C. elegans srGAP is an α-catenin M domain-binding protein that strengthens cadherin-dependent adhesion during morphogenesis

As cells rearrange and change shape during embryonic morphogenesis, adhesions between cells must be dynamically formed, broken and reformed (Pinheiro and Bellaiotache, 2018; Takeichi, 2014; Walck-Shannon and Hardin, 2014), yet they must also withstand tensile forces (Charras and Yap, 2018; Guillot and Lecuit, 2013; Leckband and de Rooij, 2014; Lecuit and Yap, 2015; Pannekoek et al., 2019; Priya and Yap, 2015). Central to meeting this fundamental challenge is the cadherin-catenin complex (CCC) (Harris and Tepass, 2010; Leckband and de Rooij, 2014; Lecuit and Yap, 2015; Niessen et al., 2011; Takeichi, 2014). The CCC mediates intercellular adhesion at adherens junctions (AJs) through homophilic interactions of transmembrane cadherins. β-Catenin binds the cadherin C terminus and the N terminus of α-catenins (Pokutta and Weis, 2007). The C terminus of α-catenin binds F-actin, linking the CCC to the cytoskeleton (Buckley et al., 2014; Maiden and Hardin, 2011; Mege and Ishiyama, 2017).

The CCC must respond to differing tension states (Charras and Yap, 2018; Guillot and Lecuit, 2013; Huveneers and de Rooij, 2013; Leckband and de Rooij, 2014; Lecuit et al., 2011; Lecuit and Yap, 2015; Mege and Ishiyama, 2017). α-Catenin is one locus for such regulation. Imaging (Kim et al., 2015), molecular dynamics simulations (Barrick et al., 2018; Ishiyama et al., 2018; Li et al., 2015), crystallographic data (Choi et al., 2012; Ishiyama et al., 2013, 2018; Kang et al., 2017) and single-molecule tweezing (Pang et al., 2019; Yao et al., 2014) suggest that the middle (M) domain of α-catenins undergoes reversible distension under tension. Tension unmasks the binding site for vinculin in vertebrate αE-catenin (Yonemura et al., 2010); however, disruption of vinculin binding sites in αE-catenin leads to weak effects (Huveneers et al., 2012), suggesting that additional tension-sensitive α-catenin binding partners may act cooperatively to strengthen the CCC.

A useful system for identifying novel, functionally relevant, interactors with the CCC is the embryonic epidermis of the Caenorhabditis elegans embryo. The epidermis is born on the dorsal side of the embryo and spreads ventrally during ventral enclosure (Williams-Masson et al., 1997). After enclosure, actomyosin-mediated constriction of epidermal cells squeezes the embryo into a vermiform shape (Chisholm and Hardin, 2005; Marston and Goldstein, 2006; Vuong-Brender et al., 2016). In C. elegans, the core CCC is composed of HMR-1/cadherin, HMP-1/α-catenin, and HMP-2/β-catenin (see Fig. S1A). Maternal loss of CCC components leads to ventral enclosure failure; zygotic loss results in failure of elongation due to defects in the junctional proximal actin network (Costa et al., 1998; Kwiatkowski et al., 2010; Maiden et al., 2013; Vuong-Brender et al., 2018).

We previously used a weak loss-of-function allele of hmp-1 to look for synergistic, lethal enhancers of morphogenetic defects during embryonic morphogenesis (Cox-Paulson et al., 2012; Lynch et al., 2012). One of these is the single C. elegans slit-robo GTPase activating protein (srGAP) homolog SRGP-1 (Zaidel-Bar et al., 2010). srGAPs have an N-terminal, lipid-binding, extended F-BAR domain (Coutinho-Budd et al., 2012; Guez-Haddad et al., 2015; Sporny et al., 2017); a GTPase-activating protein (GAP) domain that can downregulate Rho family GTPases (Foletta et al., 2002; Guerrier et al., 2009; Liang et al., 2017; Mason et al., 2011; Soderling et al., 2002; Waltereit et al., 2012; Wong et al., 2001; Yamazaki et al., 2013; Yang et al., 2006); and a C-terminal SH3 domain that binds F-actin regulators (Carlson et al., 2011; Endris et al., 2011).

