elifeeLifeeLife2050-084XeLife Sciences Publications, Ltd5792110.7554/eLife.57921Short ReportNeuroscienceRobo functions as an attractive cue for glial migration through SYG-1/NephQuZhongweihttps://orcid.org/0000-0002-6994-05301ZhangAlberthttps://orcid.org/0000-0003-1310-56801YanDonghttps://orcid.org/0000-0002-7542-1251dong.yan@duke.edu12Department of Molecular Genetics and Microbiology, Duke University Medical CenterDurhamUnited StatesDepartment of Neurobiology, Regeneration Next Initiative, Duke Center for Neurodegeneration and Neurotherapeutics, and Duke Institute for Brain Sciences, Duke University Medical CenterDurhamUnited StatesShenKangReviewing EditorHoward Hughes Medical Institute, Stanford UniversityUnited StatesCooperJonathan ASenior EditorFred Hutchinson Cancer Research CenterUnited States1911202020209e579211504202002112020© 2020, Qu et al2020Qu et alhttp://creativecommons.org/licenses/by/4.0/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.

As one of the most-studied receptors, Robo plays functions in many biological processes, and its functions highly depend on Slit, the ligand of Robo. Here we uncover a Slit-independent role of Robo in glial migration and show that neurons can release an extracellular fragment of Robo upon cleavage to attract glia during migration in Caenorhabditis elegans. Furthermore, we identified the conserved cell adhesion molecule SYG-1/Neph as a receptor for the cleaved extracellular Robo fragment to mediate glial migration and SYG-1/Neph functions through regulation of the WAVE complex. Our studies reveal a previously unknown Slit-independent function and regulatory mechanism of Robo and show that the cleaved extracellular fragment of Robo can function as a ligand for SYG-1/Neph to guide glial migration. As Robo, the cleaved region of Robo, and SYG-1/Neph are all highly conserved across the animal kingdom, our findings may present a conserved Slit-independent Robo mechanism during brain development.

Roboglial migrationSYG-1cleavageResearch organismC. eleganshttp://dx.doi.org/10.13039/100000065National Institute of Neurological Disorders and StrokeNS094171YanDonghttp://dx.doi.org/10.13039/100000065National Institute of Neurological Disorders and StrokeNS105638YanDonghttp://dx.doi.org/10.13039/100006512School of Medicine, Duke UniversityThe Holland Trice AwardsYanDongThe funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.Author impact statementAn extracellular cleavage fragment of Robo released from neurons can bind to glia-expressed SYG-1 to guide glia migration.Introduction

The evolutionarily conserved Robo family was first identified in the classic genetic studies of Drosophila CNS axonal midline crossing (Seeger et al., 1993) and belongs to the immunoglobulin (Ig) superfamily. Besides their functions in axon guidance (Dickson and Gilestro, 2006), Robo is also involved in cell migration (Killeen and Sybingco, 2008; Viveiros et al., 2011; Hinck, 2004; Grieshammer et al., 2004), organogenesis (Grieshammer et al., 2004; Mommersteeg et al., 2013), cancer development (Wang et al., 2008), and immune cell regulation (Wu et al., 2001; Branchfield et al., 2016). In these processes Robo functions as the receptor for the guidance cue molecule – Slit. However, mutations of Robo cause abnormalities in Caenorhabditis elegans (Zallen et al., 1998; Hao et al., 2001), zebrafish (Fricke et al., 2001), and cultured mammalian neurons (Hivert et al., 2002; Liu et al., 2004) that have not been observed in Slit mutants, suggesting that there may be evolutionarily conserved Slit-independent aspects of Robo function and regulation.

Glial cells often need to migrate over long distances from their birthplace to appropriate regions where they form functional units with neurons or perform other roles (Gilmour et al., 2002; Jarjour and Kennedy, 2004; Kinrade et al., 2001; Klämbt, 2009). Some guidance molecules including those in the Netrin and Semaphorin families were shown to be involved in glial migration (Kinrade et al., 2001; Jarjour et al., 2003; Sasse and Klämbt, 2016; Spassky et al., 2002; Unni et al., 2012). In Drosophila, CNS-derived glial cells can move along nerves to reach their final position, and during migration the glial expression of ROBO2 receptor is required for preventing glial breakaway from the nerve in a Slit-dependent manner (Sasse and Klämbt, 2016).

Proteolytic modifications play important roles in regulating Ig superfamily proteins. One good example is the Netrin receptor DCC. DCC can be cleaved first at the extracellular region and subsequently at the intracellular domains (Neuhaus-Follini and Bashaw, 2015). The DCC intracellular cleavage fragment can then translocate into the nucleus and equip DCC with transcriptional regulatory function (Neuhaus-Follini and Bashaw, 2015). Similarly, NCAM and EphrinA5 can go through extracellular cleavage to release adhesive interactions and cause cell detachment (Brennaman et al., 2014). Drosophila Robo can be cleaved at the FN3 (Fibronectin type-III) repeats of the extracellular region, and the extracellular cleavage fragment can recruit Son of sevenless (Sos) to mediate Slit-dependent midline repulsion (Coleman et al., 2010). However, in all these cases the cleaved fragments and the function of the receptors highly rely on their canonical ligands. It is still unknown whether the cleavage of Ig superfamily receptors can function in a fashion that is independent of their canonical ligands.

There are in total 56 glia in C. elegans, which include 50 neuroepithelial glia that ensheath sensory neuron receptive endings (Oikonomou and Shaham, 2011; Shaham, 2015). The largest sensory organ of the worm, called the amphid sensilla, is a pair of sensilla composed of 12 neurons and two glial cells each (Ward et al., 1975). These glia, called the amphid sheath (AMsh) and amphid socket (AMso) cells, form channels that ensheath the dendrites of sensory neurons in the amphid sensilla (Oikonomou and Shaham, 2011; Ward et al., 1975; Perkins et al., 1986). Furthermore, these glial cells are vital for the proper functioning of the neurons that they ensheath, where they can modulate neural activity through secreted molecules at the receptive endings and control neuron receptive ending shape (Bacaj et al., 2008; Shaham, 2010; Singhvi et al., 2016). In previous studies we show that the fate determination of C. elegans AMsh glial cells utilizes similar mechanisms as those in mammals (Zhang et al., 2020a). AMsh glia are born close to the nose of C. elegans, where their processes anchor to the nose region while the cell bodies migrate toward the nerve ring (Shaham, 2015; Heiman and Shaham, 2009), and the position of AMsh glia is important for their function (Zhang et al., 2020b). With all these advantages, AMsh glia have emerged as a powerful model to study the potential conserved mechanisms underlying glial development and function.

SAX-3/Robo regulates glial migration in a Slit-independent manner

In an unbiased genetic screen for glial migration, we isolated two loss-of-function alleles of sax-3, the only Robo receptor in C. elegans (Zallen et al., 1998), with defects in AMsh glial migration. To quantify the migration of AMsh, we used the pharyngeal terminal bulb as a reference point and found that AMsh cell bodies reside at the side of and align with the center of pharyngeal terminal bulbs in day 1 (D1) young adults (Figure 1a and b). AMsh glia fail to migrate to their final positions in both sax-3(lf) alleles, and yad10 animals display stronger phenotypes than yad147 animals and show a similar extent of defects as those in the null allele of sax-3 (ky123) (Zallen et al., 1998), suggesting that yad10 is likely a null allele of sax-3 (Figure 1a and b and Figure 1—figure supplement 1a and c). To further characterize the sax-3(lf) phenotype we examined AMsh migration at different developmental stages in a single animal and found that sax-3(lf) animals exhibited migration defects in all post-embryonic stages, suggesting that sax-3 is required for AMsh migration during development (Figure 1—figure supplement 1b).

<italic>sax-3/</italic>Robo regulates AMsh glial migration in a Slit-independent fashion.

(a) Loss-of-function in sax-3 causes defects in AMsh glial migration. Confocal images and schematic representation of AMsh glia in control and sax-3(yad10) animals expressing Pf53f4.13::GFP (yadIs48). TB: pharyngeal terminal bulb. Scale bar, 10 µm. (b) Quantification of Migration Index (MI) in control and three alleles of sax-3(lf) animals. MI is calculated as (b–a)/a × 100%. As illustrated in the (a), ‘a’ represents the distance between the tip of nose and the center of the pharyngeal terminal bulb. ‘b’ shows the distance between the tip of nose and the center of the AMsh cell bodies. (c) The function of sax-3 does not rely on slt-1. White and gray bars show the percentage of animals with defects in one AMsh or both AMsh glia respectively. ‘no vab’: animals with normal head morphology. (d) Expression of a secreted SAX-3 extracellular fragment in neurons rescue AMsh defects in sax-3(lf) animals. Data shows the percentage of animals with AMsh migration defects. Punc-33(Pneuron), Pf53f4.13(PAMsh), and Pf16f9.3(PAMsh(2)) are used as pan-neuronal and AMsh-specific promoters to express genes in neurons or AMsh glia. In (b), data are represented as mean ± SEM. One-way ANOVA test. **, p<0.01; Each point represents at least 30 worms. In (c) and (d), data are represented as mean ± SEM. Two-way ANOVA test. **, p<0.01; ns, no significant difference. Each point represents three experiments of at least 50 worms each.

Loss-of-function of <italic>sax-3/</italic>Robo causes AMsh migration defects.

(a) Diagram shows the mutations in four different sax-3 alleles examined in this study. (b) Data showing the distance between the nose tip and the center of AMsh cell bodies at different developmental stages of individual animals. Five animals were quantified for each genotype, and each animal is represented by a single line. (c) Gel pictures (top) and quantification of RT-PCR results show that both yad10 and yad147 mutations decrease the expression of sax-3 mRNA. act-1/actin was used as a control. Student’s t-test, *p<0.05. (d–e) AMsh migration defects in sax-3 mutants did not originate from a displaced nerve ring. Confocal images (d) and quantification data (e), % of animals with AMsh glia locating at the anterior of the nerve ring, show that AMsh glia are at the anterior of the displaced nerve ring. (g) Confocal images show that both PAMsh (Pf53f4.13) and PAMsh(2) (Pf16f9.3) are expressed in the AMsh glia at the 1.5-fold stage. H2B::GFP was expressed under these promoters. (f) Quantification shows the percentage of animals with ‘vab head’ defects. Punc-33 promoter is used as the pan-neuronal promoter. (h) Diagrams show all constructs used in this study. In (e) and (f), data are represented as mean ± SEM. Two-way ANOVA test. **, p<0.01; ns, no significant difference. Each point represents three experiments of at least 50 worms. Scale bar, 10 µm.

