Timely Schwann cell division drives peripheral myelination in vivo via the laminin/cAMP pathway

Schwann cells (SCs) are the myelinating glia that ensure efficient nerve impulse conduction along the nerves of the peripheral nervous system (Jessen and Mirsky, 2005; Woodhoo and Sommer, 2008; Herbert and Monk, 2017; Bouçanova and Chrast, 2020; Raphael et al., 2011; Pereira et al., 2012; Sherman and Brophy, 2005). In order to myelinate, SCs go through a series of developmental changes that include (1) migration and division along peripheral axons, (2) intensive proliferation prior to axonal ensheathment such that one SC myelinates one axonal segment and (3) substantial cytoskeletal rearrangements and changes in cell shape that allow a SC to radially sort an axon in a 1:1 ratio, a process termed radial sorting (Monk et al., 2015; Raphael and Talbot, 2011; Feltri et al., 2016). The latter enables SCs to extend their processes along a unique abaxonal-adaxonal polarity axis in order to select and sort an axon to myelinate (Tricaud, 2018). It has been shown in vitro that SCs first secrete and assemble their own basal lamina (Eldridge et al., 1989; Colognato and Tzvetanova, 2011) that would later on regulate, through receptor interactions, different signals required for SC proliferation, survival and differentiation (Chen and Strickland, 2003; Nodari et al., 2008, 2007; Court et al., 2009; Yu et al., 2005; Yamada et al., 1996). Of particular interest are the extracellular matrix (ECM) proteins laminins and collagens that play essential roles in SC development (Chernousov et al., 2008).

However, several issues regarding the characteristics of timely SC division during migration and radial sorting, as well as the coupling of division with ECM proteins and myelination in vivo, remain unresolved.

The mitotic spindle is a bipolar array of microtubules that mediates chromosome separation during cell division (Hara and Fukagawa, 2020; Petry, 2016). The organization and temporal assembly of the spindle are all important features that dictate the outcome of division (Lu and Johnston, 2013). One crucial aspect of cell division is the temporal control of mitosis; this is highlighted by the existence of a mitotic spindle checkpoint that ensures accurate mitotic spindle organization prior to anaphase (Lara-Gonzalez et al., 2012). A great number of proteins is involved in spindle organization; one such major player is Sil (also known as Stil), which is ubiquitously expressed and specifically localizes to the poles of the mitotic spindle in metaphase cells (Pfaff et al., 2007; Campaner et al., 2005). The sil gene was originally cloned from leukemia-associated chromosomal translocation and is overexpressed in several tumor types (Aplan et al., 1991, 1990). It is associated with autosomal recessive primary microcephaly (MCPH), a neurogenic mitotic disorder that results in significantly reduced brain size (Naveed et al., 2018; Zaqout et al., 2017). Studies of the zebrafish cassiopeia mutant (csp^−/−), which has a nonsense mutation in, and therefore loss of function of sil, show abnormalities in prometaphase progression within retinal neuroepithelium (Novorol et al., 2013). Thus, sil represents an ideal candidate to study for a role in mitotic synchronization and division during SC development.

Here, we used live imaging, genetics and pharmacological tools in zebrafish to monitor SC behavior during its division and unraveled for the first time an essential role for timely Schwann cell division in peripheral myelination. We identified Sil as a crucial regulator of SC radial sorting and myelination. Time-lapse imaging revealed an important role for sil in SC metaphase progression; loss of Sil resulted in delayed mitotic exit and caused a complete loss of axonal ensheathment as revealed by transmission electron microscopy (TEM). Treating csp^−/− with forskolin, which binds to adenyl cyclase and elevates the levels of cAMP, or forcing Laminin 2 expression was sufficient to fully restore peripheral myelination.

