To elucidate the role of Marcks and Marcksl1 for spinal cord development, we first analyzed mRNA expression of marcks.S, marcksl1.S, and marcksl1.L in embryos of X. laevis at stages 26 and 34 (early and late tailbud stage, respectively). Due to the 94% sequence identity between marcks.S and marcks.L mRNAs, the marcks.S probe is expected to cross-react with marcks.L, while marcksl1.S and marcksl1.L mRNA have only 84% identity and show slightly different expression suggesting that they exhibit only limited if any cross-reactivity. Whole-mount in situ hybridization showed that all of these genes have a similar pattern of expression (Figure 1A–G, Figure 1—figure supplement 1). They are expressed in the brain and spinal cord as well as in the pharyngeal arches, somites, intermediate mesoderm, and blood islands of the ventrolateral mesoderm confirming previous reports (Gawantka et al., 1998; Zhao et al., 2001). marcks.S is more broadly distributed through the entire brain, whereas marcksl1.L and marcksl1.S have a more restricted expression and are absent from the forebrain. Vibratome sections of stage 34 embryos after whole-mount in situ hybridization confirmed the expression of all genes throughout the spinal cord as well as in the brain, lens, pharyngeal arches, somites, and other mesodermal derivatives (Figure 1H, I; Figure 1—figure supplements 2–4). They also reveal the expression of all genes in the epidermis and of marcks.S in the otic vesicle and show that marcksl1.S has a weaker and more restricted expression in the brain and spinal cord than marcksl1.L.

Figure 1 with 6 supplements see all

Download asset Open asset

We next used two polyclonal antibodies raised against epitopes shared between X. laevis Marcks.S and Marcks.L protein and between Marcksl1.S and Marcksl1.L, respectively, to clarify the subcellular distribution of Marcks and Marcksl1 proteins in the developing spinal cord. These antibodies specifically recognize Marcks and Marcksl1 proteins, respectively, without cross-reactivity (Figure 1—figure supplement 5). Confocal microscopy of embryos injected with mRNAs encoding a membrane-tethered form of GFP (memGFP) and transversely sectioned at stage 26 revealed that Marcks and Marcksl1 are mostly associated with outer cell membranes. This is true for the ventricular zone, where progenitor cells are located, and the adjacent mantle zone, which harbors migrating and differentiating neurons (Figure 1K–P, Figure 1—figure supplement 6). In addition, a subset of cells in both ventricular and mantle zones shows cytoplasmic expression of Marcks and Marcksl1. Marcks, but not Marcksl1 was also occasionally seen in cell nuclei.

Because shuttling between the membrane and cytoplasm is essential for the function of Marcks and Marcksl1 in other developmental contexts (El Amri et al., 2018), their subcellular localization to membranes and cytoplasm in the spinal cord and their widespread distribution in the developing spinal cord suggests that these proteins may play hitherto unrecognized roles during normal spinal cord development. To test this hypothesis, we next used two independent approaches, injection of antisense Morpholino oligonucleotides (MOs) and CRISPR/Cas9-mediated gene editing, to determine how loss of function of these proteins affects neurite outgrowth, cell proliferation, and neuro-glial progenitor populations during spinal cord development.

We first studied the role of Marcks and Marcksl1 in neurite formation in the spinal cord individually by injecting a single blastomere in 4- to 8-cell stage embryos with either a MO blocking the translation of both marcks.L and marcks.S or with a mixture of two MOs blocking the translation of marcksl1.L and marcksl1.S, respectively. To minimize the gastrulation and axis defects previously observed after knockdown of marcks or marcksl1 (Zhao et al., 2001; Iioka et al., 2004), we injected lower amounts of MOs (9 ng) than in previous studies (16–40 ng). This resulted in the unilateral knockdown of the respective proteins on the injected side of the embryo, whereas the uninjected side developed normally. myc-GFP mRNA was co-injected to identify the injected side. The effectiveness and specificity of all MOs used were validated by in vitro transcription and translation of the various proteins in the presence or absence of the appropriate MO (Figure 2—figure supplement 1). None of these injections led to significant alterations in immunostaining for the neurite marker acetylated tubulin on the injected side of the spinal cord relative to the uninjected side in stage 34 embryos compared to an unspecific control MO (Figure 2A–C). However, when the three MOs targeting marcks.S/L, marcksl1.S, and marcksl1.L were co-injected, thus blocking all four homeologs (4M MO), acetylated tubulin staining was significantly decreased on the injected side (Figure 2D, E, G). Co-injection of mRNA constructs encoding chicken Marcks and Xenopus tropicalis Marcksl1, which were selectively mutated to prevent MO binding, resulted in complete rescue of acetylated tubulin staining, confirming the specific action of the MOs used (Figure 2F, G). This suggests that Marcks and Marcksl1 are required for neurite formation during spinal cord development but act in a largely redundant fashion. To account for these overlapping functions, we aimed to simultaneously knockdown or gene edit all four proteins (Marcks.L/S, Marcksl1.L/S) for the rest of our study.

Figure 2 with 7 supplements see all

Download asset Open asset

To verify the results of our MO knockdown experiments, we analyzed F0 embryos after CRISPR–Cas9 gene editing. For these experiments, single-guide RNA constructs targeting marcks.S, marcks.L, marcksl1.S, and marcksl1.L genes were co-injected with Cas9 as a ribonucleoprotein (RNP) and myc-GFP mRNA into one blastomere during the 1-cell stage for genotyping or the 4- to 8-cell stage for phenotypic analysis. At later embryonic stages (stages 26–42), genomic DNA was extracted from some of the embryos, sequenced, and analyzed with Synthego Inference for CRISPR Edits (ICE) to determine the frequency with which indels were generated in the targeted region (indel score) as well as the frequency of indels with frameshifts (knockout score). Sibling embryos were used for phenotypic analysis. In control experiments, we either injected Cas9 protein and myc-GFP mRNA alone or injected a sgRNA targeting slc45a2 (DeLay et al., 2018), encoding a protein that mediates melanin synthesis, Cas9 proteins, and myc-GFP mRNA. When analyzing CRISPR/Cas9-edited animals in the F0 generation, a considerable and variable extent of mosaicism has to be taken into account because different gene editing events may occur in different subsets of cells (Blitz et al., 2013; DeLay et al., 2018). We, therefore, first tested a range of different sgRNAs to ensure high overall gene editing and knockout efficiencies.

We first validated our CRISPR knockout procedure using the control gene slc45a2. Embryos injected during the 1-cell stage with slc45a2 RNPs and analyzed at stages 37–46 lost pigmentation in both eyes, whereas embryos injected during the 2- to 4-cell stage showed loss of pigmentation only in the injected side. Confirming a previous report (DeLay et al., 2018), genomic sequencing revealed a high indel score (93%) and knockout score (67–68%) in these embryos (Figure 2—figure supplement 2). Almost all surviving embryos displayed visible oculocutaneous albinism, with the majority showing almost no eye pigment (Figure 2—figure supplement 2).

