Axon ensheathment is fundamental for fast impulse conduction and the normal physiological functioning of the nervous system. Defects in axonal insulation lead to debilitating conditions, but, despite its importance, the molecular players responsible are poorly defined. Here, we identify RalA GTPase as a key player in axon ensheathment in Drosophila larval peripheral nerves. We demonstrate through genetic analysis that RalA action through the exocyst complex is required in wrapping glial cells to regulate their growth and development. We suggest that the RalA-exocyst pathway controls the targeting of secretory vesicles for membrane growth or for the secretion of a wrapping glia-derived factor that itself regulates growth. In summary, our findings provide a new molecular understanding of the process by which axons are ensheathed in vivo, a process that is crucial for normal neuronal function.
The peripheral nervous system (PNS) is responsible for the innervation of the body musculature and organs, and for relaying signals from the periphery to the brain. Timely delivery of information across body regions is crucial for adequate neuronal function and behavior. Fast and efficient conduction of the nerve impulse requires appropriate axonal insulation, known to be mediated by glial wrapping of peripheral axons. In vertebrates, this function is assured by myelinating and non-myelinating Schwann cells, which are responsible for ensheathing and supporting the axons within peripheral nerves (Barres, 2008; Chang et al., 2016; Osso and Chan, 2017; Saab and Nave, 2017). Defects in axonal insulation lead to debilitating conditions, including multiple sclerosis (Barres, 2008). In invertebrates, such as Drosophila melanogaster (hereafter referred to as Drosophila), PNS axons are similarly enwrapped by glial cells, but instead of producing myelin, they produce a membrane sheath that surrounds the axons, being highly similar to non-myelinating Schwann cells (Banerjee and Bhat, 2008; Rodrigues et al., 2010; Schirmeier et al., 2015; Silies et al., 2007). How glial and neuronal development is coordinated and how these distinct cell types develop and interact to ensure correct nervous system function and normal behavior is far from being understood.
Drosophila larval peripheral nerves are composed of motor and sensory neuron axons that are surrounded by three distinct glial layers, which are then enclosed by an acellular extracellular matrix (ECM) layer called the neural lamella. The outermost glial layer is made of perineurial glia, the function of which remains elusive. Beneath this layer are subperineurial glia, which form auto-septate junctions (SJs) that form the blood–nerve barrier (BNB), thought to prevent the potassium-rich hemolymph from contacting the axons. Below the subperineurial glia is the innermost glial layer formed by wrapping glia, the glial subtype that ensures ensheathment and insulation of axons. Unlike vertebrate myelinating Schwann cells, wrapping glia do not produce myelin, but grow tremendously during development, producing large amounts of membrane to ensure that axons are either wrapped individually or in small bundles (von Hilchen et al., 2013; Matzat et al., 2015; Silies et al., 2007; Stork et al., 2008). The formation of glial sheaths starts during embryonic development and proceeds throughout larval stages, when the larvae increase in size ∼100 times. During larval development, wrapping glia and subperineurial glia show hypertrophic growth without proliferation, whereas perineurial glia become proliferative and divide (von Hilchen et al., 2013). This means that all wrapping glia and subperineurial glia present at the end of larval stages are generated during embryogenesis and their membranes have to expand massively to accommodate larval growth (von Hilchen et al., 2008, 2013; Matzat et al., 2015). The mechanism(s) by which these growth processes are regulated remain largely mysterious, with very few genes identified as required for the development of these distinct glial subtypes.
In the Drosophila PNS, some factors have been reported to play a role in wrapping glia development (Ghosh et al., 2013; Lavery et al., 2007; Leiserson et al., 2000; Matzat et al., 2015; Schmidt et al., 2012; Xie and Auld, 2011), but although these studies revealed some of the signaling pathways, how this enormous membrane-remodeling event is specifically regulated is poorly understood. Given its nature, genes involved in membrane addition are likely required. A pathway that has recurrently been reported to orchestrate the polarized targeting of vesicles and to regulate membrane growth events is the exocyst complex, via its interaction with the small GTPase Ral (Armenti et al., 2014; Balasubramanian et al., 2010; Lee and Schwarz, 2016; Patrício-Rodrigues and Teodoro, 2018; Teodoro et al., 2013). The exocyst is an octameric protein complex conserved from yeast to human that is an effector to many GTPases, including Ral. By being able to receive regulatory information from different pathways, the exocyst can serve as a hub to regulate precisely where and when vesicles fuse with the membrane, serving as a tethering factor prior to vesicle fusion with the membrane (Martin-Urdiroz et al., 2016; Mei and Guo, 2018; Picco et al., 2017). Interaction with Ral induces the assembly and activation of this complex (Moskalenko et al., 2001). Interestingly, one of the mammalian Ral isoforms, RalA, is present in oligodendrocytes (OLs) where it is highly colocalized with proteolipid protein, a myelin constituent, leading to the suggestion that RalA could be involved in myelin membrane biogenesis (Anitei et al., 2009). Likewise, the exocyst subunits Sec6 and Sec8 (also known as Exoc3 and Exoc4, respectively) are present in OLs and in myelin, and Sec8 has been shown to participate in OL differentiation, a process characterized by rapid process extension, membrane expansion and polarization (Anitei, 2006). Additionally, through an interaction with Dlg1, Sec8 has been shown to contribute to myelin formation in culture (Bolis et al., 2009). Together, these data hint at a role for RalA and for the exocyst in the process of axonal wrapping, but whether they are required for axonal ensheathing, or whether they function in the same pathway, is unknown.