Next, we examined the function of the canonical Robo ligand, Slit, in AMsh migration and found that loss-of-function of slt-1(ok255, null allele), the only Slit in C. elegans, did not cause any defects, suggesting that Robo functions in a Slit-independent manner to regulate AMsh migration (Figure 1c). As sax-3 is involved in the regulation of nerve ring position (Zallen et al., 1998; Sasakura et al., 2005), it could be argued that the AMsh migration defect could originate from the anteriorly positioned nerve ring in sax-3(lf) mutants. To exclude this possibility, we examined the relative position of AMsh glial cells and the nerve ring. Results show that AMsh cell bodies reside posterior to the nerve ring in control animals but are misplaced anterior to the nerve ring in sax-3(lf) animals (Figure 1—figure supplement 1d and e), suggesting that the AMsh migration defect is not a side phenotype coming from a displaced nerve ring. As about 40% of sax-3(lf) animals display defects in head morphogenesis, called ‘vab head’ (Ghenea et al., 2005), it is also possible that AMsh defects could be a result of abnormal head morphology. To rule out this possibility, we examined sax-3(lf) animals without ‘vab head’ phenotypes and found that they displayed the same degree of AMsh defects as in all sax-3(lf) animals, supporting the conclusion that the AMsh migration defects in sax-3(lf) animals are independent of the ‘vab head’ phenotypes (Figure 1c).

The extracellular cleavage fragment of SAX-3/Robo functions as an attractive cue for glial migration

To understand how sax-3 regulates AMsh migration, we carried out tissue-specific rescue experiments. Unexpectedly, as a well-studied receptor, sax-3 does not function cell-autonomously in AMsh migration, and instead expression of sax-3 cDNA in neurons fully rescues sax-3(lf) AMsh phenotypes (Figure 1d, Figure 1—figure supplement 1g). It is even more striking that expression of truncated forms of SAX-3 lacking the intracellular domains (ΔC) or both intracellular and transmembrane domains (Δ[C+TM]) fully rescues sax-3 AMsh defects (Figure 1d, Figure 1—figure supplement 1h), while these transgenes did not affect the sax-3(lf) ‘vab head’ phenotypes (Figure 1—figure supplement 1f and h). A possible explanation for these observations is that a secreted extracellular fragment of SAX-3 may function as an attractive cue released by neurons to guide AMsh migration. To reach this possibility, either sax-3 should have an alternative splicing form that only contains its extracellular region or SAX-3 protein should be cleaved in the extracellular region. The C. elegans genome has been well annotated and studied (Gupta and Sternberg, 2003), and sax-3 does not have any known splicing forms containing only the extracellular region. Interestingly, Drosophila and human Robo have been shown to be cleaved at the FN3 repeats of the extracellular region (Coleman et al., 2010; Seki et al., 2010). As FN3 repeats are highly conserved between C. elegans SAX-3 and Drosophila Robo, we decided to test whether SAX-3 could also be cleaved in the extracellular region. We first generated transgenes expressing SAX-3::HA in the nervous system and collected animals at different developmental stages. As shown in Figure 2a and Figure 2—figure supplement 1a, we were able to detect a cleavage fragment around 70 kDa in different transgene lines and using different anti-HA antibodies. As the HA tag is fused to the C-terminus of SAX-3, based on the molecular weight (about 70 kDa) the cleavage site is likely to be in the FN3 repeats, which is consistent with what was found in Drosophila Robo (Coleman et al., 2010). More importantly, the extracellular cleavage happens from the embryonic 1.5-fold stage to early larval stages, when AMsh glia migrate from their birthplace to the nerve ring region (Figure 2a and Figure 2—figure supplement 1a). As the function of sax-3 in AMsh migration is independent of Slit, we asked whether the cleavage of SAX-3 relied on Slit and found that the cleavage of SAX-3 was not regulated by Slit (Figure 2b and Figure 2—figure supplement 1b). To further confirm that the cleavage site is in the FN3 repeats, we generated transgenes expressing a mini SAX-3 extracellular region that only contains the FN3 repeats, transmembrane domain, and a GFP inserted between the signal peptide and the first FN3 repeat (GFP-FN3-TM) and examined the cleavage products of this fragment. We found that FN3 repeats were sufficient for the cleavage (Figure 2c), and based on the molecular weights of free GFP (27 kDa), GFP-FN3-TM (~80 kDa), and the cleavage product (~45 kDa) we concluded that the cleavage site was in the second repeat of FN3. We then generated transgenes expressing the GFP-FN3-TM fragment lacking one of the three FN3 repeats. As deletion of the first FN3 repeat (FN3-a) caused failure of expression, we were only able to examine the effect of removing the second (FN3-b) or the third FN3 (FN3-c) repeat. We found that the cleavage was abolished if the second FN3 repeat was absent, while deletion of the third FN3 repeat did not affect the cleavage (Figure 2d). Consistent with its important role in SAX-3 cleavage, extracellular cleavage was undetectable in transgenes expressing a truncated form of SAX-3 without the second FN3 repeat (Figure 2e). These data support the conclusion that the cleavage site of SAX-3 is in the second repeat of FN3. To test whether the extracellular cleavage of SAX-3 is essential for its function in AMsh migration, we examined the rescue ability of the truncated form of SAX-3 that lacks the second FN3 repeat and found that this mutant form of SAX-3 completely lost rescue ability (Figure 2f, Figure 2—figure supplement 1c). The extracellular cleavage of SAX-3 appears to show certain specificity for AMsh migration, as the same transgene (SAX-3 ΔFN3-b) fully rescued sax-3(lf) ‘vab head’ phenotypes (Figure 1—figure supplement 1f, Figure 2—figure supplement 1c). To further confirm the role of extracellular SAX-3 cleavage in AMsh migration, we generated a sax-3(yad175) deletion allele that lacks the second repeat of FN3 (FN3-b). sax-3(yad175) decreases sax-3 mRNA to about 50% of that in control animals (Figure 2—figure supplement 1d) and appears to be a weak loss-of-function allele of sax-3, as both sax-3(ky123, presumptive null allele) and sax-3(yad10) show higher penetrance of vab head and AVM guidance defects (Zallen et al., 1999; Figure 1—figure supplement 1f and Figure 2—figure supplement 1e). Despite being a weak allele, sax-3(yad175) exhibits stronger AMsh migration defects when compared with any other sax-3 alleles examined, supporting the critical role of SAX-3 extracellular cleavage in AMsh migration (Figure 2f). To illustrate the function of the extracellular cleavage fragment, we expressed a secreted SAX-3 extracellular fragment containing Ig 1–5 domains and the first FN3 repeat (FN3-a) under pan-neuronal (Punc-33 and Prgef-1) or amphid neuron-specific (Parl-13) promoters, which mimics the cleavage product, and found that this fragment was able to fully rescue sax-3(lf) AMsh defects (Figure 2f, Figure 2—figure supplement 1f). Further experiments show that both Ig domains and the first FN3 repeat (FN3-a) may be important for the function of the SAX-3 extracellular cleavage fragment in AMsh migration (Figure 2f).

The extracellular cleavage of SAX-3 is required for AMsh glial migration.

(a) SAX-3 is cleaved in the extracellular region during development. A schematic diagram at the top panel shows that the HA tag was fused at the C-terminal of SAX-3, and the leftover of the extracellular cleavage detected by the HA antibody is about 70 kDa. Results from western blot show a cleavage fragment of SAX-3 in later embryonic and early larval stages. F: full length SAX-3::HA; C: the C-terminal fragment generated by SAX-3 extracellular cleavage. (b) Loss-of-function of slt-1/Slit does not affect the cleavage of SAX-3. F: full length SAX-3::HA; C: SAX-3 cleavage fragment. (c) The FN3 repeats mediate SAX-3 cleavage. FN3: FN3 repeats; TM: transmembrane domain; F: full length; C: cleavage fragment; G: free GFP. (d) FN3-b (the second repeat of FN3 repeats) is required for the cleavage. F: full length; C: cleavage fragment. (e) Deletion of FN3-b prevents SAX-3 cleavage. Images show results from two independent transgenic lines with FN3-b deletion (ΔFN3-b). Although having slightly different expression patterns, they both lose the extracellular cleavage fragment of SAX-3. (f) Expression of a fragment mimicking the SAX-3 extracellular fragment fully rescues AMsh defects in sax-3(lf) animals. Data are represented as mean ± SEM. Two-way ANOVA test. **, p<0.01; ns, no significant difference. Each point represents three experiments of at least 50 worms each.

The Slit-independent SAX-3 cleavage.

Images show the results from a second anti-HA antibody to confirm that SAX-3 has a developmental-stage-dependent cleavage (a), and this cleavage does not depend on slt-1 (b). (c) Deletion of the second FN3 repeats (FN3-b) did not decrease the expression or change the expression pattern of SAX-3 in transgenes. (d) Gel pictures (top) and quantification of RT-PCR results show that yad175 mutation decreases the expression of sax-3 mRNA. act-1/actin was used as a control. Student’s t-test, *p<0.05. (e) Quantification shows percentage of animals with AVM guidance defects. AVM was visualized using zdIs5 (Pmec-4::GFP) marker, and AVM guidance defects were defined as any abnormality in AVM axon growth direction. (f) Quantification data show that expression of a fragment (Ig(1-5)-FN3-a) mimicking the SAX-3 extracellular cleavage product by pan-neuronal (Punc-33, Prgef-1) or amphid neuron-specific (Parl-13) promoters can fully rescue AMsh migration defects in sax-3(lf) animals. Data are represented as mean ± SEM. Two-way ANOVA test. **, p<0.01; ns, no significant difference. Each point represents three experiments of at least 50 worms. Scale bar, 10 µm.

