ABSTRACT
The enteric nervous system (ENS), which is derived from neural crest, is essential for gut function, and its deficiency causes severe congenital diseases. Since the capacity for ENS regeneration in mammals is limited, additional complementary models would be useful. Here, we show that the ENS in zebrafish larvae at 10-15 days postfertilization is highly regenerative. After laser ablation, the number of enteric neurons recovered to ∼50% of the control by 10 days post-ablation (dpa). Using transgenic lines in which enteric neural crest-derived cells (ENCDCs) and enteric neurons are labeled with fluorescent proteins, we live imaged the regeneration process and found covering by neurites that extended from the unablated area and entry of ENCDCs into the ablated areas by 1-3 dpa. BrdU assays suggested that ∼80% of the enteric neurons and ∼90% of the Sox10-positive ENCDCs therein at 7 dpa were generated through proliferation. Thus, ENS regeneration involves proliferation, entrance and neurogenesis of ENCDCs. This is the first report regarding the regeneration process of the zebrafish ENS. Our findings provide a basis for further in vivo research at single-cell resolution in this vertebrate model.
INTRODUCTION
The enteric nervous system (ENS) is the largest peripheral autonomic nervous system. It controls gastrointestinal functions such as peristaltic movement and hormone secretion (Furness, 2006). Because of the essential roles of the ENS, congenital dysfunction of the enteric neurons often causes fatal diseases, such as Hirschsprung disease (Bondurand and Southard-Smith, 2016; Brosens et al., 2016). For the recovery of the functions of enteric neurons in patients, the regeneration of neurons from transplanted or endogenous stem cells has been discussed as a promising cure (McCann et al., 2017; Cooper et al., 2016; Fattahi et al., 2016). It is thus necessary to identify the mechanisms that control the regeneration of the ENS in model animals in which this occurs.
Regeneration studies of the ENS have been performed mainly with adult rodents (Laranjeira et al., 2011; Hanani et al., 2003; Joseph et al., 2011). New enteric neurons are not formed in the ablated area (Laranjeira et al., 2011; Joseph et al., 2011), but Laranjeira et al. (2011) demonstrated that in the area adjacent to the ablated area, new enteric neurons were formed from Sox10-expressing cells. These reports suggested that precursor cells of enteric neurons are not able to invade an ablated area, implying that adult rodents have limited regeneration ability. In human patients with serious medical conditions, it is necessary to induce neurogenesis in wide neuron-free areas in order to counteract the losses of enteric neurons in almost the entire gut area. Studies of animals with higher regenerative potential are thus necessary.
It is a challenge to observe the regeneration process in vivo using live imaging in mice, and this has hindered our understanding of the cellular behaviors during the regeneration of the ENS. Zebrafish are highly regenerative and have been used as a good vertebrate model for investigations of regeneration mechanisms for damaged zebrafish organs, including the central nervous system (CNS) (Cardozo et al., 2017; Alunni and Bally-Cuif, 2016), heart (Sehring et al., 2016; Gemberling et al., 2013), lateral line hair cells (Kniss et al., 2016) and fin (Sehring et al., 2016). The transparency of the larval intestine of zebrafish is ideal for both live imaging and the available genetic tools that can be used to visualize various cell types, including enteric neural crest-derived cells (ENCDCs) such as enteric neurons (Nikaido et al., 2018; Kuwata et al., 2019). However, the regenerative potential of the zebrafish ENS has not been tested.
In this work, we examined the zebrafish larval distal intestine as a model to study the regeneration mechanisms of the ENS, and we observed high regeneration ability after the ablation of enteric neurons. We describe the regeneration process observed, which indicated the invasion of neurites and ENCDCs into the ablated area.
RESULTS
The zebrafish enteric nervous system is highly regenerative
We first examined the regenerative ability of the zebrafish larval ENS after the removal of enteric neurons. To visualize zebrafish enteric neurons, we used a new gene trap line, Tg(SAGFF(LF)217B; uas:gfp) [referred to hereafter as Tg(217B; u:gfp)] obtained by Gal4FF gene-trap screening (Asakawa et al., 2008; Kawakami et al., 2010). To characterize the GFP-positive (GFP+) cells in Tg(217B; u:gfp) larvae, we performed immunostaining using molecular markers at 10 and 14 days postfertilization (dpf). Fig. 1A-F show expression patterns of GFP with the pan-neuronal marker HuC/D (Ganz, 2018) and Sox10, a marker for undifferentiated enteric neural crest cells (Corpening et al., 2011) (Kuwata et al., 2019) and glial cells (Hoff et al., 2008; Boesmans et al., 2015). Fig. 1A,B,D,F show that GFP+ cells in the distal intestines of three larvae at each stage of this transgenic line were HuC/D-positive (Hu+) enteric neurons. At 10 dpf, there was only one Hu-negative (Hu−) cell among 243 GFP+ cells, and there were no Hu− cells among 387 GFP+ cells at 14 dpf in distal intestines. On the other hand, we found only two Sox10-positive (Sox10+) cells out of 243 GFP+ cells (0.8%) at 10 dpf, and one among 387 GFP+ cells (0.3%) at 14 dpf. Fig. 1G-I demonstrate that GFP+ processes at 10 dpf are mostly acetylated tubulin-positive. These observations suggest that GFP+ cells in Tg(217B; u:gfp) larvae represent Hu+ enteric neurons.
Immunostaining of intestine in the transgenic line Tg(217B; u:gfp). (A-E) 10 dpf larval distal intestine stained for GFP (A), HuC/D (Hu, B) and Sox10 (C), with merge images shown (D,E). Dotted lines outline the intestine. Images are left side views, with the anterior end positioned to the left of the image. To show the overlapping expression of GFP and HuC/D, signal intensity for each marker is enhanced (A,B,D). Magenta arrowheads mark an example of a GFP−/Hu+ cell. The white asterisks in the most distal end of the intestine indicate unidentified Sox10 signals, and other asterisks around the dorsal part of intestine mark pigment cells outside of the intestine. Scale bar: 25 µm. (F) Proportions of GFP−/Hu−/Sox10+ (G−H−S+), GFP−/Hu+/ Sox10− (G−H+S−), GFP+/Hu+/Sox10− (G+H+S−) and other (Others) cell types at 10 and 14 dpf as a percentage of all cells examined (353 and 530 cells, respectively). ‘Others’ represent one GFP+/Hu+/Sox10+ and one GFP+/Hu−/Sox10+ cell at 10 dpf, and one GFP+/Hu+/Sox10+ cell at 14 dpf. Cells were counted in the distal part of the intestine, as exemplified in A-E. Numbers of examined larvae and cells at each stage are shown below the columns. (G-I) Co-expression of GFP and acetylated tublin (AcTub) in the distal intestine at 10 dpf. Markers shown are indicated in the bottom-left corners of panels. Images are left side views, with the anterior end positioned to the left of the image. Scale bar: 10 µm.