Here, we show that RalA GTPase (the sole Ral gene in Drosophila) and the exocyst are regulators of wrapping glia development in the Drosophila PNS. We demonstrate that Rala mutants have underdeveloped wrapping glia but morphologically normal subperineurial and perineurial glia, implying a specific role for RalA GTPase in the regulation of wrapping glia growth. We show that wrapping glia number is normal and the defects result from a failure to grow rather than a failure to maintain membrane integrity. Genetic analyses show that RalA is required in wrapping glia to regulate its growth and development. In addition, we demonstrate that these mutants have thicker intersegmental nerves (ISNs), often with defasciculated axons. Altogether, we provide evidence that RalA functions in wrapping glia to regulate its own growth, and has additional, more pleiotropic functions, involving other glia subtypes and neurons that result in nerve thickenings. The defects in the mutants result in abnormal locomotion, which is due to a combined contribution of neurons and glia. Furthermore, we show that exocyst mutants have similar wrapping defects and that RalA GTPase likely functions via the exocyst complex in the regulation of these developmental processes. Given the known roles of the RalA/exocyst pathway, we propose that this pathway regulates axonal wrapping directly through the regulation of the targeting of secretory vesicles for membrane addition or, alternatively, by controlling the secretion of a yet-unidentified factor, that would itself regulate growth. In summary, our findings establish RalA GTPase and the exocyst as novel players required for wrapping glia development, and axonal ensheathment in vivo, a process that is crucial for normal neural function.
Rala mutants have abnormal nerve bundles
Post-embryonic PNS development requires substantial growth to accommodate the increase in larval size. Specifically, growth of PNS axons and their surrounding glial cells must be coordinated and synchronized with that of the rest of the body to ensure the proper formation of nerve bundles (Fig. 1A). Based on the fact that RalA GTPase is enriched in myelin-producing OLs, together with its known roles in the regulation of growth and proliferation in other cellular contexts, we hypothesized that RalA could participate in Drosophila larval PNS development. To investigate this, we labeled the body wall muscles using Phalloidin to mark F-actin and horseradish peroxidase (HRP) to label the nerve outline (Jan and Jan, 1982), because anti-HRP antibodies have been shown to label several neural-specific glycoproteins present in the central and peripheral nervous system of Drosophila and other insects (Jan and Jan, 1982; Snow et al., 1987). We analyzed the overall morphology of two distinct RalA GTPase mutants, RalaG0501 and RalaEE1, and observed that the 3rd instar larval ISNs were abnormally thick in these mutants (Fig. 1B,C). This phenotype is fully penetrant with thick ISNs present in all larvae examined (n>100 for each mutant). RalaG0501 is a protein null mutant (Teodoro et al., 2013) and RalaEE1 harbors a point mutation that results in an amino acid substitution, Ser154Leu, within a conserved amino acid sequence predicted to be required for nucleotide binding (Eun et al., 2007). Using anti-HRP to label the nerve outline, we quantified the width of the ISN at muscle 4 (m4) region. Our data shows that Rala mutant larvae have thicker ISNs than the wild-type (WT, w1118) controls (w1118=5.2±0.08 µm versus RalaG0501=8.0±0.26 µm and RalaEE1=7.8±0.13 µm; Fig. 1C,D). To minimize possible biases in the location of measurement of the nerve width, we calculated the nerve area corresponding to 80 µm in length on the same region of the ISNs. In agreement with the increase in nerve width, we found that Rala mutants have increased nerve area compared with WT (Fig. 1E).
RalA GTPase is required for the development of wrapping glia
Alterations in nerve morphology can have several origins, including defects in glial development (Ghosh et al., 2013; Leiserson et al., 2000, 2011; Petley-Ragan et al., 2016; Schmidt et al., 2012; Sepp et al., 2001). To learn about the cellular origin of such defects, we analyzed the morphology of each of the cell types that compose peripheral nerves (neurons and glia), by labeling each of them with CD4GFP, expressed under the control of cell type-specific Gal4 drivers. Neurons were visualized using nSyb-Gal4, which labels all neurons (Fig. 2A); for the different glia, we used Bsg-Gal4, Moody-Gal4 and Nrv2-Gal4 to label perineurial, subperineurial and wrapping glia, respectively (see Fig. S1 for Bsg-Gal4 specificity; Xie and Auld, 2011; Fig. 2B-D). We imaged the ISNs from segments A2-A4 (Fig. 2). Imaging of CD4GFP and HRP allowed us to visualize the nerve simultaneously with each of the cell types, and assess whether the morphology was altered in the mutants. Labeling of neurons with CD4GFP revealed a certain degree of defasciculation in the mutants compared with the controls (Fig. 2A). However, careful observation of the body wall innervation showed that there were no ectopic or de-innervated muscles (Fig. 1B), indicating that neurons are present in the correct number and targeting to the correct location. Staining of axons with an antibody for Futsch, an axonal microtubule-associated protein (Roos et al., 2000), was consistent with these observations revealing that axonal tracts were often defasciculated in Rala mutants (Fig. S2). However, this phenotype was highly variable, suggesting that defasciculation per se is unlikely to be the only cause for changes in nerve bundle morphology.
By extending this approach to glia, we observed that the overall morphology of perineurial and subperineurial glia in Rala mutants was identical to that of controls (Fig. 2B,C). Normally, perineurial glia wrapped around the entire bundle, and this configuration was unchanged in Rala mutants (Fig. 2B). Likewise, subperineurial glia extend membrane processes that cover the entire length of the nerve bundle of the larvae, and this morphology was also unchanged (Fig. 2C). In contrast with perineurial and subperineurial glia, the innermost glial layer, composed of wrapping glia, was severely underdeveloped in Rala mutants (Fig. 2D). Whereas in controls wrapping glia surrounded and completely wrapped the axons, in both mutants the membranes of wrapping glia failed to grow and appeared as thin membrane extensions that failed to enclose the axons (Fig. 2D). To quantify this growth defect in wrapping glia, we measured the area of an 80 µm length of the ISN, and confirmed that wrapping glia fail to grow in Rala mutants (Fig. 2D,F). We further validated this observation by expressing mCherry-CAAX under the control of Nrv2-Gal4 as an alternative strategy to label the wrapping glia membrane (Fig. S3), and tested another Rala null allele that dies at 60 h AEL – RalaPG89 (Teodoro et al., 2013; Fig. S4).
Given that wrapping glia do not divide during larval development, it is likely that this phenotype results from a failure in cell growth and in membrane addition. As subperineurial glia also grow and do not divide during larval development and their morphology was unchanged in the mutants, we conclude that RalA plays a role specifically in wrapping glia in the regulation of membrane growth. Finally, the neural lamella was examined by analyzing the protein-trap Viking-GFP (Bainton, 2014), a collagen-IV protein, and a major ECM protein. We detected no structural defects in the lamella of Rala mutants (Fig. 2E), akin to perineurial (which secrete some of the ECM components of the lamella), and to subperineurial glia.