In order to monitor SC division, we took advantage of the Tg(foxd3:gfp) line (Gilmour et al., 2002) and imaged SCs along the posterior lateral line nerve (PLLn) during early migration, i.e. between 24 and 42 h post-fertilization (hpf) and during radial sorting (between 48 and 60 hpf) (Fig. 1A) (Lyons et al., 2005; Raphael et al., 2011). SCs were generally elongated during migration until they went through mitotic rounding, divided and then re-elongated (Fig. 1B, Movie 1). SCs divided along the anterior-posterior (AP) axis of the embryo and it took an average of 9.81±0.63 min (mean±s.e.m.) for them to complete cytokinesis. To assess the importance of this pattern of division in SC myelination, we monitored their behavior in the zebrafish cassiopeia mutant (csp^−/−). Sil is a ubiquitously expressed protein that specifically localizes to the mitotic spindle in metaphase cells and plays a crucial role in its organization (Pfaff et al., 2007; Novorol et al., 2013). SCs struggled to exit mitosis in csp^−/− with an average of 87.20±4.72 min for cytokinesis to be completed (Fig. 1B,C, Movie 2). However, re-elongation of SCs occurred normally after division in both mutants and controls (Fig. 1B). The same result was obtained during radial sorting; SCs became elongated along the axons until they went through mitotic rounding, leading to division, and then re-elongated again (Fig. 1D). Cells divided along the AP axis of the embryo and it took an average of 9.30±0.61 min for cytokinesis to complete (Fig. 1E, Movie 3). SC cytokinesis in csp^−/− took much longer to complete with an average of 90.10±14.13 min (Fig. 1D,E, Movie 4).

This result suggests that Sil deficiency leads to a significant delay in SC cytokinesis.

sil is ubiquitously expressed but highly enriched in neural cells and csp^−/− embryos show major neuronal defects, such as an increase in the percentage of mitotically arrested neuroepithelial cells in the retina that coincides with an increase in apoptosis (Novorol et al., 2013). To test whether this is the case for SCs, we first labeled SCs using the M-phase marker PH3 in the Tg(foxd3:gfp) line. As seen in the retina (Novorol et al., 2013), we observed a significant increase in the fraction of proliferating PH3⁺ SCs at 48 hpf. This coincided with a decrease in the total number of SCs and an increase in the percentage of PH3⁺ SCs relative to the total number of SCs counted within a defined region (Fig. 2). However, the number of PH3⁺ SCs and the percentage of PH3⁺ SCs relative to the total number of SCs were normal at 72 hpf (Fig. 2). The total number of SCs remained significantly lower in csp^−/− at 72 hpf in comparison with controls (Fig. 2B). Time-lapse analysis using the nuclear marker h2b:gfp pointed to a delay in prometaphase progression (Fig. S1A, Movies 5 and 6). However, in contrast to the neuronal phenotype in retina and spinal cord, most SCs exited mitosis and we detected no increase in apoptotic SCs in these mutants along the PLLn (Fig. S1B-D).

These data suggest that sil is required for the temporal control of mitotic exit of SCs; nevertheless, SCs devoid of sil do exit mitosis following a significant delay and are not apoptotic.

Having identified a delay in progression through mitosis in SCs and a decrease in the total number of SCs in csp^−/−, we wondered whether these defects have any impact on the ability of SCs to ensheath axons. TEM analysis showed a dramatic decrease in the number of myelinated axons in csp^−/− in comparison with controls. We observed an average of 5.6±0.55 myelinated axons per nerve in controls in comparison to zero myelinated axons per nerve in csp^−/− at 72 hpf [or 3 days post-fertilization (dpf)] (Fig. 3A-A″,B,E). In addition, we observed a significant decrease in the total number of axons in these mutants with an average of 55±2.85 axons in controls and 28.63±2.67 in csp^−/− PLLn (Fig. 3F). The myelination defect in csp^−/− embryos was also observed at 4 dpf (Fig. 3C-F). This analysis could not be extended to later stages because csp^−/− embryos died at 5 dpf.

Given the drastic decrease in total number of axons in these mutants, we calculated the ratio of myelinated axons relative to the total number of axons in the PLLn. Whereas we counted an average of 10.46±1.14% of myelinated axons in controls at 3 dpf and 15.44±0.87% at 4 dpf, we observed 0% in csp^−/− at 3 and 4 dpf (Fig. 3G). This myelination defect was not the result of a decrease in the number of large caliber axons (diameter ≥0.4 µm) that should be myelinated (Fig. 3H). We also analyzed the distribution of axons relative to their diameter with a 0.1 µm bin width at 3 dpf and only the percentage of a group of small caliber axons (0.2-0.3 µm) was slightly but significantly decreased in csp^−/− (Fig. 3I). We even observed a significant increase in the percentage of axons with 0.3-0.4 µm diameter in csp^−/− at 4 dpf in comparison with controls (Fig. 3J). Finally, the myelination defect coincided with a significant decrease in the number and the percentage of radially sorted axons per nerve (Fig. 3K,L).