SYG-1/Neph is the receptor for the SAX-3/Robo extracellular cleavage fragment during glial migration

With the discovery of the function of the SAX-3 extracellular cleavage fragment in AMsh glial migration, we hypothesized that the cleavage fragment is released from neurons, where sax-3 is expressed (Zallen et al., 1998), and functions as an attractive cue for glial migration. If that is the case, there should be a receptor in AMsh glia to interact with the SAX-3 cleavage fragment and mediate AMsh migration. To identify the potential receptor for the SAX-3 cleavage fragment, we generated a transgene expressing a functional FLAG::SAX-3 fusion protein, in which a FLAG tag was inserted between the signal peptide and the first Ig domain. We first confirmed that we could detect the developmental-stage-dependent SAX-3 cleavage in this transgene (Figure 3a). Then we carried out immunoprecipitation (IP) using anti-FLAG-coated beads on protein lysate collected from 44 cell and threefold stage embryos and submitted them for proteomics (Figure 3—figure supplement 1a). As the SAX-3 cleavage happens only in threefold but not in 44-cell stage embryos, we focused on membrane proteins that specifically interacted with SAX-3 in threefold stage embryos. Furthermore, Ig superfamily proteins have been shown to be able to interact with each other in many biological processes (Cameron and McAllister, 2018; Wai Wong et al., 2012), so we also focused our analysis on Ig superfamily proteins. syg-1, a nephrin family adhesion molecule (Shen and Bargmann, 2003), caught our attention (Figure 3—figure supplement 1b, Supplementary file 2) because it contains multiple Ig domains in its extracellular regions (Zallen et al., 1998; Shen and Bargmann, 2003). To confirm the interaction between SYG-1 and SAX-3, we first generated a transgene expressing SYG-1::HA in glia and FLAG::SAX-3 in neurons and then examined the interaction between SYG-1::HA and FLAG::SAX-3 by IP using anti-HA-coated beads. As shown in Figure 3b, the extracellular cleavage fragment but not the full-length SAX-3 can bind with SYG-1::HA, suggesting that SYG-1 can directly or indirectly bind with the SAX-3 extracellular cleavage fragment. With these findings, we further examined the function of syg-1 in AMsh migration and found that syg-1 was expressed in AMsh glia during migration and loss-of-function of syg-1(ok3640, null allele) caused similar AMsh migration defects as those in sax-3(lf) mutants (Figure 3c–e, Figure 3—figure supplement 2a). Furthermore, double mutants of syg-1;sax-3 displayed the same extent of defects as in sax-3 single mutants (Figure 3d and e), supporting the conclusion that syg-1 and sax-3 function in the same genetic pathway to regulate AMsh migration. As syg-1 has been shown to function together with syg-2, another Ig super family protein, in synapse formation and axon branching (Shen et al., 2004; Chia et al., 2014), we examined the function of syg-2 in AMsh migration and did not observe any abnormalities in syg-2(lf) mutants (Figure 3e). Consistent with its potential role as a receptor for the SAX-3 cleavage fragment, syg-1 cell autonomously regulates AMsh migration (Figure 3e, Figure 3—figure supplement 2b). Further genetic analysis also showed that the conserved WIRS (WAVE Regulatory receptor sequence) in the SYG-1 C-terminal (Chia et al., 2014) is essential for its function in AMsh migration (Figure 3e), and interruption of WAVE complex function by mutating gex-3, one of the key components of the WAVE complex (Chia et al., 2014), cell autonomously caused similar AMsh migration defects as observed in sax-3(lf) and syg-1(lf) mutants (Figure 3f,g). Finally, to further test the ligand/receptor function of the SAX-3 cleavage fragment and SYG-1, we tested whether AMsh glia can further migrate along the body in transgenic animals that express the cleavage product of SAX-3 in the tail neurons (Pitr-1b) or tail hypodermis (Plin-44) of sax-3(lf) animals (Ou et al., 2010). We found that in these transgenes the migration distance of AMsh was almost three times that of control animals, and loss-of-function in syg-1 completely blocked the over-migration phenotypes in these transgenes (Figure 3h and i). With this evidence we believe that the neuron-released SAX-3 extracellular cleavage fragment functions as an attractive cue, and SYG-1 functions as its receptor to mediate AMsh migration (Figure 4).

SYG-1 functions as a receptor for the SAX-3 extracellular cleavage fragment during AMsh glial migration.

(a) Image from western blot show the SAX-3 extracellular fragment by N-terminal tagging FLAG transgenes. Stars (*) label bands that are nonspecifically recognized by the anti-FLAG antibody. A schematic diagram at the top panel shows that the FLAG tag was fused at the N-terminal of SAX-3, and the extracellular cleavage fragment detected by the FLAG antibody is about 55 kDa. (b) SYG-1 specifically binds with the SAX-3 extracellular cleavage fragment. SYG-1::HA expression is driven by the pan-glial promoter Pmir-228, and FLAG::SAX-3 is expressed under the pan-neuronal promoter Punc-33. Stars (*) label bands that are nonspecifically recognized by the anti-FLAG antibody. (c) Confocal images of AMsh glia in control and syg-1(ok3640) animals. TB: pharyngeal terminal bulb. Stars (*) label AMsh cell bodies. Scale bar, 10 µm. (d) Quantification of Migration Index (MI) in control, syg-1, sax-3, and syg-1;sax-3 double mutants show that double mutants of syg-1;sax-3 display similar defects as in sax-3 single mutants. (e) syg-1 cell autonomously regulates AMsh glial migration, and the WAVE regulatory receptor sequence is required for syg-1 function. (f and g) gex-3 cell autonomously regulates AMsh glial migration. Quantification of Migration Index (f) and the percentage of animals with AMsh defects (g) show that gex-3 is required for AMsh migration. (h and i) Confocal images (h) and quantification (i) show that ectopic expression of a fragment mimicking the SAX-3 cleavage fragment in the tail neurons (Pitr-1) or hypodermis (Plin-44) causes syg-1-dependent over-migration of AMsh glia. Red arrowheads point to the center of the pharyngeal terminal bulb. Green arrowheads indicate AMsh cell body position. Scale bar, 10 µm. In d, f, and i, data are represented as mean ± SEM. One-way ANOVA test. **, p<0.01; Each point represents at least 30 worms. In (e and g), data are represented as mean ± SEM. Two-way ANOVA test. **, p<0.01; ns, no significant difference. Each point represents three experiments of at least 50 worms each.

Identification of SAX-3 interacting proteins by proteomics.

(a) A image from silver staining shows the full length (black stars) and cleaved fragment (red stars) of SAX-3. These samples were then submitted for proteomics. (b) A screenshot of the Scaffold software that shows that SYG-1 was identified with specific interactions with SYG-1 in threefold but not in 44-cell stage embryos. The green color of the asparagine(N) means that this amino acid was deamidated to either aspartic acid or isoaspartic acid.

SYG-1 is expressed in AMsh glia.

(a) Confocal images show that a 2.5 kb syg-1 promoter can drive GFP expression in AMsh (labeled by Pf53f4.13::H2B::mCherry) in embryos (left) and larva (right). Scale bar, 10 µm. (b) Expression of syg-1 in AMsh glia driven by PAMsh(2) (Pf16f9.3) can rescue syg-1(lf) phenotypes.

A model for regulation of AMsh glial migration by SAX-3 and SYG-1.Discussion

Studies in different organisms show that Robo functions as the receptor for the repulsive guidance cue Slit (Dickson and Gilestro, 2006), and loss-of-function in Robo causes many phenotypes that are shared by Slit mutants (Dickson and Gilestro, 2006). However, genetic studies also showed that some Robo-mutant phenotypes are not related to Slit (Zallen et al., 1998; Hao et al., 2001; Fricke et al., 2001; Hivert et al., 2002; Liu et al., 2004). In C. elegans, there is only one Robo, sax-3, and one Slit, slt-1. sax-3(lf) animals display many phenotypes that are not observed in slt-1 mutants, including embryonic lethality, abnormal head morphology (vab head), and misplacement of neurons (Zallen et al., 1998; Hao et al., 2001). Based on these observations, it has been long speculated that there are Slit-independent Robo regulatory mechanisms. Here we present evidence to show that neuron-expressed Robo/SAX-3 can be cleaved in the extracellular FN3 repeats during development, and the cleaved extracellular fragment can interact with glial-expressed SYG-1/Neph to mediate F-actin organization and to facilitate glial migration.

The extracellular region of Robo is evolutionarily conserved from nematodes to mammals and contains five Ig and three FN3 domains. The extracellular cleavages of Robo appear to be conserved in C. elegans (this study), Drosophila, and human (Coleman et al., 2010; Seki et al., 2010), though different metalloproteases were utilized in Drosophila and human. Recent biochemical and structural studies show that the extracellular domains of Robo can dimerize through the Ig domains, and the dimerization may be inhibited by the FN3 domain in the absence of Slit (Hivert et al., 2002; Liu et al., 2004; Zakrys et al., 2014; Aleksandrova et al., 2018; Barak et al., 2019; Yamamoto et al., 2019; Pak et al., 2020). The cleavages of the extracellular domains provide possibility for additional regulatory mechanisms of Robo. Since the extracellular SAX-3/Robo cleavage fragment lacks the inhibitory second FN3 domain, it will be expected to strongly bind with itself or with the full length SAX-3/Robo, and these interactions may play a regulatory role in the activation of Robo downstream signals. In this study we show that the SAX-3/Robo extracellular cleavage fragment can bind with the Ig domain-containing adhesion molecule SYG-1/Neph to regulate the WAVE complex. Since sax-3/Robo and syg-1/ Neph are both expressed in and functionally important for neurons and other cells, it is possible that interactions between the SAX-3 cleaved extracellular fragment and syg-1/Neph also play functions beyond glial migration. The potential dimerization of SAX-3/Robo extracellular cleavage fragments may also mediate crosstalk between syg-1/Neph and sax-3/Robo signals. The extracellular cleavage of SAX-3/Robo generated two fragments: one has all the Ig and the first FN3 domains while the other contains the rest of the FN3, transmembrane, and the intracellular domains. While the released SAX-3/Robo extracellular cleavage fragment functions as a ligand for SYG-1/Neph and has potential to interact with the full length SAX-3/Robo, the membrane-anchored fragment may also play an active role in neuronal development.

To achieve their vital roles in the nervous system and to form functional units with neurons, glia need to migrate over what are often long distances from their birthplace to the appropriate regions (Gilmour et al., 2002; Jarjour and Kennedy, 2004; Kinrade et al., 2001; Klämbt, 2009; Barres, 2008; Fields et al., 2015; Mori et al., 2005). The migration of glia relies on guidance cues and shares many similar mechanisms with neurons (Kinrade et al., 2001; Jarjour et al., 2003; Sasse and Klämbt, 2016; Spassky et al., 2002; Unni et al., 2012). Our studies uncover a surprising role of Robo receptors – an attractive cue for glial migration – and identify SYG-1/Neph as the glial receptor for the neuronal-released Robo cleavage fragment, suggesting that the function and regulation of guidance molecules are more complicated than anticipated.