Next, we designed, synthesized, and tested a total of 18 sgRNA transcripts targeting either marcks.L/S or marcksl1.L/S (sgRNAs were designed to target both homeologs of each gene) and selected the sgRNA transcripts creating the highest and most consistent number of indels and knockout scores for each pair of genes – named marcks.L/S and marcksl1.L/S sgRNA, respectively – for further experiments (Figure 2—figure supplements 3–6). The marcks.L/S sgRNA introduced indels in 88% of the targeted DNA from marcks.L exon 1 (8 embryos) and 84% from marcks.S exon 1 (8 embryos), with an average of 78% and 76%, respectively, consisting of frameshift mutations which may contribute to a functional knockout (Figure 2—figure supplement 3C and 4C). The marcksl1.L/S sgRNA introduced indels in 73% of the targeted DNA from marcksl1.L exon 2 (six embryos) and 73% from marcksl1.S exon 2 (seven embryos), with an average of 50% and 47%, respectively, being out-of-frame (Figure 2—figure supplement 5C and 6C). We also confirmed by immunohistochemistry that co-injection of marcks.L/S and marcksl1.L/S sgRNA, which is predicted to edit all four homeologs (henceforth denoted as 4M CRISPR) drastically reduced immunostaining for Marcks and Marcksl1 protein on the injected side (Figure 2—figure supplement 1B–G), indicating that protein levels are reduced in gene-edited embryos.

Similar to the knockdown of the proteins by MOs, the unilateral CRISPR-mediated knockout of all Marcks and Marcksl1 homeologs (4M CRISPR) led to a significant reduction in acetylated tubulin staining on the injected side compared to Cas9 and slc45a2 controls (Figure 2H–K). Conversely, overexpression of both proteins by injection of marcks and marcksl1 mRNA at two different doses (100 or 300 pg each) together with myc-GFP mRNA resulted in a significant increase in acetylated tubulin staining on the injected side compared to that in control embryos injected with myc-GFP mRNA alone (Figure 2—figure supplement 7). Taken together, these results suggest that Marcks and Marcksl1 play essential roles for neurite formation during normal embryonic development of the Xenopus spinal cord, with elevated levels of Marcks and Marcksl1 being able to further promote neurite outgrowth.

Next, we tested whether Marcks and Marcksl1 are required for the proliferation of neuro-glial progenitor cells during spinal cord development. After unilateral injection with a cocktail of the three MOs targeting marcks.S/L, marcksl1.S and marcksl1.L (4M MO) or with the combined marcks.L/S and marcksl1.L/S sgRNAs (4M CRISPR), we analyzed the number of cells in mitosis in the spinal cord of stage 34 embryos by immunostaining against phospho-histone H-3 (PH3). In both control and experimental animals, PH3-positive nuclei are confined to the ventricular zone at this stage, where neuro-glial progenitors reside (Figure 3A–C, E–G). The proportion of PH3-positive ventricular nuclei on the injected side (expressed as the ratio between injected/uninjected sides) was significantly reduced in both experiments compared to control injections (Figure 3A–H). Conversely, the injection of marcks and marcksl1 mRNA together with myc-GFP mRNA resulted in a significant increase in the proportion of PH3-positive ventricular nuclei on the injected side compared to control embryos injected with myc-GFP mRNA alone (Figure 3—figure supplement 1). These observations indicate that Marcks and Marcksl1 are required for normal patterns of cell proliferation in the ventricular zone of the developing spinal cord. In addition, increased levels of Marcks and Marcksl1 are sufficient to further promote cell proliferation.

Figure 3 with 1 supplement see all

Download asset Open asset

To directly test the role of Marcks and Marcksl1 for neuro-glial progenitors, we next immunostained spinal cords of stage 34 embryos for the progenitor markers Sox2 or Sox3 after unilateral injection of marcks.L/S and marcksl1.L/S sgRNAs. The number of Sox2- or Sox3-positive nuclei on the injected side (expressed as a ratio between injected/uninjected sides) was significantly reduced as compared to control embryos injected with slc45a2 sgRNAs or with Cas9 alone (Figure 3I–M). This indicates that Marcks and Marcksl1 are required for maintaining a normal pool of neuro-glial progenitor cells during spinal cord development. This may at least partly result from the role of both proteins in promoting the proliferation of these neuro-glial progenitors, as described above, but additional roles for the survival and maintenance of neuro-glial progenitors cannot be ruled out.

In spite of the reduction in progenitor proliferation after Marcks/Marcksl1 loss of function, overall cell numbers (as indicated by the numbers of DAPI+ nuclei) on the injected side were not yet significantly reduced in these embryos (Figure 3N, O) indicating that the decrease in neurite staining described in the previous section cannot be attributed to a reduction in overall cell number.

Since Marcks and Marcksl1 are known to be multifunctional proteins, we next asked which signaling pathways may mediate the effects of Marcks and Marcksl1 during spinal cord development. Our experiments described so far suggested essential functions of these proteins in both proliferation control and neurite outgrowth. We, therefore, focused on phospholipid signaling since previous studies have suggested that the main phospholipid signaling pathways, which depend on PLC and PLD, are regulated by Marcks (Glaser et al., 1996; McLaughlin et al., 2002; Morash et al., 2000; Morash et al., 1998) and have essential functions in both the regulation of mitosis and of cytoskeletal dynamics including neurite outgrowth (Rebecchi and Pentyala, 2000; Saarikangas et al., 2010; Kam and Exton, 2001; Foster and Xu, 2003; Zhang et al., 2004),. To determine whether the essential roles of Marcks and Marcksl1 for neurite formation and proliferation of neuronal progenitors are mediated by phospholipid signaling, we attempted to rescue the phenotype of 4M-CRISPants with Sphingosine-1-phosphate (S1P) dissolved in 2% dimethyl sulfoxide (DMSO), which activates both PLC and PLD signaling (Meacci et al., 2003; Siehler and Manning, 2002).

For these and subsequent pharmacological experiments, embryos were unilaterally injected with marcks.L/S and marcksl1.L/S sgRNAs (4M CRISPR), as described above, were then raised for 1 day before fixation in the pharmacological compound and were analyzed at stage 34 by immunostaining for the neurite marker acetylated tubulin and the mitosis marker PH3. 4M CRISPR-injected control embryos were raised in 2% DMSO and similarly analyzed. Because pharmacological compounds may affect the spinal cord in multiple ways, not all of which may be Marcks/Marcksl1 dependent, we present our data here as ratios of acetylated tubulin or PH3 staining in injected/uninjected sides (Figure 4). This allows us to specifically determine, whether compounds are able to rescue deficiencies in Marcks/Marcksl1-depleted spinal cords independent of their potentially additional effects on spinal cord development. For completeness, the individual effects of pharmacological treatments on the injected and uninjected sides of embryos are shown in Figure 4—figure supplement 1.