Visualization of segmental nerves (SNs) at the exit from the ventral nerve cord (VNC) revealed that, similar to the ISN, only wrapping glia are morphologically affected in Rala mutants (Fig. S5). We observed that wrapping glia did not grow like controls, resulting in underdeveloped wrapping glia in all segmental nerves of all the larvae analyzed (Fig. S6), akin to the ISN. Quantification of the area of segmental nerves in WT and Rala mutants showed that SN area was also increased in the mutants compared with controls (w1118=1252±30 versus RalaG0501=1442±40 and RalaEE1=1422±40, values in µm2; Fig. S6C). However, in contrast to the ISN, the thickening of SNs was much less pronounced: rather than being generally thicker, the SNs were uneven in Rala mutants, with thinner and thicker regions along the nerve. To quantify this asymmetry, we measured the width of SNs at the widest point within 300 µm of the VNC as previously described (Petley-Ragan et al., 2016), and concluded that Rala mutants have a maximum width higher than that of controls (w1118=6.3±0.11, RalaG0501=7.4±0.14, RalaEE1=7.5±0.13, values in µm; Fig. S6B). In summary, our structural analysis revealed that RalA GTPase plays a role specifically in the regulation of wrapping glia growth (Fig. 2G) both in SNs and at the ISNs.
During larval development, wrapping glia do not divide but their membranes need to grow to accompany the ∼100 times increase in larval size (von Hilchen et al., 2013). Given that this is a continuous developmental process, the defects observed in Rala mutants (Fig. 2D) could reflect a failure to grow during development or alternatively reflect a collapse of pre-formed membranes. To distinguish between these possibilities, we analyzed wrapping glia morphology from late 1st instar through to 3rd instar larvae (Fig. S7A-C). Our data show that the wrapping glia membrane does not grow during development. In contrast to controls, in which, by late 1st instar, a uniform membrane covering the length of the nerve was visible (Fig. S7A), in the mutants the wrapping glia membrane remained as a thin thread throughout larval development, without signs of ever growing or ensheathing the axons (Fig. S7B,C). Because this defect could arise from a reduced number of wrapping glia, we expressed a nuclear marker (Histone-2B-YFP, H2BYFP) under the control of a wrapping glia-specific Gal4 driver (Nrv2-Gal4) and observed that the number of nuclei in Rala mutants is similar to controls (average of three per hemi-segment; Fig. S8). Therefore, we conclude that RalA GTPase is necessary for the growth of the wrapping glia membrane during development.
Rala mutants have abnormal ultrastructural nerve profiles
Although it is clear that RalA is necessary for wrapping glia development, it is paradoxical that a decrease in the size of wrapping glia results in thickening of ISNs. The increase in nerve thickness has been shown to result from BNB problems, or from defective extracellular volume or ionic homeostasis regulation (Baumgartner et al., 1996; Leiserson and Keshishian, 2011; Leiserson et al., 2000, 2011; Luong et al., 2018; Yu et al., 2000). To examine the ultrastructural features of nerve bundles with respect to the presence, distribution and organization of their cell types, we used transmission electron microscopy (TEM). We dissected 3rd instar larvae keeping the brain and nerve bundles intact, and obtained ultrathin sections of the SNs between 2 and 10 µm after the exit of the VNC, as described (Matzat et al., 2015). TEM images of WT and Rala mutant larvae revealed that the inner structure of the bundle in the mutants had a high incidence of large electron-transparent regions (ETRs), occupying an average of 6.55±1.49% in RalaG0501 and 10.28±2.31% in RalaEE1 of the nerve cross-section area compared with 0.93±0.33% in controls (Fig. 3, Fig. S9). These ETRs are highly reminiscent of what has been observed in fray and Ncc69 mutations where the regulation of the extracellular volume by subperineurial glia was disrupted (Leiserson et al., 2000, 2011). This suggests that RalA may also be an important regulator of ionic homeostasis or of the extracellular volume. In addition to the presence of ETRs, the ultrastructural analysis of the nerves supports the notion that wrapping glia are underdeveloped in Rala mutants (Fig. 3A,B, cyan), in agreement with the immunofluorescence data. We also counted the number of axonal profiles present in controls and in mutants and observed that these were identical between genotypes, and correspond to what has been reported previously (Leiserson et al., 2000; Matzat et al., 2015), which is consistent with normal neural innervation pattern. A defective BNB allows ions and solutes to exchange with the hemolymph also resulting in extracellular volume and ion homeostasis defects (Leiserson and Keshishian, 2011), raising the possibility that RalA may additionally play a role in BNB development or function.
In conclusion, our TEM images suggest that, in addition to defects in wrapping glia growth, Rala mutants show signs of defective subperineurial glia function, independent of membrane growth. These defects in the regulation of extracellular fluid are most likely the cause of the thickening of nerves, rather than the failure of wrapping glia to grow, but this would have to be confirmed by doing additional experiments, including performing TEM at the ISNs, which is technically more challenging.
RalA GTPase is required in wrapping glia to promote their growth
We identified RalA GTPase as necessary for wrapping glia development, but whether it is required cell-autonomously cannot be concluded from our mutant analysis. To test this, we used a knockdown strategy using RNAi (IR) to reduce the levels of RalA GTPase in wrapping glia, while simultaneously expressing CD4GFP under Nrv2-Gal4 to assess morphology and area. As in the mutants, expressing RalA-IR in wrapping glia resulted in reduced growth of this cell type (Fig. 4A,B). We observed a decrease in area comparable to that of Rala mutants (compare Figs 4B and 2F), indicating that RalA is not only required cell-autonomously in wrapping glia, but the effect observed with RalA-IR fully accounts for the defects seen in the mutants.
