Single neuromast electroablation locally ablates all components of the PLL system
Our aim was to understand the cellular mechanisms involved in the activation and differentiation of sensory organ precursor cells after complete neuromast ablation. To this end, we have taken advantage of a simple methodology recently developed in our laboratory, electroablation [35]. This technique allowed us to induce a localized tissue injury by applying an electrical pulse directly to the neuromasts, which are superficially located, to completely remove them.
We decided to ablate the third neuromast (L3) of the PLL in larvae 3 days post fertilization (dpf) because, at this stage, the primary PLL is completely laid down and innervated [18]. Also, the region where L3 is located is easily recognizable and no intercalary neuromasts appear normally in its vicinity at the stages examined [21], avoiding potential misinterpretation of the results or interference by normal developmental processes in our observations. Furthermore, the secondary primordium (PrimII) does not travel as far caudally as L3 and does not generate secondary neuromasts near it.
In order to ablate the L3 neuromast, we applied two 8 μA pulses for 2 s each directly over the neuromast in tg(cxcr4b:mCherry) transgenic larvae; in this line, PLL neuromasts and INCs express membrane-tagged red fluorescent protein (RFP) [36]. As shown in Fig. 1a, intact trunk neuromasts (L2, L3, and L4 are shown) have a rosette-like structure and are interconnected by INCs (Fig. 1a, arrows). Figure 1b shows the trunk of the same larva 4 hpi. As was previously reported [35], our electroablation protocol creates a gap of 55.8 ± 26.3 μm between remaining INCs at the position where L3 was located, with no remaining primordium-derived cells in the gap (Fig. 1b, asterisk; 1c). In every experiment, we confirmed that this was the case before proceeding. Adjacent neuromasts L2 and L4, as well as most of the surrounding INCs, remained intact. After 24 hpi, surviving lateral line cells migrated and converged midway to seal the gap (Fig. 1d). After 72 hpi, a new L3 regenerated at the same position as the original L3 in 58.2 ± 4.4 % of cases (Fig. 1e, yellow arrowhead; Additional file 1). Thus, electroablation can be used to inflict localized—yet complete—damage to a single neuromast and to study its regeneration.
To further characterize the impact of electroablation on the underlying lateral line components, such as the PLLn, we repeated the same experiment in double transgenic tg(neurod:GFP; cxcr4b:mCherry) larvae that express cytoplasmic green fluorescent protein (GFP) in the axons of the lateral line afferent neurons and mCherry in all cells derived from PrimI (Fig. 1f). Ablation of the L3 neuromast produced a complete interruption of the PLLn (Fig. 1g). Between 3 and 9 hpi, the distal nerve degenerated (Additional file 2, white arrowheads) as previously described [37, 38]. As a consequence of distal nerve degeneration, caudal neuromasts (L4–L8) were temporarily deprived of afferent innervations, whereas the neuromasts located rostral to the damaged site (L1, L2) remained innervated. After 9 hpi, the axons began to regenerate (Additional file 2). As a consequence of the sprouting behavior of the regenerating nerve, a few afferent axons defasciculated and their growth became arrested at the gap left by electroablation (Fig. 1h; Additional file 2, yellow arrowhead). Most axons, however, grew along the myoseptum (Additional file 2C, F, yellow arrow), reaching the tip of the tail and reinnervating neuromasts caudal to L3 (Additional file 2C, F, white arrow). Nerve regeneration ended after 30 hpi, and INCs had sealed the breach created by ablation of the L3 neuromast.