Figure 4 with 1 supplement see all

Download asset Open asset

Compared to DMSO-treated controls, incubation in the phospholipid signaling activator S1P significantly increased the intensity of acetylated tubulin staining (Figure 4A) as well as the proportion of PH3-positive cells (Figure 4B) on the CRISPR-injected side relative to the uninjected side suggesting that Marcks and Marcksl1 promote neurite outgrowth and progenitor proliferation at least partly by a phospholipid-dependent mechanism. However, after incubation in the compound M-3M3FBS, which specifically activates PLC-dependent phospholipid signaling (Bae et al., 2003), only the proportion of PH3-positive cells was significantly increased compared to the controls (Figure 4D), but acetylated tubulin was not (Figure 4C), suggesting that PLC signaling may mediate some of the effects of Marcks and Marcksl1 on progenitor proliferation but not on neurite formation. In contrast, neither the specific PLD inhibitor FIPI (Monovich et al., 2007; Su et al., 2009), nor the specific PLC inhibitor U-73122 (Bleasdale et al., 1989) had any significant effect on neurite formation or progenitor proliferation in Marcks/Marcksl1-depleted spinal cords (Figure 4A–D).

Because phospholipid signaling downstream of Marcks has previously been suggested to be PIP₂ dependent (Morash et al., 2000; Sundaram et al., 2004), we next tested whether enhancement or inhibition of PIP₂ synthesis is also able to modulate the effects on spinal cord development in our CRISPR-injected embryos. For PIP₂ biosynthesis enhancement and inhibition, we incubated 4M CRISPR-injected embryos in N-methyllidocaine iodide (NMI) and ISA-2011B, respectively (Lee et al., 1995; Semenas et al., 2014). Compared to DMSO-treated controls, embryos in which NMI enhanced PIP₂ biosynthesis showed a significant increase in the intensity of acetylated tubulin staining (Figure 4E) as well as in the proportion of PH3-positive cells (Figure 4F) on the CRISPR-injected side relative to the uninjected side. In contrast, the PIP₂ synthesis inhibitor ISA-2011B had no significant effect (Figure 4E, F). This indicates that the stimulation of neurite formation and progenitor proliferation by Marcks and Marcksl1 is indeed PIP₂ dependent. While PIP₂ is generally thought to act upstream of PLC and PLD, additional experiments will be needed to confirm this for the developing spinal cord.

The ability of Marcks to promote PIP₂ signaling has been previously suggested to be linked to the subcellular localization of the protein, which depends on its phosphorylation status, with unphosphorylated Marcks being mostly membrane-bound and phosphorylated Marcks being largely cytoplasmic (Kim et al., 1994; McLaughlin and Aderem, 1995). In turn, the phosphorylation status of Marcks is regulated by PKC (McLaughlin and Aderem, 1995). Although Marcks and Marcksl1 proteins are severely depleted in our CRISPR-injected embryos, clones of cells with at least partially functional Marcks and Marcksl1 proteins are expected to be present due to the mosaicism of CRISPR gene editing effects in our F0 embryos. Assuming that subcellular localization of Marcks and Marcksl1 is functionally important in the spinal cord, we should thus be able to partially rescue the deficiencies of our CRISPR embryos by modulating PKC activity.

To test this hypothesis, we used the potent PKC activator Phorbol 12-myristate 13-acetate (PMA) and the broad-spectrum PKC inhibitor Go6983 (Liu and Heckman, 1998; Shen, 2003). Compared to DMSO-treated controls, embryos treated with the PKC inhibitor Go6983 showed a significant increase in the intensity of acetylated tubulin staining (Figure 4G) as well as in the proportion of PH3-positive cells (Figure 4H) on the CRISPR-injected side relative to the uninjected side. In contrast, the PKC activator PMA had no significant effect (Figure 4G, H). These findings suggest that the unphosphorylated, presumably membrane-bound fraction of Marcks and Marcksl1 is primarily required for neurite formation and progenitor proliferation in the spinal cord. Our other findings suggest that membrane-bound Marcks and Marcksl1 may promote these processes via the activation of PIP₂-mediated PLC and/or PLD signaling. However, the hierarchy of these interactions needs to be confirmed in future studies.

After complete transection of the spinal cord, tadpoles of X. laevis have previously been shown to close the injury gap with cells and regenerating neurites at around 1o days post transection (DPT), with almost full functional recovery at 20 DPT (Lee-Liu et al., 2014; Muñoz et al., 2015; Edwards-Faret et al., 2018). Because Marcks and Marcksl1 were identified in a proteomic screen for proteins differentially upregulated after SCI in regeneration-competent tadpoles (Kshirsagar, 2020), we next focused on the role of these proteins during spinal cord regeneration in Xenopus.

For these experiments, marcks.L/S and marcksl1.L/S sgRNAs (4M CRISPR) were injected into 1-cell stage embryos, which were then raised to stage 50 tadpoles (Figure 5A). After removing some dorsal skin, muscle, and meninges, the spinal cord was transected entirely at a mid-thoracic level. Sham-injured tadpoles were treated the same way but without transection of the spinal cord. Tadpoles were screened before SCI and at 2-, 5-, and 10-DPT for swimming capacity and some tadpoles at each time point were fixed, cryosectioned and evaluated by assessing the distribution of neurites (acetylated tubulin+) and measuring spinal cord gap closure length as well as percentages of proliferating cells (EdU+) and neuro-glial progenitor cells (Sox2+) (Figure 5A).

Figure 5 with 2 supplements see all

Download asset Open asset

We first analyzed the distribution of Marcks and Marcksl1 in the uninjured spinal cord and at 2-, 5-, and 10-DPT in normal (uninjected) and 4M CRISPR-injected tadpoles. In uninjected, uninjured tadpoles, Marcks protein was detected primarily in the meningeal layer surrounding the spinal cord, with relatively little immunostaining in the spinal cord itself (Figure 5B–D). Barely any Marcks protein was detectable in uninjured 4M CRISPR tadpoles (Figure 5N, O). At 2 DPT, the levels of Marcks protein in the meninges and spinal cord remain relatively unchanged in both normal (Figure 5E–G, V) and CRISPR-injected (Figure 5P, Q, V) tadpoles. However, additional Marcks staining was observed in normal tadpoles in cells within the injury gap (Figure 5E–G). At 5 DPT, normal tadpoles showed an apparent upregulation of Marcks in all layers of the spinal cord adjacent to the injury site, including the ventricular zone, where neuro-glial progenitors reside, the mantle zone, which harbors differentiated neurons, and the marginal zone, occupied by neurites (Figure 5H–J). Marcks also continues to be elevated in cells in the region of the injury gap, which is about to close (Figure 5H–J). This pattern of elevated Marcks staining was maintained at 10 DPT (Figure 5K–M, V). In contrast, 4M CRISPR tadpoles showed no significant increases of Marcks immunostaining in the spinal cord or adjacent tissues at 5- and 10-DPT (Figure 5R–U).

