Centrally-projecting DRG axons in zebrafish exhibit regenerative period closure

To develop a model to study the genetic mechanisms of regenerative periods, we first determined if developing zebrafish exhibit reduced regenerative capacity with age. We investigated the regenerative capacity of centrally-projecting DRG axons because of their established regenerative period in mammals and clinical relevance to brachial plexus injuries in humans. To visualize the DRG and its central axon bundle we expressed GFP using ngn1 promoter elements and visualized the glial limitans (spinal cord boundary) with mCherry driven by the gfap promoter in Tg(ngn1:GFP);Tg(gfap:NTR-mCherry) animals (Fig 1A and 1B) [49,50]. We assayed re-invasion and re-entry with a published analysis protocol, orthogonal displacement quantifications, where the dorsal tip (growth cone) of the GFP⁺ regenerating axon, if present, was located in the orthogonal image relative to the mCherry⁺ glial limitans (Fig 1A) [43]. Plotting the fluorescent intensity profile at the tip of the GFP⁺ axon allowed us to determine if the regenerated axon was medial or lateral to the mCherry⁺ glial limitans (Fig 1C). Utilizing our confocal laser-pulse lesioning system, we axotomized the central axon bundle of a single DRG per animal in 10–12 animals at 2, 3 and 5 dpf. We then re-imaged these DRG 24 hours post-injury (hpi) (Fig 1B) to perform the orthogonal displacement quantifications (Fig 1C) and establish when the regenerative period closes. This lesioning system targets select diffraction-limited z-planes with scalable laser pulses to minimize injury to surrounding tissues. We confirmed that such injuries cause a complete transection of the centrally-projecting nerve while limiting damage to the surrounding area (Fig 1D). Zebrafish DRG pioneer axons extend, invade and bifurcate in the spinal cord between 40–48 hpf, while also adding additional axons in the nerve after the initial pioneer axon invades. Injured DRG axons must re-navigate to the dorsal root entry zone (DREZ) and then re-invade across the spinal cord boundary to re-enter the CNS and re-connect with spinal circuits. From the orthogonal displacement quantification analyses, we observed that 80% (n = 10) of axons injured at 2 dpf were medial to the glial limitans by 24 hpi, indicating successful re-entry/regeneration into the spinal cord. In contrast, only 10% (n = 10) of axons injured at 3 dpf had regenerated across the glia limitans (Fig 1E). We could not detect regenerated axons medial to the glial limitans after injuries performed at 5 dpf (n = 12). Consistent with previous reports, once an axon failed to re-invade it retracted to the cell soma [42]. These data indicate that zebrafish centrally-projecting DRG axons experience a regenerative period that rapidly closes in the 24 hours between 2 and 3 dpf and thereby establishes zebrafish as a model to investigate the mechanisms that close the regenerative period.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Centrally-projecting DRG axons experience a decline in regenerative capacity between 2 and 3 dpf.

A) A schematic of orthogonal displacement quantifications. A line is drawn across where the distal tip of the axon and spinal cord boundary meet to quantify the fluorescent profiles of each marker. This quantification is used to determine if previously axotomized DRG central axons regenerate by scoring re-entry into the spinal cord. B) Max z-projection images and of 2, 3, and 5 dpf Tg(ngn1:GFP) DRG axons pre-injury (0 hpi), post-injury (0.1 hpi) and 24 hpi (scale bar = 5 µm). Yellow circles on the pre-injury image indicate where the laser lesioning system was targeted. C) Orthogonal sideview images from the 24 hpi captures from each age group (injured at 2, 3 and 5 dpf). These images depict the right side of the spinal cord (pink) and the tip of the regenerative axon (cyan). The adjacent graphs are the orthogonal displacement quantifications, plotting the fluorescent profiles of the axon and spinal cord boundary, displaying if the axon has re-entered the spinal cord boundary (axon fluorescent peak is left or inside of the spinal cord boundary fluorescent peak) or failed to re-enter (axon peak is right/outside of spinal cord boundary peak). White arrowheads point to the distal tip of the axon where these fluorescent profiles were obtained. Pink and cyan arrowheads point to the fluorescent peak of the spinal cord boundary and axon, respectively. D) Representative max projection images displaying the DRG and spinal cord (glial limitans) before and after axotomy. Post-injury images display that axotomy limits excessive damage to surrounding tissues (scale bar = 10 µm). E) 24 hpi outcomes of the DRG central axons when injured at 2 dpf (n = 10 animals, 1 central axon per animal), 3 dpf (n = 10 animals), and 5 dpf (n = 12 animals). Comparison of the percentages of re-entry and failed re-entry were made between age groups using Fisher’s Exact tests. Raw data information for this figure can be found in the S1 Data.

https://doi.org/10.1371/journal.pgen.1012033.g001

Glial cells re-organize during regenerative period closure

To determine the cellular mechanisms that close the regenerative period, we first explored if there were structural changes in the tissue between 2 and 3 dpf. Several glial cell populations, including Schwann cells, satellite glial cells, oligodendrocyte progenitors and astrocytes have been shown to respond to injury [51–55]. To determine the differential response of glia between 2 and 3 dpf, we used Tg(ngn1:GFP);Tg(sox10:nls-Eos) animals to visualize the GFP⁺ DRG neurons and centrally-projecting axons and photoconverted sox10⁺ glial nuclei surrounding the DRG (Fig 2A) [56]. In timelapse imaging after axotomy, we calculated the migration distances of the axon (based on growth cone position) and the dorsal-most glial nucleus surrounding the DRG soma (Fig 2B and S1–S2 Movies). These migration distance measurements started when regenerative axons began their re-extension at 11.6 hours (5 min intervals, 140 timepoints, Fig 2B). The glial nuclei in 3 dpf animals were consistently positioned more dorsal to the DRG compared to 2 dpf animals (Fig 2A–2B). Furthermore, glia migrated dorsally throughout the timelapse in 3 dpf animals, with an average position of 10.85 ± 2.801 µm (SEM, n = 7) dorsal from the dorsal border of the DRG neuron over the 140 timepoints following axon re-extension (Fig 2B–2C and S2 Movie). In stark contrast, the average glial nuclear position in 2 dpf animals was 5.27 ± 3.049 µm dorsal to the DRG (SEM, n = 7) (Fig 2C). This indicates that glia surrounding the DRG differentially re-organize and undergo a significant dorsal migration in 3 dpf animals.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Glial cells surrounding the DRG re-organize during regenerative period closure.