Even though the ISN-thickening phenotype observed in Rala mutants is unlikely to be solely caused by defects in wrapping glia development (Fig. 3), it can be used as readout for developmental alterations. To test the contribution of each cell type to this phenotype, we used the same Gal4 lines as in Fig. 2 to express RalA-IR. As controls, we crossed UAS-RalA-IR and the Gal4 lines with WT flies. Expression of RalA-IR in all glia, using Repo-Gal4, led to widespread thickening of the ISN with clear signs of defasciculated axons (Fig. 4C-E). Likewise, using the wrapping glia-specific driver (Nrv2-Gal4) to reduce RalA also resulted in an ISN significantly thicker than controls (Fig. 4C-E), indicating that RalA in wrapping glia also contributes to the regulation of nerve bundle morphology. Interestingly, even though both RalA-IR of all glia and RalA-IR of wrapping glia resulted in ISN thickening, we also observed the appearance of swellings along the ISN, but only in the ‘all glia’ knockdown (Fig. S10). This result corroborates the notion that RalA, in addition to regulating wrapping glia development, also plays additional roles in other glial subtypes.
To our surprise, we found that ISN thickness was also increased when we knocked down RalA in neurons using the pan-neuronal driver nSyb-Gal4 (Fig. 4C-E), suggesting that RalA in neurons mediates other aspects of ISN development. Given that mutants showed some defasciculation (Fig. 2A, Fig. S2), we examined the organization of axons in the neuronal knockdown of RalA. Staining of axons with anti-Futsch revealed axonal defasciculation in the RalA-IR larvae (Fig. S11). Altogether, these data indicate that the ISN thickening results from a combination of factors, whereas the regulation of wrapping glia growth is cell-autonomous.
To validate further the cell type-specific findings from the RNAi experiments, we asked whether re-insertion of RalA in wrapping glia could rescue the developmental defects observed in Rala mutants. For this, we expressed UAS-HA-tagged RalA (RalA-HA) under the control of the wrapping glia-specific driver Nrv2-Gal4, in the background of each of the Rala mutants. Because RalA is known to localize to the plasma membrane irrespective of its nucleotide binding state (Teodoro et al., 2013), and given that the rescue construct has an HA-tag, we extrapolated the morphology of wrapping glia by staining RalA-HA. Our results show that expression of RalA-HA in wrapping glia leads to the rescue of wrapping glia area and morphology (Fig. 5A,B).
To quantify nerve thickness as an alternative readout for the capacity of each cell type to rescue Rala phenotypes, we expressed UAS-RalA-HA in the mutant backgrounds using cell type-specific Gal4 drivers. However, we observed that introduction of the UAS-RalA-HA transgene in the background of the hypomorphic mutant RalaEE1 (but not in the null RalaG0501) was able to rescue nerve thickness alone, even without any Gal4 driver (Fig. 5E), implicating that this construct is leaky. This result supports the notion of RalaEE1 being a hypomorphic allele, where low levels of RalA are sufficient to rescue nerve thickness, but it does limit the conclusions that can be made from the cell type-specific rescues. Still, we decided to test the effects of expressing RalA-HA using each of the Gal4 drivers (Fig. S12). We observed that re-insertion of RalA-HA in neurons was unable to rescue nerve thickness in the null RalaG0501 mutant, whereas in RalaEE1 background it did rescue, but not above the leakiness of RalA-HA (compare RalaEE1 versus RalaEE1;nSyb>RalA-HA with RalaEE1;RalA-HA versus RalaEE1;nSyb>RalA-HA; Fig. 5C,E). However, analysis of the area of the nerve upon nSyb-Gal4>RalA-HA expression, which is more sensitive to smaller changes, revealed that neuronal expression of RalA rescued nerve area in both mutants (Fig. 5D,F), suggesting that RalA in neurons also contributes to ISN morphology. When RalA-HA was expressed in all glia using Repo-Gal4, there was a rescue of both nerve thickness and nerve area in the two Rala mutants, compatible with a glial function in the regulation of nerve morphology (Fig. 5C-F). Wrapping glia-specific expression of RalA-HA also rescued nerve thickness in both Rala mutants, but, akin to the neuronal rescue, in RalaEE1 the rescue of nerve area was not above the leakiness of RalA-HA (area of RalaEE1;RalA-HA not different from RalaEE1;Nrv2>RalA-HA; Fig. 5C-F). Altogether, our results suggest that RalA is required in glia to regulate nerve thickness and morphology. Additionally, although to a lesser degree, neurons also contribute to ISN morphology. Finally, given that RalA-HA is leaky, we cannot exclude the possibility that the rescue of wrapping glia growth (Fig. 5A,B) has a contribution from other cell types, but it is clear that re-introducing RalA in the mutant background leads to a complete rescue of wrapping glia morphology, confirming that RalA GTPase is required for wrapping glia development (Fig. 5A,B).
RalA is required for normal locomotion
Alterations in nerve development and function often result in abnormal larval crawling behavior (Schmidt et al., 2012). As wrapping glia are responsible for insulating axons, defects in wrapping glia membrane growth should lead to reduced axonal insulation, altered electrical conduction and, ultimately, locomotor problems. To assess whether this was the case in Rala mutants, we performed a larval locomotion assay in which wandering 3rd instar larvae were filmed for 5 min while freely wandering in an agar arena, and their positions were tracked with IdTracker Software (Perez-Escudero et al., 2014). WT larvae displayed the expected bimodal crawling behavior, alternating between active crawling and reorientation events, and crawled on average 134.6±7.7 mm in 5 min, with an average speed of 0.52±0.03 mm/s (Fig. 6A,B). Rala mutants also displayed the bimodal pattern, but crawled less and at slower speed than controls: during the 5 min analyzed, RalaG0501 and RalaEE1 larvae crawled on average 92.5±5.7 mm and 88.2±4.4 mm, at an average speed of 0.37±0.02 mm/s and 0.39±0.0 2 mm/s, respectively (Fig. 6A,B). Interestingly, we observed that, at times, Rala mutant larvae seemed to drag their posterior body-half, as the peristaltic contractions appeared incomplete, losing power along the anterior-posterior axis (compare WT and mutant larvae in Movies 1-3). To uncover if the differences between WT and mutants were due to the mutants crawling more slowly in the active crawling phase or spending more time in reorientation events, we analyzed the speed of long uninterrupted forward crawls. WT larvae had an average active crawling speed of 0.67±0.07 mm/s, whereas RalaG0501 and RalaEE1 mutants performed fewer long forward runs and crawled at significantly lower speeds: 0.38±0.02 mm/s and 0.38±0.04 mm/s, respectively (Fig. 6B). These data are consistent with our hypothesis of signal dissipation in axons, suggesting that motor axons in peripheral nerves lacking fully developed wrapping glia fail to propagate action potentials efficiently, leading to locomotor deficits. However, given that this assay was performed in Rala mutant larvae, in which the gene is mutated in all tissues, we cannot conclude whether this defective behavior derives from RalA being required in glia or elsewhere.