Compared to Marcks, Marcksl1 showed slightly stronger immunostaining in the spinal cord and weaker staining in the meninges of uninjected uninjured tadpoles but less pronounced upregulation after SCI (Figure 5—figure supplement 1A–L, U). As described for Marcks, Marcksl1-immunopositive cells also accumulated within the injury gap from 2 DPT onwards (Figure 5—figure supplement 1D–L). Again, almost no Marcksl1 immunostaining was detectable in 4M CRISPR tadpoles in and around the spinal cord before the injury, and no significant changes were observed following SCI (Figure 5—figure supplement 1M–T, U). Patterns of Marcks and Marcksl1 immunostaining for sham-operated tadpoles resembled those observed in uninjected and uninjured tadpoles and did not change at days 2–10 post-injury (Figure 5—figure supplement 2). Taken together, these observations suggest that Marcks and Marcksl1 are present at low levels in the tadpole spinal cord but become upregulated in all layers of the spinal cord next to the injury site after SCI (in particular, Marcks), while Marcks- and Marcksl1-immunopositive cells infiltrate the lesion site.

To further elucidate the role of Marcks and Marcksl1 in spinal cord regeneration, we next used a behavioral test to analyze whether functional recovery after SCI is impaired in 4M CRISPR tadpoles compared to normal (uninjected) tadpoles. Tadpoles were put into 6-well plates equipped with an intermittently vibrating motor to stimulate movement and were video recorded for 1.5 min. Their total swimming distances were then calculated from the videos. The distances covered by uninjected and 4M CRISPR-injected uninjured tadpoles were not significantly different (Figure 6A). This, together with our histological data described below, suggests that while the CRISPR editing of Marcks and Marcksl1 perturbs neurite formation and cell proliferation during normal development, ultimately, these tadpoles develop a relatively normally functioning spinal cord. Following SCI, there is a significant decrease in swimming distances at 2 and 5 DPT with no significant difference between uninjected and CRISPR animals (Figure 6A, Figure 6—figure supplement 1A).

Figure 6 with 1 supplement see all

Download asset Open asset

However, at 10 DPT, the average swimming distance of uninjected tadpoles had recovered to a level not significantly different from pre-injury levels. In contrast, the average swimming distance of 10 DPT CRISPR tadpoles was significantly shorter (Figure 6, Figure 6—figure supplement 1). Animals with sham surgeries showed no significant changes in swimming distances compared to uninjected and uninjured tadpoles over time (Figure 6—figure supplement 1B), suggesting that spinal cord transection alone accounts for these behavioral differences. This indicates that the CRISPR editing of Marcks and Marcksl1 significantly delays the capacity of tadpoles to recover functionally after SCI, supporting the vital role of these proteins for spinal cord regeneration.

We next used immunostaining with acetylated tubulin and DAPI nuclear staining to analyze how marcks and marcksl1 gene editing affects the regrowth of neurites and injury gap closure (measured as the distance between the DAPI+ ventricular zones of rostral and caudal stump) after SCI. Horizontal sections of the spinal cords of uninjured 4M CRISPR-injected animals resembled those of uninjured animals with prominent and well-distinguished ventricular, mantle, and marginal zones (Figure 6B, F). This demonstrates that the CRISPR-induced defects during spinal cord development described above only delay but ultimately do not prevent the formation of an adequately organized spinal cord, allowing the use of these animals for spinal cord regeneration studies.

After SCI, the length of the injury gap is initially similar for uninjected and 4M CRISPR-injected tadpoles (Figure 6C, G, J). However, at 5 DPT, the first axonal bridges seem to form and connect the rostral and caudal stumps in the uninjected animals, which is delayed until 10 DPT in the CRISPR tadpoles (Figure 6D, E, H, I). Moreover, while the injury gap decreased over time in both uninjected and 4M CRISPR-injected tadpoles, the injury gap at 5- and 10-DPT in 4M CRISPR-injected remained significantly larger than in uninjected embryos, indicating a delay in regenerative closure (Figure 6D, E, H, I, J).

To determine the effects of Marcks and Marcksl1 loss of function on cell proliferation during spinal cord regeneration, EdU was injected into the tadpoles 16 hr before the analysis timepoint. DAPI-stained nuclei and EdU+ nuclei were then counted in a defined window across the injury site (i.e., the area shown in each of the panels of Figure 7). There was no significant difference between uninjured uninjected and 4M CRISPR-injected animals (Figure 7 A1, B1, A2, B, I). In response to SCI, then a significant increase in the proportion of EdU+ cells was observed in uninjected tadpoles at 2- and 5-DPT, which returned to pre-injury levels at 10 DPT (Figure 7A1–H, I, Figure 7—figure supplement 1). An increase in EdU+ cell numbers was seen in the ventricular zone and cells in and adjacent to the injury gap (Figure 7C1–F1). Sham-operated animals showed no significant changes in EdU+ cells over time (Figure 7—figure supplement 2), indicating that these changes occur only after spinal cord transection. In contrast, there was no significant change in the proportion of EdU+ cells in the 4M CRISPR-injected tadpoles after SCI (Figure 7A2–H2, Figure 7—figure supplement 1).

Figure 7 with 2 supplements see all

Download asset Open asset

To test whether Marcks and Marcksl1 also affect the distribution and numbers of neuro-glial progenitors during spinal cord regeneration, we then visualized and quantified Sox2-immunopositive cells after SCI. Again, the distribution of Sox2+ cells in the ventricular zone was similar, and there was no significant difference between uninjured uninjected and 4M CRISPR-injected animals (Figure 8 A1, B1, A2, B2, I).

Figure 8 with 2 supplements see all

Download asset Open asset

In uninjected tadpoles at 2 and 5 DPT, we observed an increase in Sox2+ cell numbers rostral to the injury site of the transected spinal cord (Figure 8C1–F1). However, near the injury, Sox2+ cells were reduced (particularly at 2 DPT). This was reflected in a slight decrease in the number of Sox2+ cells counted in our standard window around the injury site (Figure 8I). However, at 10 DPT, Sox2+ cells were noticeably and significantly increased in the entire ventricular zone of the regenerating spinal cord, both rostral and caudal of the injury site (Figure 8, Figure 8—figure supplement 1). Again, animals with sham surgeries did not show any significant changes in the population of Sox2+ cells (Figure 8—figure supplement 2), indicating that the observed changes occur only after spinal cord transection.