A) Max projection images of DRG and centrally-projecting axons (cyan), and glial nuclei (pink) of Tg(ngn1:GFP);Tg(sox10:nls-Eos) animals pre-injury/pre-mock injuries (0 hpi), following injury/mock injury (0.3 hpi), and during the 24 hpi timelapse of 2 dpf, 3 dpf, and mock ablated 3 dpf animals. Yellow dashed boxes indicate the location of the DREZ. White arrowheads indicate the regenerating central axons’ location, and white dashed boxes indicate the location of the mock injury (scale bars = 10 µm). B) Line graphs displaying the migration distances (± SEM) of the axon (cyan) and the dorsal-most glial nuclei (pink) surrounding the DRG for 140 timepoints (5 min intervals) after axon re-extension begins or at 4 hour post mock ablation (n = 7 axons and glial nuclei per group in injured animals, n = 6 for mock group). C) Quantification of the average distance of the axon and dorsal-most glial nucleus (± SD) over the 140 analyzed timepoints, measured from the top of the DRG. Comparisons between average axon and glial migration distances were made with Mann-Whitney T-Tests (n = 7 per injured group and n = 6 for the mock group). D) Total migration distances of glial nuclei per group (± SEM), calculated by subtracting the x, y positions of the dorsal-most glial nuclei at t140 from t0 over the 140 analyzed timepoints per animal (n = 7 per injured group and n = 6 for mock group). The migration distances were compared with Kruskal-Wallis One-Way ANOVA (p = 0.05). E) Representative images of 3 dpf Tg(ngn1:GFP);Tg(sox10:nls-Eos) animals where the neighboring and dorsal glial nuclei surrounding the DRG were ablated prior to central axon axotomy. Panels show pre-injury and ablation (0 hpi), post ablation and axotomy (0.3 hpi), and 24 hpi with a regenerated central axon (scale bar = 10 µm). F) Post-injury outcomes in 3 dpf Tg(ngn1:GFP);Tg(sox10:nls-Eos) animals where glial cells were ablated, not ablated, or mock ablated (lesioning targeted outside the nuclei). Fisher’s Exact tests compared percentages of re-entry and failed re-entry outcomes between groups. Raw data information for this figure can be found in S1 Data.

https://doi.org/10.1371/journal.pgen.1012033.g002

This glial re-organization could be a result of axotomy or a part of an innate developmental process. To distinguish between these possibilities, we performed mock injuries (outside the glial nuclei and central axon, Fig 2A) in 3 dpf animals and observed that glia were an average of 10.94 ± 3.150 µm dorsal to the DRG (SEM, n = 6, Fig 2C) over the 140 measured timepoints. This migration was comparable to axotomized 3 dpf animals, consistent with the idea that axotomy itself is not required for glial re-organization. Quantifications of the total dorsal migration distances (t0-t140) in 3 dpf animals that were either mock-injured (7.209 ± 1.617 µm (n = 6)) or injured (8.081 ± 0.8504 µm (n = 7)) showed that, when compared with 2 dpf injured animals, a similar glial re-organization does not occur at 2 dpf (2.119 ± 0.9025 µm (n = 7))(Fig 2D). The simplest explanation for this data is that glia surrounding the DRG inherently re-organize at 3 dpf along the central DRG axon, which intriguingly coincides with regenerative period closure.

To address if glia surrounding the DRG and central axon impact the regenerative period, we used the same laser pulse lesioning settings to ablate neighboring/dorsal glial nuclei in 3 dpf animals prior to central axon axotomy (1 DRG per animal) (Fig 2E). We observed that 33.33% (n = 27 animals) of DRG central axons regenerated by 24 hpi in glia-ablated animals, while only 11.11% (n = 18 animals) regenerated in animals that did not undergo glial ablations (p = 0.0003, Fisher’s Exact test) (Fig 2F). Furthermore, to rule out the possibility that non-specific photoablation, rather than specific glial ablation, induced regeneration, we performed mock glial ablations targeted outside the glial nuclei prior to central axon axotomy. These mock ablations had no effect on regenerative capacity (12.5% re-entered n = 8 animals, Fig 2F), similar to non-glial ablated animals (p > 0.9999, Fisher’s Exact test). These collective results support a hypothesis in which the re-organization of glia surrounding the DRG in 3 dpf animals takes part in closing the regenerative period for centrally-projecting DRG axons.

netrin1b increases in glial cells dorsal to DRG during regenerative period closure

To identify molecular mechanisms that regulate regenerative period closure, we searched for transcripts expressed in glia that surround the DRG neurons. It was previously demonstrated that the invasion of developmental pioneer DRG central axons across the DREZ is regulated by the DCC receptor and its downstream signaling cascade involving cAMP and Rac1 [44]. By assaying the DCC ligand, Netrin-1, with hybridization chain reaction (HCR), we surprisingly detected that netrin1b was enriched in cells surrounding the DRG neurons (Fig 3A and 3B). We also detected netrin1b in the spinal cord floor plate, consistent with its well-established role in axon guidance in the spinal cord (Fig 3C and 3D) [57,58]. To address if netrin1b is differentially expressed during regenerative period closure, we visualized netrin1b expression in Tg(ngn1:GFP) animals at 2 and 3 dpf. Quantitative analysis showed an increase in netrin1b expression dorsal to DRG neurons at 3 dpf compared to 2 dpf, coincident with the glial re-organization and regenerative period closure (Fig 3B). In contrast, the expression of netrin1b in the spinal floorplate was not different between 2 and 3 dpf animals (Fig 3C and 3D). To determine if netrin1b changes after injury, we also assayed its expression 10 hpi in animals injured at 3 dpf and 4 hpi in animals injured at 2 dpf; these times represent the respective midpoint of regeneration attempts (when axons are approximately 50% through the regenerative process) at these ages. This quantification revealed that elevated levels of netrin1b expression dorsal to the DRG were maintained 10 hpi in 3 dpf animals, but expression levels remained low 4 hpi in 2 dpf animals (Fig 3B). These data indicate higher netrin1b availability directly in the regenerative axon’s path when the regenerative period is closed in 3 dpf animals. To confirm that glial cells were responsible for the robust netrin1b expression we observed around the DRG, we assayed its expression in Tg(ngn1:GFP);Tg(sox10:nls-Eos) animals after sox10⁺ nuclei surrounding one DRG were ablated in 3 dpf animals (Fig 3E). Prior to fixing the animals 4 hours post ablation (hpa), we confirmed glial cell death by the long-term loss of nuclear fluorescence and/or presentation of debris. HCR and quantitative analysis in these animals revealed a significant loss of netrin1b transcripts around DRG that underwent glial ablation compared to non-glia ablated DRG in the same animals (Fig 3F), indicating that glia are responsible for this robust expression.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Netrin-expressing glia re-organize and increases netrin1b dorsal to the DRG during closure of the regenerative period.

A) Max projection images of 2 and 3 dpf Tg(ngn1:GFP) animals after hybridization chain reaction (HCR) staining for netrin1b (scale bar = 10 µm). White dashed tracings indicate how the dorsal expression of netrin1b was quantified – traced from the top of the DRG to the DREZ signified by the dorsal longitudinal fasciculus. B) Quantifications of netrin1b expressed dorsal to the DRG in animals at 2 dpf, 2 dpf – 4 hpi, 3 dpf, and 3 dpf – 10 hpi (± SEM). The mean grey value of netrin1b puncta dorsal to the DRG minus background was compared between groups with Kruskal-Wallis One-Way ANOVA (n = 6 animals per group). Red bars on graph indicate DRG 4 was axotomized. (DRG number represents the position of the DRG from anterior to posterior. Anterior-most DRG is 1.) C) Quantification of the mean grey values of netrin1b in the spinal floor plate adjacent to each side of the DRG between 2 and 3 dpf animals (± SEM) (n = 6 animals per group). Mean grey values of floor plate expression was obtained with a 200 µm² box for each measurement. Comparisons were made with Kruskal-Wallis One-Way ANOVA. D) Representative z-cropped max projection images (scale bar = 10 µm) displaying the netrin1b expression in the spinal floor plates of 2 and 3 dpf animals. Dashed white boxes indicate where the floor plate expression was obtained with 200 µm² boxes. E) Images of netrin1b HCR in glia ablated and non-glia ablated DRG in 3 dpf animals 4 hours post ablation (hpa) (scale bar = 10 µm). F) Quantification of netrin1b expressed dorsal to DRG minus background signal (± SEM). Red bar indicates DRG 5 had glia ablated. Kruskal-Wallis One-Way ANOVA (n = 6). G) Representative images of dcc HCR staining in DRG of Tg(ngn1:GFP) animals at 2 dpf and an axotomized DRG at 3 dpf – 10 hpi (scale bars = 5 µm). H) Quantification of the number of dcc puncta in DRG neurons divided by DRG area (± SEM) in 2 dpf (n = 20), 3 dpf (n = 20), and 3 dpf – 10 hpi (n = 13) animals. Red bar indicates DRG 4 was axotomized. Kruskal-Wallis One-Way ANOVA. I) Representative images of 2 dpf Tg(ngn1:GFP);Tg(sox10:gal4 + myl7:GFP) animal injected with UAS:dcc-tdTomato pre-injury (0 hpi) as well as images throughout the timelapse (scale bar = 5 µm). White arrowheads indicate the DCC-tdTomato puncta in the DRG as well as the maintained position of the axon at the DREZ. J) Quantification of regeneration rates of DCC-tdTomato⁺ DRG (n = 5 animals, 1 DRG per animal) and uninjected animals injured at 2 dpf and assayed 24 hpi (n = 8 animals, 1 DRG per animal). Fisher’s Exact test was used to compare re-entry and failed re-entry rates between groups. Raw data information can be found in S1 Data.