Here, we use a combination of genetic, biochemical and imaging approaches in C. elegans to demonstrate a role for SRGP-1 in binding a closed conformation of the HMP-1 M domain to stabilize adherens junctions under mechanical tension during embryonic morphogenesis.

SRGP-1 is a bona fide srGAP based on homology and structure-function analyses ((Neukomm et al., 2011; Zaidel-Bar et al., 2010; see Fig. 1C). SRGP-1 colocalizes with the CCC, and when SRGP-1::GFP is overexpressed outward membrane bends form that contain CCC proteins but not DLG/AJM complex proteins (Zaidel-Bar et al., 2010). We also identified SRGP-1 in an HMP-1 co-immunoprecipitation/mass spectrometry screen (Callaci et al., 2015). We verified this result (Fig. 1A) and used a yeast two-hybrid assay to show that full-length SRGP-1 can interact directly with full-length HMP-1 but not HMP-2/β-catenin (Fig. 1B). We confirmed this interaction via protein pull-down assays using His-SUMO-HMP-1 (Callaci et al., 2015) and GST-SRGP-1 in bacteria. Using a series of deletions (for domain maps, see Fig. 1C), we found that the HMP-1M domain appears to be responsible for the interaction with amino acids 685-1059 of SRGP-1 (hereafter the ‘C terminus’; Fig. 1D, Fig. S1B).

We next assessed the necessity of the SRGP-1 C terminus for SRGP-1 colocalization with HMP-1 in vivo. SRGP-1(ΔC)::GFP exhibits a marked increase in cytoplasmic versus junctional signal compared with full-length SRGP-1 (Fig. 1E,F). Both proteins contain an intact F-BAR domain, and so, as expected, readily associate with and follow the curvature of the membrane. Full-length SRGP-1::GFP and HMP-1::mScarlet-I closely followed the curvature of the membrane at junctions. In contrast, HMP-1 frequently no longer followed SRGP-1(ΔC)::GFP into membrane bends (Fig. 1F, insets), as reflected in a change in overlap at junctions in 1.5-fold embryos (Manders overlap coefficient M2 with thresholding; GFP overlap with mScarlet-I=0.82 for SRGP-1::GFP, n=57 junctions; M2=0.46 for SRGP-1(ΔC)::GFP, n=36). These data indicate that the SRGP-1 C-terminus is necessary for restricting SRGP-1 to junctions and for normal association with HMP-1. Because some SRGP-1ΔC can localize to junctions, we infer that additional interactions outside the C terminus also promote accumulation of SRGP-1 at junctions.

We next determined whether the C terminus is sufficient to target SRGP-1 to the CCC. Full-length SRGP-1::GFP expressed in the srgp-1(gk441841) null background was robustly targeted to cell-cell junctions throughout morphogenesis (Fig. 2A). In contrast, SRGP-1(C term)::GFP failed to localize in the early embryo; SRGP-1(C term)::GFP remained much more cytoplasmic in later embryos, but localized to junctions during ventral enclosure and elongation (Fig. 2B, yellow arrows).

We next assessed the effects of HMP-1 depletion on SRGP-1 junctional recruitment. SRGP-1::GFP was not depleted from induced membrane tubulations in hmp-1(RNAi) embryos (Fig. 2C, right, yellow arrowheads), but junctional SRGP-1 was greatly reduced (see Fig. 2E for quantification), indicating that HMP-1 is required for normal junctional recruitment of SRGP-1 during morphogenesis. In contrast, hmp-1(RNAi) led to complete loss of junctional SRGP-1(C term) at all stages (Fig. 2D,E), indicating that interaction of the SRGP-1 C terminus with the CCC has a stringent requirement for HMP-1.