In the 4M CRISPR-injected tadpoles, there was a slight overall decrease in Sox2+ cells at 2- and 5-DPT, similar to the uninjected tadpoles (Figure 8C2–F2, I, Figure 8—figure supplement 1). However, in contrast to the uninjected tadpoles, the CRISPR tadpoles did not show any elevation of Sox2+ immunostaining in the rostral spinal cord, and there was no overall increase in Sox2+ cell numbers at 10 DPT, with Sox2+ counts significantly decreased compared to uninjected tadpoles (Figure 8G2, H2, I, Figure 8—figure supplement 1). Taken together, our findings show that after SCI Marcks and Marcksl1 are required for multiple processes implicated in spinal cord regeneration (Lee-Liu et al., 2013; Alunni and Bally-Cuif, 2016; Rodemer et al., 2020), including neurite outgrowth and the proliferation and possibly maintenance of neuro-glial progenitors in the ventricular zone.

Since we found phospholipid signaling to be involved in mediating the effects of Marcks and Marcks1 on neurite formation and proliferation of neuro-glial progenitors during normal spinal cord development, we next asked whether the role of Marcks and Marcks1 for spinal cord regeneration may also be modulated by phospholipid signaling. 4M CRISPR-injected tadpoles with SCI were exposed to the activator of phospholipid signaling S1P, the PLD inhibitor FIPI, or only DMSO in control embryos for 7 days. At 7 DPT, the tadpoles were then compared for behavioral recovery, injury gap closure, and EdU+ and Sox2+ cell numbers (Figure 9). The swimming recovery test revealed that tadpoles exposed to S1P swam significantly longer distances than control animals (Figure 9P). Conversely, tadpoles exposed to the PLD inhibitor FIPI traveled a shorter distance than the DMSO controls, but this difference was insignificant (Figure 9P). Immunostaining for acetylated tubulin combined with nuclear DAPI staining showed that injury gap length was significantly reduced in animals incubated in S1P. In contrast, it was significantly higher in the animals incubated in the PLD inhibitor FIPI (Figure 9A–C, Q). The proportion of proliferative EdU+ cells (Figure 9D–I, R) and of Sox2+ neuro-glial progenitor cells (Figure 9J–O, S) was also significantly increased in the tadpoles incubated in S1P and significantly decreased in those exposed to the PLD inhibitor FIPI compared to DMSO controls. Thus, in all parameters investigated, 4M CRISPR tadpoles treated with an activator of phospholipid signaling adopted a phenotype resembling normal tadpoles after SCI. In contrast, those treated with a PLD inhibitor showed more severe loss of function phenotypes than untreated CRISPR tadpoles. In distinction to embryos, where FIPI has no effect, the latter finding suggests that residual PLD activity must be present in CRISPR tadpoles either because some Marcks/Marcksl1-independent PLD activity remains or because Marcks/Marcksl1 have not completely been knocked out in the entire spinal cord.

Figure 9

Download asset Open asset

Taken together our findings show that activation of phospholipid signaling can rescue the delaying effects of Marcks and Marcksl1 loss of function on functional recovery, injury gap closure, and proliferation of neuro-glial progenitors after SCI, while PLD inhibition further exacerbates them. This suggests that the functions of Marcks and Marcksl1 during spinal cord regeneration are at least partially mediated by phospholipid signaling, similar to their roles in normal spinal cord development.

Frogs (X. laevis) were obtained from a commercial breeder (Xenopus One Corp, Michigan, USA). All experiments were done in accordance with Irish and European legislation under licenses AE19125/P079 and AE19125/P098 from the Health Products Regulatory Authority (HPRA).

Translation-blocking MOs, and a standard control MO (5′-CCTCTTACCTCAGTTACAA TTTATA-3′) were obtained from Gene Tools. X. laevis is an allotetraploid (Session et al., 2016) and its genome contains two homeologs (denoted .L and .S, respectively) of many genes, including marcks and marcksl1. Therefore, in our functional studies, we designed MOs to block both homeologs. The marcks MO (5′-GGTCTTGGAGAATTGGGCTCCCA TT-3′) (Iioka et al., 2004) target base pairs –16 to +9 of both marcks.S and marcks.L. The marcksl1.L MO (5′-GGATTCTATACTACCCATTTTCCGC-3′) targets base pairs –7 to +18 of marcksl1.L, while the marcksl1.S MO (5′- GGACTCTACGCTACCCATTGTGACT-3′) (Zhao et al., 2001) targets base pairs –2 to +23 of marcksl1.S.

The efficacy of all MOs was verified in western blots following in vitro transcription and translation (TNT-coupled Wheat Germ Extract kit, Promega) of pCMV-Sport6-marcks.S, pCMV-Sport6-marcksl1.S, and pCMV-Sport6.ccdb-marcksl1.L (1 μg/50 μl reaction) with and without MO (20 pg marcksl1.L MO or 40 pg of marcksL1.S or marcks.S MOs/50 μl reaction) as previously described (Ahrens and Schlosser, 2005). 2 µl biotinylated Transcend tRNA (Promega) was included in the reaction mix resulting in biotinylation of lysine residues. Polyvinylidene difluoride (PVDF) membranes were subsequently incubated with Streptavidin Alkaline Phosphatase (1:5000; Promega) to visualize proteins with Transcend Chemiluminescence Substrate (Promega). The specificity of all MOs was verified by co-injection with the rescue mRNAs (see below).

Plasmids encoding membrane-GFP (pCS2+-memGFP) and c-myc-GFP (pCMTEGFP) were provided by John Wallingford (Wallingford et al., 2000) and Doris Wedlich, respectively. cDNA clones for X. laevis marcks.S (clone accession: BC041207), marcks.L (clone accession: BC084888), and marcksl1.S (clone accession: BC157454) were purchased from Source Bioscience, provided in pCMV-Sport6 backbone vectors and pCMV-Sport6.ccdb for marcksl1.L. Due to a nucleotide sequence identity of 94% between marcks.L and marcks.S, only the cDNA clone for marcks.S was used. All plasmids were verified by sequencing and linearized with ClaI before mRNA synthesis using the SP6 Ambion Message Machine Kit (Invitrogen).