https://doi.org/10.1371/journal.pgen.1012033.g003

To address if the expression of DCC, a receptor for Netrin, was changing during these ages, we performed HCR for dcc transcripts (Fig 3G). We fixed Tg(ngn1:GFP) animals at 2 and 3 dpf, as well as 10 hpi in 3 dpf animals. Quantification of dcc expression in DRG neurons revealed no statistical difference in dcc expression between 2 and 3 dpf animals (Fig 3H). However, in the injured context, particularly in the DRG that had undergone central axon axotomy, neuronal dcc expression was significantly increased 10 hpi in 3 dpf animals (Fig 3G–3H). These findings, coupled with our conclusion that glia dorsally migrate/re-organize and therefore increase the glial expression of netrin1b dorsal to the DRG, support a model in which both dcc and netrin1b are spatiotemporally organized in the tissue to close the regenerative period at 3 dpf.

One possible hypothesis to explain these results is that differential upregulation of dcc after injury closes the regenerative period. This hypothesis would predict that artificially increasing dcc expression would prematurely close the central axon’s regenerative period. We tested this by overexpressing dcc in the DRG and assaying regeneration in animals injured at 2 dpf, when the regenerative period is normally still open. To do this, we injected UAS:dcc-tdTomato into Tg(ngn1:GFP);Tg(sox10:Gal4 + myl7:GFP) animals to enable UAS-driven overexpression [44,59]. We previously demonstrated this approach expresses a functional DCC tagged with tdTomato [44]. We specifically axotomized tdTomato⁺ DRG and assayed regeneration within 24 hpi (Fig 3I). Each of these DRG central axons regenerated within 24 hpi (n = 5), inconsistent with the hypothesis that DCC upregulation is sufficient to close the regenerative period (Fig 3J). Since netrin1b-expressing glia do not re-organize until 3 dpf, the premise of overexpressing netrin at 2 dpf to prematurely close the regenerative period is also inconsistent with such a hypothesis. These results led us to test the hypothesis that limiting DCC signaling could re-open the regenerative period in 3 dpf animals, after the regenerative period has closed and when netrin1b-expressing glia have re-organized dorsally (surrounding the regenerative axon).

Antagonizing DCC-cAMP-Rac1 signaling re-opens the regenerative period

To determine if the DCC signaling axis is part of a genetic mechanism that closes the regenerative period of DRG central axons at 3 dpf, we targeted each step of the DCC-cAMP-Rac1 signaling process and scored regeneration. We hypothesized that antagonistic manipulations of this DCC signaling axis would enhance central axon regeneration because during development it modulates invadopodia that are essential for pioneer axon invasion [44]. To test this, we used Tg(ngn1:GFP);Tg(gfap:NTR-mCherry) animals to perform central axon axotomy on a single DRG per animal at 3 dpf and used genetic or molecular manipulations of each step of the DCC-cAMP-Rac1 signaling process for 24 hours after the injury (Fig 4A). At 24 hpi, the injured DRG central axon was re-imaged to assess regeneration into the spinal cord via orthogonal displacement quantifications (Figs 4C–4F and S1A1–S1F). In WT animals treated with only the vehicle control (1% DMSO), 0% of the central axons regenerated by 24 hpi (n = 10) (Fig 4B). However, in heterozygous dcc mutants (dcc^+/–) (ZM130198 [60]) bathed in 1% DMSO, 55% of axons regenerated back into the spinal cord by 24 hpi (n = 16) (p < 0.0001, Fisher’s Exact test comparing re-entry and failed re-entry rates, Fig 4B). This ZM130198 allele incorporates a 5.2 kb retrotransposon insertion into the 5’ UTR of dcc (106 bp upstream of the start codon) and was selected because it results in >90% reduction of dcc mRNA in homozygous mutant zebrafish [60]. Additionally, treatment at the time of injury with 25 µM rp-cAMP [61], an antagonistic cAMP analogue, or 1 µM NSC23766 [62], a Rac1 inhibitor, resulted in 50% (n = 10) and 55% (n = 20) of axons to regenerate into the CNS, respectively.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Antagonistic manipulations of the DCC-cAMP-Rac1 signaling process re-opens the regenerative period.

A) Schematic of the agonistic (blue) and antagonistic (green) manipulations of each step of the DCC-cAMP-Rac1 signaling pathway. B) Quantifications of the percentage of outcomes 24 hpi under each manipulation. See key below 4A. Comparisons were made on the basis of central axon re-entry and failed re-entry using Fisher’s Exact tests between treatment groups and DMSO vehicle controls. DMSO n = 10, dcc^+/– n = 16 (p < 0.0001); rp-cAMP n = 10 (p < 0.0001); NSC23766 n = 20 (p < 0.0001); rp-cAMP + NSC23766 n = 12 (p < 0.0001), sp-cAMP + NSC23766 n = 14 (p < 0.0001); dcc^+/– + sp-cAMP + NSC23766 n = 12 (p < 0.0001); sp-cAMP n = 19 (p = 0.0734); dcc^+/– + sp-cAMP n = 16 (p = 0.0012). C and E) Max projections of Tg(ngn1:GFP) animals pre-injury (0 hpi), post-injury (0.05-1.0 hpi) and 10-24 hpi in DMSO control and antagonistic treatment groups (scale bar = 10 µm). Treatment conditions for each are listed above each panel. Yellow arrows indicate bifurcated axons in panels where bifurcation is visible. D and F) DRG axons’ orthogonal images and orthogonal displacement quantifications. White arrowheads indicate the axons’ location in the orthogonal images. Cyan and pink arrowheads indicate the fluorescent peak of the axon and spinal cord boundary, respectively. G) Representative quantification of axon length (distance of growth cone from the DRG) during 22 hpi timelapses during failed (DMSO treated) and H) successful regeneration (rp-cAMP treated) attempts. I) Max projections of Tg(ngn1:GFP) 5 dpf animals pre-injury, post-injury (0.5 hpi) and 24 hpi treated with rp-cAMP + NSC23766 (scale bar = 10 µm). J) DRG axons’ orthogonal image and orthogonal displacement quantification. White arrowheads indicate where the fluorescent profile of the axon and spinal cord boundary were measured. Cyan and pink arrowheads on the graph indicate the fluorescent peak of the axon and spinal cord boundary, respectively. K) Quantification of the percentage of outcomes 24 hpi in rp-cAMP + NSC23766 (n = 10) versus DMSO (n = 7) control treated groups. Fisher’s Exact test compared central axon re-entry and failed re-entry rates. L) Target graph of the x, y positions of the dorsal-most Sox10 IHC stained DRG glia relative to the dorsal border of the DRG (X at the center). WT Uninjured 3 dpf (72 hpf) animals (n = 8), 12 hpi WT DMSO treated animals (n = 10), 12 hpi dcc^+/- DMSO treated animals (n = 7), 12 hpi rp-cAMP treated animals (n = 7), and 12 hpi NSC23766 treated animals (n = 7). The graph displays dashed (grey) lines in 5 µm increments. Raw data information can be found in S1 Data.