Immunostaining of intestine in the transgenic line Tg(217B; u:gfp). (A-E) 10 dpf larval distal intestine stained for GFP (A), HuC/D (Hu, B) and Sox10 (C), with merge images shown (D,E). Dotted lines outline the intestine. Images are left side views, with the anterior end positioned to the left of the image. To show the overlapping expression of GFP and HuC/D, signal intensity for each marker is enhanced (A,B,D). Magenta arrowheads mark an example of a GFP−/Hu+ cell. The white asterisks in the most distal end of the intestine indicate unidentified Sox10 signals, and other asterisks around the dorsal part of intestine mark pigment cells outside of the intestine. Scale bar: 25 µm. (F) Proportions of GFP−/Hu−/Sox10+ (G−H−S+), GFP−/Hu+/ Sox10− (G−H+S−), GFP+/Hu+/Sox10− (G+H+S−) and other (Others) cell types at 10 and 14 dpf as a percentage of all cells examined (353 and 530 cells, respectively). ‘Others’ represent one GFP+/Hu+/Sox10+ and one GFP+/Hu−/Sox10+ cell at 10 dpf, and one GFP+/Hu+/Sox10+ cell at 14 dpf. Cells were counted in the distal part of the intestine, as exemplified in A-E. Numbers of examined larvae and cells at each stage are shown below the columns. (G-I) Co-expression of GFP and acetylated tublin (AcTub) in the distal intestine at 10 dpf. Markers shown are indicated in the bottom-left corners of panels. Images are left side views, with the anterior end positioned to the left of the image. Scale bar: 10 µm.
Meanwhile, some Hu+ cells were GFP-negative (GFP−) (14% of 281 Hu+ cells at 10 dpf, and 7% of 418 Hu+ cells at 14 dpf). To test whether GFP−/Hu+ cells are a particular set of enteric neurons, we also performed immunostaining assays using anti-Nitric oxide synthase 1 (NOS1) and anti-serotonin (5HT) antibodies and observed GFP−/Hu+ neurons, finding that GFP−/Hu+ cells were a mixture of NOS1+, 5HT+ and NOS1−/5HT− neurons (Fig. S3), as enteric neurons normally are (Uyttebroek et al., 2010).
Double immunostaining for HuC/D and Sox10 expression shown in Fig. 1E indicates that Sox10 is hardly expressed in Hu+ enteric neurons. The percentage of Sox10+ cells among Hu+ cells was 0.4% at 10 dpf and 0.2% at 14 dpf (Fig. 1). The ratio of Sox10+ cells to Hu+ neurons was about 1:4 at 10 and 14 dpf (Fig. 1F), suggesting there is one Sox10+ cell per four neurons in larval distal gut.
Using Tg(217B; u:gfp) larvae at 10-15 dpf, we laser ablated one-somite-width of GFP+ enteric neurons in the intestine at the second-somite level from its end (Fig. 2A). For this, we mounted a transgenic larva under an objective and ablated GFP+ cells by irradiating with a 1/30 s infrared laser pulse (see Materials and Methods for detail). After the ablation, we observed the same larva at 1, 3 and 5 days post-ablation (dpa) (Fig. 2B-G). The series of pictures in Fig. 2C-G show the regeneration process of a representative larva among the five ablated larvae. After the removal of GFP+ cells (Fig. 2C,D), new GFP+ cells appeared in the ablated area, and their number increased over 1 to 5 dpa (Fig. 2E-G). By 5 dpa, new GFP+ cells became detectable in all operated larvae (Fig. 2H). These data demonstrate that the zebrafish ENS is highly regenerative, and that laser ablation using Tg(217B; u:gfp) is a useful method to remove enteric neurons and observe the subsequent regeneration process in living zebrafish larvae.
Enteric neurons marked with GFP are regenerated after ablation. (A) Schematic of the ablation equipment. An anesthetized zebrafish larva of Tg(217B; u:gfp) (10-15 dpf) was mounted in agar under a coverglass with its left side facing up toward the objective. The neurons at the second-somite level from the anus were removed by irradiating an infrared (IR) ray through the objective (see Materials and Methods for details). (B) Timeline of the experiments. After ablation, each larva was cultured and live imaged at 1, 3 and 5 dpa. (C-G) Recovery of GFP+ cells after ablation in the same larva. Cells before ablation are marked with magenta arrowheads in the posterior intestine (C), and they were removed as shown in (D). Imaging at 1 dpa (E), 3 dpa (F) and 5 dpa (G) revealed new GFP+ cells (yellow arrowheads). Numbers in each panel mark identified cells before and after regeneration. Cells marked with asterisks were not counted for the data presented in H, because these cells are located near the ventral midline of the gut tube and are likely to have been hidden behind the gut when ablating. White arrowhead in G indicates the identified neuron 1. Intestines are outlined with dotted lines. Images are left side views of the distal intestine, with the anterior end positioned to the left of the image. Scale bar: 50 µm. (H) Numbers of new GFP+ cells in the ablated areas of five larvae examined at 1, 3 and 5 dpa during regeneration.
Enteric neurons marked with GFP are regenerated after ablation. (A) Schematic of the ablation equipment. An anesthetized zebrafish larva of Tg(217B; u:gfp) (10-15 dpf) was mounted in agar under a coverglass with its left side facing up toward the objective. The neurons at the second-somite level from the anus were removed by irradiating an infrared (IR) ray through the objective (see Materials and Methods for details). (B) Timeline of the experiments. After ablation, each larva was cultured and live imaged at 1, 3 and 5 dpa. (C-G) Recovery of GFP+ cells after ablation in the same larva. Cells before ablation are marked with magenta arrowheads in the posterior intestine (C), and they were removed as shown in (D). Imaging at 1 dpa (E), 3 dpa (F) and 5 dpa (G) revealed new GFP+ cells (yellow arrowheads). Numbers in each panel mark identified cells before and after regeneration. Cells marked with asterisks were not counted for the data presented in H, because these cells are located near the ventral midline of the gut tube and are likely to have been hidden behind the gut when ablating. White arrowhead in G indicates the identified neuron 1. Intestines are outlined with dotted lines. Images are left side views of the distal intestine, with the anterior end positioned to the left of the image. Scale bar: 50 µm. (H) Numbers of new GFP+ cells in the ablated areas of five larvae examined at 1, 3 and 5 dpa during regeneration.