Materials and methods

Further information and requests for resources should be directed to and will be fulfilled by the Lead Contact, Dong Yan (dong.yan@duke.edu).

Experimental model and subject details<italic>C. elegans</italic> genetics

C. elegans strains were maintained on nematode growth media (NGM) plates using Escherichia coli OP50 as a food source. Animals were grown according to standard methods (Brenner, 1974) at 20°C unless otherwise stated. Wild-type worms were of the Bristol N2 strain. Only hermaphrodite animals were used for experiments. All transgenes and strains are described in Supplementary file 1. yadIs48 (Pf53f4.13::GFP) was used to visualize AMsh cells. The null alleles used in genetic analyses were sax-3(yad10), slit-1 (ok255), syg-1(ok3640), syg-2(ky671), and gex-3(zu196) unless specific alleles were mentioned. gex-3(zu196) homozygous animals are sterile, and we analyzed the AMsh phenotypes in first generation homozygotes of gex-3(zu196), which may not represent null allele phenotypes due to maternal effects. All other data were acquired in animals that have been homozygous for genotypes of interest for at least three generations. The recessive sax-3 alleles yad10 and yad147 were isolated from a visualized EMS mutagenesis screen of over 4000 haploid genomes and were two of five mutants with reduced migration defects. Genetic studies show that those five mutants belong to three complementation groups, and yad10 and yad147 are in one complementation group. Based on their X-linkage and the vab head phenotypes, we tested whether they were alleles of sax-3 and found that they both failed to complement sax-3(ky123) null alleles. yad10 and yad147 mutations were finally confirmed through sequencing and rescue experiments. The ‘vab’ head phenotype in sax-3 mutants was defined as any abnormality of head morphology.

Method detailsCloning and constructs

All DNA expression constructs were generated using Gateway cloning technology (Invitrogen, Carlsbad, CA) and subsequently sequenced. sax-3, syg-1, and gex-3 cDNA were amplified from C. elegans yk cDNA clones gifted from Dr. Yuji Kohara’s lab. Pf53f4.13 (PAMsh), Pf16f9.3 (PAMsh 2), Punc-33 (Pneuron), and Prgef-1(Pneuron) promoters were used to express genes of interest in AMsh and all neurons. Parl-13 was used to drive gene expression in amphid neurons. In embryos and other stages, f53f4.13 and f16f9.3 are exclusively expressed in AMsh glia, arl-13 is exclusively expressed in amphid neurons, and unc-33 and rgef-1 are broadly expressed in neurons (Packer et al., 2019). A complete list of DNA constructs used is included in Supplementary file 1. In general, plasmids used in this study were injected at a concentration of 1–50 ng/µL with a Pttx-3::RFP co-injection marker injected at a concentration of 50 ng/µL.

Microscopy

Representative images were acquired with a Zeiss LSM700 confocal microscope using a Plan-Apochromat 40×/1.4 objective. Worms were immobilized using 1.5% 1-phenoxy-2-propanol (TCI America, Portland, OR) in M9 buffer and mounted on 5% agar slides. 3D reconstructions were done using Zeiss Zen software. A Zeiss Axio Imager two microscope equipped with Chroma HQ filters was used to score AMsh migration defects and take images to measure Migration Index. Experiments were conducted in triplicates consisting of at least 50 day one adult animals each.

Protein analysis

All protein analyses were carried out using transgenes made in this study (see details in Supplementary file 1). Synchronous animals were grown on NGM plates to reach adult stages, and animals were bleached to collect embryos. The freshly isolated embryos were mostly between the 16-cell and 44-cell stages. Embryos were incubated in M9 buffer and collected at different embryonic stages or at the L1 stage. L1 animals were then cultured in NGM plates with O.P. 50 as a food source until they reached the L4 stage, where they were subsequently collected for protein analysis. For western blotting, protein was directly extracted in SDS sample buffer containing 1 mM DTT by freeze-thawing for 20–50 cycles between dry ice/ethanol and a 37°C water bath, and then denatured by heating to 95°C for 5 min. Blots were probed with anti-HA (Sigma, H9658; abcam, ab9110), anti-GFP (Sigma, G1544), and anti-FLAG (Sigma, F3165) antibodies and visualized with Amersham HRP-conjugated anti-rabbit or anti-mouse secondary antibodies at a 1:5000 dilution (GE Healthcare Life Sciences) using the SuperSignal West Femto kit (Pierce, Rockford, IL). For IP experiments, embryos were collected at the 44-cell and 1.5- to 3-fold stages and subsequently lysed with IP lysis buffer (25 mM Tris-HCl pH 7.4, 150 mM NaCl, 1% NP-40, 1 mM EDTA, 5% glycerol) containing protease inhibitor cocktail (Thermo Fisher, 87785, added before use) and sonicated by an ultrasonic cell crusher. Lysates were centrifuged for 10 min at 12,000 rpm at 4°C and supernatant was used for Co-IP using the anti-Protein A/G magnetic beads (Thermo Fisher, 78608) or anti-HA Magnetic beads (Thermo Fisher, 88836) separately according the manufacturer’s protocol. Briefly, beads were equilibrated before use and added to the supernatant and mixed gently. The mixture was incubated on a rotator at 4°C overnight. Beads were collected with a magnetic stand and washed twice with phosphate buffered saline. Supernatant was discarded and proteins were eluted from the beads using SDS-sample buffer at 95°C for 10 min.

Collection of supernatant for western blotRT-PCR

To measure sax-3 mRNA levels in N2, yad10, and yad147 animals, total RNA were isolated from mixed-stage animals. Briefly, worms were collected in the tube and washed with M9 buffer on an end-over-end rotator three times for 90 min in total. RNA was then isolated by using TRI Reagent (Invitrogen). Total RNA (1 μg) of each genotype was reverse transcribed using an iScript Reverse Transcription Supermix (Bio-Rad Laboratories). Relative sax-3 mRNA levels were determined by PCR with 1 μL cDNAs as templates, and act-1/actin was used as internal control. Primers for sax-3, Forward: 5ʹ CTGTTTGATTGTCGTGTGACTGG-3ʹ; Reverse: 5ʹ-AAAGCTGCCTTGCTCAACG-3ʹ. Primers for act-1, Forward: 5ʹ-TTGCCGCTCTTGTTGTAGAC-3ʹ; Reverse: 5ʹ-TTCGTAGATTGGGACGGTG-3ʹ. The PCR products were run in the agarose gel, and the density was calculated using Image J software.

CRISPR-Cas9-based homology recombination approach

The sax-3 FN3-b deletion allele, yad175, was constructed by using CRISPR-Cas9-based homology recombination. Briefly, two arms containing about 1000 bp on both sides of FN3-b of sax-3 were amplified from genomic DNA of wild-type animals, and the fusion fragment of the two arms was then cloned into the PCR8 vector. Two sgRNAs of sax-3 FN3-b deletion were cloned into pDD122 plasmid (Addgene, #47550). Finally, the recombination template and the two sgRNAs were injected into young adult of yadIs48 worms simultaneously. yad175 were isolated from the F2 generation, and the deletion was confirmed by sequencing. sax-3 sgRNA1: GGATGGAGAGTCAACATG; sequence of sax-3 sgRNA2: GGATGTGCGAATCCGTAT.

Mass spectrometric analysis

To identify the potential proteins binding with SAX-3, embryos from FLAG::SAX-3::GFP expressing worms were collected at the 44-cell and 1.5- to 3-fold stages. In the same experiment, 1.5- to 3-fold stage embryos of FLAG::INX-3::GFP transgenes were also collected as a control to further preclude unspecific proteins. Next, embryos were subsequently lysed with IP lysis buffer (25 mM Tris-HCl pH 7.4, 150 mM NaCl, 1% NP-40, 1 mM EDTA, 5% glycerol) containing protease inhibitor cocktail (Thermo Fisher, 87785, added before use) and sonicated by an ultrasonic cell crusher. Lysates were centrifuged for 10 min at 12,000 rpm at 4°C and supernatant was used for Co-IP to pull down all the proteins that bind to SAX-3 by using the anti-FLAG magnetic agarose beads (Thermo Fisher, A36798) according to the manufacturer’s protocol.

IP products were sent to the Duke Center for Genomic and Computational Biology for mass spectrum analysis. Briefly, these in-solution samples were brought to 4% SDS and then subjected to S-trap (Protifi) trypsin digestion and sample cleanup using manufacturer recommended protocols. Digested peptides were lyophilized to dryness and resuspended in 15 µL of 0.2% formic acid/2% acetonitrile. Each sample was subjected to chromatographic separation on a Waters NanoAquity UPLC equipped with a 1.7 µm HSS T3 C18 75 µm I.D. × 250 mm reversed-phase column (NanoFlow data). The mobile phase consisted of (A) 0.1% formic acid in water and (B) 0.1% formic acid in acetonitrile. Of this, 3 µL was injected and peptides were trapped for 3 min on a 5 µm Symmetry C18 180 µm I.D. × 20 mm column at 5 µL/min in 99.9% A. The analytical column was then switched in-line and a linear elution gradient of 5% B to 40% B was performed over 30 min at 400 nL/min for in-gel band analysis and over 90 min at 400 nL/min for Co-IP studies. The analytical column was connected to a Fusion Lumos mass spectrometer (Thermo) through an electrospray interface operating in a data-dependent mode of acquisition. The instrument was set to acquire a precursor MS scan from m/z 375–1500 at R = 120,000 (target AGC 2e5, max IT 50 ms) with MS/MS spectra acquired in the ion trap (target AGC 5e3, max IT 100 ms). For all experiments, HCD energy settings were 30 v and a 20 s dynamic exclusion was employed for previously fragmented precursor ions.

Raw LC-MS/MS data files were processed in Proteome Discoverer (Thermo Scientific) and then submitted to independent Mascot searches (Matrix Science) against a C. elegans protein database containing both forward (4043 entries) and reverse entries of each protein. Search tolerances were five ppm for precursor ions and 0.8 Da for product ions using trypsin specificity with up to two missed cleavages. Carbamidomethylation (+57.0214 Da on C) was set as a fixed modification, whereas oxidation (+15.9949 Da on M) and deamidation (+0.98 Da on QN) were considered dynamic mass modifications. All searched spectra were imported into Scaffold (v4.4, Proteome Software) and scoring thresholds were set to achieve a peptide false discovery rate of 1% using the PeptideProphet algorithm.