These results show that sil is required for SC radial sorting and myelination.

To test whether the myelination defect observed here is specific to SCs or secondary to neuronal defects, we first examined several aspects of posterior lateral line ganglia (PLLg) and PLLn development. csp^−/− embryos presented no major defects in PLLn growth and SC migration at 48 hpf (Fig. S2A). A few csp^−/− embryos (<30%, n=7/24) showed an axonal growth arrest but posterior to the yolk extension and SCs were found at the tip end of these axons. We next analyzed axonal transport along the PLLn using a mito:gfp construct, which labels mitochondria. We observed no significant difference in the speed of mitochondrial movement along the PLLn with an average of 1.92±0.14 μm/s and 1.85±0.06 μm/s in controls and csp^−/−, respectively (Fig. S2B,C, Movies 7 and 8). We next counted the number of neurons within the PLLg and observed a significant decrease in csp^−/− embryos (Fig. S2D,E) that coincided with the reduction in the number of axons in the PLLn.

These results suggest a role for sil in the development of PLLg, but not in axonal growth or transport.

To test directly whether sil has an autonomous function in SC myelination, we forced the expression of sil specifically in SCs under the control of the sox10 promoter. csp^−/− embryos were injected with pTol2-sox10:sil-P2A-mCherry-CaaX and tol2 transposase mRNA at the one-cell stage and selected at 3 dpf for continuous mCherry expression in SCs at the level of yolk extension. By forcing the expression of sil specifically in SCs, we significantly increased the number of myelinated axons in csp^−/− embryos at 3 dpf (Fig. 4A-C″). The percentage of myelinated axons per nerve in injected csp^−/− embryos was similar to that observed in controls (Fig. 4D). We observed no significant difference in the distribution of axons relative to their diameter with 0.1 µm bin width at 3 dpf between the three groups (Fig. 4E).

We also processed control and csp^−/− larvae for immunohistochemistry to detect Mbp at 3 dpf. We observed a significant decrease in Mbp expression in csp^−/− embryos (Fig. 4F-G″). We then injected csp^−/− embryos with pTol2-sox10:sil-P2A-mCherry-CaaX and tol2 transposase mRNA at the one-cell stage and selected embryos for mosaic mCherry⁺ clonal analysis. We analyzed four mutant larvae that had sox10:mCherry⁺ SCs along mutant PLLn; the mCherry⁺ clones appeared to express normal levels of Mbp (Fig. 4H-I″).

Altogether, these results suggest a specific requirement for sil within SCs to initiate axonal wrapping and Mbp expression.

M phase is characterized by chromatin condensation and temporal ejection of transcription factors (Ma et al., 2015). Given that sil is implicated in spindle checkpoint and mitotic progression during M phase, it is possible that the delayed mitotic exit observed in csp^−/− embryos could result in prolonged ejection of transcription factors required for myelination in SCs. Alternatively, it is possible that transcriptional activity is effective after mitosis, but SCs in csp^−/− embryos present defective upstream signaling activity. To distinguish between these possibilities, we first analyzed the expression of krox20 (egr2b) and mbp (mbpa) in controls and csp^−/− embryos at 52 hpf and 3 dpf, respectively. csp^−/− embryos showed a sharp decrease in the expression of peripheral myelin genes (Fig. 5A,B,D,E). This coincided, as shown before, with a total lack of radial sorting and axonal ensheathment (Fig. 5H). Second, we decided to treat embryos with forskolin (FSK), which interacts with adenyl cyclase and elevates the levels of cAMP. When csp^−/− embryos were treated with FSK during radial sorting, the expression of krox20 and mbp was restored (Fig. 5C,F), as well as the ratio of myelinated axons relative to the total number of axons (Fig. 5I,J). The distribution of axons relative to their diameter with 0.1 µm bin width at 3 dpf between the two groups of csp^−/− and csp^−/−+FSK embryos was not altered, even though some significant differences were observed between controls and csp^−/−+FSK embryos (Fig. 5K). We also labeled SCs with the M-phase marker PH3 and observed a significant decrease in the total number of SCs between FSK-treated and non-treated (control) groups (Fig. 5L), but the decrease was not significant between csp^−/− and csp^−/−+FSK embryos. However, the number of PH3⁺ SCs and the percentage of PH3⁺ SCs were significantly decreased in csp^−/−+FSK in comparison with csp^−/− embryos (Fig. 5M,N). We also used time-lapse imaging to track SC division in csp^−/−+FSK embryos and we could only detect two dividing cells from six different embryos (36 h recording in total) because proliferation was almost blocked in these embryos. However, dividing SCs took as long as 121 min to complete cytokinesis, highlighting a defective progression through mitosis as observed in csp^−/−.