To count the number of GFP+ cells recovered in the ablated area during the regeneration process, we fixed ablated larvae at three time points (1, 5 and 10 dpa) after ablating at 13 dpf, and we performed immunostaining with an anti-GFP antibody (Fig. 3), because it was difficult to count the number of GFP+ cells in the living intestine, which moves. The average number of GFP+ cells in the one-somite-width of the ablated areas increased during regeneration and reached ∼50% of the number of GFP+ cells in the control area (the intestine at the fifth-somite level from its end) by 10 dpa. The ratio of the mean number of regenerated GFP+ cells in the ablated area to the mean number in the control area in the same larva (internal control) was 14.8:27.8 cells/somite. In the internal control area, the number of enteric neurons did not increase significantly as development progressed (1 dpa versus 10 dpa; unpaired two-tailed t-test, P=0.2) (Fig. 3E). The number of enteric neurons increased as development progressed under normal rearing conditions (14 h light and 10 h dark cycle) (Fig. S2). This difference may be due to the culture conditions in the regeneration experiment, where ablated and control larvae were cultured in the dark, a condition in which the larvae might not be able to find food well.
The number of GFP+ cells in the ablated area increases after ablation. (A) Timeline of the experiments. After ablation at 13 dpf, larvae were cultured and fixed at 1, 5 and 10 dpa for immunostaining. (B-D) Posterior intestines stained for GFP at 1 dpa (B) 5 dpa (C), and 10 dpa (D). The arrows and enclosed areas indicate the one-somite-width of intestine used for counting the GFP+ cells in the ablated area (magenta; ablated) and unablated (blue; internal control) area of the same larva at the second- and the fifth-somite level, respectively (see Materials and Methods for details). Images are left side views of the mid-distal intestines, with the anterior end positioned to the left of the image. Scale bar: 50 µm. (E) Number of GFP+ cells in ablated and internal control areas at 1, 5 and 10 dpa. Filled symbols indicate the numbers in individual larvae, and outline symbols indicate the mean±s.e.m. In the ablated areas, the number of GFP+ cells increased and reached ∼50% of the control level within 10 days, whereas a significant change in the number of cells was not observed in the control area during the same period (1 dpa versus 10 dpa; two-tailed unpaired t-test, P=0.2).
The number of GFP+ cells in the ablated area increases after ablation. (A) Timeline of the experiments. After ablation at 13 dpf, larvae were cultured and fixed at 1, 5 and 10 dpa for immunostaining. (B-D) Posterior intestines stained for GFP at 1 dpa (B) 5 dpa (C), and 10 dpa (D). The arrows and enclosed areas indicate the one-somite-width of intestine used for counting the GFP+ cells in the ablated area (magenta; ablated) and unablated (blue; internal control) area of the same larva at the second- and the fifth-somite level, respectively (see Materials and Methods for details). Images are left side views of the mid-distal intestines, with the anterior end positioned to the left of the image. Scale bar: 50 µm. (E) Number of GFP+ cells in ablated and internal control areas at 1, 5 and 10 dpa. Filled symbols indicate the numbers in individual larvae, and outline symbols indicate the mean±s.e.m. In the ablated areas, the number of GFP+ cells increased and reached ∼50% of the control level within 10 days, whereas a significant change in the number of cells was not observed in the control area during the same period (1 dpa versus 10 dpa; two-tailed unpaired t-test, P=0.2).
Taking these results together, Fig. 3 shows that the number of enteric neurons in the ablated area increased at a significantly higher rate than those in the internal control area. The data also showed that 10 days was not long enough to restore the number of enteric neurons to the level in the control area.
The entrance of enteric neural crest-derived cells and bridging with neurites in the ablated area
To observe the temporal changes in the behavior of ENCDCs during the regeneration process, we prepared another transgenic line, Tg(sox10:cre; EF1alpha:loxP-gfp-loxP-dsred), referred to as Tg(sox10:cre; EF1a:G/R) (Rodrigues et al., 2012). In this transgenic line, cells that have expressed Sox10 are labeled with DsRed permanently, because Sox10 is expressed in all neural crest cells just after segregation from the neural tube during early embryogenesis (Dutton et al., 2001). In the larvae obtained by intercrossing Tg(217B; u:gfp) and Tg(sox10:cre; EF1a:G/R), enteric neurons are labeled with GFP, and ENCDCs (including enteric neurons) are labeled with DsRed, although there is some mosaicism in both transgenic lines. The GFP signal driven by the EF1a promoter has also been detected in non-neuronal cell types, but the GFP expression in Tg(217B; u:gfp) larvae was strong enough to distinguish enteric neurons and their neurites.
Using these quadruple transgenic larvae, we laser-ablated GFP+ cells in the intestine at the second-somite level from its end at 10-15 dpf, and we took a series of live images daily until 5 dpa (Fig. 4A). Fig. 4A–G present maximum-projection images and corresponding schematic drawings of DsRed-positive (DsRed+) cells, GFP+ cells and neurites in the ablated area observed in a representative example among 12 ablated larvae. Drawings are presented because the continuous movement of the zebrafish intestine made it difficult to demonstrate cells and neurites clearly in a single projection image. Z-slices used to create these drawings are shown in Fig. S4, and all z-slices are shown in Movie 1.