To assess the requirement for RalA in a cell type-specific manner, we performed the same locomotor assay, followed by tracking with IdTracker (Heras et al., 2019), in larvae expressing RalA-IR under the control of the cell type-specific Gal4 drivers used before. Controls were done by crossing the UAS-RalA-IR or the Gal4 with WT. RalA knockdown significantly reduced locomotor activity in neurons (nSyb=148.8±0.5.18 mm versus nSyb>RalA-IR=118.2±6.34 mm), all glia (Repo=147.7±3.93 mm versus Repo>RalA-IR=98.7±5.59 mm), and to a lesser degree in wrapping glia (Nrv2=138.6±5.32 mm versus Nrv2>RalA-IR=116.7±4.95 mm; Fig. 6C). Altogether, as in previous sections, our results support a combined contribution of the different cell types to locomotion.
RalA GTPase regulates wrapping glia and nerve bundle development via the exocyst complex
Several different pathways have been reported to act downstream of RalA GTPase, including the exocyst complex, RalBP1 and filamin (Moghadam et al., 2017; Shirakawa and Horiuchi, 2015). RalA interaction with RalBP1 is mostly involved in endocytosis, filamin and RalA act as scaffold/synaptic organizers (Lee and Schwarz, 2016), and the RalA/exocyst pathway is required for membrane addition of postsynaptic membranes and for the polarized trafficking of vesicles to exocytic places (Armenti et al., 2014; Balasubramanian et al., 2010; Hase et al., 2009; Teodoro et al., 2013; Wang et al., 2004). Of these interactors, the exocyst is an excellent candidate to mediate RalA-dependent wrapping glia growth. If the exocyst complex regulates wrapping glia and nerve bundle morphology, mutants should have a phenotype similar to Rala mutants.
The exocyst is required for many intracellular functions, including targeting and tethering of secretory vesicles to the plasma membrane. This complex is usually ubiquitously expressed given its roles in the regulation of several key intracellular trafficking steps (Fukuda, 2019; Lepore et al., 2018). By carrying out antibody staining at the ISN, we showed that the exocyst subunit Sec5 is ubiquitously distributed, and that a tagged version of exocyst subunit Sec8 (Sec8HA), expressed in wrapping glia, had some enrichment at the plasma membrane (Fig. S13). To test if the exocyst is required for wrapping glia development, we expressed RNAi against the subunit Sec5, which is known to bind directly to RalA and has been reported to promote membrane growth in other systems (Teodoro et al., 2013). We expressed RNAi against Sec5 (Sec5-IR) in wrapping glia together with CD4GFP, using Nrv2>Sec5-IR,CD4GFP, to visualize the morphology of this glial subtype and observed that reduction of Sec5 in wrapping glia impaired their growth and the total wrapping glia area, akin to Rala mutants (Figs 7A,B, compare with 2D,F). Rather than completely ensheathing the axons, wrapping glia in Nrv2>Sec5-IR,CD4GFP did not fill the entire nerve area and were thinner and less regular than those of controls.
To assess the contribution of the exocyst to overall nerve development, we knocked down Sec5 in the other cell types. When Sec5-IR was expressed in neurons, using nSyb-Gal4, there was no effect on the morphology of the bundle (Fig. 7C,D; nSyb=5.92±0.25 µm, nSyb>Sec5-IR=5.41±0.17 µm). However, expression of Sec5-IR in all glia resulted in a very thick nerve, with extreme signs of axonal defasciculation, highly reminiscent of Rala mutants (Fig. 7C,D, Fig. S14; Repo=4.64±0.10 µm, Repo>Sec5-IR=8.02±0.38 µm). When Sec5-IR was expressed exclusively in wrapping glia, using Nrv2>Sec5-IR, the nerve was also significantly thicker than that of Gal4 controls (Fig. 7C,D; Nrv2=6.40±0.25 µm, Nrv2>Sec5-IR=7.48±0.19 µm), resembling RalA-IR in wrapping glia (Fig. 4A). In Drosophila, most exocyst mutants die as 1st instar larvae, but a mutation induced by the insertion of a P-element in the Sec8 locus (Sec8P1) results in a hypomorphic mutant that survives to 3rd instar stages (Liebl et al., 2005). We analyzed the thickness of the ISN in this mutant and in a control line in which the P-element was precisely excised (Sec8revs). Our data show that control larvae (Sec8revs) have an ISN that is 5.67±0.13 µm, which is significantly different from Sec8P1, where the ISN is thicker (7.08±0.13 µm; Fig. 7E,F). Therefore, like Rala mutants, Sec8P1 larvae have thicker ISNs than do larvae of the control line. In summary, similar to RalA GTPase, the exocyst is involved in wrapping glia growth and in the regulation of nerve morphology, resembling Rala mutants.