At 3 dpf, the PLLn is lined by SCs that will form the myelin sheath [29]. We presumed that the loss of the neuromast and nerve caused by electroablation would be accompanied by a local loss of SCs in the damaged zone. Previous work by ourselves and others has shown that SCs undergo significant changes (dedifferentiation and/or cell death) when the nerve they are interacting with disappears [38–44]. We examined the fate of SCs in our conditions by using double transgenic tg(foxd3:GFP; cxcr4b:mCherry) larvae that express GFP in the SCs and RFP in the lateral line components (Fig. 1i). As we observed with the PLLn, SCs located under the L3 neuromast were completely eliminated, generating a temporary gap that was filled after 24 hpi (Fig. 1i–k). Thus, glial cells and INCs sealed the injury zone at approximately the same time. However, the differentiation process of SCs was reversed after temporary denervation, as was previously reported in several animal models including zebrafish [39, 40, 45, 46]. Whereas control 3 dpf zebrafish larvae started to express myelin basic protein (MBP), which is a differentiation marker of myelinating cells [44] (Additional file 3A, arrowheads), in injured larvae we observed a fragmentation and then loss of MBP expression caudal to L3 (compare Additional file 3C, D with I–J). By 24 hpi, there was a complete absence of the MBP marker from the injury point to the caudal fin (Additional file 3B; also compare Additional file 3E, F with K, L). The loss of MBP expression, but not of the Foxd3:GFP signal, after 24 hpi suggests changes in the differentiation state of SCs rather than death of these cells. After 48 hpi, we observed a partial recovery of MBP expression in the caudal zone, as was previously reported [38], suggesting that the myelination process had restarted after nerve regeneration.
Thus, we have shown that the continuity of the PLLn, SCs, and INCs along the PLL is interrupted by electroablation at the position of neuromast L3. Further, we show that all three components (neuromast/INCs, axons, and SCs) reestablish continuity of the PLL after a few days, indicating a coordinated regeneration process. We then sought to examine the behavior of these cells and determine how they contribute to the regeneration of L3 after its ablation.
Identification of the cells that contribute to a regenerated neuromast
In both axolotl and zebrafish, remaining neuromasts contribute to the regeneration of the lateral line after tail fin amputation [9, 47]. To evaluate if this mechanism could account for neuromast regeneration in our model, we looked for changes in the number of cells in neighboring neuromasts after damage to L3. To that end, we counted the hair cells in the L2 and L4 neuromasts by in vivo observation of tg(brn3c:GAP43-GFP) L3 electroablated larvae 2, 24, 48, and 72 hpi. We also fixed ablated and control larvae at several time points after injury (6, 24, 48, and 72 hpi) and carried out immunohistochemistry to identify and quantify additional cell types in the L2 and L4 neuromasts (Additional file 4). Total cell numbers were quantified by TO-PRO-3 (nuclear staining) while ET20:GFP transgenic larvae in combination with Sox2 immunolabeling were used to recognize mantle cells (GFP+Sox2+) and the progenitor cell population (GFP−Sox2+) [33, 48]. We could not detect any significant differences in the cell composition of the L2 and L4 neuromasts when compared to the control and injured animals. We therefore concluded that neighboring neuromasts were not affected by L3 damage and that progenitor cells residing in neighboring neuromasts likely do not contribute to L3 regeneration.
It has been reported that INCs have the ability to form intercalary neuromasts during late larval development in zebrafish [16, 25, 26]. Here, these cells respond shortly after damage by sealing the gap left by a localized injury. To determine if the INCs are in fact responsible for regeneration of the L3 neuromast, we electroablated L3 in a double transgenic line that labels INCs with GFP and all lateral line cells with an RFP tag: tg(cxcr4b:mCherry;et20:GFP). Time-lapse imaging of the damaged site after ablation revealed movement of GFP+ cells on both sides of the gap that resulted in their reconnection (Additional file 5). In no instance did we observe GFP− cells migrating into the injury zone (n = 15). This suggests that cells with INC identity are the only cells that populate the injury site after electroablation.