Entrance of the ENCDCs into the ablated area during the regeneration process. (A) Timeline of the experiments. (B-G) Maximum projection images of the intestine at 0 dpa (B), 1 dpa (C), 2 dpa (D), 3 dpa (E), 4 dpa (F) and 5 dpa (G), showing GFP+ cells (green) and DsRed+ cells (magenta). In the Tg(217B;u:gfp)×Tg(sox10:cre; EF1a:G/R) transgenic fish shown, enteric neurons are labeled with GFP and ENCDCs are labeled with DsRed. Schematic drawings are shown below each image. Green circles outlined with magenta, GFP+/DsRed+ cells; green circles, GFP+/DsRed− cells; magenta circles, GFP−/DsRed+ cells (referred to here as ‘Red Cells’); circles with dots represent cells identified during imaging based on their relative position and color balance of GFP and DsRed; green lines represent neurites. Some neurons are GFP+/DsRed−, probably because of mosaicism of the transgenic fish. Intestines are outlined with white lines. Images are left side views, with the anterior end positioned to the left of the image. Areas marked by brackets contain after-images of cells due to intestinal movement during confocal scanning along the z-axis. Boxes indicate regions shown enlarged in panels below. (D*) Image of the area enclosed by the yellow box in D. A growth cone-like structure at the end of a neurite is shown (white arrowhead). (D′-G′) Images of the areas enclosed dashed boxes in D (D′), E (E′), F (F′) and G (G′). Two Red cells became GFP+ (marked by arrows). Scale bars: 5 µm (D*), 10 µm (D′-G′).
Entrance of the ENCDCs into the ablated area during the regeneration process. (A) Timeline of the experiments. (B-G) Maximum projection images of the intestine at 0 dpa (B), 1 dpa (C), 2 dpa (D), 3 dpa (E), 4 dpa (F) and 5 dpa (G), showing GFP+ cells (green) and DsRed+ cells (magenta). In the Tg(217B;u:gfp)×Tg(sox10:cre; EF1a:G/R) transgenic fish shown, enteric neurons are labeled with GFP and ENCDCs are labeled with DsRed. Schematic drawings are shown below each image. Green circles outlined with magenta, GFP+/DsRed+ cells; green circles, GFP+/DsRed− cells; magenta circles, GFP−/DsRed+ cells (referred to here as ‘Red Cells’); circles with dots represent cells identified during imaging based on their relative position and color balance of GFP and DsRed; green lines represent neurites. Some neurons are GFP+/DsRed−, probably because of mosaicism of the transgenic fish. Intestines are outlined with white lines. Images are left side views, with the anterior end positioned to the left of the image. Areas marked by brackets contain after-images of cells due to intestinal movement during confocal scanning along the z-axis. Boxes indicate regions shown enlarged in panels below. (D*) Image of the area enclosed by the yellow box in D. A growth cone-like structure at the end of a neurite is shown (white arrowhead). (D′-G′) Images of the areas enclosed dashed boxes in D (D′), E (E′), F (F′) and G (G′). Two Red cells became GFP+ (marked by arrows). Scale bars: 5 µm (D*), 10 µm (D′-G′).
At 0 dpa, there are neither GFP+ nor DsRed+ cells in the ablated area (Fig. 4B; Fig. S5). At 1 dpa, GFP+ neurites were found to extend into the ablated area (Fig. 4C). By 2 dpa, DsRed+ cells without GFP expression (referred to as ‘Red cells’ hereafter to distinguish from GFP+/DsRed+ double-positive cells) were first observed in the central part of the ablated area (Fig. 4D,D′). Neurites from the flanking unablated area extended further and became denser, resulting in formation of a bridge across the ablated area (‘neurite bridging’) (Fig. 4D). Some neurites had growth cone-like structures (Fig. 4D*). Fig. S6A,B show other examples of the growth cone-like structures in different larvae. To observe how neurites extend in the ablated area, we performed time-lapse imaging at 1 dpa (5 dpf) for 3 h (Movies 2,3). We used 5 dpf larvae for this experiment because 10 dpf larvae could not survive long continuous time-lapse imaging after ablation. GFP+ neurites had expanded tips that extended during observation (white arrowheads in Fig. S6C). At 3 dpa, more Red cells were observed in the central part of the ablated area (compare Fig. 4D with 4E, and 4D′ with 4E′), and some of the Red cells became neurons expressing GFP by 4 dpa (Fig. 4E-G′; Fig. S7).
To quantify the regeneration processes of neurites, we classified the 12 ablated larvae into three categories (stages I-III) based on the extent of neurite elongation (Fig. 5A), and counted the numbers of larvae of each category daily after ablation (Fig. 5B). At stage I, GFP+ neurites are not found in the central part of the ablated area. At stage II, neurites extend into the central area and, at stage III, they connected the anterior and posterior sides of the ablated area, forming a neurite bridge (Fig. 5A). To quantify the behavior of ENCDCs and neurogenesis during the regeneration process, we counted the number of larvae that had DsRed+ cells (ENCDCs) in the center of the ablated area (Fig. 5B, magenta circles) and GFP+/DsRed+ cells (enteric neurons) in the same area (Fig. 5B, green triangles). We also plotted the average numbers of DsRed+ and GFP+/DsRed+ cells in the center of the ablated area (Fig. 5C), both of which increased during the period examined. Raw cell numbers plotted in Fig. 5C are shown in Table S1.
Timecourses of the appearance of axons and ENCDCs in the ablated area during regeneration. (A) The appearance of neurites in the ablated area was classified into three stages (stages I-III) based on the degree of entry into ablated areas, as described below each schema. (B) The number of larvae at each stage for 12 operated larvae across the six timepoints examined (0 to 5 dpa). Magenta circles show the number of larvae with DsRed+ cells in the central 50% of the ablated area. Green triangles with magenta lines show the number of larvae with GFP+/DsRed+ cells in the same area. Neurite bridging was completed by 3 dpa for all operated larvae, and at that time, DsRed+ cells were also found in the central part of the ablated gut in 11 of the 12 larvae. GFP+/DsRed+ cells in the central part of the ablated area were found in most larvae (9/12) at 4 dpa. (C) The average number of DsRed+ and GFP+/DsRed+ cells in 12 operated larvae at the six time points examined (0 to 5 dpa). Magenta circles show the number of DsRed+ cells in the central 50% of the ablated area. Green triangles with magenta lines show the number of GFP+/DsRed+ cells in the same area. Data are presented as the mean±s.e.m. (D,E) Two examples of Red cells that entered into the ablated area. The Red cell in D extends its processes along the GFP+ processes (indicated by magenta arrowheads with green lines), whereas the Red cell(s) in E has/have DsRed+ processes extending independently of GFP+ neurites (magenta arrowheads). Scale bar: 25 µm. (F) Stages of GFP+ and DsRed+ processes at each day after ablation for 12 larvae (1-12). GFP+ and DsRed+ processes reached the same stage with the same timing for more than half (8/12) of larvae, while DsRed+ processes showed slight delay for four samples, suggesting there is no evidence that non-neuronal processes entered the ablated area to facilitate subsequent neurite extension.