Our previous work showed that expression of a construct lacking most of the HMP-1M domain [HMP-1(ΔVH2)::GFP] can rescue viability, albeit weakly, in embryos carrying a C-terminal HMP-1 truncation, hmp-1(zu278) (Maiden et al., 2013). It remained possible, however, that other parts of the HMP-1M domain retained in zu278 mutants recruit other interactors. We confirmed that the same M domain deletion can partially rescue hmp-1 CRISPR null (jc48) homozygotes (Shao et al., 2019) from 100% to 35.3% embryonic lethality (Fig. 2G).

We next examined the effects of deletion of the HMP-1M domain on SRGP-1 recruitment. srgp-1::mScarlet-I; hmp-1(jc48); Ex[hmp-1(ΔVH2)::gfp] embryos retained detectable membrane-localized SRGP-1::mScarlet-I in lateral epidermal cells (seam cells) during elongation but at greatly reduced levels (Fig. 2F,H,I). Thus, the HMP-1M domain appears to be important for recruiting or maintaining SRGP-1 at the CCC during embryonic elongation.

Our previous structural work (Kang et al., 2017) demonstrated that HMP-1 possesses conserved salt bridges that are structural analogs of salt bridges in mammalian αE-catenin (Barrick et al., 2018; Li et al., 2015). Two arginines (R551 and R554) in the MIII subdomain are predicted to stabilize the internal structure of the HMP-1M domain, resisting its complete extension by forming salt bridges with two residues (D386 and D497, respectively) in the MII subdomain. We used CRISPR/Cas9 mutagenesis to mutate these arginines to alanines. Limited proteolysis has been used to assess the conformational openness of α-catenin (Heier et al., 2021; Pokutta and Weis, 2000; Xu et al., 2020); limited proteolysis of HMP-1 similarly indicates that the D386A and D497A mutations led to an open conformation of the HMP-1M domain (Fig. S2A,B). hmp-1(R551/554A)::mScarlet-I mutants were viable and did not display overt morphological defects under laboratory conditions. We then constructed srgp-1(gk441841);hmp-1(R551/554A)::mScarlet-I strains expressing srgp-1::gfp constructs. Whereas full-length HMP-1::mScarlet-I colocalized precisely with full-length SRGP-1::GFP (Fig. 3A, top), HMP-1(R551/554A) did not (Fig. 3A, second row), reflected in a statistically significant decrease in colocalization at junctions [Pearson's r=0.53 for HMP-1::mScarlet-I versus SRGP-1::GFP pixels above threshold, n=25 junctions; r=0.49 for HMP-1(R551/554A)::mScarlet-I versus SRGP-1::GFP, n=25; P<0.0001, unpaired Student's t-test]. The SRGP-1 C terminus is enriched at seam-seam junctions (Fig. 3A, blue dashed line) compared with seam-non-seam junctions (Fig. 3A, yellow dashed line) during elongation. Weak RNAi against let-502/ROCK, an activator of actomyosin contractility, resulted in increased recruitment of SRGP-1(C-term)::GFP to seam-non-seam junctions (Figs S2C,D). Strikingly, junctional enrichment of SRGP-1(C term)::GFP was completely abrogated in hmp-1(R551/554A)::mScarlet-I embryos (Fig. 3A bottom row; quantified in Fig. 3B). These results indicate that junctional recruitment of the SRGP-1 C terminus requires the HMP-1M domain to be in a partially closed conformation. Weak knockdown of let-502/ROCK presumably leads to a decrease in the fraction of HMP-1 protein in a fully extended state at seam-non-seam junctions, which are under high tension during elongation (Costa et al., 1998; Cox-Paulson et al., 2012; Vuong-Brender et al., 2017, 2018), thereby allowing recruitment of the SRGP-1 C terminus.