We also fixed control and injured larvae at different times points after electroablation (2 or 6, 24, 48, and 72 hpi) with the aim of describing the temporal appearance of the different cell types of the neuromast until the end of the regeneration process. We used the transgenic lines tg(et20:GPF) for labeling INCs and mantle cells, tg(brn3c:Gap43-GFP
+) for labeling hair cells, and an anti-Sox2 antibody for detecting progenitor cells [33]. We first confirmed that Sox2 protein is in fact expressed in neuromast progenitor cells and not in INCs. This was indeed the case at all ages analyzed (4, 5, and 6 dpf), shown by the fact that INCs and mantle cells were labeled with GFP in tg(et20:GFP) transgenic fish whereas only neuromast progenitors were immunostained with anti-Sox2 antibody in the same animals (Additional file 6). As shown in Fig. 2, at 6 hpi the electroablation gap remained devoid of any cells (Fig. 2a, e) but by 24 hpi, INCs (GFP+Sox2−) from both sides of the gap had connected. At this stage, all the fish examined had a newly established line of INCs (n = 40) and some (~29 %) also displayed an accumulation of GFP+ cells (Fig. 2b). In the cases where INCs accumulated, some cells began to express the Sox2 progenitor marker (Fig. 2h). At this time point (24 hpi), none of these protoneuromasts had developed hair cells (Brn3c:Gap43-GFP+). However, by 48 hpi, 29.2 ± 2.2 % (n = 100) of injured larvae exhibited regeneration of the organ (presence of at least two hair cells in the neuromast); this percentage doubled after 72 hpi to 58.2 ± 4.4 % (n = 100; see Additional file 1).
In regenerating neuromasts, at 48 hpi, ET20:GFP+ cells (also positive for the Sox2 marker) formed the characteristic ring of mantle cells (Fig. 2c) while a group of centrally distributed cells lost the expression of ET20 and expressed only Sox2 (Fig. 2d, h). This was reflected by the recorded decrease in the number of ET20+ cells between 24 and 48 hpi (Fig. 2d). Hair cells arose from the central-apical region of the new neuromast, first evident by the loss of ET20:GFP/Sox2 expression and the appearance of the mature hair cell markers (Fig. 2g, yellow arrowhead). After the emergence of hair cells in the regenerating neuromast, we performed time-lapse imaging in a tg(cxcr4b:mCherry;brn3c:gap43-GFP) larva (Additional file 7; see also Fig. 2i). We quantified hair cell number by in vivo observation of injured larvae during the 3 days following electroablation (Fig. 2j). The number of hair cells increases over time, although there was still a significant difference with respect to controls at at 72 hpi (Fig. 2j). These experiments showed that all of the cell types that compose the mature neuromast develop after the accumulation and posterior differentiation of INCs. However, the lack of appropriate and specific markers to efficiently differentiate accumulating INCs from progenitor or mantle cells prevented us from assigning a direct progenitor role for these cells or determining whether they must first undergo a transition through an intermediate fate.
As described above, we consistently found a proportion of electroablated larvae in which the INC stripe reconnected but that never achieved neuromast regeneration (~40 % of fish). In tg(et20:GFP) transgenic larvae that did not regenerate the L3 neuromast, INCs continued to express GFP but not the Sox2 protein (see Additional file 6). Thus, achieving INC accumulation and expression of Sox2 can be considered a robust predictor of regenerative success in this context.
Two INCs are sufficient to form a new neuromast after electroablation
To elucidate whether INCs are multipotent progenitors able to reconstitute an entire neuromast, including all of its cell types, we generated mosaic animals by transplanting cells (at the high blastula stage, 3 h 20 min) from a tg(ubi:RFP) donor embryo that ubiquitously expresses a cytoplasmic RFP into a tg(et20:GFP) host. We screened and selected transplanted larvae 48 hours post fertilization (hpf) that had only one or very few red-labeled INCs (we discarded those larvae that had red mantle cells). In these fish, we were able to follow individual INCs and analyze their behavior during the regeneration process. At 3 dpf, we electroablated the neuromast nearest to the implanted INC in chimeric larvae (Fig. 3a, b, asterisk) and recorded the regeneration process until 72 hpi. As we have shown previously, electroablation generates local damage that is circumscribed to the neuromast without affecting the neighboring INCs. At 24 hpi, INCs accumulated at the injury site. In the example shown, a single red-labeled cell (Fig. 3a) divided into two daughter cells (Fig. 3c), indicating that this accumulation is due to both the migration of INCs into the gap (Additional file 5) and the local proliferation of these cells. From 24 to 48 hpi, these cells continued to increase in number and also began to organize into a rosette-like structure (Fig. 3e, f). At 72 hpi, the number of cells had continued to increase and the newly formed neuromast was apparent (Fig. 3g, h). We also observed labeled daughter cells that remained within the INC population, suggesting self-renewal of the INCs (Fig. 3g, h, yellow arrowhead). Finally, in all cases analyzed (n = 10), the regenerated neuromast was composed of a combination of red fluorescent-labeled and ET20:GFP+-labeled cells, suggesting that the regenerated sensory organs come from at least two different INCs, most likely the two INCs (rostral and caudal) flanking the ablation gap. In support of this conclusion, we observed that only the most proximal INCs were responsive to neuromast electroablation, given that INCs located in more distal positions remained quiescent and immotile during the regeneration process (Additional file 8, white arrowheads).