https://doi.org/10.1371/journal.pgen.1012033.g004

To determine if these molecules function in a similar signaling pathway to regulate the regenerative period, we performed double manipulations and determined genetic epistasis (Figs 4B–4F and S1A–S1F). Treatment with both rp-cAMP and NSC23766 resulted in 75% re-entry (n = 12), which was similar to either treatment alone (rp-cAMP vs rp-cAMP + NSC23766 p = 0.6020, Fisher’s Exact test, Figs 4B and S1A–S1B). This supports the hypothesis that cAMP and Rac1 function in the same molecular pathway. Treating with 100 µM sp-cAMP [63], a cAMP agonist, resulted in 5.26% axon regeneration (n = 19) (Figs 4B and S1C–S1D). Co-treating with sp-cAMP and NSC23766 however, increased central axon regeneration to 57% (n = 14), indicative of Rac1 epistatically functioning downstream of cAMP. Combined manipulations of sp-cAMP, the cAMP agonist, in dcc^+/– mutants reduced regeneration rates to 12.5% (n = 16) (Figs 4B and S1E–S1F), while co-treating with sp-cAMP and NSC23766 in dcc^+/– animals improved regeneration back to 50% (n = 12) (p < 0.0001, Fisher’s Exact test, Fig 4B–4F), indicating that the regenerative period is regulated by this signaling pathway in the order of DCC-cAMP-Rac1.

To test if such antagonistic manipulations of DCC-cAMP-Rac1 can re-open the regenerative period beyond 3 dpf, rather than simply extending it, we performed single central axon injuries in 5 dpf Tg(ngn1:GFP);Tg(gfap:NTR-mCherry) animals. These animals were either co-treated with rp-cAMP and NSC23766 or with 1% DMSO alone as a vehicle control for 24 hpi. This combination treatment was selected because it produced the highest regeneration rates at 3 dpf (Fig 4B). Additionally, it enabled us to add the pharmacological agents at 5 dpf, allowing the regenerative period to fully close before we tested if we could re-open it and promote regeneration. After 24 hpi, re-imaging and orthogonal displacement quantifications (Fig 4I–4J) confirmed that rp-cAMP and NSC23766 co-treatment (n = 10) resulted in 30% central axon re-entry compared to 0% in DMSO controls (n = 7) (Fig 4K). These findings are consistent with the idea that antagonistic manipulation of DCC-cAMP-Rac1 signaling can re-open the regenerative period.

One potential hypothesis for how these antagonistic manipulations are affecting the regenerative period is that they affect glial organization, which could influence regeneration independent of their effects on the regenerative axon. To test if antagonistic manipulations are affecting glial organization, we fixed WT uninjured 3 dpf (72 hpf) animals, as well as axotomized 3 dpf animals 12 hours after axotomy and treatment with 1% DMSO, 25 µM rp-cAMP, or 1 µM NSC23766. We also did this 12 hpi in dcc^+/– animals treated with DMSO. We then performed immunohistochemistry (IHC) against Sox10 [64] to label glial nuclei surrounding the DRG. The x, y positions of the dorsal border of the DRG and the center of the dorsal-most Sox10 stained glial nuclei were obtained to quantify glial organization. The distance between the DRG and glial nuclei was calculated and plotted on target graphs (Figs 4L and S1G–S1I). These data demonstrated that neither dcc mutation nor treatment with rp-cAMP or NSC23766 affected glial re-organization compared to controls (S1I Fig), inconsistent with the hypothesis that the DCC-cAMP-Rac1 pathway modulates the regenerative period by altering glial organization. Taken together, these data indicate that this signaling pathway functions in the order of DCC-cAMP-Rac1 to close the regenerative period, and that antagonistically targeting this signaling pathway can re-open the period and promote regeneration.

Antagonizing the DCC-cAMP-Rac1 signaling pathway increases invadopodia stability

Centrally-projecting DRG axons must navigate back to the DREZ to regenerate, during which filopodia are the predominant growth cone structure guiding its migration [44,65]. Once at the DREZ, invasion across the spinal cord boundary requires specialized actin-based growth cone structures called invadopodia, marked by robust actin accumulations and orthogonal basal protrusions [43–45,48,65]. The DCC-cAMP-Rac1 signaling pathway has been shown to spatiotemporally limit invadopodia formations during pioneer central axon development, until the pioneer axon growth cone reaches the DREZ and stable invadopodia assemblies are required for invasion into the CNS [44]. Based on our observations that glial re-organization and increased netrin presentation dorsal to the DRG coincides with regenerative period closure, we hypothesized that the regenerative period closes because DCC-cAMP-Rac1 signaling prevents the formation of stable invadopodia [42,65]. To determine the underlying mechanism by which DCC signaling closes the regenerative period, we manipulated each molecule in the pathway and assayed invadopodia. To visualize and quantify growth cone actin dynamics as a proxy for invadopodia formation and stability, we used Tg(sox10:Gal4 + myl7:GFP);(UAS:LifeAct-GFP) animals, in which LifeAct-GFP fluorescently labels filamentous actin in DRG cells, including the growth cone [66]. In 3 dpf animals, we axotomized the centrally-projecting axon of a single DRG per animal and then performed timelapse imaging (22–24 hpi, 5 min intervals) and measured the fluorescence intensity of actin at the center of the regenerative growth cone in each capture (S3–S7 Movies). We characterized invadopodia stability according to the duration of the fluorescent peaks in actin (above the average fluorescence, calculated over the entire timelapse). Peaks of 20–35 min were transient, unstable invadopodia, 40–55 min were intermediate invadopodia, and long duration peaks of 60–70 + mins were stable invadopodia (see key above Fig 5A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Antagonistic manipulations of the DCC-cAMP-Rac1 cascade stabilizes invadopodia at the DREZ.

A) Max projections of Tg(sox10:Gal4 + myl7:GFP);(UAS:LifeAct-GFP) DRG post-axotomy (0.1 hpi) and the regenerative growth cones during 24 hpi timelapses. White dashed boxes indicate the growth cone positioned at the DREZ, yellow arrows point to bifurcated axons. Treatment conditions for each are listed at the top of each panel. Quantifications of growth cone (GC) LifeAct intensity measurements in 3 representative growth cones under DMSO control and antagonistic manipulations over the 22-24 hpi timelapses. The grey portions of these line graphs indicate when the growth cone is navigating, and the black portions indicate when the growth cone was at the DREZ. Colored brackets under peaks of LifeAct fluorescence represent their duration. See key above. B) Quantifications of LifeAct peak durations only when the growth cone was at the DREZ (± SEM). C) Quantification of all LifeAct peak durations throughout the timelapses. In B & C, comparisons between groups were made with Brown-Forsythe and Welch ANOVA tests. D) Representative images of Tg(sox10:Gal4 + myl7:GFP);(UAS:LifeAct-GFP) with orthogonal side views of the growth cones filopodia and invadopodia. White brackets in these side view images indicate orthogonal basal widths. E) Quantification of the orthogonal basal widths of filopodia and invadopodia (± SD) (25 growth cone measurements for each). Mann-Whitney T-test. F) Single z-plane images and orthogonal views of Tg(sox10:Gal4 + myl7:GFP);(UAS:LifeAct-GFP) growth cones 10 hpi treated with NSC23766 stained for Cortactin (IHC). G) Schematic of agonistic (blue) and antagonistic (green) manipulations at each step of the DCC-cAMP-Rac1 pathway regulating invadopodia stability. All scale bars = 10 µm. Raw data information for this figure can be found in S1 Data.