Timecourses of the appearance of axons and ENCDCs in the ablated area during regeneration. (A) The appearance of neurites in the ablated area was classified into three stages (stages I-III) based on the degree of entry into ablated areas, as described below each schema. (B) The number of larvae at each stage for 12 operated larvae across the six timepoints examined (0 to 5 dpa). Magenta circles show the number of larvae with DsRed+ cells in the central 50% of the ablated area. Green triangles with magenta lines show the number of larvae with GFP+/DsRed+ cells in the same area. Neurite bridging was completed by 3 dpa for all operated larvae, and at that time, DsRed+ cells were also found in the central part of the ablated gut in 11 of the 12 larvae. GFP+/DsRed+ cells in the central part of the ablated area were found in most larvae (9/12) at 4 dpa. (C) The average number of DsRed+ and GFP+/DsRed+ cells in 12 operated larvae at the six time points examined (0 to 5 dpa). Magenta circles show the number of DsRed+ cells in the central 50% of the ablated area. Green triangles with magenta lines show the number of GFP+/DsRed+ cells in the same area. Data are presented as the mean±s.e.m. (D,E) Two examples of Red cells that entered into the ablated area. The Red cell in D extends its processes along the GFP+ processes (indicated by magenta arrowheads with green lines), whereas the Red cell(s) in E has/have DsRed+ processes extending independently of GFP+ neurites (magenta arrowheads). Scale bar: 25 µm. (F) Stages of GFP+ and DsRed+ processes at each day after ablation for 12 larvae (1-12). GFP+ and DsRed+ processes reached the same stage with the same timing for more than half (8/12) of larvae, while DsRed+ processes showed slight delay for four samples, suggesting there is no evidence that non-neuronal processes entered the ablated area to facilitate subsequent neurite extension.
In the central part of the ablated area, we found that at 1 dpa, most of the larvae had neurite extensions (7/12, stage II), and some already had neurite bridging (2/12, stage III). At the same time, DsRed+ cells were detected in the central part of the ablated area in four larvae (4/12). At 2 dpa, the number of larvae with neurite bridging and DsRed+ cells increased. By 3 dpa, neurite bridging was completed in all of the larvae (12/12) and DsRed+ cells were present in almost all of these (11/12). On the other hand, GFP expression (neurogenesis) began to be observed among the DsRed+ cells (ENCDCs) at 3 dpa, 2 days later than the invasion of DsRed+ cells, and this GFP expression occurred in most of the larvae (11/12) by 5 dpa.
Because entrance of GFP+ neurites and DsRed+ cells occurred almost simultaneously, we looked at the relationship between them. We found two types of Red cells (GFP−/DsRed+ cells) in the ablated area: one that extended processes along the GFP+ processes (Fig. 5D) and another that extended processes independently of the GFP+ processes (Fig. 5E), although it is possible that there are guiding processes without fluorescence due to some mosaicism in the Tg(217B; u:gfp)×Tg(sox10:cre; EF1a:G/R) larvae.
Formation of a pioneering glial cell bridge (Goldshmit et al., 2012; Mokalled et al., 2016) and facilitation of axon pathfinding by Schwann cells (Ceci et al., 2014; Villegas et al., 2012) have been reported in the regeneration of central and peripheral nervous systems, respectively, in zebrafish. Because DsRed+ cells in Tg(217B; u:gfp)×Tg(sox10:cre; EF1a:G/R) larvae marks neural crest derivatives, including non-neuronal cells, we further analyzed DsRed+ processes in the ablated area as for GFP+ neurites (Fig. 5A) to assess which appeared first in the ablated area. Fig. 5F shows the progression from stage I to III of GFP+ and DsRed+ processes for individual larvae during 0-5 dpa. This demonstrates that GFP+ and DsRed+ processes developed into the ablated area with the same timing in more than half of the larvae (8/12), while DsRed+ processes showed slight delay in the other four (4/12). The result shown in Fig. 5F suggests that formation of a non-neuronal cell bridge prior to neurite bridging is unlikely in enteric neuron regeneration in the larval zebrafish gut.
Newly appeared enteric neurons and Sox10+ ENCDCs in the ablated area were mostly generated via cell proliferation
We next examined whether enteric neurons and Sox10+ ENCDCs were generated through cell division during regeneration, by performing a BrdU incorporation assay (Fig. 6). To count the number of GFP+ and Sox10+ cells formed by cell division, we added BrdU to the medium after ablation and refreshed it daily. Ablated larvae were fixed for immunostaining at four time points (1, 3, 5 and 7 dpa), and unablated stage-matched control larvae were also fixed at the same time (Fig. 6A).
GFP+ enteric neurons and Sox10+ cells in the ablated area regenerate through induction of cell proliferation. (A) Timeline of the experiments. Ablated larvae were fixed at four time points (1, 3, 5 and 7 dpa) for immunostaining. Unablated control larvae were also fixed at the same times. (B-G) Projection images showing the staining pattern of the intestine of control unablated larvae at the level of the second somite from the end of the intestine at the 5 day timepoint. GFP+ cells (green arrowheads) and Sox10+ cells (magenta arrowheads) were BrdU−. (H-M) The staining pattern of the intestine at 5 dpa, in which enteric neurons were ablated at the level of the second somite from the end of the intestine. A Sox10+ cell (blue triangles outlined with magenta) was BrdU+. A GFP+/BrdU+ cell (blue triangles outlined with green) was observed near the Sox10+/BrdU+ cell. Single slice images are shown for ablated areas because many BrdU+ cells in the mucosal layer made it difficult to identify BrdU+ cells in the plexus layer. In B-M, markers examined are shown at the top right corner of each image (BF, brightfield). Images are left side views, with the anterior end positioned to the left of the image. Scale bar: 25 µm. (N) Ratios of GFP+/BrdU+ cells to GFP+ cells (left) and Sox10+/BrdU+ cells to Sox10+ cells (right). Ratios measured in the ablated areas are shown by circles, ratios measured at the second-somite level from the end of the intestine in unablated larvae (unablated) are shown by triangles and ratios measured at the fifth-somite level from the end of the intestine in ablated larvae (internal) are shown by squares. Numbers of larvae examined are shown below each timepoint. Data are presented as mean±s.e.m. (O) Bar graphs showing the proportion of GFP+/BrdU+ cells (left) and Sox10+/BrdU+ cells (right) at 5 dpa in control (unablated and internal) and ablated areas, expressed as a percentage of the total number of GFP+ and Sox10+ cells, respectively, in all larvae examined.