We next examined whether SRGP-1 is required for normal junctional morphology. We could not detect a statistically significant effect on junctional tortuosity (total path length/Feret length) in srgp-1(gk441841) embryos (wild-type ratio=1.24, n=43; gk441841 ratio=1.23, n=33; P>0.76, two-tailed Student's t-test). We did find, however, that in srgp-1(gk441841) mutants the average junctional intensity of an endogenous HMP-1::GFP knock-in (Marston et al., 2016) was measurably decreased compared with wild type (Fig. 3C). HMP-1(R551/554A)::GFP showed a similar decrease in junctional intensity (Fig. 3C). We next analyzed mobility of HMP-1 at junctions in 1.5- to 1.75-fold embryos using fluorescence recovery after photobleaching (FRAP; Fig. 3D). Loss of SRGP-1 led to a statistically significant decrease in recovery half-life (t_1/2 for wild type=25.0 s; t_1/2 for srgp-1=30.5 s; P<0.04, unpaired two-tailed Student's t-test, n=11 for each), indicating increased mobility of HMP-1 in the absence of SRGP-1. These results suggest a reciprocal role for SRGP-1 in organizing and stabilizing HMP-1 at junctions.

srgp-1(RNAi) in the hmp-1 weak loss-of-function mutant hmp-1(fe4[S823F]) causes many embryos to arrest with defects in gastrulation cleft closure and subsequent ventral enclosure (Zaidel-Bar et al., 2010). We analyzed the remaining hmp-1(fe4); srgp-1(RNAi) embryos that survived beyond ventral enclosure. hmp-1(fe4); srgp-1(RNAi) embryos arrested as they approached the 2-fold stage of elongation with pronounced dorsal folds (100% of embryos, n=21 embryos; see Fig. 3E; Movie 1); thus, SRGP-1 also plays key roles during early stages of embryonic elongation.

We were unable to create srgp-1(gk441841); hmp-1(fe4) double mutants for rescue experiments, likely because fe4 is a maternal effect allele (J. Pettitt, personal communication) instead, we used homozygotes for a hypomorphic allele of hmp-2/β-catenin, hmp-2(qm39) (Lockwood et al., 2008; Zaidel-Bar et al., 2010) as a sensitized background. hmp-2(qm39) embryos appeared largely wild type during embryonic development, but later developed slight body morphology defects that resolved during cuticle molts (Fig. 4A). In contrast, double-homozygous srgp-1(gk441841); hmp-2(qm39) embryos displayed 57.1% embryonic lethality, arresting with dorsal humps during elongation (Fig. 4B).

We next performed rescue experiments using SRGP-1 fragments in srgp-1(gk441841); hmp-2(qm39) embryos. SRGP-1(full-length)::GFP and SRGP-1(ΔGAP)::GFP were capable of strong rescue, although full-length SRGP-1::GFP rescued more effectively (Fig. 4C,D,G). Importantly, srgp-1(gk441841); hmp-2(qm39); Ex[srgp-1(ΔC)::gfp] embryos were indistinguishable from srgp-1(gk441841); hmp-2(qm39) alone (Fig. 4E,G), indicating that the C terminus of SRGP-1 is functionally important during elongation. The C terminus alone was not sufficient to rescue lethality, however (Fig. 4F,G). Taken together with its inability to support normal junctional localization (see Figs 1 and 2), these data suggest that a key functional role for the SRGP-1 C terminus is to target SRGP-1 to and maintain it at the CCC.

Here, we demonstrated the importance of the SRGP-1 C terminus in binding the HMP-1 M domain, and, unexpectedly, that a partially closed conformation of the M domain is necessary for junctional maintenance of SRGP-1 mediated via its C terminus. Junctional SRGP-1 is in turn crucial for epidermal morphogenesis in sensitized backgrounds.

Although roles for srGAPs outside of neural tissues are beginning to be recognized (reviewed by Lucas and Hardin, 2017), roles of srGAPs during epithelial morphogenesis are poorly understood. In cultured mammalian cells, srGAP1 is recruited to junctions by tyrosine non-phosphorylated cortactin, where it downregulates Rho activity via its GAP domain (Liang et al., 2017). There is no clear cortactin homolog in C. elegans (J.H., unpublished) and the SRGP-1 GAP domain is relatively unimportant for its roles in morphogenesis (Zaidel-Bar et al., 2010; present work). A parsimonious model for SRGP-1 recruitment to the CCC during embryonic elongation involves direct HMP-1 binding.