In summary, destruction of all cells in a neuromast is followed by convergent migration of at least two INCs into the injury zone. Once there, they proliferate and differentiate, giving rise to all of the cell types of the mature regenerated neuromast.
We next asked whether INCs located only to one side of the injury zone could be responsible not only for neuromast regeneration, but also for the entire lateral line system after more severe damage, as occurs, for example, after tail fin amputation. To examine this question, we electroablated the L3–L8 neuromasts of a 3 dpf tg(et20:GFP) larvae and mechanically removed all of the INCs posterior to L3. The remaining INCs located near the L3 position migrated caudally (Fig. 4a–d; Additional file 9). These cells started to proliferate and accumulated in the region where L3 was located. At this position, the INCs reorganized into a new neuromast, as we previously described. Then, more distally located INCs that did not belong to the prospective neuromast continued to migrate caudally, iterating the process and reconstituting the canonical row of INCs and neuromasts at the myoseptum (n = 20). We never observed the formation of a primordium (collective migration of cells) during this type of lateral line regeneration. We conclude that INCs are progenitor cells that have the capacity, on their own, to restore the entire lateral line system after severe damage. Intriguingly, the progeny of these cells were able to correctly position the new sensory organs to maintain appropriate spacing between them.
We also sought to know if the appearance of a new sensory organ was restricted to the previous location of the damaged neuromast or whether the interruption of the row of INCs by the local application of current at any point could elicit neuromast reconstitution. To investigate this, we electroablated midway between the L2 and L3 neuromasts: 48 hpi we observed the formation of a new neuromast at the ablation point in 28 ± 2.45 % of injured larvae, and at 72 hpi in 56.75 ± 10.05 %, indistinguishable from neuromast ablations (Fig. 4e–g). This result shows that generating a discontinuity in the PLL (including the PLLn, SCs, and INCs) is sufficient to induce the formation of a new neuromast.
SCs are key regulators of the neuromast regeneration process
As we have previously shown, 60 % of the electroablated larvae regenerate the L3 neuromast after 72 hpi by local activation and differentiation of INCs. However, we were curious why the remaining 40 % of the larvae failed to regenerate a neuromast even though the INCs became activated and migrated to seal the gap between them (Additional file 2). Based on previous knowledge on the development of this sensory system [16, 25, 26], we hypothesized that SCs might be responsible for the control of INC behavior during neuromast regeneration.
As described above, both INCs and SCs seal the gap created by electroablation at approximately the same time (24 hpi). To evaluate whether the ability of INCs to regenerate a neuromast depends on their reconnection and local activation before SCs seal the gap, we decided to measure the distance between surviving cells after electroablation in tg(cxcr4b:mCherry;foxd3:GFP) fish; in this double transgenic line both INCs and SCs are labeled [36, 49]. We correlated the size of the gap created between INCs and between SCs at 2 hpi with the regenerative outcome of injured larvae after 72 hpi. As shown in Fig. 5, we observed that regeneration success was independent of the size of the SC gap. However, the size of the INC gap or, rather, the INC/SC gap size ratio had a significant impact on the regenerative capacity. Fish that failed to regenerate exhibited a higher ratio compared to fish that regenerated the neuromast (Fig. 5c). This suggests that SCs could be interacting with INCs during the early steps of the regeneration process.