GFP+ enteric neurons and Sox10+ cells in the ablated area regenerate through induction of cell proliferation. (A) Timeline of the experiments. Ablated larvae were fixed at four time points (1, 3, 5 and 7 dpa) for immunostaining. Unablated control larvae were also fixed at the same times. (B-G) Projection images showing the staining pattern of the intestine of control unablated larvae at the level of the second somite from the end of the intestine at the 5 day timepoint. GFP+ cells (green arrowheads) and Sox10+ cells (magenta arrowheads) were BrdU−. (H-M) The staining pattern of the intestine at 5 dpa, in which enteric neurons were ablated at the level of the second somite from the end of the intestine. A Sox10+ cell (blue triangles outlined with magenta) was BrdU+. A GFP+/BrdU+ cell (blue triangles outlined with green) was observed near the Sox10+/BrdU+ cell. Single slice images are shown for ablated areas because many BrdU+ cells in the mucosal layer made it difficult to identify BrdU+ cells in the plexus layer. In B-M, markers examined are shown at the top right corner of each image (BF, brightfield). Images are left side views, with the anterior end positioned to the left of the image. Scale bar: 25 µm. (N) Ratios of GFP+/BrdU+ cells to GFP+ cells (left) and Sox10+/BrdU+ cells to Sox10+ cells (right). Ratios measured in the ablated areas are shown by circles, ratios measured at the second-somite level from the end of the intestine in unablated larvae (unablated) are shown by triangles and ratios measured at the fifth-somite level from the end of the intestine in ablated larvae (internal) are shown by squares. Numbers of larvae examined are shown below each timepoint. Data are presented as mean±s.e.m. (O) Bar graphs showing the proportion of GFP+/BrdU+ cells (left) and Sox10+/BrdU+ cells (right) at 5 dpa in control (unablated and internal) and ablated areas, expressed as a percentage of the total number of GFP+ and Sox10+ cells, respectively, in all larvae examined.
For example, in the (unablated) controls (Fig. 6B–G), GFP+ cells and Sox10+ cells were BrdU-negative (BrdU−). In the ablated area, there was a Sox10+/BrdU+ double-positive cell (Fig. 6H–M) and a GFP+/BrdU+ double-positive cell nearby. It is possible that these two cells might be produced via a cell division. Daughter cells after division move apart from each other by ∼20-70 µm in the intestine during normal development through 89-107 hours postfertilization (Kuwata et al., 2019). We next quantified the ratios of GFP+/BrdU+ and Sox10+/BrdU+ cells to GFP+ cells and Sox10+ cells, respectively (Fig. 6N). The average numbers of GFP+ and Sox10+ cells per one-somite area are shown in Table S2. In the control areas, almost no GFP+/BrdU+ and Sox10+/BrdU+ cells were found. The proportions of GFP+/BrdU+ cells and Sox10+/BrdU+ cells increased in the ablated areas after 3 dpa, reaching ∼80% of the GFP+ and ∼90% of the Sox10+ cells, respectively, at 7 dpa. These findings demonstrated that GFP+ cells in the ablated areas were formed mostly by cell proliferation in response to ablation. These data showed that the number of Sox10+ cells also increased via cell division during regeneration.
Fig. 6O shows the percentages of GFP+/BrdU+ cells and Sox10+/BrdU+ cells at 5 dpa in the control and ablated areas. Most of the GFP+ cells and Sox10+ cells in the ablated areas were BrdU+. We also observed that ∼20% of the GFP+ cells in the ablated areas were BrdU− (Fig. 6N,O), indicating that these cells did not become proliferative over the assay time. It is possible that these cells differentiated from precursor cells directly without cell division. Alternatively, some of those GFP+/BrdU− cells might have been GFP− when ablation was performed due to mosaicism.
DISCUSSION
The results of our experiments show that the ENS of zebrafish larvae has a high regenerative capacity. We live imaged the regeneration process after ablating enteric neurons in detail for the first time by visualization using transgenic fish. Our findings show that neurite bridging occurred across the ablated areas, and that ENCDCs entered the ablated areas, proliferated and differentiated into enteric neurons in response to ablation.
In zebrafish larvae, we observed neurites bridging across the ablated area at 1 dpa in some of the ablated larvae. By 3 dpa, all of the ablated larvae had completed neurite bridging. In mice, enteric neurons have been removed by applying a detergent (benzalkonium chloride) to exteriorized gut at the adult stage, and the subsequent histochemical analysis demonstrated that new neurites are located in the ablated area from 7 dpa, and that their growth continues until 30-60 dpa to cover the most of the ablated area (Hanani et al., 2003). These data suggest that covering the ablated area with neurites occurs in both mice and zebrafish during regeneration.
Our results also demonstrate that enteric neurons are generated by the entrance and proliferation of ENCDCs in the ablated area in zebrafish larvae because: (1) some Red cells in Tg(217B; u:gfp)×Tg(sox10:cre; EF1a:G/R) transgenic larvae appeared in the ablated area and started to become neurons (indicated by the expression of GFP) there during regeneration, and (2) most of the GFP+ neurons in the ablated area were BrdU+. In adult mice, new neurons are formed in the area adjacent to the ablated area in response to ablation, but not in the ablated area (Laranjeira et al., 2011). The ENCDCs in zebrafish larvae may be more capable of moving, or guiding molecules may be deposited, released or expressed in the ablated area, which would allow the ENCDCs to enter the ablated area.
It is possible that the difference in results obtained in zebrafish and mice may not reflect a difference in the regeneration capability of enteric neurons, because the regeneration studies in mice were performed in adults, whereas our studies were performed in larval stages (10-20 dpf), during which various organs including the ENS are still growing. Differences observed in regeneration capacities between these two species may be due to differences in the developmental stages examined. Nevertheless, functional analyses of the motility and attractants involved in the mechanism that directs ENCDCs to the ablated area in zebrafish larvae may help achieve regeneration of the ENS in mammals.