Given when C-terminal fragments of SRGP-1∷GFP localize to junctions, increased interaction of the SRGP-1 C terminus with HMP-1 appears to occur at a time when adherens junctions are under increased, but not maximal, tension (Fig. 4H). This contrasts with vertebrate vinculin. Vinculin is recruited to the CCC via α-catenin under tension by exposing a binding site within α-catenin that is normally buried within its M domain (Choi et al., 2012; Huveneers and de Rooij, 2013; Yonemura et al., 2010). Maximal vinculin binding to mammalian αE-catenin is favored by complete unfurling of the αE-catenin M domain (Yao et al., 2014); salt bridge mutations in the M domain of αE-catenin thus activate it for vinculin binding (Choi et al., 2012; Huveneers et al., 2012; Ishiyama et al., 2013; Yonemura et al., 2010). In contrast, we have shown here that similar mutations in the HMP-1M domain lead to loss of SRGP-1 junctional recruitment via its C terminus. SRGP-1 may therefore bind the HMP-1M domain in an intermediate conformation under tension that precedes complete unfurling of its M domain. At higher tension states, when the M domain is fully extended, HMP-1 may recruit other binding partners. Rigorous testing of such a mechanism awaits further biophysical experiments to interrogate the possibility of such a ‘conformational clutch’.

In contrast to the SRGP-1 C terminus, full-length SRGP-1 is retained at junctions in hmp-1 salt bridge mutants, but it is lost when HMP-1 is absent or when the HMP-1M domain is deleted. Multiple domains of SRGP-1 may therefore recognize different conformational states of the HMP-1M domain. Alternatively, another junctional protein with which SRGP-1 interacts may be recruited in an HMP-1-dependent manner and is retained even when the HMP-1M domain is more fully extended.

We showed previously that N- and C-terminal fragments of SRGP-1 can bind in vitro, suggesting head-to-tail autoinhibition. Moreover, overexpression of a C-terminal truncation of SRGP-1 leads to excessive membrane tubulations (Zaidel-Bar et al., 2010). Here, we have shown that the SRGP-1 C terminus alone can associate with junctions, but is insufficient for function, and that the SRGP-1 F-BAR domain is necessary. An irenic model that harmonizes these data is that HMP-1 recruits SRGP-1, leading to highly local, restrained activation of the membrane binding activities of SRGP-1 when embryonic elongation begins. SRGP-1 may therefore be similar to vinculin and talin, which undergo autoinhibition in the cytosol that is thought to be relieved by interactions with binding partners at adhesion sites (Bakolitsa et al., 2004; Dedden et al., 2019).

In summary, our results identify previously unknown roles for SRGP-1 during cadherin-dependent epithelial morphogenetic events. It will be interesting to determine whether vertebrate srGAPs can interact similarly with a component of the CCC. Future experiments should continue to clarify how this multifunctional family of proteins contributes to cadherin-dependent adhesion during morphogenesis.

C. elegans strains were maintained using standard methods (Brenner, 1974). Bristol N2 was used as wild type. Ten outcrosses to N2 were performed to remove background mutations from the Million Mutation Project strain, VC30226, which carries the srgp-1(gk441841) allele (Thompson et al., 2013). See Table S2 for a full list of strains used in this study.

Four-dimensional (4D) DIC movies were collected on either a Nikon Optiphot-2 connected to a QiCAM camera (QImaging) or an Olympus BX50 connected to a Scion CFW-1512M camera (Scion Corp.) using Micro-Manager software (v. 1.42) (Edelstein et al., 2010, 2014). ImageJ plugins (https://worms.zoology.wisc.edu/research/4d/4d.html) were used to compress and view DIC movies.

SRGP-1::GFP; HMP-1::mScarlet-I embryos were dissected from adult hermaphrodites and mounted onto 10% agar pads in M9 solution and imaged essentially as described (Zaidel-Bar et al., 2010). For fluorescence imaging, a Dragonfly 500 spinning disc confocal microscope (Andor), mounted on a Leica DMi8 microscope, equipped with a Zyla camera and controlled by Fusion software (Andor) was used to collect images using 0.5 μm slices with a 63×/1.3 NA glycerol Leica objective at 20°C.