For rescue experiments, chicken marcks (clone ID: OGa33947C) and Xenopus tropicalis marcksl1 (clone ID: OXa00945C) ORF clones were purchased from GenScript (GenBank accession numbers: NM_001044437.1; XM_015284506.2). The clones were received in pET-23a(+) vectors, containing a His-tag and T7-tag, flanking the cloning sites on each terminal. There were 3/25 base-pair mismatches between Xl-Marcks.S MO and the pET-23a(+)-marcks_OGa33947C sequence; 9/25 mismatches between Xl-Marcksl1.L MO and pET-23a(+)-marcksl1_OXa00945C; and 10/25 mismatches between Xl-Marcksl1.S MO and pET-23a(+)-marcksl1_OXa00945C. To increase the number of mismatches between the MOs and the mRNA rescue constructs, seven and eight additional substitutions were introduced into pET-23a(+)-marcks_OGa33947C and pET-23a(+)-marcksl1_OXa00945C_plasmids, respectively, using the Q5 site-directed mutagenesis kit (New England Biolabs). These substitutions did not alter protein sequences. For mutagenesis, forward (5′-TTTAGTAAAACAGCTGCGAAGGGCGAAGCC-3′) and reverse (5′- CTGTGCACCCATGGATCCGCGACCCATTTGC-3′) primers for pET-23a(+)-marcks_OGa33947C were annealed at 70°C, while forward (5′- ATCGAGAGTAAGTCTAAGAGTGCAGATATTAGC-3′) and reverse (5′-GGAGCCCATGGATCCGCGACCCATTTG-3′) primers for pET-23a(+)-marcksl1_OXa00945C were annealed at 63°C. Mutagenized plasmids were verified by sequencing and linearized with AdeI before mRNA synthesis using the T7 Ambion Message Machine Kit (Invitrogen), followed by polyadenylation using the Poly(A) Tailing Kit (Invitrogen).

marcks.L/S and marcksl1.L/S genomic sequences were derived from the most recent version of the X. laevis genome on Xenbase (https://www.xenbase.org/) and NCBI (https://www.ncbi.nlm.nih.gov/genome/). Two single-guide RNA primers targeting the first exon of marcks.L/S (5′-CTAGCTAATACGACTCACTATAGGCTGCGGTCTTGGAGAATTGTTTTAGAGCTAGAAATAGCAAG-3′) and second exon of marcksl1.L/S (5′-CTAGCTAATACGACTCACTATAGGTTGGGGTCCCCGTTGGCGGTTTTAGAGCTAGAAATAGCAAG-3′), respectively (target-specific region underlined), were designed using CRISPRscan (https://www.crisprscan.org). These gene-specific sgRNA primers contained a T7 promoter site (TAATACGACTCACTATA) with five additional nucleotides in the 5′ end for enhanced transcriptional output, two guanines for in vitro transcript yield efficiency, 17–18 nucleotides encoding the target-specific region (underlined), and a 23-nucleotide tail for annealing to the 80-nucleotide universal reverse primer (5′-AAAAGCACCGACTCGGTGCCACTTTTTCAAGTTGATAACGGACTAGCCTTATTTTAACTTGCTATTTCTAGCTCTAAAAC-3′). As a control, an sgRNA targeting solute carrier family 45 member 2 (slc45a2) 5′-CTAGCTAATACGACTCACTATAGGTTACATAGGCTGCCTCCAGTTTTAGAGCTAGAAATAGCAAG-3′, encoding a protein that mediates melanin synthesis, was used (DeLay et al., 2018). In X. laevis, slc45a2 is present as only one homeolog, slc45a2.L. Off-target, and editing efficiency probabilities were assessed using InDelphi (https://indelphi.giffordlab.mit.edu/), GGGenome (https://gggenome.dbcls.jp/), and CRISPRscan.

sgRNAs were synthesized (Bhattacharya et al., 2015) using PCR with Q5 High-Fidelity DNA Polymerase (NEB), 1 µM each of forward and reverse primer and the following cycling conditions: 98°C for 30 s; 10 cycles of 98°C for 10 s, 62°C for 20 s, 72°C for 20 s; 25 cycles of 98°C for 10 s, 72°C for 30 s; and 72°C for 5 min. PCR products were purified using the DNA Clean & Concentrator-5 (Zymo Research, D4014) and in vitro transcribed using the MEGAshortscript T7 Transcription Kit (Invitrogen). After a 4-hr incubation at 37°C and subsequent TURBO DNase treatment for 15 min, the sgRNA was purified using the GeneJET RNA Cleanup and Concentration Micro Kit (Thermo Scientific) in accordance with the manufacturer’s protocol with the following modifications: The transcription reaction mixture was first adjusted to 50 µl with nuclease-free water, followed by the addition of 250 µl of Binding Buffer. After elution in nuclease-free water, the concentration of sgRNAs was adjusted to 500 ng/µl, and aliquots were stored at −80°C.

Embryos of X. laevis were staged according to Nieuwkoop and Faber, 1967 and injected according to standard procedures (Sive et al., 2000). mRNAs or MOs were injected into single blastomeres at the 4- to 8-cell stage, giving rise to the dorsal ectoderm. To visualize membranes in confocal microscopy, mem-GFP mRNA was injected (125 pg). For the overexpression study, 100 or 300 pg of marcks or marcksl1 mRNA was co-injected with 125 pg myc-GFP mRNA to identify the injected side. As a control, only myc-GFP mRNA (125 pg) was injected. MOs (see above) were injected singly or as a cocktail (9 ng each) with 125 pg c-myc-GFP mRNA. 27 ng of the standard control MO was used in control injections. For MO rescue experiments, 300 pg mutated chicken marcks mRNA and 350 pg mutated Xenopus tropicalis marcksl1 mRNA was co-injected with 9 ng of each MO. For CRISPR/Cas9 experiments, embryos were injected into single blastomeres at the 1- to 8-cell stage (targeted to the dorsal blastomere in injections at the 4- to 8-cell stage) with a ribonucleoprotein (RNP) mix containing 500 pg of each sgRNA, 1 ng TrueCut Cas9 Protein v2 (Invitrogen, A36497), and 125 pg myc-GFP mRNA. The RNP mix was incubated at room temperature for at least 10 min before injection.