We have previously shown that Sox10+ cells are present in the intestine at ∼3.5-4 dpf (Kuwata et al., 2019) using an anti-Sox10 antibody (Barske et al., 2018; Williams et al., 2018). In this work, we showed that Sox10+ cells were also present in the intestine of 10 and 14 dpf (Fig. 1) larvae, as well as in the adult (Fig. S8). Recent work reported that cells expressing mRNA of the sox10 gene are not found in the gut at 3.5 dpf (El-Nachef and Bronner, 2020). The discrepancy may be due to the difference in method used to detect Sox10.
Some Sox10+ cells were located adjacent to the soma of neurons, and others were not (Fig. 1B,C,E). Both of these populations of Sox10+ cells could be candidates for enteric neuronal stem cells in the zebrafish intestine. In mice, new enteric neurons are generated from Sox10+ cells after ablation (Laranjeira et al., 2011). In addition, enteric neural stem cells have been prepared in vitro by using Sox10 as a marker in mice (Kawaguchi et al., 2010) and humans (Fattahi et al., 2016; Barber et al., 2019). Examinations of the molecular characteristics and behaviors of Sox10+ cells after ablation will provide further information about the role(s) of Sox10+ cells during regeneration.
One possible explanation is that Sox10+ cells divide and invade to become neurons. On the other hand, we have previously identified Sox10−/Hu− cells among ENCDCs (Kuwata et al., 2019), which could be another candidate for non-dividing neuronal precursors or proliferative neurogenic progenitor cells. Further analysis on the molecular nature of Sox+ cells and Sox10−/Hu− cells possibly involved in ENS regeneration is needed in future studies.
We could not rule out the effects of non-neuronal neighboring cells, such as smooth muscle cells and mucosal cells, on the regeneration process described in this study, because we could not ablate target cells specifically without affecting neighbors. In future, to examine the specific role of neuronal interactions in migration, proliferation and neurogenesis of ENCDCs, it will be important to remove only neurons, for example by using the nitroreductase (NTR) system, as well as to remove only non-neuronal cells located close to enteric neurons, and see the response of ENCDCs. It would also be interesting to test whether a specific neuronal cell type alone is regenerated when it is lost. Such experiments require some technical improvements, such as creating new transgenic lines to label a subtype of enteric neurons, and cell type-specific genetic manipulation.
There are some GFP−/Hu+ neurons in Tg(217B; u:gfp). Thus, as shown in Fig. S1, these cells could remain after ablation. We could not examine the possible role of these unablated neurons in the regeneration process.
Our results provide the first fundamental information for future research into the regeneration of enteric nervous systems using zebrafish larvae. Further analyses of cellular behavior during regeneration and the identification of signaling molecules required for regeneration in zebrafish will shed light on efficient neuronal regeneration in the mammalian gut.
MATERIALS AND METHODS
All experimental procedures were carefully done according to the guidelines issued by the Committee for Animal Experiments of the University of Hyogo. We made maximum efforts to reduce the number of fish used and the suffering.
Fish strains and maintenance
We used zebrafish (Danio rerio) transgenic lines, Tg(SAGFF(LF)217B; uas:gfp) obtained by Gal4FF gene-trap screening (Asakawa et al., 2008; Kawakami et al., 2010) and Tg(sox10:cre; EF1alpha:loxP-gfp-loxP-dsred) (a gift from Dr Robert N. Kelsh, University of Bath, UK Rodrigues et al., 2012), to visualize the enteric neurons and ENCDCs, respectively. Fish strains were maintained in our fish facility at 28.5°C under a 14 h light/10 h dark cycle. Embryos were obtained by natural spawning, and we cultured them in embryo water [60 mg of artificial sea salt (Marinetech, Osaka, Japan; Sea Life) per liter] at 28.5°C for experiments. 1 mg/l of Methylene Blue (Wako Pure Chemicals, Osaka Japan; 133-06962) was usually added as an antiseptic. Transgenic embryos were selected for operation by observation of patterns of fluorescence under a stereo-fluorescence microscope (MZ16F, Leica, Germany). Developmental staging was carried out according to time after fertilization (Kimmel et al., 1995).
Laser ablation of enteric neurons
For the ablation experiments, we cultured transgenic zebrafish larvae until 10-15 dpf. For the clear visualization of enteric neurons, feeding was stopped 24 h before ablation in order to prevent the formation of a bolus. For the experiments, a single larva anesthetized with 0.016% (w/v) ethyl 3-aminobenzoate methanesulfonate (Tricaine; MS-222; cat. #A5040; Sigma, St. Louis, MO) was mounted in low-melting-point agarose (#A9045, Sigma) under a coverslip with its left side facing up toward the objective.
For the ablation of enteric neurons, we used an infrared laser-evoked gene operator (IR-LEGO) system (Sigmakoki, Tokyo) (Kamei et al., 2009; Deguchi et al., 2009; Itoh et al., 2014) equipped with an upright microscope (model BX-51; Olympus, Tokyo) and a 1480 nm infrared laser (FiberLabs, Saitama, Japan). An infrared laser beam (operation current, 573 mA; laser power under the objective, 70 mW) through a 20× objective (numerical aperture=0.75) was irradiated to enteric neurons marked by green fluorescent protein (GFP) for 1/30 s. In this study, unless specified otherwise (see the next paragraph) we removed neurons in the gut at the second-somite level from the end of the intestine. In our current setup, we cannot ablate neurons alone without affecting neighboring tissues. Thus, other cell types such as epidermis, smooth muscle cells and mucosal cells are possibly damaged in addition to loss of enteric neurons. After the ablation, the larvae were released from the agar and cultured until use in E3 medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl2, 0.33 mM MgSO4, 5 mM HEPES, pH 7.2) in an incubator without light. The ablated larvae were fed Paramecia from 1 dpa.
For counting new neurons in the ablated area, we ablated the GFP+ cells not only from the gut adjacent to the second somite from the end of the intestine, but also from the gut of a half-somite-width extended on both sides of the second somite. After immunostaining, we counted the number of GFP+ cells in the one-somite-width area of the ablated area. For a negative control, we counted the number of GFP+ cells in the one-somite-width unablated area at the fifth-somite level from the end of the intestine of the ablated larvae (internal control).