https://doi.org/10.1371/journal.pgen.1012033.g005

We first confirmed that WT axons do not form stable invadopodia after injury in 3 dpf animals, as previously reported (DMSO vehicle controls, Fig 5A and S3 Movie) [42]. We then repeated the experiments while dcc, cAMP and Rac1 were manipulated (Fig 5A). Quantification of invadopodia revealed that reduction of DCC (in dcc^+/– animals), antagonism of cAMP (rp-cAMP), and inhibition of Rac1 (NSC23766) all increased invadopodia stability compared to the appropriate controls (Fig 5A and S4–S6 Movies). Since we hypothesized that regeneration across the DREZ requires the formation of stable invadopodia, probing if these manipulations alter invadopodia stability (actin peak durations) specifically at this boundary is essential. Notably, the shift in stability from each manipulation was particularly evident when comparing only the actin peak durations produced when the growth cone was at the DREZ and stable invasive structures would be required for regeneration (Fig 5B). While DMSO controls (n = 8) averaged invadopodia with a duration of 41.04 ± 4.777 min at the DREZ, invadopodia in dcc^+/– animals averaged 69.44 ± 9.020 min (n = 7) (Fig 5A–5B and S4 Movie). Further, rp-cAMP treatment, the cAMP antagonist, enabled invadopodia durations that averaged 76.58 ± 9.272 min (n = 6) at the DREZ (S5 Movie), while the cAMP agonist, sp-cAMP, averaged 31.90 ± 2.544 min (n = 7) (Figs 5A–5B and S2A). Moreover, in NSC23766, the Rac1 inhibitor, invadopodia averaged 68.33 ± 9.804 min (n = 7) in duration at the DREZ (Fig 5A–5B and S6 Movie). To determine if these manipulations regulated invadopodia globally rather than at the precise spatial location of invasion (DREZ), we displayed the distribution and frequency of all actin peak durations and compared invadopodia durations throughout the entire navigation process, including at the DREZ, between groups (Fig 5C). These results demonstrated that the stabilizing effects on invadopodia were present at the DREZ (Fig 5B) and not throughout the entire navigation process (Fig 5C). In contrast to the antagonistic manipulations of the pathway, agonistic manipulations of DCC signaling recapitulated the WT phenotype and produced primarily transient invadopodia formations (Figs 5A–5C and S2A).

To test if these molecules function in a common signaling process, we again performed combination treatments for epistasis analysis while measuring invadopodia during regeneration. Co-treatment with sp-cAMP and NSC23766 in WT animals resulted in invadopodia durations averaging 80.28 ± 16.30 min (n = 5) at the DREZ, indicating Rac1 functions downstream of cAMP in regulating invadopodia stability (Fig 5A–5B). Similar to sp-cAMP alone (S7 Movie), sp-cAMP treatment in dcc^+/– animals reduced invadopodia durations at the DREZ, averaging 31.55 ± 2.708 min (n = 7), indicating cAMP functions downstream of DCC (Figs 5B and S2A); however, co-treating with both sp-cAMP and NSC23766 in dcc^+/– animals increased invadopodia durations at the DREZ to 59.17 ± 7.413 min (n = 7) (Fig 5A–5B). Together, these findings indicate that invadopodia stability is regulated by this signaling pathway in the order of DCC-cAMP-Rac1 during regeneration, corresponding with our re-entry analyses.

In timelapse movies, we could detect the key morphological differences between a filopodia-based growth cone and an invadopodia-based growth cone (Fig 5D) [42–44,65]. During invadopodia formations, the robust, long duration, actin accumulations at the center of the regenerative growth cone were accompanied by orthogonal basal protrusions toward the DREZ, which measured significantly wider in orthogonal width measurements compared to filopodia (Fig 5D–5E) (n = 25 growth cone measurements for both filopodia and invadopodia across 5 biological replicates). While previous work has shown that long duration LifeAct-GFP peaks with basal-projecting structures are indicative of invadopodia assembly, we used a complementary approach to identify mature invadopodia [43,44,46]. Cortactin is enriched at invadopodia in growth cones and cancer cells, acting as a key regulator of invadopodia assembly and maturation, with important functions in scaffolding and signaling [67,68]. We therefore axotomized a DRG central axon of 3 dpf Tg(sox10:Gal4 + myl7:GFP);(UAS:LifeAct-GFP) animals and fixed them 10 hpi to perform IHC against Cortactin. Prior to fixing, these animals were treated with the Rac1 inhibitor (NSC23766), as this treatment reliably produced stable invadopodia and maximized the likelihood of observing invadopodia post fixation. From this, we observed Cortactin localized to invadopodia-like structures in the growth cone (Fig 5F). Taken together, these data support the hypothesis that DCC signaling closes the regenerative period by actively reducing the stability of invadopodia formations (Fig 5G).

Cell-specific manipulation of DCC and Rac1 alter invadopodia stability

Based on HCR expression analysis of dcc and netrin, we next hypothesized that DCC signaling functions autonomously in DRG. To test this, we first manipulated DCC cell-specifically using the functional DCC construct, UAS:dcc-tdTomato, injected at the one-cell stage of Tg(sox10:Gal4 + myl7:GFP);(UAS:LifeAct-GFP); dcc^+/– embryos to increase expression of dcc specifically in DRG cells [44]. Since a specific driver that expresses reliably in DRG neurons vs glia has not been identified in zebrafish, this experimental paradigm tested the specific requirement in DRG cells. To provide more specificity, we selected mosaic animals that express tdTomato in neurons but we could not completely rule out glial expression. The dcc^+/– animals exhibit increased invadopodia stability and regenerative outcomes, so we predicted that increasing dcc expression would reduce invadopodia stability and regenerative potential to WT levels compared to uninjected dcc^+/– animals. At 3 dpf, the central axon of a DCC-tdTomato⁺ DRG was axotomized and then time lapsed for 24 hours to record and measure invadopodia stability (S8–S10 Movies). We then scored if these regenerative axons either maintained their position at the DREZ or subsequently retracted to discern if regenerative outcomes were correspondingly affected.

We first confirmed that we could detect tdTomato puncta in a DRG of injected animals prior to injury, post-injury and throughout the timelapse (Fig 6A–6C). In the dcc^+/– animals, DRG expressing DCC-tdTomato displayed reduced invadopodia stability (S8 Movie), averaging 31.46 ± 2.472 min (n = 4) at the DREZ, recapitulating invadopodia durations seen in injected WT animals (31.90 ± 2.372 min, n = 4, S9 Movie) and uninjected WT animals (29.35 ± 2.058 min, n = 4) (Fig 6A, 6B and 6D). These results were distinct from uninjected dcc^+/– animals without DCC supplementation when comparing invadopodia durations at the DREZ (averaging 57.00 ± 8.162 mins at the DREZ, n = 3, Fig 6D), but this distinction was lost when compared cumulatively (S3A Fig). These data demonstrate that DCC functions, specifically in DRG cells, to destabilize invadopodia. While the cell autonomy of cAMP could not be determined, we tested if manipulating cAMP during this cell-specific manipulation of DCC altered invadopodia. To do this we treated dcc^+/– animals expressing DCC-tdTomato with rp-cAMP, the cAMP antagonist (Fig 6C and S10 Movie). This treatment confirmed that rp-cAMP is capable of reducing downstream signaling of DCC, increasing invadopodia durations to 54.07 ± 7.478 min (n = 5) at the DREZ (Fig 6D).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Cell-specific manipulations of DCC and Rac1 indicate cell-autonomous regulation of invadopodia stability.