Approximately 5 days after the injection of RNPs, genomic DNA was extracted from embryos using the Wizard Genomic DNA Purification Kit (Promega) according to the manufacturer’s Animal Tissue protocol. Upon extraction, the DNA was rehydrated in 25 µl of DNA Rehydration Solution overnight at 4°C. CRISPR-targeted regions in marcks.L/S and marcksl1.L/S were PCR-amplified using the EmeraldAmp GT PCR Master Mix (Takara Bio). For every gene, each 40 µl reaction contained 20 µl of 2× EmeraldAmp Master Mix, 1 µl of the forward and reverse primer mix diluted to 10 µM, 100 ng of genomic template DNA, and PCR-grade H₂O for volume adjustment. The cycling conditions were 98°C for 30 s, 30 cycles of 98°C for 10 s, 60°C for 30 s, 72°C for 1 min per kilobase of PCR product, and 72°C for 5 min. The following forward (F) and reverse (R) primer pairs were used: marcks.L (F: 5′-ATCACCTGATGGACGCATGG-3′, R: 5′-CCCCCACATCTAAAGCGGAG-3′), marcks.S (F: 5′-AGTGTCATGAATCAGCGGGG-3′, R: 5′-GTGTGTCTATTAGCGGCGGA-3′), marcksl1.L (F: 5′-GCTAGGAAGAAGCGAGTCCC-3′, R: 5′-ACCCGTTAACATGAGCAGCA-3′), marcksl1.S (F: 5′-AGAAAGAAGTGAGCCTAGAGTGATT-3′, R: 5′-ATCCCTCCAAGGGTGACAGG-3′), slc45a2.L (F: 5′-GTTCCCTTCGCTCATACAATG-3′, R: 5′-GCCAGAAAGGGGTTTATTGC-3′). 5 µl of each PCR product was run in a 1% agarose gel and the remaining 35 µl were purified using the QIAquick PCR Purification Kit (QIAGEN) before being sent for sequencing (sequencing primers: marcks.L: 5′-CCCATGCTGTCTGTCTTTGA-3′, marcks.S: 5′-GTGTGTCTATTAGCGGCGGA-3′, marcksl1.L: 5′-TCTTCTTCTGGCGCCTGC-3′, marcksl1.S: 5′-AGTTTAGGGAGGCAGGGTTG-3′, slc45a2.L: 5′-GTTCCCTTCGCTCATACAATG-3′). Sequencing raw data from wild-type, slc45a2, and marcks/marcksl1-targeted embryos were analyzed for the frequency of insertion/deletion mutations using the Synthego ICE algorithm (https://ice.synthego.com/) (Conant et al., 2022).

Affinity-purified polyclonal antibodies against Xenopus Marcks and Marcksl1 proteins were generated by the company GenScript. For anti-Marcks, a His-tagged fusion protein with the following sequence was expressed in bacteria: MHHHHHHASPAEGEPAEPASPAEGEPAAKTEEAGSTSTPSTSNETPKKKKKRFSFKKSFKLSGFSFKKNKKENSEGAENEGA. Using the Protein BLAST NCBI tool, this immunogen (including the 6-His-tag) shared a 98.68% identity (E value: 3e−43) with Marcks.L (NCBI Reference Sequence XP_018118779.1) and 98.67% (E value: 3e−42) with Marcks.S (XP_018119725.1). No other sequences in X. laevis (taxonomy id:8355) returned any significant alignments. The protein was purified in Nickel and Superdex 75 columns and, after liquid chromatography–mass spectrometry validation, was used to immunize rabbits. The antibody was then antigen-affinity-purified from sera and tested by indirect ELISA by Genscript. For anti-Marcksl1, keyhole limpet hemocyanin (KLH)-tagged peptides (CGDPKPEDPPGKQAK) were used to immunize rabbits and were affinity-purified. This peptide had a 100% sequence identity (E value: 6e−08) with both Marcksl1.L (NP_001108274.1) and Marcksl1.S (NP_001081982.2). The next closest sequence alignments returned insignificant E values greater than 0.34.

Embryos and tadpoles were fixed in 4% paraformaldehyde (PFA) in 0.1 M phosphate buffer (PB) overnight, followed by two washes in PB for 10 min, before cryoprotection in 20% sucrose/PB overnight at 4°C. The specimens were then positioned in rubber forms and embedded in O.C.T. (VWR) for 30 min before shock-freezing in 2-methylbutane cooled in liquid nitrogen. Immunohistochemistry on 20 µm cryosections was performed as previously described (Schlosser and Ahrens, 2004). Primary antibodies were diluted, as indicated in Table 1. The following secondary antibodies were used at 1:500 dilution supplemented with DAPI (100 ng/µl): Alexa Fluor 488- and Alexa Fluor 594-conjugated goat anti-mouse (Invitrogen A11001 and A11005, respectively); Alexa Fluor 488- and Alexa Fluor 594-conjugated goat anti-rabbit (Invitrogen A11008 and A11012, respectively); and Alexa Fluor 594-conjugated donkey anti-mouse and anti-rabbit (Invitrogen A21203 or A21207, respectively).

Incorporation and histochemical detection of 5-ethynyl-2′-deoxyuridine (EdU) was used to label proliferating cells before and after SCI. EdU was initially diluted to 5 mg/ml in 0.8× phosphate-buffered saline (PBS), followed by 5 µl intracoelomic injections into tadpoles 16 hr before the desired time point (Edwards-Faret et al., 2017). In tadpole cryosections immunostained for Sox2, EdU was detected after secondary antibody incubation using the Click-iT EdU Cell Proliferation Kit for Imaging and Alexa Fluor 488 dye (Invitrogen).

After fixation in 4% PFA, whole-mount in situ hybridization was carried out as previously described (Schlosser and Ahrens, 2004). pcMV-Sport6-marcks.S, pCMV-Sport6-marcksl1.S, and pCMV-Sport6.ccdb-marcksl1.L were linearized with EcoRV followed by synthesis of anti-sense RNA probes using the digoxigenin RNA Labeling Kit (Sigma-Aldrich) and SP6 polymerase. Vibratome sections (40 μm) were cut after the whole-mount in situ hybridization.

Uninjected and RNP-injected embryos (see above) were incubated in various chemical compounds 3 days post-fertilization for 24 hr at 14°C in a dark incubator. All compounds were solubilized in DMSO and diluted in 0.1× Modified Barth's Solution (MBS) (Sive et al., 2000). Control embryos were incubated in 2% DMSO/0.1× MBS. The medium was changed once after 12 hr of incubation in a 12-well plate (6 embryos per well). To determine the proper dosage, a dose-titration screening was first conducted for all chemicals using dilutions of 0.01, 0.1, 1, and 10 µM in DMSO. The highest dose that produced comparable mortality rates to the DMSO control embryos was selected for further studies. S1P (1 µM, TOCRIS) and 5-Fluoro-2-indolyl des-chlorohalopemide (FIPI) (10 µM, TOCRIS) were used to activate and inhibit PLD, respectively. NMI (10 µM, TOCRIS) and ISA-2011B (10 µM, MedChemExpress) enhanced or inhibited phosphatidylinositol biosynthesis, respectively. Finally, PMA (0.01 µM, TOCRIS) and Go 6983 (10 µM, TOCRIS) were used to activate or inhibit PKC, respectively.