This result suggests that SCs in csp^−/− are still capable of activating the transcriptional program for terminal myelination. It also shows that FSK treatment significantly decreases the number of proliferating SCs in csp^−/− while prompting differentiation.

The radial sorting defect and lack of axonal ensheathment by SCs observed in this mutant might also be linked to a defective cytoskeletal re-arrangement. To test this hypothesis, we forced the expression of Rac1, a small GTPase protein that interacts with SC cytoskeleton (Nodari et al., 2007) and is able to restore peripheral radial sorting defects in zebrafish (Mikdache et al., 2020; Boueid et al., 2020). csp^−/− embryos injected with a constitutive active form of Rac1 were comparable to non-injected csp^−/− embryos and showed a total lack of radial sorting and axonal ensheathment (Fig. S3A,B).

This result suggests that csp^−/− myelin defects do not ensue from defective Rac1 activity.

Previous studies have shown that Laminin 211 promotes myelination via a GPR126/cAMP-dependent pathway (Petersen et al., 2015); given that FSK treatment of csp^−/− embryos is sufficient to re-establish myelination in these mutants, we hypothesized that Laminin 211 expression and/or function might be impaired in these mutants. To test this, we first analyzed laminin expression at 48 hpf in controls and csp^−/− embryos by immunostaining. Indeed, laminin expression was significantly reduced along the PLLn of csp^−/− embryos (Fig. 6A-C). We next wondered whether increasing extracellular Lama2 levels would rescue the myelination defect. To this end, we took advantage of a lama2 overexpression construct that expresses membrane EGFP as well as secreted mCherry-tagged Lama2 under the control of muscle-specific skeletal-alpha actin (acta1) promoter (denoted pacta1) (Sztal et al., 2012; Petersen et al., 2015) and injected 20 pg of this construct into csp^−/− mutants. Embryos that showed significant expression of EGFP within muscles surrounding the PLLn and mCherry-tagged laminin within the PLLn itself (Fig. 6D,E) were then selected and analyzed by TEM at 3 dpf. Injected csp^−/− embryos showed a significant increase in the number of myelinated axons in comparison with non-injected csp^−/− larvae and were comparable to controls (wild-type lama2-injected embryos) with respect to the percentage of myelinated axons relative to the total number of axons (Fig. 6F-J). The distribution of axons relative to their diameter with 0.1 µm bin width at 3 dpf between csp^−/− and csp^−/− injected with pacta1-lama2 showed a significant increase in groups of small caliber axons (0.1-0.3 µm) but axons of other sizes were not altered (Fig. 6K).

These results strongly suggest defective laminin expression and function within the PLLn of csp^−/− mutant, which is responsible for the radial sorting and myelination defect.

Laminin 2 is known to control several aspects of SC development, such as proliferation and differentiation (Chernousov et al., 2008). We therefore wondered whether Lama2 overexpression impacts the timely division and/or proliferation of SCs to mediate myelination or whether the radial sorting/myelination process is triggered by Lama2 regardless of the number of SCs. To investigate this, we used the M-phase marker PH3 and time-lapse imaging to monitor dividing SCs. We observed a significant reduction in the number of SC in control and csp^−/− embryos injected with pacta1-lama2 in comparison with their respective non-injected controls as well as the percentage of PH3⁺ SCs in csp^−/− injected with pacta-lama2 in comparison with non-injected csp^−/− (Fig. 7A-C). Time-lapse analysis showed a delay in SC mitotic exit similar to that observed in csp^−/− embryos during migration and prior to axonal ensheathment (Fig. 7D,E).

This result strongly suggests that the myelination defect observed in csp^−/− is due to a defective laminin pathway and that a reduction in SC numbers is not essential for myelination to proceed.