A-C) Max projections of DCC-tdTomato⁺ puncta (white arrowheads) in Tg(sox10:Gal4 + myl7:GFP);(UAS:LifeAct-GFP) DRG pre-injury, the DRG post-injury, and images of the growth cone during 24 hpi timelapses in (A) dcc^+/– animals, (B) WT siblings and (C) dcc^+/– animals treated with rp-cAMP. The genotype and treatment condition are listed above each panel. The white dashed boxes indicate the growth cone positioned at the DREZ. Yellow arrows point to bifurcated axons. Quantification of growth cone LifeAct intensity measurements in 3 growth cones under each condition. The grey portion of these line graphs indicate that the growth cone is navigating, and black portions indicate it is at the DREZ. Colored brackets beneath LifeAct fluorescent peaks indicate their duration (see key below 6C). D) Quantification of LifeAct peak durations only when the growth cone was at the DREZ (± SEM). Kruskal-Wallis One-Way ANOVA test. E) Quantification of the proportion of regenerative axons that remained at the DREZ or retracted by the end of the 24 hpi timelapse between DCC-tdTomato⁺ DRG of dcc^+/– (n = 7), wildtype (n = 6) and dcc^+/– animals treated with rp-cAMP (n = 5). Chi-Squared tests. F-H) Max projections of pa-Rac1-mCherry⁺ puncta in Tg(sox10:Gal4 + myl7:GFP);(UAS:LifeAct-GFP) DRG (white arrow heads) pre-injury, the DRG post-injury, and images of the growth cone during 24 hpi timelapses in injected and (H) uninjected animals treated with rp-cAMP. The injection status and exposure conditions to the activating 445 nm light are listed above each panel. The white dashed boxes indicate the growth cone positioned at the DREZ. Yellow arrows point to bifurcated axons. Quantification of growth cone LifeAct intensity measurements in 3 growth cones under each condition. The grey portion of these line graphs indicate when the growth cone is navigating, and black portions indicate when it is at the DREZ. Colored brackets under the LifeAct fluorescent peaks indicate their duration (see key below 6B). I) Quantifications of the actin peak durations while growth cones were at the DREZ (± SEM). Kruskal-Wallis One-Way ANOVA test. J) Quantification of the proportions of regenerative axons that remained at the DREZ or retracted by the end of the 24 hpi timelapse between pa-Rac1-mCherry⁺ DRG of rp-cAMP treated animals that were exposed (n = 7) or were not (n = 6) exposed to 445 nm light and uninjected animals treated with rp-cAMP and exposed to 445 nm light (n = 6). Chi-Squared tests. Raw data information for this figure can be found in S1 Data.

https://doi.org/10.1371/journal.pgen.1012033.g006

To next determine if DRG-specific manipulations also altered the regenerative capacity, we scored if these regenerative axons either maintained their position at the DREZ or subsequently retracted by the end of the 24 hour timelapse. For technical reasons because of fluorescent reporter conflict (e.g., DCC-tdTomato vs gfap:mCherry), we could not perform orthogonal displacement quantifications with the glial limitans. While an axon maintaining its position at the DREZ 24 hpi is indicative of regeneration, axonal retraction is a direct indicator of regenerative failure. Therefore, the proportions of these outcomes were compared between groups to quantify regenerative potential. These measurements showed that untreated animals with DCC-tdTomato-expressing DRG had significantly more axons that subsequently retracted (Fig 6E). Untreated dcc^+/– animals (n = 7) only had 28.57% of axons remain at the DREZ, and WT siblings (n = 6) had 50% remain (Fig 6E). These findings are particularly impactful given our previous demonstration that dcc mutation reduces central axon regenerative failure. In contrast, dcc^+/– animals treated with rp-cAMP (n = 5) had 80% of their axons remain at the DREZ (Fig 6E). These results are consistent with the idea that DCC functions, specifically in the DRG cells, to close the regenerative period by destabilizing invadopodia.

To test if Rac1 activity, specifically in the DRG, controls the regenerative period, we injected a photoactivatable Rac1 construct (UAS:pa-Rac1-mCherry) into Tg(sox10:Gal4 + myl7:GFP);(UAS:LifeAct-GFP) embryos, allowing us to again manipulate the signaling cascade in DRG cells [69]. As we previously published, pa-Rac1 is activated by 445 nm light [43,44,69]. Therefore, uninjected animals, as well as injected animals without 445 nm light exposure, could be used as controls. Since our previous experiments showed that rp-cAMP stabilizes invadopodia dynamics during regeneration and that Rac1 functions downstream of cAMP in the signaling process, we hypothesized that pa-Rac1-mCherry⁺ DRG exposed to 445 nm light would reverse rp-cAMP-induced invadopodia stability back to WT/untreated levels. We also predicted that photoactivated Rac1 would correspondingly reduce regenerative potential. Uninjected animals treated with rp-cAMP and exposed to 445 nm light (every 5 min) maintained the ability to form long-duration, stable, invadopodia at the DREZ (averaging 55.54 ± 6.162 min, n = 6) (Fig 6H and 6I), which was comparable to the pa-Rac1-mCherry⁺ DRG treated with rp-cAMP unexposed to 445 nm light (averaging 49.26 ± 4.284 min at the DREZ, n = 6) (Fig 6F and 6I and S11 Movie). However, pa-Rac1-mCherry⁺ DRG treated with rp-cAMP that were exposed to 445 nm light displayed destabilized invadopodia dynamics (averaging 33.61 ± 2.463 mins at the DREZ, n = 7, Fig 6G and 6I and S12 Movie), which was distinct from the durations quantified at the DREZ in the unexposed group (Fig 6F-6I). These exposed and unexposed pa-Rac1-mCherry⁺ DRG were not significantly different on a cumulative basis (S3B Fig).

To test the cell-autonomous effects of Rac1 activity on regenerative potential, we again scored the proportions of regenerative axons that either maintained its position at the DREZ or retracted by the end of the 24 hpi timelapse (Fig 6J). All animals in this analysis were treated with rp-cAMP. We predicted that photoactivation of pa-Rac1 (exposure to 445 nm light) would negatively affect regenerative potential compared to unexposed animals and uninjected animals. Here, we observed that only the animals/DRG that had pa-Rac1 photoactivated had significantly higher retraction rates (57.14%, n = 7, Fig 6J). This was in stark contrast with unexposed/unactivated pa-Rac1⁺ DRG (n = 6), in which only 16.67% of their axons retract (83.33% remained at the DREZ, Fig 6J). Furthermore, uninjected animals (n = 6) that were exposed to 445 nm light had 100% of their axons remain at the DREZ (Fig 6J). This indicates that 445 nm light exposure does not inherently affect regenerative axons, but photoactivation of pa-Rac1 does. Collectively, these data support the model that Rac1 functions autonomously in DRG cells to reduce the stability of invadopodia, which correspondingly hampers regenerative potential.