Immunostained cryosections and vibratome sections were visualized using an Olympus BX51 fluorescence microscope and recorded with an Olympus DP72 camera. Some representative images were obtained by confocal microscopy (Olympus Fluoview 1000 Confocal Microscope). For each epitope, images were acquired using identical exposure settings for signal threshold determination and balancing. Representative images were flipped and/or cropped with contrast adjustments to highlight biological relevance using ImageJ (National Institutes of Health).

To quantify acetylated tubulin integrated density, the injected and uninjected areas of every second consecutive spinal cord section were manually selected using the freehand selection tool and measured in ImageJ. At least eight transverse 20 µm sections were analyzed per animal, spanning the length of more than 300 µm in developing spinal cords. The integrated density for the injected and uninjected sides was averaged separately for every animal (mean integrated density). Then, a ratio for these averages – the integrated density ratio – was calculated for every animal. To determine Sox2/3-positive cell numbers, individual cells from at least eight alternating serial sections were counted on the injected and uninjected sides. For every animal, the ratio for the average number of cells in the injected to uninjected sides was calculated. The percentage of PH3+ cell nuclei out of all ependymal DAPI-positive nuclei was calculated for quantification of cell division on the injected and uninjected sides. For every animal, the ratio of the percentage of PH3+ cell nuclei as well as the ratio of DAPI+ nuclei in the injected to uninjected sides was calculated. After SCI, at least five horizontal spinal cord 20 µm sections were obtained per animal and condition. The length between the regenerating stumps was calculated using ImageJ, and averaged for every animal. To quantify Marcks/Marcksl1 integrated density, the entire area of every second consecutive spinal cord section was measured in ImageJ. Six to ten horizontal 20 µm sections were analyzed per animal, spanning 300 µm across the spinal cord and extending for 440 µm along the anteroposterior axis across the injury site and including the rostral and caudal stumps (i.e., the area depicted in each of the panels of Figure 5, Figure 5—figure supplement 1), seeking to include all cell types within this area during the injury response. Total cell counts from DAPI, EdU, and Sox2 images were also determined using ImageJ. DAPI-stained nuclei and EdU+ nuclei were counted in at least five horizontal sections per animal. Within each section, the entire population of cells in the spinal cord in a window extending 330 µm across the spinal cord and extending for 440 µm along the anteroposterior axis across the injury site and including the rostral and caudal stumps was analyzed (i.e., the area depicted in each of the panels of Figure 7). After binary conversion and particle analysis in ImageJ, the percentage of EdU+ or Sox2+ cells over the entire nuclear cell count was calculated.

Several statistical tests were employed for this study using GraphPad Prism (San Diego, California). A one-way ANOVA was chosen to compare how a single variable, MOs or RNPs, could affect neurite formation, cell proliferation, numbers of nuclei, or levels of Marcks or Marcksl1. The mean of each condition was then compared using the Tukey test, which compares every mean with every other mean. For pharmacological treatments on developing embryos, a one-way ANOVA was also used with a Dunnett test to compare every mean to a control mean, which was the DMSO condition. A two-way ANOVA was used to compare how CRISPR-gene disruption and injury time points can affect gap closure length after SCI. Here, an uncorrected Fisher’s least significant difference test was used for pairwise comparisons of two specific groups (uninjected and CRISPR) against each other at every time point. Additionally, the mean percentage of EdU+ and Sox2+ cells in CRISPR and uninjected animals were compared at different time points using a two-way ANOVA with Sidak’s test to compare every mean independently. Results are represented in box plots or bar plots indicating the range between the 25th and 75th percentiles, with all data points shown. Single data points represent the average value of a given measurement per animal, with error bars representing minimum to maximum values.

Before surgery, stage 50 tadpoles were placed in 0.1% MS222 in 0.1× MBS for approximately 30 s until the reflex response to touch was absent. The animals were then immobilized in a plasticine-coated Petri dish and rinsed with transplantation solution (88 nM NaCl, 1 mM KCl, 1 mM MgSO₄·7H₂O, 5 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), 3 mM CaCl₂, 100 units penicillin/ml, 0.1 mg/ml streptomycin), removing any excess solution. Dorsal skin and adjacent muscle tissues from the mid-thoracic level were incised using fine forceps, exposing the spinal cord. After removing a small portion of the meninges, a tungsten wire was inserted below the spinal cord to lift it slightly, facilitating subsequent transection with a 25-gauge injection needle. For sham injury, only the skin and muscle tissues were incised. Following surgery, the tadpoles were placed in a transplantation buffer for 30 min before being transferred to 0.1× MBS supplemented with antibiotics (400 mg/l penicillin, 400 mg/l streptomycin, and 25 mg/l gentamicin) for the duration of the study. The animals were housed in individual wells in a six-well plate in 0.1× MBS and supplemented with antibiotics, and the media was changed twice daily for the first 3 days post-injury and once daily after that. Feeding the animals began 2 days post-surgery, consisting of Micron Growth Food (Sera, 00720) in 0.1× MBS. For pharmacological treatments, tadpoles were incubated post-surgery in DMSO, FIPI, and S1P, using the aforementioned concentrations, and solutions were changed once daily for 7 days. Tadpoles were terminally anesthetized during the appropriate time points by prolonged exposure (20 min) in 0.1% MS222. Following euthanasia, the tadpoles were rinsed twice for 5 min in 1× PBS and fixed in 4% PFA overnight at 4°C.

A quantitative recovery test was designed by recording and measuring vibration-induced swimming to analyze the recovery of tadpoles following SCI. For this, tadpoles were placed in a custom-made six-well plate attached to a 10-mm Pico Vibe vibration motor (Precision Microdrives), connected via USB to a Mega 2560 MCU Board (Arduino). The vibration’s duration was programmed using Arduino’s Integrated Development Environment. Tadpoles were placed in the recording chamber for a 1-min equilibration period and subsequently recorded for 90 s, during which they were subjected to six 3-s stimulation pulses at equal time intervals. After recording, total swimming distances were measured using EthoVision XT video tracking software (Noldus) and analyzed in GraphPad Prism.

The funders had no role in study design, data collection, and interpretation, or the decision to submit the work for publication.

This study was funded by Research Ireland and the European Regional Development Fund (ERDF) under grant number 13/RC/2073_P2. The authors acknowledge the use of the facilities of the Centre for Microscopy and Imaging at the University of Galway, Galway, Ireland, a facility that is co-funded by the Irish Government’s Programme for Research in Third Level Institutions, Cycles 4 and 5, National Development Plan 2007–2013. During the course of this work MEA held a Hardiman fellowship from University of Galway and a Government of Ireland PhD Scholarship from the Irish Research Council (GOIPG/2018/500).

All experiments were done in accordance with Irish and European legislation under licenses AE19125/P079 and AE19125/P098 from the Health Products Regulatory Authority (HPRA).

© 2024, El Amri et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.