It remained important to determine the source of laminin within the PLLn and whether SC-specific expression of laminin would rescue myelination autonomously in sil mutants. To investigate this, we first incubated embryos with AG1478, which specifically blocks ErbB signaling in zebrafish and inhibits SC migration along the PLLn (Lyons et al., 2005). Embryos treated with AG1478 between 24 and 48 hpf presented no SCs along the PLLn and showed a significant decrease in laminin expression at 48 hpf (Fig. 8A-C).

Second, we designed a pUAS-lama2-EGFP-mCherry construct and took advantage of the tg(sox10:Gal4VP16) line to allow a specific expression of Lama2 within SCs (EGFP⁺ labeling); Lama2 is also found secreted within the PLLn (mCherry⁺ labeling) (Fig. 8D-E‴). We injected 20 pg of plasmid in sox10:Gal4VP16 and csp^−/−/sox10:Gal4VP16 embryos at the one-cell stage and proceeded with EGFP and Mbp immunostaining for clonal analysis at 3 dpf. Injected csp^−/− embryos that showed expression of EGFP in SCs along the PLLn expressed normal levels of Mbp, whereas those that were negative for EGFP labeling showed a significant decrease in Mbp expression similar to that in non-injected csp^−/− larvae (Fig. 8F-H″).

These results strongly suggest that timely SC division is largely responsible for laminin expression within the PLLn. This constitutes a crucial step for timely myelination.

SCs migrate and divide along peripheral axons and the intimate contact between the two contributes to SC proliferation, mainly through the neuregulin/ErbB axis. Other signals provided by ECM proteins are also important for their proliferation (Lyons et al., 2005; Bunge et al., 1982; 1990; Michailov et al., 2004; Nave and Salzer, 2006). Here, we investigated the characteristics of SC division and its impact on myelination in zebrafish.

SCs must regulate their numbers in order to myelinate or ensheath all axons that are supposed to be myelinated by the end of radial sorting (Jessen and Mirsky, 2005), so one would predict that reducing the number of SCs in the nerve would result in a reduction of myelinated axons. Indeed, several molecules that control SC proliferation have been identified, such as Cdc42, focal adhesion kinase and several extracellular matrix proteins of the laminin family such as laminins 2, 8 and γ1 (Benninger et al., 2007; Grove et al., 2007; Yang et al., 2005). Specific ablation of these molecules in SCs causes a reduction in the number of myelinated axons.

But, how important is the timely division of SC? Are insufficient SC numbers per se a cause of radial sorting and myelination defects?

Here, we show that timely SC division is essential for peripheral radial sorting and myelination. Delaying SC division during development is sufficient to halt radial sorting and axonal ensheathment (Fig. 9). We provide evidence that laminin expression is significantly reduced along the PLLn in embryos with delayed SC division. Providing laminin, through surrounding muscles or from within SCs, which reduces even further the number of SCs and their proliferation, is enough to initiate radial sorting and myelination (Fig. 9). Several studies have pointed to a role for laminin in controlling SC proliferation, but this is the first study that implicates SC division in laminin expression and subsequent myelination. The temporal control of SC division is therefore an important step towards initiating the myelination program through the laminin/cAMP pathway, a role that stretches beyond simply increasing the pool of SCs required to continuously engulf the nerve as it grows. Consequently, is it important to generate enough SCs for myelination to proceed? Our findings reveal that the myelination defects observed in csp^−/− embryos do not ensue from the reduction in SC numbers per se, because laminin overexpression can fully rescue the percentage of myelinated axons within the PLLn even if SC numbers remain significantly reduced. Indeed, laminin overexpression and FSK treatment restore myelination in csp^−/− mutants while reducing even further the number of SCs and/or their proliferation. This points to a role for laminin/cAMP in prompting myelination by interrupting proliferation and that this pathway does not drive proliferation in order to myelinate. Nonetheless, it is important to note that different levels of laminin/cAMP expression or activity may coordinate SC proliferation and differentiation during development.

Given that csp^−/− embryos show a radial sorting defect, it remains possible that the myelination defects result, at least partially, from defective cytoskeletal re-arrangement. A group of genes that regulate this aspect of SC behavior has been identified and includes elmo1, dock1, β1 integrin genes and ILK, which link extracellular signaling to rac1 (Mikdache et al., 2020; Cunningham et al., 2018; Feltri et al., 2002). Our analysis strongly suggests that the myelination defects observed in csp^−/− do not result from defective Rac1 activity as forcing the expression of Rac1 does not rescue this defect, even partially.