Quantification and statistical analysesStatistical analysis

Data were analyzed using Student’s t-test and one-way or two-way ANOVA followed by Tukey’s post-hoc test in Graphpad Prism (Graphpad Software, La Jolla, CA). Statistical details are included in the figure legends. In general, a p-value cutoff of 0.05 was considered statistically significant. The data are represented by mean ± SEM. For phenotype penetrance experiments, three biological replicates of at least 50 animals each were used unless otherwise stated.

Acknowledgements

We thank Dr. Erik Soderblom, Tricia C Ho and Dr. Greg Waitt at Duke Proteomics and Metabolomics Core Facility for helping on sample preparation, mass spectrum analyses, and data interpretation, and Dr. Yuji Kohara for syg-1, sax-3, and gex-3 cDNAs. Some strains used in this study were provided by the Caenorhabditis Genetics Center (CGC), which is funded by the NIH Office of Research Infrastructure Programs (P40 OD010440). Ms. Chia-hui Chen characterized some preliminary observations of this study. This project was supported by the Holland Trice Awards. QZ, AZ, and DY were supported by the NIH R01 (NS094171 and NS105638 to DY).

Additional informationCompeting interests

No competing interests declared

Author contributions

Data curation, Formal analysis, Validation, Investigation, Methodology, Writing - original draft, Writing - review and editing

Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing

Conceptualization, Supervision, Funding acquisition, Validation, Writing - original draft, Writing - review and editing

Additional filesStrains and plasmids used in this study.A list of protein identified by proteomics show specific interactions with SAX-3 in 1.5- to 3-fold stage embryos.Data availability

All data generated or analysed during this study are included in the manuscript and supporting files.

ReferencesAleksandrova NGutsche IKandiah EAvilov SVPetoukhov MVSeiradake EMcCarthy AA2018Robo1 forms a compact Dimer-of-Dimers assemblyStructure2632032810.1016/j.str.2017.12.00329307485Bacaj TTevlin MLu YShaham S2008Glia are essential for sensory organ function in C. elegansScience32274474710.1126/science.116307418974354Barak RYom-Tov GGuez-Haddad JGasri-Plotnitsky LMaimon RCohen-Berkman MMcCarthy AAPerlson EHenis-Korenblit SIsupov MNOpatowsky Y2019Structural principles in robo activation and Auto-inhibitionCell17727228510.1016/j.cell.2019.02.00430853216Barres BA2008The mystery and magic of Glia: a perspective on their roles in health and diseaseNeuron6043044010.1016/j.neuron.2008.10.01318995817Branchfield KNantie LVerheyden JMSui PWienhold MDSun X2016Pulmonary neuroendocrine cells function as airway sensors to control lung immune responseScience35170771010.1126/science.aad796926743624Brennaman LHMoss MLManess PF2014EphrinA/EphA-induced ectodomain shedding of neural cell adhesion molecule regulates growth cone repulsion through ADAM10 metalloproteaseJournal of Neurochemistry12826727910.1111/jnc.1246824117969Brenner S1974The genetics of Caenorhabditis elegansGenetics7771944366476Cameron SMcAllister AK2018Immunoglobulin-Like receptors and their impact on wiring of brain synapsesAnnual Review of Genetics5256759010.1146/annurev-genet-120417-03151330212237Chia PHChen BLi PRosen MKShen K2014Local F-actin network links synapse formation and axon branchingCell15620822010.1016/j.cell.2013.12.00924439377Coleman HALabrador JPChance RKBashaw GJ2010The adam family metalloprotease kuzbanian regulates the cleavage of the roundabout receptor to control axon repulsion at the midlineDevelopment1372417242610.1242/dev.04799320570941Dickson BJGilestro GF2006Regulation of commissural axon pathfinding by slit and its robo receptorsAnnual Review of Cell and Developmental Biology2265167510.1146/annurev.cellbio.21.090704.15123417029581Fields RDWoo DHBasser PJ2015Glial Regulation of the Neuronal Connectome through Local and Long-Distant CommunicationNeuron8637438610.1016/j.neuron.2015.01.014Fricke CLee JSGeiger-Rudolph SBonhoeffer FChien CB2001Astray, a zebrafish roundabout homolog required for retinal axon guidanceScience29250751010.1126/science.105949611313496Ghenea SBoudreau JRLague NPChin-Sang ID2005The VAB-1 eph receptor tyrosine kinase and SAX-3/Robo neuronal receptors function together during C. elegans embryonic morphogenesisDevelopment1323679369010.1242/dev.0194716033794Gilmour DTMaischein HMNüsslein-Volhard C2002Migration and function of a glial subtype in the vertebrate peripheral nervous systemNeuron3457758810.1016/S0896-6273(02)00683-912062041Grieshammer UMa LPlump ASWang FTessier-Lavigne MMartin GR2004SLIT2-mediated ROBO2 signaling restricts kidney induction to a single siteDevelopmental Cell670971710.1016/S1534-5807(04)00108-X15130495Gupta BPSternberg PW2003The draft genome sequence of the nematode Caenorhabditis briggsaeGenome Biology423810.1186/gb-2003-4-12-238Hao JCYu TWFujisawa KCulotti JGGengyo-Ando KMitani SMoulder GBarstead RTessier-Lavigne MBargmann CI2001C. elegans slit acts in Midline, dorsal-ventral, and anterior-posterior guidance via the SAX-3/Robo receptorNeuron32253810.1016/S0896-6273(01)00448-211604136Heiman MGShaham S2009DEX-1 and DYF-7 establish sensory dendrite length by anchoring dendritic tips during cell migrationCell13734435510.1016/j.cell.2009.01.05719344940Hinck L2004The versatile roles of "axon guidance" cues in tissue morphogenesisDevelopmental Cell778379310.1016/j.devcel.2004.11.00215572123Hivert BLiu ZChuang CYDoherty PSundaresan V2002Robo1 and Robo2 are homophilic binding molecules that promote axonal growthMolecular and Cellular Neuroscience2153454510.1006/mcne.2002.119312504588Jarjour AAManitt CMoore SWThompson KMYuh SJKennedy TE2003Netrin-1 is a chemorepellent for oligodendrocyte precursor cells in the embryonic spinal cordThe Journal of Neuroscience233735374410.1523/JNEUROSCI.23-09-03735.200312736344Jarjour AAKennedy TE2004Oligodendrocyte precursors on the move: mechanisms directing migrationThe Neuroscientist109910510.1177/107385840326075115070484Killeen MTSybingco SS2008Netrin, slit and wnt receptors allow axons to choose the Axis of migrationDevelopmental Biology32314315110.1016/j.ydbio.2008.08.02718801355Kinrade EFBrates TTear GHidalgo A2001Roundabout signalling, cell contact and trophic support confine longitudinal Glia and axons in the Drosophila CNSDevelopment12820721611124116Klämbt C2009Modes and regulation of glial migration in vertebrates and invertebratesNature Reviews Neuroscience1076977910.1038/nrn272019773781Liu ZPatel KSchmidt HAndrews WPini ASundaresan V2004Extracellular ig domains 1 and 2 of robo are important for ligand (Slit) bindingMolecular and Cellular Neuroscience2623224010.1016/j.mcn.2004.01.00215207848Mommersteeg MTAndrews WDYpsilanti ARZelina PYeh MLNorden JKispert AChédotal AChristoffels VMParnavelas JG2013Slit-roundabout signaling regulates the development of the cardiac systemic venous return and pericardiumCirculation Research11246547510.1161/CIRCRESAHA.112.27742623255421Mori TBuffo AGötz M2005The novel roles of glial cells revisited: the contribution of radial Glia and astrocytes to neurogenesisCurrent Topics in Developmental Biology69679910.1016/S0070-2153(05)69004-716243597Neuhaus-Follini ABashaw GJ2015The intracellular domain of the frazzled/DCC receptor is a transcription factor required for commissural axon guidanceNeuron8775176310.1016/j.neuron.2015.08.00626291159Oikonomou GShaham S2011The Glia of Caenorhabditis elegansGlia591253126310.1002/glia.2108421732423Ou CYPoon VYMaeder CIWatanabe SLehrman EKFu AKPark MFu WYJorgensen EMIp NYShen K2010Two Cyclin-Dependent kinase pathways are essential for polarized trafficking of presynaptic componentsCell14184685810.1016/j.cell.2010.04.01120510931Packer JSZhu QHuynh CSivaramakrishnan PPreston EDueck HStefanik DTan KTrapnell CKim JWaterston RHMurray JI2019A lineage-resolved molecular atlas of C. elegans embryogenesis at single-cell resolutionScience365eaax197110.1126/science.aax197131488706Pak JSDeLoughery ZJWang JAcharya NPark YJaworski AÖzkan E2020NELL2-Robo3 complex structure reveals mechanisms of receptor activation for axon guidanceNature Communications11148910.1038/s41467-020-15211-132198364Perkins LAHedgecock EMThomson JNCulotti JG1986Mutant sensory cilia in the nematode Caenorhabditis elegansDevelopmental Biology11745648710.1016/0012-1606(86)90314-32428682Sasakura HInada HKuhara AFusaoka ETakemoto DTakeuchi KMori I2005Maintenance of neuronal positions in organized ganglia by SAX-7, a Caenorhabditis elegans homologue of L1The EMBO Journal241477148810.1038/sj.emboj.7600621Sasse SKlämbt C2016Repulsive epithelial cues direct glial migration along the nerveDevelopmental Cell3969670710.1016/j.devcel.2016.11.01627997826Seeger MTear GFerres-Marco DGoodman CS1993Mutations affecting growth cone guidance in Drosophila: genes necessary for guidance toward or away from the midlineNeuron1040942610.1016/0896-6273(93)90330-T8461134Seki MWatanabe AEnomoto SKawamura TIto HKodama THamakubo TAburatani H2010Human ROBO1 is cleaved by metalloproteinases and gamma-secretase and migrates to the nucleus in Cancer cellsFEBS Letters5842909291510.1016/j.febslet.2010.05.00920471383Shaham S2010Chemosensory organs as models of neuronal synapsesNature Reviews Neuroscience1121221710.1038/nrn274020029439Shaham S2015Glial Development and Function in the Nervous System of Caenorhabditis elegansCold Spring Harbor Perspectives in Biology7a02057810.1101/cshperspect.a020578Shen KFetter RDBargmann CI2004Synaptic specificity is generated by the synaptic guidepost protein SYG-2 and its receptor, SYG-1Cell11686988110.1016/S0092-8674(04)00251-X15035988Shen KBargmann CI2003The immunoglobulin superfamily protein SYG-1 determines the location of specific synapses in C. elegansCell11261963010.1016/S0092-8674(03)00113-212628183Singhvi ALiu BFriedman CJFong JLu YHuang X-YShaham S2016A Glial K/Cl Transporter Controls Neuronal Receptive Ending Shape by Chloride Inhibition of an rGCCell16593694810.1016/j.cell.2016.03.026Spassky Nde Castro FLe Bras BHeydon KQuéraud-LeSaux FBloch-Gallego EChédotal AZalc BThomas JL2002Directional guidance of oligodendroglial migration by class 3 semaphorins and netrin-1The Journal of Neuroscience225992600410.1523/JNEUROSCI.22-14-05992.200212122061Unni DKPiper MMoldrich RXGobius ILiu SFothergill TDonahoo A-LSBaisden JMCooper HMRichards LJ2012Multiple slits regulate the development of midline glial populations and the corpus callosumDevelopmental Biology365364910.1016/j.ydbio.2012.02.004Viveiros RHutter HMoerman DG2011Membrane extensions are associated with proper anterior migration of muscle cells during Caenorhabditis elegans embryogenesisDevelopmental Biology35818920010.1016/j.ydbio.2011.07.02621820426Wai Wong CDye DECoombe DR2012The role of immunoglobulin superfamily cell adhesion molecules in Cancer metastasisInternational Journal of Cell Biology20121910.1155/2012/340296Wang LJZhao YHan BMa YGZhang JYang DMMao JWTang FTLi WDYang YWang RGeng JG2008Targeting Slit-Roundabout signaling inhibits tumor angiogenesis in chemical-induced squamous cell carcinogenesisCancer Science9951051710.1111/j.1349-7006.2007.00721.x18201275Ward SThomson NWhite JGBrenner S1975Electron microscopical reconstruction of the anterior sensory anatomy of the nematode Caenorhabditis elegans.?2uuThe Journal of Comparative Neurology16031333710.1002/cne.9016003051112927Wu JYFeng LPark HTHavlioglu NWen LTang HBacon KBJiang ZHZhang XCRao Y2001The neuronal repellent slit inhibits leukocyte chemotaxis induced by chemotactic factorsNature41094895210.1038/3507361611309622Yamamoto NKashiwagi MIshihara MKojima TMaturana ADKuroda SNiimi T2019Robo2 contains a cryptic binding site for neural EGFL-like (NELL) protein 1/2Journal of Biological Chemistry2944693470310.1074/jbc.RA118.00581930700556Zakrys LWard RJPediani JDGodin AGGraham GJMilligan G2014Roundabout 1 exists predominantly as a basal dimeric complex and this is unaffected by binding of the ligand Slit2Biochemical Journal461617310.1042/BJ2014019024673457Zallen JAYi BABargmann CI1998The conserved immunoglobulin superfamily member SAX-3/Robo directs multiple aspects of axon guidance in C. elegansCell9221722710.1016/S0092-8674(00)80916-29458046Zallen JAKirch SABargmann CI1999Genes required for axon pathfinding and extension in the C-elegans nerve ringDevelopment1263679369210409513Zhang ANoma KYan D2020aRegulation of gliogenesis by lin-32/Atoh1 in Caenorhabditis elegansG3: Genes, Genomes, Genetics1040154710.1534/g3.120.401547Zhang AAckley BDYan D2020bVitamin B12 regulates glial migration and synapse formation through Isoform-Specific control of PTP-3/LAR PRTP expressionCell Reports303981398810.1016/j.celrep.2020.02.1133220946110.7554/eLife.57921.sa1Decision letterShenKangReviewing EditorHoward Hughes Medical Institute, Stanford UniversityUnited StatesShenKang ReviewerHoward Hughes Medical Institute, Stanford UniversityUnited States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