FRAP analysis was conducted on z-projections of 4D confocal movies composed of four z-positions spaced 0.18 μm apart. Embryos were aged between 1.5- and 1.75-fold. 4D movies were generated by photobleaching a region of interest for 3 s and images were collected at 2 s intervals thereafter. Photobleaching was performed using an Andor Mosaic DMD and imaging was performed on a Leica Dmi8 microscope with a Dragonfly spinning disk (Andor). Images were collected on a Zyla CMOS camera using Fusion software (Andor). Recovery curves were analyzed via single-exponential curve fitting using custom Jython scripts written for Fiji. Scripts are available at https://worms.zoology.wisc.edu/research/microscopy/4d.html.

To compare intensity, mean HMP-1 junctional signal was measured by tracing HMP-1::GFP junctional signal on z-projections of four focal planes spaced 0.18 µm apart in embryos between 1.5- and 1.75-fold. Intensity profiles were collected by drawing a 1-pixel-wide line between nuclei during elongation. Colocalization of HMP-1 and SRGP-1 was performed in Fiji using the Just Another Colocalization Plugin (JACoP; https://imagej.nih.gov/ij/plugins/track/jacop.html) (Bolte and Cordelieres, 2006). Four focal planes from 25 junctional segments were combined into single stacks for each genotype. Maximum intensity z-projections were obtained, and automated Costes thresholding within JACoP was visually confirmed in each case.

HMP-1::mScarlet-I and SRGP-1::mScarlet-I worms were generated by plasmid-based CRISPR/Cas9 (Dickinson et al., 2015) using repair templates cloned using SapTrap cloning (Schwartz and Jorgensen, 2016). Small substitution mutations were made by marker-free genome editing (Arribere et al., 2014). Guides, homology arms primers and single-stranded repair templates for all CRISPR/Cas9 editing can be found in Table S1.

For transgenic lines, 10 ng/μl of the transgene of interest, in addition to 20 ng/μl non-coding DNA (F35D3) and 80 ng/μl rol-6(su1006), was injected into the gonads of srgp-1(gk441841) young adults, as described previously (Mello and Fire, 1995).

Injection RNAi was performed as previously described (Walston et al., 2004). dsRNAs were generated using an Ambion T7 and/or T3 MEGAscript kit; the template used was the srgp-1 feeding clone from a feeding library (Kamath et al., 2003), and the template used for hmp-1 (yk1315a06) was obtained from Dr Yuji Kohara (NEXTDB, http://nematode.lab.nig.ac.jp/). ‘Weak’ injection RNAi was accomplished by injecting young adults with 1 µg/µl dsRNA into the pseudocoelom and imaging embryos 18 h after injection.

GST- and SUMO-His-tagged proteins were expressed in BL21-Gold(DE3) Escherichia coli cells and purified as described (Maiden et al., 2013; Mayers et al., 2011). Cells were induced with 0.1 mM IPTG at 18°C for 16 h. Wash and elution buffers were as follows: GST wash (1× PBS, 500 mM NaCl, 0.1% Tween-20 and 1 mM DTT), GST elution (50 mM, Tris pH 8.0, 0.3% glutathione, 150 mM NaCl), His wash (50 mM sodium phosphate, pH 8.0, 300 mM NaCl, 0.1% Tween-20, 10 mM imidazole) and His elution (250 mM imidazole, 100 mM NaCl, 10% glycerol, 50 mM HEPES, pH 7.6).

DNA encoding wild-type full-length HMP-1 (aa1-927) and M region (aa270-646) and DNA encoding 551/554A mutant proteins was cloned into pMAL-c6t to make MBP-tagged fusion proteins. Fusion proteins were expressed in Novagen Rosetta DE3 competent E. coli cells (Sigma-Aldrich, 70954) and purified as described by Kwiatkowski et al., (2010) and Hansen et al. (2013). MBP-tagged proteins were bound to amylose-conjugated beads. Bound beads were then equilibrated in cleavage buffer (20 mM Tris, pH 8.0, 150 mM NaCl, 2 mM EDTA, 10% glycerol and 1 mM DTT) and incubated with tobacco etch virus protease overnight at 4°C to cleave proteins from the tag. Proteins were then purified by Mono Q ion-exchange chromatography at 4°C, followed by S200 gel-filtration chromatography at 4°C. Purified proteins were eluted in 20 mM Tris, pH 8.0, 150 mM NaCl, 10% glycerol and 1 mM DTT, concentrated to working concentrations using a Millipore column concentrator and flash-frozen in liquid nitrogen.