To test the role of SCs in the regenerative capacity of neuromasts, we took advantage of a pharmacological ablation tool that is specific for this cell type. We treated zebrafish embryos with 5 μM of the drug AG1478 from 10 hpf until 58 hpf. This treatment, which blocks ErbB signaling, completely inhibits SC migration along the lateral line nerve during early development [38, 44] without affecting the development of other cellular components of the system. At 3 dpf, after confirming that no SCs were present in the myoseptum, we ablated the L3 neuromast. As shown in Fig. 6e, 100 % of drug-treated larvae showed neuromast regeneration after 48 hpi, compared to only 29 % of the control-injured larvae.
Next, we incubated zebrafish larvae with 5 μM of AG1478 from 2 hours before injury (hbi) until 72 hpi. Treatment at this time does not interfere with early SC migration (when these cells migrate caudally together with the growing PLLn) but impairs their ability to continue to differentiate into mature (myelinating) cells once the drug is added to the medium [44]. In this case, we observed that 92.67 ± 3.21 % (n = 75) of injured larvae showed neuromast regeneration after 48 hpi. These results strongly suggest that the presence of differentiated SCs impairs regeneration due to local inhibition of INCs after damage.
In order to effectively test this hypothesis, we decided to damage lateral line neuromasts without affecting the underlying SCs. To this aim, we treated 3 dpf tg(cxcr4b:mCherry;et20:GFP) larvae with 100 μM CuSO4 for 2 h. This treatment leads to complete neuromast loss [13, 31], which was revealed by a discontinuity of the INC line in each of the locations where neuromasts were located (Fig. 6b, yellow arrowhead). Importantly, in this context, neither SCs nor INCs were affected (Additional file 10). After 96 hpi, copper-treated larvae failed to regenerate damaged neuromasts (Fig. 6e; n = 100). Surprisingly, when we then electroablated these fish at a point within the reconnected INC row 24 hours post CuSO4 treatment (hpt), a new neuromast appeared only where electroablation was done (Fig. 6c, white square). The percentage of larvae that regenerated a neuromast after 48 hpi was 21.21 ± 3.47 % (n = 60), similar to that observed in our first electroablation experiment (Fig. 6c–e).
Our results show that the interaction between SCs and INCs is key to control the balance between neuromast formation (regeneration) and replacement with INCs (repair) after damage to the lateral line. The appearance of a new neuromast depends of the temporal and spatial interaction between the two cell types. Finally, the results obtained from AG1478-treated larvae suggests that the interaction between the two cell types in a regeneration context is dependent on the ErbB signaling pathway, as has previously been shown during development of the lateral line system [25].
Lateral line and single neuromast regeneration in adult zebrafish
Finally, we wished to know whether the cellular mechanisms used in the larval stage to regenerate the neuromast are maintained in adult fish, where SCs (and all other components of the system) are fully mature. To solve this question, we electroablated an area of approximately 300 μm that usually spans two or more neuromasts in the caudal fin lateral line of a tg(et20:GFP) adult zebrafish (6 months of age) (Fig. 7). We monitored the regeneration process daily during the following 20 dpi (Fig. 7). We observed the appearance of a regenerated neuromast in the ablated region in 18.33 ± 2.08 % (n = 60) of the cases. As seen before in larvae where we ablated all INCs and neuromasts caudal to L3, adult tail fin INCs migrated into the electroablation gap. However, during the time we observed the ablated animals, accumulation of cells to form a new neuromast occurred exclusively on the rostral side of the gap. INCs continued to migrate caudally (Fig. 7b-d, yellow arrowheads) but we did not observe additional neuromasts forming at more posterior positions. On the caudal side of the gap, the last remaining neuromast became gradually disorganized and disappeared 12 dpi (Fig. 7c–f, red arrowheads). This result suggests that the mechanisms employed at larval stages for neuromast regeneration are conserved in adulthood, although the regenerative capacity of neuromasts ablated in this fashion occurs in a lower percentage of individuals.