Your manuscript identified a new function of the SAX-3/ROBO guidance receptor in regulating glial cell migration. The finding that the ROBO ectodomain is secreted and interacts with a Ig adhesion receptor to guide glial cell migration broadens our understanding of receptor-ligand interactions in developmental neurobiology.

Decision letter after peer review:

Thank you for submitting your article "Robo functions as an attractive cue for glial migration through SYG-1/Neph" for consideration by eLife. Your article has been reviewed by three peer reviewers, including Kang Shen as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by Jonathan Cooper as the Senior Editor.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

As the editors have judged that your manuscript is of interest, but as described below that additional experiments are required before it is published, we would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). First, because many researchers have temporarily lost access to the labs, we will give authors as much time as they need to submit revised manuscripts. We are also offering, if you choose, to post the manuscript to bioRxiv (if it is not already there) along with this decision letter and a formal designation that the manuscript is "in revision at eLife". Please let us know if you would like to pursue this option. (If your work is more suitable for medRxiv, you will need to post the preprint yourself, as the mechanisms for us to do so are still in development.)

Summary:

All three reviewers found that this work is interesting for several reasons. For example, one reviewers wrote "Dong and colleagues described a slit independent function of the well-known axon guidance receptor ROBO/SAX-3 in the migration of glial cells in C. elegans. They found that both ROBO and an IgSF protein SYG-1 were required for the glial cell migration. Through structure-function analyses, they found the extracellular domain of ROBO was sufficient to rescue the phenotype when overexpressed and they proposed that the extracellular domain of ROBO functions as a secreted ligand to attract SYG-1 expressing glial cell migration. The novelty of this manuscript lies in two folds. First, the development of glial cells are poorly understood. Second that this is the first paper to assign a secreted function of ROBO through interaction with SYG-1. The majority of the genetic data supports each other and is consistent with the model they proposed."

Essential revisions:

However, all three reviewers found that there are important experiments that need to be performed in order to fully support the model. They mostly fall into three areas. First, the use of mis-expression constructs using exogenous promoters is potentially risky in drawing conclusions about cellular locus of SAX-3 in this case. Second, inclusion of cleaner genetic experiments such as a knockin will strengthen the model. Three, there are several concerns about the masspec and biochemistry data presented. The detailed comments are as following:

1) Cell autonomy

Misexpression of SAX-3 and SYG-1 high-copy transgenes under control of Punc-33 for neurons and Pf53f4.13 for the amphid sheath cell is used to infer cell autonomy. There are three issues with this – first, these promoters are not specific, in particular Punc-33 is expressed prominently in the amphid socket which could also affect amphid sheath migration (Altun-Gultekin et al. Development 2001). Pitr-1 used for the over-migration experiment is also expressed in the amphid socket. Second, promoters can be expressed differently in embryos than larvae, so images of the expression pattern of Pf53f4.13 in 1.5 fold embryos in particular should be provided. Third, it is risky to base such a major claim on misexpression experiments especially since promoters can be expressed transiently in unexpected places, so these experiments should really include mosaic analysis as well. Alternatively, the authors can use additional sets of promoters to strengthen the arguments. For example, an amphid neuron promoter might be able to tested, as well as neurons that do not associate with the glial cells but still has a process in the nerve ring such as PVQ or AVA. There are promoters available for those experiments.

Figure 3H and I and crucial for the model. These are quite compelling results but I am not entirely sure about the interpretation of these results. The DA9 expression of itr-1 promoter does not turn on consistently until L3-L4. Do the sheath cells continue to grow in the sax-3 single mutants expressing the transgene? To mimic more endogenous situation, the authors should use the lin-44 promoter which expresses embryonically.

2) Overexpression issues.

All rescue experiments are based on overexpression. At least one knockin should be done to show that the FN3(2) is required. For example, if FN3(2) is deleted using crispr, they can try to test if the it is required for slit function and for this new function of sax-3.

3) Binding to SYG-1

The syg-1 phenotype is convincing but the SAX-3 binding data is less convincing. First, the list of hits from the mass spec screen is not shown. How many other hits were there, what were they, and how did the representation of syg-1 compare? Second, there are no specificity controls (for the antibody or for the SAX-3-SYG-1 binding itself) in the IP experiments. This is especially important because the SAX-3 fragment is running at the wrong size in this experiment compared to all other gels in the paper.

Related to this issue, A. The mass spectrometric analysis is not explained. Presumably, the authors identified hundreds of peptides, many of which might have been non-specific – a common issue with such analysis. The fact that only two peptides covering 5% of SYG-1 were identified by mass spec (buried in supplement and not mentioned in text) implies that the confidence level for this identification was likely not high. This is generally fine; the authors followed up this finding with additional experiments. But a proper report of total number of peptides identified, cutoffs, and how SYG-1 stands out from the list of peptides is warranted. I suggest authors contact their mass spec facility for details and statistical analysis.

Additionally, the supplement shows one Asn labeled green in the second peptide by the mass spec analysis software. Does that imply a chemical modification/artifact that affected the mass of the peptide? The supplement, which is essential, could be explained better. Currently, it is a screen snapshot from an unknown piece of software that is not mentioned or cited.

B) The authors should report exact boundaries of their deletion constructs (ΔC, ΔC+TM, Ig(1-5)FN3(1), etc.) presented in Figures 1, 2 and 3, especially since they would be relevant for judging whether their manipulations (domain deletions or inclusions) could inadvertently cause misfolding of proteins, or where exactly the cleavage might be happening. It would help narrow down a binding site on the sequence for the proposed SYG-1 interaction. In addition, we'd know if they removed the long linker sequences shown to be important between some of the IG and FN domains (Barak et al., 2019; Aleksandrova et al., 2018; Pak et al., 2020), when they removed the domains. It would be of interest to see if proteolysis reported by Qu et al. matches that observed by Pak et al., 2020, between the first and second FN domains of human Robo3. On that remark, I cannot find how they draw their schematics for Figure 4 degradation product (containing two FN domains).