SCs in csp^−/− struggle to progress through the M phase of the cell cycle. This phase is characterized by chromatin condensation and by ejection of transcription factors and chromatin-binding proteins (Ma et al., 2015). It is possible that SC transcription factors required for the initiation of myelination are permanently ejected in csp^−/− and therefore unable to bind the DNA. Another possibility is defective upstream signaling activity in csp^−/− embryos that blocks SC differentiation, assuming that chromatin accessibility is re-established after division. Our analysis suggests that it is rather defective signaling activity that is responsible for the peripheral lack of myelin observed in csp^−/− embryos and that transcription activity is responsive to upstream signaling cues.

A particularly interesting observation is the difference in the sil deficiency phenotype between retina and Schwann cells. Sil is a ubiquitously expressed protein and is involved in the fundamental process of mitosis but is highly enriched in the nervous system. Neurons might in this case be more sensitive to Sil defects leading to massive apoptosis within the central nervous system; however, it is Schwann cell differentiation that is impaired in csp^−/− with no increase in apoptosis. It is therefore possible that different cells might respond differently to sil defects and our results suggest that genes involved in MCPH have a broader requirement during development that is not solely confined to neuronal cells in the central nervous system.

What is the molecular link between SC division, its temporal control and the cAMP pathway? Several studies have now established an essential role for the Gpr126 (adgrg6) in driving cAMP activity in SC through G proteins (Mogha et al., 2013) and this has recently been attributed to an interaction of Gpr126 with the ECM, more specifically with Laminin 211 (Petersen et al., 2015). It is therefore possible that sil-deficient SCs are incapable of secreting extracellular proteins and/or responding to them to drive the ECM/Gpr126/cAMP pathway (Paavola et al., 2014; Petersen et al., 2015). Our work strongly suggests that timely SC division is required for timely laminin expression. Laminin, in return, with increasing levels and polymerization, would drive radial sorting and myelination through cAMP activation, at least partially via Gpr126 (Petersen et al., 2015).

Finally, we show that SCs are a major source of laminin within peripheral nerves in zebrafish. Our clonal analysis shows that forcing laminin expression autonomously within SCs is enough to lead to secretion of laminin into the extracellular space and restore normal Mbp expression within the vicinity of these cells in csp^−/− embryos.

In summary, our work provides evidence of a previously unappreciated mechanism that links temporal control of SC division to radial sorting and axonal wrapping via the laminin/cAMP pathway in vivo.

Zebrafish (Danio rerio) of the AB strain were used. Embryos were staged and cared for according to standard protocols (https://zfin.org/zf_info/zfbook/cont.html). The stable transgenic line Tg(foxd3:gfp), which labels SCs, was used in this study (Gilmour et al., 2002). The Tg(sox10:Gal4VP16)^el159 line was kindly provided by Gage Crump (Center for Stem Cell and Regenerative Medicine, Keck School of Medicine, University of Southern California, Los Angeles, CA, USA). Csp^+/− embryos (stil^cz65) were purchased from the Zebrafish International Resource Center (ZIRC). All animal experiments were conducted with approved protocols at Inserm by DDPP Val de Marne, France, under license number F 94-043-013.

The sil-P2A-mCherry cassette, which allows simultaneous expression of sil and membrane-localized mCherry separated by the self-cleaving P2A peptide, was generated by PCR amplification.

5′-CGATTCACTAGTATGAACCGTGTACAAGTGGATTTTAAAGG-3′ (forward) and 5′-GAAGTTCGTGGCTCCGGATCCAAAGAGCTTGGGGAGCTGCCGTAACC-3′ (reverse) primers were used on the pDNR-lib-STIL cDNA (Horizon Discovery, MDR1734-202835545) and the 5′-GGTTACGGCAGCTCCCCAAGCTCTTTGGATCCGGAGCCACGAACTTC-3′ (forward) and 5′-CTATGACCATGATTACGCCAAG-3′ (reverse) primers on the pCS2-mCherry-CAAX plasmid.