If RalA and the exocyst complex are in the same genetic pathway, heterozygous mutations of these genes could genetically interact and give rise to defective wrapping glia or a thick nerve phenotype. We postulated that mutating a single exocyst component would probably be insufficient to reveal an interaction between this eight-protein complex and RalA GTPase. Therefore, we tested if heterozygous mutations in Rala genetically interacted with double-heterozygous mutations in two exocyst components: Sec5 and Sec6. For this experiment, we used a line in which null mutations in the loci of Sec5 (Sec5E10) and Sec6 (Sec6Ex15) were recombined and tested in heterozygosity. This double mutant was tested in combination with heterozygous Rala mutations, RalaG0501 and RalaEE1. To assess genetic interactions, we measured wrapping glia area and ISN thickness in the different heterozygous combinations, which was labeled by expressing CD4GFP with Nrv2-Gal4. We found that Sec5, Sec6 and RalaG0501 interact genetically for the regulation of wrapping glia growth and development, whereas RalaEE1 does not (Fig. 8A,B). We observed that the triple-heterozygous RalaG0501/+; Sec5, Sec6/+ have significantly reduced wrapping glia compared with the other genotypes (Fig. 8A,B), a situation similar to what was observed in the mutants (or RNAi). On the other hand, the lack of interaction with RalaEE1 is consistent with this allele being a hypomorph, which is in agreement with our rescue data (Fig. 5), where RalA-HA leakiness rescued RalaEE1, but not RalaG0501. In other words, it is likely that there was enough RalA function in the heterozygous RalaEE1 to sustain wrapping glia growth, contrary to what occurs in the null allele.
Analysis of the ISN width shows that, in contrast to the single and double mutants, triple-heterozygous mutants of Rala, Sec5 and Sec6 exhibit a significantly thicker nerve (compare Figs 1C and 8C,D). Heterozygous RalaG0501 and RalaEE1 were indistinguishable from WT (w1118). Double heterozygous mutations of Sec5 and Sec6 result in a small but significant increase in nerve thickness compared with WT, but that is strongly potentiated by the introduction of one mutated allele of the Rala gene (Fig. 8B). In this assay, both Rala alleles interacted with the exocyst, again in agreement with the more pleiotropic nature of the thick nerve. In conclusion, our data demonstrate that RalA GTPase regulates wrapping glia and nerve bundle development via the exocyst complex.
Axonal ensheathment is a fundamental biological process that allows fast impulse conduction and the correct functioning of the nervous system (Saab and Nave, 2017; Seidl, 2014). Despite its importance, the molecular players responsible for the regulation of this process are ill-defined. Our study identifies RalA GTPase and the exocyst complex as novel factors required for wrapping glia growth and development in vivo (Fig. 8E). Through mutant analysis, cell type-specific RNAi and rescue experiments, we discovered that RalA GTPase plays a specific role in the regulation of wrapping glia growth. We also discovered that Rala mutants have locomotion defects, abnormal extracellular volume in the SNs and defasciculated axon bundles. The defects observed in exocyst mutants and RNAi suggest that, at the ISN, glia-dependent functions are mediated via the exocyst, whereas neuronal fasciculation functions are not. Based on our findings, we propose a mechanism whereby activation of RalA GTPase in wrapping glia induces recruitment of the exocyst, probably via Sec5 interaction, targeting vesicles to the membrane. These vesicles can directly contribute to membrane growth, but can also carry a wrapping glia-derived secreted factor, which can signal wrapping glia growth (Fig. 8E). Alternatively, a growth signal from another cell type might fail to be perceived by wrapping glia. Our study identifies a new in vivo mechanism by which glia regulate axonal ensheathing, providing novel insights into the molecular basis of wrapping glia and nerve development.
In this study, we tested two mutations in the Rala locus that, despite having different strengths (null versus hypomorphic), result in similar defects in wrapping glia and ISN morphology. Still, we show that the hypomorphic mutation can more easily be rescued (RalA function is restored by RalA-HA leakiness and this mutant does not interact with Sec5, Sec6 in the regulation of wrapping glia growth). These data indicate that low levels of RalA GTPase are sufficient for function, but that if a certain threshold is not achieved, the resulting defects are identical to null mutations. This observation is compatible with the known role of small GTPases, including RalA, as molecular switches, whereby tight regulation of the levels and activity of these proteins are crucial for function. For example, although the absence of RalA results in reduced wrapping glia growth (this study), it is known that mutations that render RalA overactive give rise to peripheral nerve sheath tumors in mice (Bodempudi et al., 2009). It will be interesting to explore further how RalA expression and activation are regulated and the impact of this on axonal ensheathing in vivo.
We identified RalA and the exocyst as factors required for wrapping glia membrane growth. This pathway has been implicated in membrane addition in other cellular contexts, including postsynaptic growth (Patrício-Rodrigues and Teodoro, 2018; Teodoro et al., 2013), nanotube formation (Hase et al., 2009) and in growth phases during cell division (Giansanti et al., 2015; Holly et al., 2015; Murthy et al., 2010). Although the RalA/exocyst pathway is clearly a mechanism widely used for growth regulation, our results indicate that it is not simply a default growth signal, as subperineurial glia size was unchanged in the mutants and this glia subtype also depends on hypertrophic growth during larval development. Therefore, one (or more) yet-unknown factor(s) must confer this specificity.
Previous studies have identified molecules required for wrapping glia development, including Vein (Matzat et al., 2015), the homolog of mammalian neuregulin 1, and Akt (Lavery et al., 2007) – two proteins that also play a role in mammalian myelination (Domenech-Estevez et al., 2016; Michailov, 2004). Additionally, mutations in fray (a serine/threonine kinase) (Leiserson et al., 2000), Khc (Schmidt et al., 2012), lace (a serine palmitoyltransferase) (Ghosh et al., 2013), integrin signaling (Xie and Auld, 2011) and Laminin (Petley-Ragan et al., 2016) result in defective wrapping glia development. Some of these pathways can potentially intersect with RalA/exocyst function. Namely, the neuregulin 1 homolog Vein has been shown to be required in wrapping glia to mediate its own development, and non-cell-autonomously to induce SJ formation in subperineurial glia. Knowing that the exocyst is required for the correct trafficking and secretion of a different EGFR ligand, Gurken (Murthy and Schwarz, 2004), and that RalA has been implicated in EGFR activity (Lu et al., 2000), it is conceivable that the RalA/exocyst pathway could participate in the secretion of Vein.