C) Authors present glial migration phenotypes for a set of truncations and deletions of SAX-3. The data provided is self-consistent with the biochemistry they present (which is sparse but enough for an initial study of this kind). However, they do not present data that shows that these deletions (such as IG(1-5) or FN3(1) in Figure 2, which did not rescue) are successfully expressed, make it to through the secretory machinery and are presented on/secreted from neurons. One exception is the FN3(2)Δ constructs, where they did show protein production, but not necessarily successful display on cells. These are useful controls that actually allow them to support their conclusions. It would also be preferable to see if the expression levels of these deletions are comparable to wild-type, so that the observed phenotypes are not due to underexpression of or overproduction of SAX-3. Due to the current circumstances with the coronavirus pandemic, the authors may be given a dispensation from completing these experiments, as long as they clarify the possibility that some of these constructs might not be produced and secreted/displayed in vivo, and hence the lack of rescue.

D) I do not understand the comment about FN domains being mostly structural, while IG domains are functional. If this is a general comment about IG/FN proteins such as Robo, one of the most prominent IG/FN receptors DCC/UNC-40 binds its Netrin ligand with the FN domains, not IG. If this is a comment about the FN domains of Robos not being "functional", that's also not appropriate: While Slits bind the Robo IG domain 1, the other known extracellular ligand of Robos, the NELL proteins, bind the first FN domain of vertebrate Robos (Pak et al., 2020 and Yamamoto et al., 2019), the same domain identified by the authors for SYG-1 binding, and NELLs don't touch the IG domains.

In fact, the authors may want to discuss the relevance of the FN domains for Robo function, which was recently resolved in a burst of publications (the Pak et al., Yamamoto, et al., Barak et al. and Aleksandrova et al. manuscripts). The cleavage described here might actually act to open Robo receptors for binding other ligands, such as NELLs (not present in C. elegans) or SYG-1.

E).The authors do not show that the interaction is direct at any point in the manuscript. The pull-downs can simply be the result of a third protein mediating this interaction; this is not an uncommon occurrence. This does not make the paper less interesting, but the possibility has to be clearly mentioned across the manuscript.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Robo functions as an attractive cue for glial migration through SYG-1/Neph" for further consideration by eLife. Your revised article has been evaluated by Jonathan Cooper (Senior Editor) and Kang Shen as a Reviewing Editor.

The manuscript has been improved and but there are some remaining issues that need to be addressed before acceptance, as outlined below: Can you address these points by providing more information and clarification in your writing when providing the revision.

Reviewer #2:

1) Cell autonomy experiments still depend exclusively on promoters whose expression and timing in the embryo, relative to AMsh migration, have not been determined. The images of AMsh-specific promoters (Figure 1—figure supplement 1G) appear to show different cells, one much more anterior than the other. Mosaic analysis was not performed.

2) In the mass spectrometry experiments, it appears that SYG-1 was one of the weakest hits detected, ranking below 1000 in the overall list. Even with aggressive filtering, SYG-1 is among a few dozen hits and appears at the level of background noise, consistent with the presence of a "DECOY" hit with identical results. This is not the impression given in the manuscript.

3) The relative position of the nerve ring and the AMsh in Figure —figure supplement 1D strongly suggests that the AMsh is tracking with the displaced nerve ring, contrary to the authors' interpretation. Due to COVID, the authors could not obtain videos of glial migration.

4) yad10 is expected to encode a protein similar to the Ig(1-5)FN(1-3) fragment used for rescue. One possibility was that the mRNA is not expressed but the authors now show that is not the case. The yad10 phenotype contradicts the overexpression results in which the N-terminal fragment of SAX-3 is sufficient for rescue.

Reviewer #3:

The authors responded to comments satisfactorily as much as I can judge. My geneticist colleagues should review the validity of the new experiments with the new promoters and the CRISPR deletion of the second FN3 domain.

I am not sure the statement "Recent biochemical and structural studies show that the extracellular domains of Robo can dimerize through the fourth Ig domain, and this dimerization is inhibited by the second FN3 domain in the absent of Slit" is strictly true. It may be safest to define this as "FN3 domains", since both are involved FN3 domains are involved in a hairpin-like structure. Also, there is plenty of evidence that Robo1 is a dimer even when the FN3 domains are there, so I would say "the dimerization may be inhibited" to allow for multiple possibilities. Also see the typo, "in the absent of".

Also, the last statement appears to be false. NELLs bind the first FN3 domain, not second, so they would not be binding the membrane-anchored fragment. NELLs probably should not be mentioned here at the end of the text at all. My initial comment was that the first FN3(1) domain is not a non-functional/only structural domain as implied by the original draft. NELLs do bind the first FN3 domain, and can turn on a repulsive Robo response.

I should also remind the authors that they still use inconsistent naming of the Fibronectin domains. Figure 1—figure supplement 1 has FN3-a and FN1 for the same domains in panels F and H. Or there is FN1 in Figure 3I, but refers to this domain as FN3-a in Figure 2F. Minor issue, but no reason to confuse readers.

10.7554/eLife.57921.sa2Author response

Essential revisions:

However, all three reviewers found that there are important experiments that need to be performed in order to fully support the model. They mostly fall into three areas. First, the use of mis-expression constructs using exogenous promoters is potentially risky in drawing conclusions about cellular locus of SAX-3 in this case. Second, inclusion of cleaner genetic experiments such as a knockin will strengthen the model. Three, there are several concerns about the masspec and biochemistry data presented. The detailed comments are as following:

1) Cell autonomy

Misexpression of SAX-3 and SYG-1 high-copy transgenes under control of Punc-33 for neurons and Pf53f4.13 for the amphid sheath cell is used to infer cell autonomy. There are three issues with this – first, these promoters are not specific, in particular Punc-33 is expressed prominently in the amphid socket which could also affect amphid sheath migration (Altun-Gultekin et al. Development 2001). Pitr-1 used for the over-migration experiment is also expressed in the amphid socket. Second, promoters can be expressed differently in embryos than larvae, so images of the expression pattern of Pf53f4.13 in 1.5 fold embryos in particular should be provided. Third, it is risky to base such a major claim on misexpression experiments especially since promoters can be expressed transiently in unexpected places, so these experiments should really include mosaic analysis as well. Alternatively, the authors can use additional sets of promoters to strengthen the arguments. For example, an amphid neuron promoter might be able to tested, as well as neurons that do not associate with the glial cells but still has a process in the nerve ring such as PVQ or AVA. There are promoters available for those experiments.

Figure 3H and I and crucial for the model. These are quite compelling results but I am not entirely sure about the interpretation of these results. The DA9 expression of itr-1 promoter does not turn on consistently until L3-L4. Do the sheath cells continue to grow in the sax-3 single mutants expressing the transgene? To mimic more endogenous situation, the authors should use the lin-44 promoter which expresses embryonically.

In the revised manuscript we repeated the key rescue experiments using a well established Pneuronal promoter, Prgef-1, an amphid neuron specific promoter, Parl-13, and another AMsh specific promoter Pf16f9.3, and all results supported our original conclusions that sax-3 and syg-1 function in neurons and AMsh, respectively, to regulate AMsh migration (Figure 1D, Figure 2—figure supplement 1D, and Figure 3—figure supplement 1B). We also provided new data to show that both Pf53f4.13 (PAMsh) and Pf16f9.3 (PAMsh(2)) can drive expression in AMsh glia of 1.5-fold stage embryos (Figure 1—figure supplement 1G). To further confirm that ectopic expression of the SAX-3 extracellular cleavage fragment can attract AMsh migration in a syg-1 dependent manner, we expressed SAX-3(Ig1-5-FN3a) under the Plin-44 promoter, as suggested by the reviewer, and found that AMsh glial displayed the same syg-1 dependent AMsh migration defects as that in Pitr-1 transgenes (Figure 3H and 3I).

2) Overexpression issues.

All rescue experiments are based on overexpression. At least one knockin should be done to show that the FN3(2) is required. For example, if FN3(2) is deleted using crispr, they can try to test if the it is required for slit function and for this new function of sax-3.

In the revised manuscript, we generated a deletion allele of sax-3(yad175) by CRISPR-mediated recombination. yad175 has an in frame deletion that removes the coding sequence of the second repeat of FN3 (FN3-b). yad175 displays much stronger AMsh migration defects than sax-3(ky123, suggested null allele) and sax-3(yad10), while having relatively weaker vab head and AVM guidance defects than those in ky123 and yad10 animals (Figure 1—figure supplement 1F, and Figure 2—figure supplement 1E). The vab head and AVM guidance defects are likely associated with the low sax-3 mRNA level in yad175 (about 50% of that in control animals, Figure 2—figure supplement 1D), which is likely caused by the deletion itself. In summary, even though being a weak allele, sax-3(yad175) has stronger defects in AMsh migration, further arguing the importance of the second FN3 repeat/SAX-3 cleavage in AMsh migration. The stronger AMsh migration defect displayed in yad175 also suggests that ky123 may not be a completely null allele of sax-3.

3) Binding to SYG-1

The syg-1 phenotype is convincing but the SAX-3 binding data is less convincing. First, the list of hits from the mass spec screen is not shown. How many other hits were there, what were they, and how did the representation of syg-1 compare? Second, there are no specificity controls (for the antibody or for the SAX-3-SYG-1 binding itself) in the IP experiments. This is especially important because the SAX-3 fragment is running at the wrong size in this experiment compared to all other gels in the paper.

Related to this issue, A. The mass spectrometric analysis is not explained. Presumably, the authors identified hundreds of peptides, many of which might have been non-specific – a common issue with such analysis. The fact that only two peptides covering 5% of SYG-1 were identified by mass spec (buried in supplement and not mentioned in text) implies that the confidence level for this identification was likely not high. This is generally fine; the authors followed up this finding with additional experiments. But a proper report of total number of peptides identified, cutoffs, and how SYG-1 stands out from the list of peptides is warranted. I suggest authors contact their mass spec facility for details and statistical analysis.

In the revised manuscript, we provided a list of all SAX-3 binding proteins that were identified only in the 1.5-3 fold stage embryos (Supplementary file 2). In this experiment, we compared the proteins identified from IP-Mass spec of FLAG::SAX-3 in embryos in 16-44-cell and 1.5-3 fold stages, and identified proteins with specific interactions with SAX-3 only in 1.5-3 fold stage embryos. We also ran a similar experiment using IP of FLAG::INX-3(innexin/gap junction) in 1.5-3 fold stage embryos as a negative control for FLAG IP. All peptides identified were beyond cutoff and have a high confidence as mentioned in the new Materials and methods section. We have also added the detailed description of proteomics and data analyses in the updated Materials and methods section.