Opening the regenerative period by antagonizing DCC signaling restores somatosensory circuits

Regenerative period closure prevents the functional recovery of somatosensory circuits between the DRG and spinal cord after injury. To determine if re-opening the regenerative period with manipulations of DCC signaling restores somatosensory circuits, we utilized the activity of a genetically encoded calcium indicator expressed in the DRG and spinal neurons, GCaMP6s, as a proxy for neuronal activity upon exposure to a sensory stimulus. DRG neurons exhibit Ca²⁺ transients that are detected by increases in GCaMP6s fluorescence [70–72]. To assess the restoration of the somatosensory circuit, we determined if spinal neurons exhibited GCaMP6s transients in a synchronous manner with DRG neurons upon a cold-water stimulus (4°C) (S13 Movie) [71]. We hypothesized that the antagonistic manipulations of DCC signaling that re-opens the regenerative period would also enable the re-establishment of somatosensory circuits. In such a case, synchronized neuronal activity between DRG and spinal neurons would be observed. This assay is limited to the temporal resolution of GCaMP6s dynamics, which fluoresces quickly upon neuronal activation and has a delayed decay of fluorescence intensity [73]. We observed that an absence of synchrony between DRG neurons and spinal neuron transients was exhibited by a considerable delay of neuronal activation (~2–3 seconds) in the spinal neuron population. Z-scores above 2.0 were used to identify active neurons [72].

To determine if somatosensory-induced behaviors are being restored from our manipulations of the DCC signaling pathway, we measured circuit function in animals that had 8 consecutive DRG axons axotomized (DRG #4–11). This published assay creates a discernable difference in circuit and behavior dynamics after exposure to sensory stimuli [42,74]. We first performed series axotomies to 8 DRG (DRG #4–11) in 3 dpf Tg(NeuroD:Gal4 + myl7-GFP);(UAS:GCaMP6s) and Tg(sox10:mRFP) animals [46,75,76]. Animals were treated for 24 hours and then recovered untreated for an additional 24 hours (Fig 7A–7B). At 48 hpi, we performed the Ca²⁺ imaging assay with an evoked cold-water stimulus to assess the synchrony of previously axotomized DRG and spinal neuron activity as an indication of functionally restored sensory circuits (Fig 7C). All animals included in this assay displayed at least one DRG neuron per ganglia that was rapidly active upon cold water exposure (Fig 7D). In uninjured animals, an average of 97.78 ± 2.222% (n = 7) of spinal neurons displayed activity that occurred rapidly, synchronized with the activity of the DRG (Fig 7E and S13 Movie). However, the injured DMSO vehicle treated and untreated control animals exhibited a significant reduction of this synchrony, averaging 71.90 ± 8.724% (n = 7) and 64.24 ± 11.69% (n = 7) of active spinal neurons in synchrony with the DRG, respectively (Fig 7E and S14 Movie). Similarly, injured animals treated with the cAMP agonist, sp-cAMP, averaged 68.94 ± 11.370% (n = 5) of spinal neuron activity synchronized with the DRG (Fig 7E). This was in stark contrast to the antagonistically manipulated groups (Fig 7E). In dcc^+/– animals, 89.41 ± 3.293% (n = 6) of spinal neurons were synchronized with the DRG (Fig 7E and S15 Movie). In rp-cAMP and NSC23766 treated animals, 92.67 ± 4.760% (n = 5) and 93.59 ± 2.722% (n = 5) of spinal neurons were synchronized with the previously injured DRG, respectively (Fig 7E and S16 Movie). While the number of DRG neurons responding per DRG was variable and dependent on GCaMP6s expression, no difference was detected in the propensity for spinal neurons to exhibit synchronized activity between cases of one DRG neuron versus multiple neurons being activated. Together this data is consistent with the idea that re-opening the regenerative period of centrally-projecting DRG axons enables the restoration of somatosensory circuits.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Antagonistically manipulating DCC signaling promotes the recovery of synchronous activity between previously axotomized DRG neurons and spinal neurons.

A-B) Timeline of the sensory circuit assay, wherein following series axotomy (B) the animals were treated with manipulations of the DCC-cAMP-Rac1 pathway for 24 hpi and then allowed to recover for another 24 hrs. C) Representative max projection images and quantification of the GCaMP6s calcium indicator transients after room temperature (23°C) and cold (4°C) water additions. White dashed boxes indicate the DRG. Individual traces (lines) represent GCaMP6s transients in a single neuronal soma over 120 seconds. DRG neurons are plotted at the bottom of the graph and spinal neurons above. The color of the trace indicates the characterized activity based on z-score change (see key below 7A). The vertical line in front of “z-score = 2.0” indicates the vertical increase in the individual traces (lines) needed to reach a z-score of 2.0 and be considered an activated neuron. D) Quantification of the activity response of DRG neurons per animal in each treatment group. Comparisons were made based on the proportions of Delayed Activity and Rapid Activity between groups with Fisher’s Exact tests. E) Quantification of spinal neuron responses in each treatment group immediately following cold water exposure. Comparisons were made with Fisher’s Exact tests based on the proportions of Delayed Activity and Synchronous Activity with the DRG between groups. For 7D and 7E Uninjured animals n = 7, injured No Treatment (Trx) n = 7, injured DMSO n = 7, injured dcc^+/– n = 6, injured rp-cAMP n = 5, injured NSC23766 n = 5, injured sp-cAMP n = 5. In 7E, Fisher’s Exact tests were used to compare the proportions of Delayed versus Synchronous spinal neuron activity, and when compared to Uninjured animals: injured No Trx p < 0.0001; injured DMSO p < 0.0001; injured dcc^+/– p = 0.2305, injured rp-cAMP p = 0.4895; injured NSC23766 p = 0.4922; injured sp-cAMP p < 0.0001. Raw data information for this figure can be found in S1 Data.

https://doi.org/10.1371/journal.pgen.1012033.g007

Opening the regenerative period recovers somatosensory-induced behaviors

Functional recovery of somatosensory circuits after injury would also be expected to restore behavioral responses [42]. When exposed to cold (4°C) water, larval animals display a shivering behavior consisting of head and tail movements without forward locomotion. This behavior depends on intact DRG circuits in zebrafish both in regenerative and developmental contexts [42,44]. At 3 dpf, Tg(ngn1:GFP);Tg(gfap:NTR-mCherry) animals underwent the series axotomy and 24 hpi treatments to manipulate DCC-cAMP-Rac1 (Fig 8A-8B). After 24 hpi treatment, the animals were allowed to recover for another 24 hours with no additional treatment (Fig 8A). After 48 hpi, DRG #4–11 were re-imaged in each animal to perform orthogonal displacement quantifications and determine how many of the 8 injured central axons regenerated (Fig 8D), and then animals were subjected to the behavioral assay. In the assay, animals were first immersed in room temperature water (23°C) and then cold water (4°C), and 600 frame movies were collected (50 ms intervals) in each water condition. To compare groups, the number of frames in which animals displayed the shivering behavior (described above) were measured. We hypothesized that antagonistic manipulation of DCC signaling would re-open the regenerative period and enable functional recovery of somatosensory circuits, causing the animals to display longer shivering times. Uninjured controls displayed shivering in ~50% of the frames (tan region of Fig 8C and 8E). However, at 48 hpi, injured DMSO treated controls averaged 28.18 ± 4.697% (n = 11) of frames spent shivering (pink region of Figs 8C and S4). In these animals, 25 ± 2.919% of axotomized DRG axons had regenerated (pink region of Fig 8D). Similarly, agonistic treatments of sp-cAMP, sp-cAMP in dcc^+/– animals, and DMSO control treatment in WT siblings had central axon regeneration rates of 29.61 ± 2.901%, 17.86 ± 2.845%, and 28.57 ± 4.024%, respectively (blue region of Fig 8D). Correspondently, these animals had reduced shivering times, shivering in 33.63 ± 3.659% (n = 19), 35.11 ± 4.331% (n = 14), and 36.68 ± 6.266% (n = 11) of frames, respectively (blue region of Figs 8C and S4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Antagonistically manipulating DCC signaling enables the recovery of somatosensory-induced behaviors.