Despite not being required for subperineurial glia growth, given the observed disruption of the extracellular volume in Rala mutants, it is likely that RalA functions in this glia subtype to regulate other aspects of development and function. In fact, RalA, via the exocyst, has been shown to be required for normal tight junction assembly and barrier function in MDCK epithelial cells (Hazelett et al., 2011), making this pathway a candidate for regulating SJ, and possibly BNB, formation in the Drosophila PNS.
Even though invertebrates do not produce myelin, it is possible that the genes important for accomplishing such an important function as axonal ensheathing are conserved, even if with some layers of adaptation. The fact that in vertebrates RalA and exocyst are present in myelin (Anitei, 2006; Anitei et al., 2009) and that Sec8 has been implicated in the secretory events that promote myelin sheath growth (Bolis et al., 2009) support this idea. In this work, we found that cell-autonomous manipulation of the levels of RalA GTPase in wrapping glia leads to severe defects in axon ensheathment and that Rala mutants have locomotor deficits, supporting an important physiological role of RalA GTPase for the control of movement. Interestingly, two recent papers reported that RalA GTPase and the exocyst are required in Schwann cells during axonal sorting and in nerve repair by regulating Schwann cell process formation, migration and myelination (Galino et al., 2019; Ommer et al., 2019). These studies support our data concerning the involvement of RalA in glial regulation of axonal ensheathment. Furthermore, our study together with Ommer et al. (2019) and Galino et al. (2019), highlight the conservation of the molecular mechanisms responsible for axonal insulation between vertebrates and invertebrates.
Defects in axonal wrapping are the main underlying cause of the symptoms present in patients with multiple sclerosis (Pan and Chan, 2017; Saab and Nave, 2017). Although it has long been appreciated that myelination in the PNS is required for movement, recent studies have shown that in neurological disorders of the central nervous system (CNS), such as Alzheimer's disease, there are signs of myelin loss in the white matter, even in pre-symptomatic phases of the disease (Saab and Nave, 2017), making the understanding of the mechanism by which ensheathing is regulated ever more relevant. Currently, no therapies are available that can induce any adjustment to the levels of axonal wrapping, which precludes the treatment of disorders that have either too little or too much myelin. Identification of the specific molecules that regulate axonal wrapping in the PNS and CNS might help identify genes that could be used as a target for new drugs or novel therapies. Our study defines a novel pathway that specifically regulates wrapping glia growth, providing a new mechanistic insight into the process by which axons in the PNS are ensheathed in vivo. RalA GTPase has successfully been used as a target for the design of small molecules that interfere with its function (Yan et al., 2014). Therefore, if the molecular players required for this process are conserved, Drosophila PNS nerves could be used as an in vivo system in which to test genetic interactions and/or new drugs by easily assessing the degree of ensheathment, and this information could potentially be extended to vertebrates.
MATERIALS AND METHODS
Fly stocks and husbandry
All the flies used in this study were kept at 25°C, except the RNAi crosses that were performed at 29°C. The following fly strains were used: w1118 [Bloomington Drosophila Stock Center (BDSC), 3605]; RalaG05010 (BDSC, 12283); RalaEE1 (BDSC, 25095); RalaPG89 (Teodoro et al., 2013); Sec8PI (Liebl et al., 2005); Sec8revs (Liebl et al., 2005); Sec5E10 (Murthy et al., 2003), Sec6Ex15 (Murthy et al., 2005); nrv2-GAL4 (BDSC, 6800); repo-GAL4 (BDSC, 7415); nSyb-GAL4 (BDSC, 19183); Bsg-GAL4 (Kyoto Drosophila Genetic Resource Center, 105188); Vkg-GFP [Vienna Drosophila Resource Center (VDRC), 318167]; UAS-RalA-HA (Lee and Schwarz, 2016); UAS-CD4-GFP (BDSC, 35836); UAS-CD8-GFP; UAS-CAAX-mCherry. RNAi knockdown strains used were UAS-RalA-IR (BDSC, 29580) and UAS-Sec5-IR (VDRC, 28873).
Description of RalA mutations
RalaG0501 is a protein null mutant (Teodoro et al., 2013) with P-element inserted upstream of the Rala 5′ UTR region, and RalaEE1 harbors a point mutation that results in an amino acid substitution, Ser154Leu, within a conserved amino acid sequence predicted to be required for nucleotide binding (Eun et al., 2007). The Rala gene is located on the X chromosome and the mutants analyzed in this study are either late 3rd instar/early pupae lethal (RalaG0501) or survive to adulthood; adult males are sterile and die a few hours post-eclosion (RalaEE1). Because Rala is located on the X chromosome, we always tested male larvae. Using these two independent mutants ensures that our observations are not due to second site mutations.
Larvae of the desired stage were dissected exposing the body walls, the brain and the peripheral nervous system. Dissections were performed on Sylgard plates, or Sylgard-covered slides, in a drop of 1× PBS and fixed in 4% formaldehyde for 20 min at room temperature. Antibody staining was performed in PBS containing 0.3% Triton X-100 and 5% normal goat serum. Larvae were incubated overnight at 4°C in primary antibodies, washed for at least 1 h, blocked for 30 min to 1 h and incubated for 2 h at room temperature in secondary antibodies, diluted in blocking solution. Larval fillets were mounted in mounting media DABCO (1,4-diazabicyclo [2.2.2] octane, Sigma-Aldrich).
The following primary antibodies were used: rabbit anti-HA (C29F4, Cell Signaling Technology, 1:500); mouse anti-Futsch (22C10, Developmental Studies Hybridoma Bank, 1:100); mouse anti-Sec5 (Murthy et al., 2003). Secondary antibodies used were the following: Alexa 488 anti-mouse (715-545-150, Jackson ImmunoResearch, 1:500) and DyLight 647 anti-mouse (715-605-151, Jackson ImmunoResearch, 1:500). Cyanine 3 (Cy3)- or Alexa 488-conjugated goat anti-HRP (123-165-021 and 123-545-021, respectively, Jackson ImmunoResearch, 1:500) was used to label neuronal membranes. Texas Red-conjugated Phalloidin (T7471, Molecular Probes/Invitrogen, 1:500) was used to to label F-actin. All reactions were performed using secondary antibodies with minimal cross-reactivity. Endogenous fluorescence was used in the experiments with GFP and mCherry.