In our assay, we counted only GFP+ cells for assessment of the regeneration process. A few GFP−/Hu+ neurons remain after ablation in some of our ablation experiments (two out of six larvae), as shown in Fig. S1. One larva had a Hu+ cell, and the other (Fig. S1C) had two in the ablated area (one-somite width). We also found that four of the six larvae had a Sox10+ cell in each ablated area (Fig. S1C), while the unablated six larvae had 6.5 Sox10+ cells on average in the control area (Fig. S1A).
BrdU incorporation assay
After the ablation of enteric neurons described above, the medium for larvae was replaced each day with 0.3 mg/ml of freshly prepared 5-bromo-2′-deoxyuridine (BrdU; #B9285; Sigma) and 0.5% DMSO in E3 medium. For protection of BrdU from the light, we incubated larvae in an incubator without light. For immunostaining to detect GFP, Sox10 and BrdU, the BrdU-treated larvae were anesthetized with 0.016% (w/v) of MS-222 and fixed with 4% paraformaldehyde.
For the assessment of BrdU incorporation in GFP+ cells and in Sox10+ cells, we examined the ratio of BrdU+ cells to GFP+ or Sox10+ cells in the one-somite-width area of the ablated area. For a negative control, we examined GFP+ or Sox10+ cells in the gut at the second-somite level from the end of the intestine in unablated larvae (unablated control) and in the one-somite-wide unablated area at the fifth-somite level from the end of the intestine in the ablated larvae (internal control).
Immunostaining
Immunostaining was performed according to our previous protocol (Moly and Hatta, 2011), with the addition of three times 1 h incubation in distilled water at room temperature. Fixed larvae were permeabilized by storing in 100% methanol at −20°C overnight or longer. The primary antibodies used were mouse anti-GFP (Thermo Fisher Scientific, Waltham, MA, USA; A11120; 1:500 dilution), chicken anti-GFP (Aves Labs, Tigard, OR, USA; GFP-1010; 1:500 dilution), rat anti-BrdU (Abcam, Cambridge, UK; ab6326; 1:500 dilution), rabbit anti-Sox10 (Gene Tex, Irvine, CA, USA; GTX128374; 1:500 dilution), rabbit anti-RFP (MBL, Nagoya, Japan; PM005; 1:500 dilution) and mouse anti-HuC/D (Thermo Fisher Scientific; A21271; 1:500 dilution). The secondary antibodies used were goat anti-mouse IgG-Alexa Fluor 488 (Thermo Fisher Scientific; A11001; 1:1000 dilution), goat anti-chicken IgY-FITC (Aves labs; F-1005; 1:1000 dilution), goat anti-rat IgG-Alexa Fluor 405 (Abcam; ab175673; 1:1000 dilution), goat anti-rabbit IgG-Alexa Fluor 555+ (Thermo Fisher Scientific; A32732; 1:1000 dilution) and goat anti-mouse IgG-Alexa Fluor 647 (Thermo Fisher Scientific; A21235; 1:1000 dilution). Stained samples were observed using a Leica SP8 inverted confocal microscope installed with Leica LASX software. Images were processed using ImageJ (ver. 1.52n; U.S. National Institutes of Health, Bethesda, MD).
Live imaging
For taking a serial images during regeneration in the transgenic line Tg(SAGFF(LF)217B; uas:gfp), we anesthetized larvae with 0.016% (w/v) MS222 and mounted them in 1% low-melting-point agarose on the bottom of a glass-bottom dish (Hatta et al., 2006). Mounted samples were observed under an upright Zeiss AxioPlan2 microscope equipped with an Olympus DP72 camera. For live imaging of Tg(SAGFF(LF)217B; uas:gfp)×Tg(sox10:cre; EF1alpha:loxP-gfp-loxP-dsred), mounted samples were observed using a Leica SP8 inverted confocal microscope, and images were processed with LASX and ImageJ software. For the time-lapse imaging of regenerating larvae, we additionally used a stage-top incubator (Tokai hit, Shizuoka, Japan) equipped on the SP8 to maintain the temperature in the chamber at 28.5°C. Because ablated 10 dpf larvae did not survive for a long time, we used 5 dpf larvae for this purpose. The ablation was performed as in 10 dpf larvae, as described above.
Estimating the number of enteric neurons in the intestine of larvae and adult fish
We used HuC/D antibodies to detect the enteric neurons. Whole zebrafish intestines were dissected out, and they were processed for immunoreaction. The stained intestine was mounted between a slide glass and a coverslip to make it as flat as possible, and then observed using a Leica SP8 microscope. For middle and distal intestine at 5, 10, 15 and 20 dpf, the average density of HuC/D+ cells from three sites (see Fig. S2 for an example) was examined. For proximal intestine, because HuC/D+ cells were distributed unevenly, all cells were counted. For the intestine of 1, 2 and 4 months postfertilization (mpf) fish, we examined the density of HuC/D+ cells at one site per each segment of tri-fold intestine, and calculated the average HuC/D+ numbers. For all samples, we counted HuC/D+ cells only on a half side and multiplied with the area of the flattened intestine. The total numbers of HuC/D+ cells in an intestine were estimated by doubling the counts obtained above.
Acknowledgements
We thank Dr Robert N. Kelsh for providing us with a transgenic line, Tg(sox10:cre; EF1alpha:loxP-gfp-loxP-dsred).
Footnotes
Author contributions
Conceptualization: M.N., K.H.; Validation: M.N.; Formal analysis: M.O., M.N., K.H.; Investigation: M.O., M.N., N.H., K.H.; Resources: K.K.; Writing - original draft: M.N.; Writing - review & editing: M.N., K.H.; Visualization: M.O., M.N., N.H., K.H.; Supervision: M.N., K.H.; Project administration: M.N., K.H.; Funding acquisition: M.N., K.K., K.H.
Funding
This work was supported by Sumitomo Foundation to M.N., Japan Society for the Promotion of Science KAKENHI 20HP8021, NBRP and NBRP/Fundamental Technologies Upgrading Program from the Japan Agency for Medical Research and Development (AMED) to K.K., and Japan Society for the Promotion of Science KAKENHI 16K06998 to K.H.
Peer review history
The peer review history is available online at https://dev.biologists.org/lookup/doi/10.1242/dev.195339.reviewer-comments.pdf
References
Competing interests
The authors declare no competing or financial interests.