The resulting PCR fragment was digested by SpeI and NotI before sub-cloning into the −4725Sox10-cre vector (a gift from Robert Kelsh, Centre for Regenerative Medicine, Developmental Biology Programme, School of Biology and Biochemistry, University of Bath, UK) to obtain the pSox10:sil-P2A-mCherry plasmid.

pT2i5uasbiH2bChe-GFP (a gift from Michel Volovitch, Centre Interdisciplinaire de Recherche en Biologie, Collège de France, Paris, France) was used to amplify the 5′ UAS sequence using 5′-GTGGGCCCTGCGTCTAGAGTC-3′ (forward) and 5′-CTCACCATATGGGCGACCGGTGG-3′ (reverse) primers. Then, in order to replace the acta1 promoter, the PCR product was subcloned into the pacta1:lama2-mCherry-T2A-EGFPcaax vector [a gift from Peter Currie (Australian Regenerative Medicine Institute, Monash University, Clayton, Australia) and Kelly Monk (The Vollum Institute, Oregon Health & Science University, Portland, OR, USA)] digested with ApaI and NdeI enzymes.

Mito:gfp (a gift from Dominik Paquet, Institute for Stroke and Dementia Research, and SyNergy, Ludwig-Maximilians University, Munich, Germany) and h2b::gfp (a gift from Jon Clarke, Department of Developmental Neurobiology, King's College London, UK) mRNAs were synthesized using the SP6 mMessage mMachine System after linearization with NotI and injected at 200 pg per embryo. pTol2:sox10-sil-2A-mCherry-Caax was injected at 10 pg per embryo along with 50 pg of tol2 transposase mRNA (a gift from David Lyons, Centre for Discovery Brain Sciences, University of Edinburgh, UK). Rac1V12 (constitutively active Rac1, kindly provided by Nicolas David, Laboratory for Optics and Biosciences, Institut Polytechnique de Paris, Palaiseau, France) mRNA was synthesized using the SP6 mMessage mMachine System after linearization with NotI and injected at 2-10 pg per embryo. pacta1-lama2 (kindly provided by Kelly Monk and Peter Currie) was injected at 20 pg per embryo along with 100 pg of Tol2 transposase mRNA. pUAS-lama2 was injected at 20 pg per embryo.

In situ hybridization was performed following standard protocols using krox20 and mbp probes (Fontenas et al., 2016). Embryos were raised in egg water with phenylthiourea (0.003%) to avoid pigmentation and were fixed overnight in 4% paraformaldehyde (PFA) at 52 hpf and 3 dpf. Embryos were dehydrated the next day and kept in methanol at −20°C. Embryos were then rehydrated and treated with proteinase K and incubated overnight with the corresponding probe. Embryos were then washed and the staining was revealed using anti-digoxigenin antibody (Roche; 11093274910).

The following antibodies and dilutions were used: mouse anti-acetylated tubulin (Sigma-Aldrich; 1/500; T7451), mouse anti-HuC/HuD neuronal protein (Invitrogen; 1/500; A21271), rabbit anti-laminin (Sigma-Aldrich, 1/200; L9393), mouse anti-phospho histone 3 (Ser 10) (Millipore; 1/500; 06-570), rabbit anti-myelin basic protein (custom produced; Tingaud-Sequeira Angèle et al., 2017; gift from Patrick Babin, Laboratoire Maladies Rares: Génétique et Métabolisme, Université de Bordeaux, France; 1/500), mouse anti-EGFP (Millipore; 1/500; MAB3580). Primary antibodies were detected with appropriate secondary antibodies conjugated to either Alexa 488, Alexa 568 or Alexa 647 (Invitrogen; A11036, A11034, A11001 and A31571) at 1/500. For immunostaining, embryos were fixed in 4% PFA in 1× PBS overnight at 4°C, then washed with 1× PBS and dehydrated in methanol for at least 6 h at −20°C. Samples were then rehydrated, digested with proteinase K and blocked with 0.5% Triton X-100 in PBS and 10% sheep serum and then incubated with primary antibody overnight at 4°C (diluted in PBS with 2% sheep serum). Larvae were then washed with PBS for a few hours and then incubated with secondary antibody in 0.5% Triton X-100 in PBS and 2% sheep serum overnight at 4°C. Stained larvae were then imaged with a Leica SP8 confocal microscope. For laminin and Mbp staining, embryos were fixed in 4% PFA for 30 min and treated for 7 min with acetone at −20°C. Samples were then blocked with 0.5% Triton X-100 in PBS and 10% sheep serum and then incubated with primary antibody overnight at 4°C (diluted in PBS with 2% sheep serum). Larvae were then washed with PBS for a few hours and then incubated with secondary antibody in 0.5% Triton X-100 in PBS and 2% sheep serum for 3 h at room temperature. For laminin fluorescence intensity quantification, the same parameters (excitation/emission, gain for detectors, lasers intensity) were applied for image acquisition in controls and csp^−/− embryos, and fluorescence intensity was measured using ImageJ (‘Analyze/measure’).

