-actin, but stained immediately below (Fig. 1J). Considering that Atp6v1a is part of the cytoplasmic domain of the H+ pump, our results did not exclude a possible plasma membrane location of the whole pump.

Overall, these three different approaches are in agreement with a specific upregulation of V-ATPase during regeneration. Both V-ATPase expression and H+ efflux intensity seem stronger during blastema formation and decrease thereafter, suggesting a causal relation.

To assess the functional significance of the V-ATPase during fin regeneration, we used concA to specifically inhibit the pump's activity. Translation of atp6v1e1b, a V-ATPase subunit essential for the pump activity, was also blocked using gene specific fluo-MOs. Half regenerating fin was treated with one inhibitor whereas the other half received the corresponding control (File S3). All V-ATPase inhibitors decreased the regenerate area for at least 48 h compared to the corresponding control, despite high phenotypic variability (Fig. 2A). This suggests a role for this H+ pump in the regenerative process. In fact, V-ATPase inhibition seemed to affect regeneration rate more that the regenerative ability itself, since, notwithstanding the reduced area, regeneration still progressed in a delayed fashion.

To investigate if V-ATPase was related to regeneration rate, we took advantage of the fact that proximal stumps have higher regeneration rate than distal ones [37] [38], and compared V-ATPase expression after proximal versus distal (PD) amputation. In proximal wounds, atp6v1e1b was first visible around 12 hpa in the first ray segment below the amputation plane and, to a smaller extent, in the interray (Fig. 2B). Expression at the distal stump only became evident at 24 hpa, as described (Fig. 1F, Fig. 2C, E). By that time atp6v1e1b domain in proximal stumps was much stronger and wider (Fig. 2D). By 48 hpa the differences between proximal and distal atp6v1e1 expression had faded (Fig. 2F, G). Accordingly, at the protein level, Atp6v1a in proximal stumps was present in almost twice the length than in the regions amputated distally by 24 hpa (proximal:distal length ratio mean ± s.e.m = 1.84±0.17) (Fig. 2H, I, M). At 48 hpa, the protein domain was still 1.57±0.08 fold longer in proximal wounds (Fig. 2J, K, M). From the above, V-ATPase seems to have a dynamic expression according to the level of amputation: in the proximal, fast regenerating regions, the pump expression starts earlier and the expression domain is larger than in the slower regenerating distal wounds.

We have previously suggested V-ATPase as a mediator of the regeneration-specific H+ efflux. If so, the dynamic expression of V-ATPase along the PD axis should be accompanied by a concordant H+ efflux pattern. Thus, we measured H+ flux throughout regeneration after PD amputation, using SIET (Fig. 2L). The efflux peaked always at 24 hpa. However, in proximal stumps H+ efflux started earlier (3 hpa instead of 12 hpa) and was significantly higher than in distal regions for each time-point (p<0.05, paired T-test). This agrees with the different onset and magnitude of V-ATPase expression at proximal and distal stumps, reinforcing the correlation between H+ efflux and V-ATPase and extending such correlation to the position-dependent regeneration rate.

To deepen the relation between V-ATPase activity and H+ efflux, we analysed the effect of V-ATPase inhibition on H+ flux. For that, we used atp6v1e1bhi577aTg/+ (AB) fish to perform an atp6v1e1b knockdown using vivo-MO as described in materials and methods, at 2 h after proximal and distal amputation. Then, H+ fluxes were measured at main regeneration stages using SIET (Fig. 3A, B). Aside from the higher efflux intensity proximally, distal and proximal wounds had a similar flux pattern, as follows: in the controls, H+ efflux intensity was maximal at 24 hpa and then decreased gradually, as previously detected (Fig. 2L). After atp6v1e1b knockdown, H+ efflux at 24 hpa decreased significantly (p<0.05, paired T-test), confirming the causal relation between V-ATPase activity and H+ efflux. Surprisingly, by 48 hpa the efflux pattern reverted compared to the control: as the latter began its descendent path, the former increased significantly (p<0.01, paired T-test) to a level that resembled the control efflux at 24 hpa for the same amputation plane. Similarly, at 72 hpa the H+ efflux intensity was higher than in the control (p<0.01, paired T-test), and was similar to the control efflux at 48 hpa, as if there was a 24 h delay on flux intensity due to the transient vivo-MO effect. Thus, the level of flux seems to be controlled by V-ATPase activity and relevant for position- dependent regeneration.

To further investigate the role of V-ATPase and its association with regeneration rate, we performed vivo-MO mediated atp6v1e1b knockdown in atp6v1e1bhi577aTg/+ (AB) fish, 2 or 16 h after proximal and distal amputation. In all experimental sets, atp6v1e1b knockdown decreased the regenerate area compared to the control (Fig. 3C, C’, D, D’). However, proximal regenerates were more affected by vivo-MO delivery at 2 hpa (Fig. 3C, C’) whereas distal stumps exhibited a more significant area reduction when atp6v1e1b was inhibited at 16 hpa (Fig. 3D, D’). Importantly, this correlates with the different onset of both V-ATPase expression and H+ efflux, and again it links to regeneration rate. Besides, the overall decrease in the regenerate area was steeper proximally than at distal stumps (24 hpa, mean±s.e.m. = 35.6±8.7% and 11.7±4.9%, respectively, for delivery at 2 and 16 hpa), suggesting that higher regeneration rate have a stronger dependence on V-ATPase activity.

The knockout of one V-ATPase subunit becomes lethal around 6 dpf [23], long enough to study the effects of the gene absence in the larval fin fold regeneration. In fact, many evidences suggest that regeneration of adult and larval caudal appendage follow similar mechanisms [5] [28]. Given that, we amputated the fin fold of atp6v1e1hi577aTg/- mutants and AB WT fish at 2 dpf and the regenerated area was compared with non-amputated fins of the same genotype 3 days later. Intact mutants were underdeveloped and had a smaller fin fold than the wild type (Fig. 4A, C, E). After amputation, there were no additional phenotypic differences between the two fish lines (Fig. 4B, D), and the regenerated fin fold area was similar to the control, for either genotypes (Fig. 4E). These results show a normal larval regeneration process in the absence of V-ATPase activity, contrary to what was observed in the adult caudal fin. Hence, regeneration of the adult and larval caudal appendages do not seem to operate through the same mechanisms.

Reduced regeneration upon V-ATPase inhibition could be due to increased cell death or decreased cell proliferation. To investigate that, we knocked down atp6v1e1b in the caudal fin of atp6v1e1bhi577aTg/+ (AB) fish by delivering vivo-MO at 2 hpa. Fins were collected at different regeneration timepoints and immunostained for active-Caspase-3 and Phospho-Histone-3 (H3P), apoptotic and proliferative markers respectively. Apoptosis was similar in both control and treated fish, suggesting that reduced regeneration is not due to increased apoptosis (File S4). As for proliferation, 24 h after proximal amputation, most H3P-positive cells in control fins were found in the intra-ray mesenchyme 1–2 ray segments below the proximal amputation plane, with only occasional cells stained above the amputation plane (Fig. 5A); whereas at 48 hpa, many proliferating cells had accumulated in the mature blastema (Fig. 5B). Likewise, atp6v1e1b knockdown didn't affect the number or location of proliferating cells by 24 hpa (Fig. 5C). However, at 48 hpa there were significantly less blastema cells positive for H3P than in the control (less 40.1%±11.58) (Fig. 5D, E). Similar results were found in distal regenerates (File S5). These results show that V-ATPase is required for normal cell proliferation in the blastema.