A-B) A timeline of the behavioral assay, wherein following series axotomy (B) the animals were treated for 24 hpi to manipulate the DCC-cAMP-Rac1 signaling pathway then allowed to recover for 24 hours. At 48 hpi DRG were re-imaged and then animals underwent the behavioral assay. C) Quantification of the average percentage of time spent shivering in uninjured and injured animals under each manipulation (± SEM). Kruskal-Wallis One-Way ANOVA. D) Quantification of central axon regeneration rates per animal (± SEM). Kruskal-Wallis One-Way ANOVA. For 8C and 8D Uninjured WT n = 17, Uninjured dcc^+/– n = 16, Uninjured DMSO n = 16, Uninjured rp-cAMP n = 17, Uninjured NSC23766 n = 16, Uninjured sp-cAMP n = 15, DMSO n = 11, dcc^+/– n = 20, dcc^+/– + rp-cAMP n = 14, rp-cAMP n = 15, NSC23766 n = 15, sp-cAMP + NSC23766 n = 15, WT siblings n = 11, dcc^+/– + sp-cAMP n = 14, sp-cAMP n = 19. E) 2-sec overlay examples of the shivering phenotype in uninjured WT, injured WT and injured dcc^+/– animals (48 hpi) in cold (4°C) water. F) Representative traces (red lines) of animal locations (tracked by head displacement) 48 hpi in 23°C and 4°C water. G) Quantification of head displacement between the two water conditions and genotypes. Mann-Whitney T-test compares swimming distances between WT and dcc^+/– animals in 23°C water 48 hpi (p = 0.6857). Paired T-tests compares head displacement per frame between the two water conditions in WT animals (p = 0.0049) and dcc^+/– animals (p = 0.0154) (n = 4 in each group). Raw data information can be found in S1 Data.

https://doi.org/10.1371/journal.pgen.1012033.g008

These results are in stark contrast with the antagonistic manipulations (green regions of Figs 8C, 8D, and S4). The dcc^+/– animals had central axon regeneration rates of 45.45 ± 3.728% and shivered 49.83 ± 4.513% (n = 20) of the time (Fig 8C–8D). Furthermore, rp-cAMP treatment alone, as well as rp-cAMP in dcc^+/– animals, resulted in regeneration rates of 53.33 ± 5.652% and 50.78 ± 3.864%, respectively, and shivered 52.47 ± 4.593% (n = 15) and 59.20 ± 6.879% (n = 16) of the frames, respectively (Fig 8C–8D). Rac1 inhibition with NSC23766 alone, as well as in co-treatment with sp-cAMP resulted in regeneration rates of 50.83 ± 4.307% and 55.00 ± 3.819%, respectively (Fig 8D). These animals also displayed longer shivering times, 59.00 ± 6.489% (n = 15) and 44.40 ± 3.752% (n = 15) of frames, respectively (Fig 8C). While many of the animals spent much of their time in the 23°C water stationary, all animals were of good health and were capable of swimming. In animals that were stationary in the 23°C, a nose touch was used to elicit an escape response to confirm the capability of swimming. Analysis of WT and dcc^+/– animals 48 hpi between water conditions demonstrate that the shivering phenotype is a distinct response that is not caused by an inability to swim (Fig 8E–8G). Taken together, these results indicate that antagonistic manipulations of DCC-cAMP-Rac1 signaling re-opens the regenerative period and enhances DRG central axon regeneration. These manipulations correspondently recover somatosensory-induced behaviors, indicative of improved functional regeneration of sensory circuits between previously axotomized DRG and the CNS.

Limitations

Our quantifications of glial re-organization were completed by tracking the migration of glial cell nuclei which is indicative but not resolute in establishing the scale of migration for glia. While our transgenic labeling of glial nuclei is also unable to distinguish specific glial cell types, based on our data and the location in the tissue, it is likely that both satellite glia and Schwann cells are the relevant sources of netrin that contribute to regenerative period closure. Observations of such peripheral glial populations expressing netrin have been made [104,105]. While we do not have the tools to manipulate netrin in a cell-specific manner and thus cannot fully test the source of Netrin, ablation of glial cells (nuclear ablation) surrounding the DRG significantly reduced the amount of netrin1b transcripts we observed at 3 dpf. These ablations to surrounding glia alone also significantly enhanced DRG central axon regeneration in 3 dpf animals. While outside the scope of this paper, expressing netrin ectopically would definitively test the hypothesis that it closes the regenerative period. The analysis in cell-autonomous manipulations was limited to scoring if axons maintained their position at the DREZ, which correlates with regeneration, but does not confirm CNS re-entry directly. Cell-specific experiments also did not distinguish between neuronal and glial molecular function. We were also unable to interrogate the cell-autonomy of cAMP because the genetic modulator is not known. While we demonstrated that the stabilizing effects on invadopodia by rp-cAMP treatment were capable of overcoming cell-specific DCC-induced destabilization and could be reversed by cell-specific activity of Rac1, further characterizations of cAMP in DRG would be beneficial in understanding its role in regenerative period closure.

Ethics Statement

Experimental procedures complied with the NIH guide for care and use of laboratory animals. The University of Notre Dame Institutional Animal Care and Use Committee (IACUC) approved all experiments (protocol 19-08-5464), which is guided by the United States Department of Agriculture, the Animal Welfare Act (USA) and the Assessment and Accreditation of Laboratory Animal Care International (AALAC).

Experimental model and subject details

Animal specimens.

Zebrafish used in this study were: Tg(ngn1:GFP) [49], Tg(gfap:NTR-mCherry) [50], Tg(sox10:nls-Eos) [56], Tg(sox10:Gal4 + myl7:GFP) [59], Tg(UAS:LifeAct-GFP) [66], Tg(NeuroD:Gal4 + myl7:GFP) [46], Tg(UAS:GCaMP6s) [75], and Tg(sox10:mRFP) 76. Stable mutant lines were generated by crossing dcc^zm130198 [60] into relevant transgenic backgrounds. All embryos were generated via pairwise matings and grown at the recommended 29°C in our satellite fish facility incubator. After 24 hpf, zebrafish embryos were maintained in PTU (0.0003%) to reduce pigmentation for in vivo imaging. The age of our animals for experiments were determined by developmental stages and hpf (hours post fertilization) [106] (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Key resources table.

https://doi.org/10.1371/journal.pgen.1012033.t001

Statistical analysis

Statistical analyses were completed with Prism. No statistical methods were used to predetermine sample sizes, but sample sizes were similar to previous publications. Statistical tests were performed with biological replicates (animals) rather than technical replicates. No data points were excluded for these analyses. Characteristics for inclusion for analyses were predetermined for calcium imaging, actin dynamics, and glia ablations (DRG central axon regeneration) as discussed in those sections. Prior to all experiments, healthy animals were selected at random. Each experiment in each condition was repeated at least once to confirm results. All data collected and analyzed are presented in this study.

Software

Slidebook, Prim, ImageJ, Microsoft Excel, and Adobe Illustrator were used to acquire, analyze and compile figures. Microsoft Word was used in the construction of this manuscript. Google Gemini was used to edit the manuscript. Mendeley was used to create and organize citations.