Yeast two-hybrid assays were performed as described previously (Lynch et al., 2012). Either full-length HMP-1 or HMP-2 yeast two-hybrid plasmids (Kwiatkowski et al., 2010) were transformed into Y2H Gold yeast singly or with a plasmid encoding full-length SRGP-1 that was cloned into the pGBKT7 vector (Clontech). Positive single transformants were tested for autoactivation, and double transformants were patched onto SD/-Leu/-Trp/X-α-gal/AurA plates. Colonies that grew and turned blue were considered positive for a direct interaction.

C. elegans used for co-immunoprecipitations were grown in liquid culture as previously described (Stiernagle, 2006). Co-immunoprecipitations were carried out as previously described (Cox-Paulson et al., 2012). HMP-1 and HMP-2 were co-immunoprecipitated using a rabbit anti-HMP-1 antibody (Zaidel-Bar et al., 2010) as a control. Western blots were performed as described previously (Zhang et al., 2016), blotting with rabbit anti-GFP or rabbit anti-HMP-1 antibodies and using chemiluminescence to detect proteins using an Odyssey Fc imaging system (LI-COR Biosciences).

L4 worms were transferred to plates seeded with E. coli expressing dsRNA from let-502 or L4440 (control) vectors overnight. The following day, worms were cut open and their embryos were mounted, aged to 1.5 fold, and imaged.

Pull-downs between GST-SRGP-1 and His-SUMO-HMP-1 fragments were performed as described previously (Takahashi et al., 2015). Two replicates were performed in each case. When testing GST-SRGP-1 (C term) against His-SUMO-HMP-1 fragments, purified GST-SRGP-1 (C term) was incubated individually with His-SUMO-HMP-1 fragments still immobilized on Ni-NTA agarose resin for 1 h at 4°C. After washing five times with buffer (50 mM HEPES, pH 7.4; 1 mM EDTA, pH 8.0; 0.1 M KCl; 1 mM MgCl₂; 12.5% w/v glycerol), His elution was then carried out. Finally, eluates were run on SDS-PAGE gels and stained with Coomassie Brilliant Blue to visualize proteins.

Proteins were diluted to 12 μM in 20 mM Tris, pH 8.0, 150 mM NaCl and 1 mM DTT and incubated at room temperature in 0.05 mg/ml sequencing grade trypsin (Roche Applied Science). Digestions were stopped with 2× Laemmli sample buffer followed by boiling. Samples were run by SDS-PAGE and then stained with 0.1% Coomassie Brilliant Blue R-250, 40% ethanol and 10% acetic acid. Gels were scanned on a LI-COR Odyssey FC scanner.

Fisher's Exact Test (http://www.graphpad.com/quickcalcs/contingency1/) was used to determine significance between rescue groups. Differences in cytoplasmic fluorescence levels were determined using one-way ANOVA with Tukey post-hoc testing for multiple comparisons or unpaired two-tailed Student's t-tests for single comparisons after measuring average junctional and cytoplasmic signals in ImageJ. Significant difference in Pearson's R for colocalizations was assessed using the online Z calculator available at https://www.calculator.net/z-score-calculator.html.

Some strains were provided by the C. elegans Genetics Center, which is funded by the NIH Office of Research Infrastructure Programs (P40 OD010440). A cDNA clone for hmp-1 (yk1315a06) was provided by Yuji Kohara (National Institute of Genetics, Japan). We are grateful to Seth Blair for the use of his LI-COR system.

The peer review history is available online at https://journals.biologists.com/dev/article-lookup/doi/10.1242/dev.200775.reviewer-comments.pdf.