In Figure 2, the HA tag was fused at the end of SAX-3 C-terminal, and the “cleavage” band detected by the anti-HA antibody is the leftover fragment of the extracellular cleavage (about 70 kDa). In Figure 3, the FLAG tag was fused at the beginning of SAX-3 N-terminal (right after the signal peptide), and the cleavage band detected by the anti-FLAG antibody is the extracellular cleavage product (about 50-55 kDa). We apologize for the confusion, and have highlighted the position of HA and FLAG tags and the predicted size of each band in the updated figures. We also included additional controls (IgG control and transgene control) in the updated Figure 3B, and ran the IP products and inputs of SAX-3 in the same gel in which the SYG-1-Co-IPed SAX-3 extracellular fragment is about 55 kDa.

Additionally, the supplement shows one Asn labeled green in the second peptide by the mass spec analysis software. Does that imply a chemical modification/artifact that affected the mass of the peptide? The supplement, which is essential, could be explained better. Currently, it is a screen snapshot from an unknown piece of software that is not mentioned or cited.

The screen shot was the mass spec results analyzed by software called “Scaffold”. The green color of the Asparagine(N) means this amino acid was deamidated to either aspartic acid or isoaspartic acid. We added those details in the updated figure legend and Materials and methods sections. Thanks!

B) The authors should report exact boundaries of their deletion constructs (ΔC, ΔC+TM, Ig(1-5)FN3(1), etc.) presented in Figures 1, 2 and 3, especially since they would be relevant for judging whether their manipulations (domain deletions or inclusions) could inadvertently cause misfolding of proteins, or where exactly the cleavage might be happening. It would help narrow down a binding site on the sequence for the proposed SYG-1 interaction. In addition, we'd know if they removed the long linker sequences shown to be important between some of the IG and FN domains (Barak et al., 2019; Aleksandrova et al., 2018; Pak et al., 2020), when they removed the domains. It would be of interest to see if proteolysis reported by Qu et al. matches that observed by Pak et al., 2020, between the first and second FN domains of human Robo3. On that remark, I cannot find how they draw their schematics for Figure 4 degradation product (containing two FN domains).

In the revised manuscript, we added a description of exact boundaries of their deletion constructs in Figure 1—figure supplement 1H. The linker sequences were maintained in the deletion constructs. We revised the model and the SAX-3 extracellular cleavage fragments now only contain the first FN3 repeat. We apologize for the oversight.

C) Authors present glial migration phenotypes for a set of truncations and deletions of SAX-3. The data provided is self-consistent with the biochemistry they present (which is sparse but enough for an initial study of this kind). However, they do not present data that shows that these deletions (such as IG(1-5) or FN3(1) in Figure 2, which did not rescue) are successfully expressed, make it to through the secretory machinery and are presented on/secreted from neurons. One exception is the FN3(2)Δ constructs, where they did show protein production, but not necessarily successful display on cells. These are useful controls that actually allow them to support their conclusions. It would also be preferable to see if the expression levels of these deletions are comparable to wild-type, so that the observed phenotypes are not due to underexpression of or overproduction of SAX-3. Due to the current circumstances with the coronavirus pandemic, the authors may be given a dispensation from completing these experiments, as long as they clarify the possibility that some of these constructs might not be produced and secreted/displayed in vivo, and hence the lack of rescue.

In the revised manuscript, we provided new data to show that SAX-3(ΔFN3-b)::GFP has similar expression pattern and level as those of the full length of SAX-3::GFP (Figure 2—figure supplement 1C). We also examined the expression of IG(1-5)::GFP and FN1::GFP. However, these fragments are secreted into extracellular space and have very different expression patterns when compared with the full length of SAX-3::GFP, and we were not able to compare their expression with SAX-3 full-length. Therefore we toned down our conclusions regarding IG(1-5) and FN3(1). Thanks!

D) I do not understand the comment about FN domains being mostly structural, while IG domains are functional. If this is a general comment about IG/FN proteins such as Robo, one of the most prominent IG/FN receptors DCC/UNC-40 binds its Netrin ligand with the FN domains, not IG. If this is a comment about the FN domains of Robos not being "functional", that's also not appropriate: While Slits bind the Robo IG domain 1, the other known extracellular ligand of Robos, the NELL proteins, bind the first FN domain of vertebrate Robos (Pak et al., 2020 and Yamamoto et al., 2019), the same domain identified by the authors for SYG-1 binding, and NELLs don't touch the IG domains.

We apologized for the confusion and have removed the comment about the FN domains in the revised manuscript.

In fact, the authors may want to discuss the relevance of the FN domains for Robo function, which was recently resolved in a burst of publications (the Pak et al., Yamamoto, et al., Barak et al. and Aleksandrova et al. manuscripts). The cleavage described here might actually act to open Robo receptors for binding other ligands, such as NELLs (not present in C. elegans) or SYG-1.

We have revised the Discussion section and highlighted the possibilities of other regulations mediated by the SAX-3 cleavage fragment including NElls. Thanks!

E) The authors do not show that the interaction is direct at any point in the manuscript. The pull-downs can simply be the result of a third protein mediating this interaction; this is not an uncommon occurrence. This does not make the paper less interesting, but the possibility has to be clearly mentioned across the manuscript.

We have updated the manuscript and added a sentence to summarize the IP results as “suggesting that SYG-1 can directly or indirectly bind with the SAX-3 extracellular cleavage fragment”. Thanks

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

The manuscript has been improved and but there are some remaining issues that need to be addressed before acceptance, as outlined below: Can you address these points by providing more information and clarification in your writing when providing the revision.

Reviewer #2:

1) Cell autonomy experiments still depend exclusively on promoters whose expression and timing in the embryo, relative to AMsh migration, have not been determined. The images of AMsh-specific promoters (Figure 1—figure supplement 1G) appear to show different cells, one much more anterior than the other. Mosaic analysis was not performed.

As our data shown in Figure 1—figure supplement 1G, both Pf53f4.13, and Pf16f9.3 can drive expression of GFP in AMsh glia in embryos, which is consistent with a recent study published in Science ( “A lineage-resolved molecular atlas of C. elegans embryogenesis at single-cell resolution”), in which they showed that both f53f4.13, and f16f9.3 mRNA were exclusively expressed in AMsh glia in embryos. In the same manuscript, they also presented data to show that in embryos arl-13 was exclusively expressed in amphid neurons, and unc-33 was broadly expressed in neurons. Based on this evidence, we believe that it is appropriate to reach our conclusion using these promoters, and we have cited the Science manuscript in the Materials and methods section to support the use of these promoters. The pictures of Pf53f4.13 and Pf16f9.3 reporters in embryos are not at identical developmental stages, which causes the appearance of slightly different positions of AMsh glia.

2) In the mass spectrometry experiments, it appears that SYG-1 was one of the weakest hits detected, ranking below 1000 in the overall list. Even with aggressive filtering, SYG-1 is among a few dozen hits and appears at the level of background noise, consistent with the presence of a "DECOY" hit with identical results. This is not the impression given in the manuscript.

We agree with the reviewer that SYG-1 was not abundantly detected in our proteomic studies, but as described in the Materials and methods section their interactions were true signals and were beyond the level of “background noise”. In this experiment, we had two control groups: one was the same transgene at the “non-cleavage” stage, and the other one was a membrane channel protein INX-3 at the same developmental stage, and we didn’t detect any SYG-1 peptides in those groups. The proteomic experiments were done twice, and the detection of SYG-1 peptides in 1.5-3 fold stage embryos but not in control groups were observed both times.

3) The relative position of the nerve ring and the AMsh in Figure 1—figure supplement 1D strongly suggests that the AMsh is tracking with the displaced nerve ring, contrary to the authors' interpretation. Due to COVID, the authors could not obtain videos of glial migration.

In the control animals AMsh glia migrate from their birth place, the nose region of the animals, toward the nerve ring, and terminate their migration at the region just past the nerve ring. There were no control adult animals we observed with any AMsh glia cell body anterior to the nerve ring. If the “AMsh is tracking with the displaced nerve ring”, we would not expect to see any change in the relative position of AMsh glia and the nerve ring in sax-3(lf) animals. However, that is not case as shown in Figure 1—figure supplement 1D. In sax-3(lf) animals about 60% of adult animals had at least one AMsh glia presiding at a region anterior to the nerve ring and closer to their birth place, supporting our conclusion that the AMsh migration defects are not a secondary effect of the displaced nerve ring. We apologize for the possible confusion, and we have revised this part in the manuscript to make it clearer.

4) yad10 is expected to encode a protein similar to the Ig(1-5)FN(1-3) fragment used for rescue. One possibility was that the mRNA is not expressed but the authors now show that is not the case. The yad10 phenotype contradicts the overexpression results in which the N-terminal fragment of SAX-3 is sufficient for rescue.

How a mutation affects gene expression is complicated, and detectable sax-3 mRNA does not mean the protein can be expressed, folded, and released in a correct manner. It is still to be determined that how yad10 affects sax-3, but the case like yad10 is not unusual (at least in C. elegans).

Reviewer #3:

The authors responded to comments satisfactorily as much as I can judge. My geneticist colleagues should review the validity of the new experiments with the new promoters and the CRISPR deletion of the second FN3 domain.

I am not sure the statement "Recent biochemical and structural studies show that the extracellular domains of Robo can dimerize through the fourth Ig domain, and this dimerization is inhibited by the second FN3 domain in the absent of Slit" is strictly true. It may be safest to define this as "FN3 domains", since both are involved FN3 domains are involved in a hairpin-like structure. Also, there is plenty of evidence that Robo1 is a dimer even when the FN3 domains are there, so I would say "the dimerization may be inhibited" to allow for multiple possibilities. Also see the typo, "in the absent of".

Corrected.

Also, the last statement appears to be false. NELLs bind the first FN3 domain, not second, so they would not be binding the membrane-anchored fragment. NELLs probably should not be mentioned here at the end of the text at all. My initial comment was that the first FN3(1) domain is not a non-functional/only structural domain as implied by the original draft. NELLs do bind the first FN3 domain, and can turn on a repulsive Robo response.

We have removed the description about the potential interactions between NELLs and the membrane cleavage product.

I should also remind the authors that they still use inconsistent naming of the Fibronectin domains. Figure 1—figure supplement 1 has FN3-a and FN1 for the same domains in panels f and h. Or there is FN1 in Figure 3I, but refers to this domain as FN3-a in Figure 2F. Minor issue, but no reason to confuse readers.

Revised to FN3-a and FN3-b in all figures.