Embryos were incubated in fish water containing 20 µM FSK diluted in DMSO between 45 and 52 hpf. Controls were incubated in the equivalent amount of DMSO solution (0.1%) during the same period.

Embryos were incubated in fish water containing 4 µM AG1478 diluted in DMSO between 24 and 48 hpf. Controls were incubated in the equivalent amount of DMSO solution (0.1%) during the same period.

Embryos were anesthetized with 0.03% tricaine and incubated in fish water containing 5 μM Acridine Orange for 20 min in the dark. After washing, they were embedded in 1.5% low melting point agarose, and imaged with a Leica SP8 confocal microscope. Transmitted light imaging was used to identify the PLLn and the spinal cord for counting apoptotic corpses within these structures. Fluorescent corpses were counted at the level of the yolk extension.

Tg(foxd3:gfp) embryos, csp^−/−/Tg(foxd3:gfp) as well as treated and injected ones were analyzed. Schwann cells and PH3⁺ SCs were counted from the most anterior to the end of the yolk extension along the AP axis (≈900 µm) after immunostaining. Schwann cells as well as PH3⁺ SCs were counted blindly by two independent researchers using ImageJ.

Total neuronal cell numbers (HuC⁺) were counted from a confocal z-series of the PLLg using ImageJ (Mikdache et al., 2021). Cells were counted blindly by two independent researchers.

Embryos were anesthetized with 0.03% tricaine and embedded in 1.5% low melting point agarose. For mito:GFP tracking experiments, the PLLn was examined at 48 hpf from a lateral view. A series of 10 min time-lapses recordings were made. Recordings were performed at 27°C using a Leica SP8 confocal microscope. For Tg(foxd3:gfp), Tg(foxd3:gfp)/csp^−/− as well as controls and csp^−/− injected and treated embryos, the PLLn was examined at different time points and recordings were acquired for up to 12 h. Transmitted light imaging was used to identify the PLLn for h2b:gfp and mito:GFP recordings.

At 3 and 4 dpf, embryos were fixed in a solution of 2% glutaraldehyde, 2% PFA and 0.1 M sodium cacodylate, pH 7.3, overnight at 4°C. This was followed by a post-fixation step in cacodylate-buffered 1% osmium tetroxide (Serva) for 1 h at 4°C and in 2% uranyl acetate for 1 h at room temperature. The tissue was then dehydrated and embedded in epoxy resin. Sections were contrasted with saturated uranyl acetate solution and were examined with a 1010 electron microscope (JEOL) and a digital camera (Gatan).

Means and standard errors of the mean were calculated with Graph Pad Prism 7. All data were first tested for normal distribution using the Shapiro–Wilk normality test combined with D'Agostino & Pearson normality test. All experiments with only two groups and one dependent variable were compared using an unpaired t-test with Welch's correction if they passed normality test; if not, groups were compared with the nonparametric Mann–Whitney test. Statistically significant differences were determined using one-way ANOVA for all experiments with more than two groups but only one dependent variable if data were normally distributed; if not, groups were compared using Kruskal–Wallis test. Error bars depict s.e.m. In figures, the following notation is used to represent significance levels: ns, not significant (P>0.05); *P≤0.05; **P≤0.01; ***P≤0.001; ****P≤0.0001. n represents the number of embryos.

We would like to thank Peter Currie and Kelly Monk for the pacta1-lama2 plasmid, Michel Volovitch and Sophie Vriz for the pUAS construct and for discussing cloning strategy, Robert Kelsh for the sox10 promoter, Gage Crump for Tg(sox10:Gal4VP16)^el159 embryos, Patrick Babin for providing the Mbp antibody and Jon Clarke for his critical reading of the manuscript.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.200640.reviewer-comments.pdf.