Transmission electron microscopy (TEM)
For TEM, larvae fillets from 3rd instar larvae were dissected and fixed in 2% formaldehyde, 2.5% glutaraldehyde and 0.03% of picric acid in PHEM buffer at room temperature for 2 h as described by Matzat et al. (2015). Dissection pins were then removed and larvae were fixed overnight at 4°C with the same solution freshly made. The specimens were subsequently fixed with 2% tannic acid for 30 min on ice and post-fixed with 2% osmium tetroxide reduced with 0.8% potassium ferricyanide for 1 h on ice and 2% uranyl acetate for 30 min on ice. Larvae were dehydrated in an ascending ethanol series, and were then treated with propylene oxide to improve resin infiltration due to the presence of the larval cuticle. The larval fillets were then embedded in EPON resin. Transverse ultrathin sections (85-90 nm) were cut 2-10 µm after the end of the ventral nerve cord. TEM images were acquired using Hitachi H-7650 microscope. For statistical analysis, we measured the area of the ETR that is present immediately after the subperineurial glia layer (and within the wrapping glia layer) and divided this value by the total nerve area.
To assess larvae behavior, an open field arena crawling assay was used. Five 3rd instar larvae were placed in a 3% agarose (MultiPurpose Agarose, tebu-bio) circular arena with a 10% agarose salt concentrated barrier on the edges of the arena. After acclimatization to the arena, larvae were filmed for 5 min at 5 frames/second. The larvae were tracked in each frame using IdTracker software (Perez-Escudero et al., 2014), which provides coordinates that can be used to calculate total distance and medium velocity. To represent the trajectories, the initial coordinates were plotted in a smooth line dispersion graph.
Locomotor behavior protocol: analyses and corrections
Confocal imaging and quantification of nerve parameters
All images were acquired using an LSM710 confocal microscope (Zeiss). For the images in Fig. 1B, we used a 10× objective and optical sections of 10 µm; images for Fig. S2 with segmental nerves were acquired with a 40× objective with optical sections of 0.7 µm. All ISN images were acquired using a 63× lens, with optical sections of 0.7 µm. All images were acquired using the same settings within each experiment. Images were analyzed using a maximum z-stack projection in ImageJ. Images of whole fillets were obtained by tiling overlapping 10× images of the same larvae.
Quantification of nerve thickness
Nerve thickness was measured in the ISN at the region of muscle 4 innervation, segments A2-A4. Measurements were performed in a region in the maximum projection image that is representative of the ISN and that is not at the motor neuron exit point and does not include any glial nuclei, since these can represent thicker points in the nerve. The measure tool in ImageJ was used to acquire the thickness (in µm).
Quantification of 80 µm-length nerve area
ImageJ was used to trace a rectangle that was 80 μm in length and 20 μm in height to ensure that the entire region of interest (ROI) was captured. Thresholding followed by conversion to 'binary', to outline the entire ROI in the HRP channel (HRP ROI) was done; this labels the whole HRP area in each genotype. An area measurement using the ImageJ ‘measure’ function, limited to the HRP ROI, was done to determine the total nerve area within 80 μm length.
Statistical analysis was conducted using Excel (Microsoft Corporation, 2011) and/or GraphPad Prism 6 software (GraphPad Software). Datasets were tested for normality and if all samples passed this test we used parametric tests; otherwise, non-parametric tests were used. For pairwise comparisons, statistical significance was determined using a Mann–Whitney test. All multiple comparisons were performed using the following criteria: (1) if values normally distributed, a one-way analysis of variance (ANOVA) with Tukey's multiple comparisons test, or (2) if values not normally distributed, Kruskal–Wallis with Dunn's multiple comparisons test. Data are represented as box plots, which comprise the 25-75% confidence interval, as well as the median. Whiskers represent the 90% confidence interval. In the legends and text, values are given as mean±s.e.m.; sample sizes (n) are given either in the figure legends or in the Results section. In Fig. 3C, measurements are shown as mean±s.e.m. All statistical details related to the figures are provided in Table S1.
We would like to thank César Mendes for help with the behavioral experiments and for critical discussions throughout, and Ann Goldstein, Alisson Gontijo, Catarina Homem and Lara Carvalho for critical reading of the manuscript. We thank Dr Edgar Gomes for his help. We thank Telmo Pereira from the Microscopy Facility for technical support. We thank Marta Santos and the Fly Facility at CEDOC; CONGENTO: consortium for genetically tractable organisms; and Erin Tranfield from the Electron Microscope Facility at IGC. We thank Thomas Schwarz, the Developmental Studies Hybridoma Bank, the Drosophila Genomics Resource Center, and the Bloomington Drosophila Stock Center for antibodies and fly stocks.
Conceptualization: J.F.S.-R., C.F.P.-R., V.d.S.-X., P.M.A., A.C.F., R.O.T.; Methodology: J.F.S.-R., C.F.P.-R., V.d.S.-X., P.M.A., A.C.F., A.R.F., R.O.T.; Formal analysis: J.F.S.-R., C.F.P.-R., V.d.S.-X., P.M.A., A.C.F., R.O.T.; Investigation: J.F.S.-R., C.F.P.-R., J.P.M., V.d.S.-X., P.M.A., A.C.F., R.O.T.; Writing - original draft: J.F.S.-R., C.F.P.-R., V.d.S.-X., P.M.A., R.O.T.; Writing - review & editing: J.F.S.-R., C.F.P.-R., V.d.S.-X., P.M.A., R.O.T.; Visualization: J.F.S.-R., C.F.P.-R., A.R.F., J.P.M., R.O.T., V.d.S.-X., P.M.A.; Supervision: R.O.T.; Funding acquisition: R.O.T.
This work was supported by H2020 Marie Skłodowska-Curie Actions [H2020-GA661543-Neuronal Trafficking to R.O.T.], Fundo Regional para a Ciência e Tecnologia [IF/00392/2013/CP1192/CT0002 to R.O.T.] and iNOVA4Health (UID/Multi/04462/2013) (co-funded by FCT-FEDER-PT2020).
The authors declare no competing or financial interests.