ABSTRACT
During embryonic morphogenesis, cells and tissues undergo dramatic movements under the control of F-actin regulators. Our studies of epidermal cell migrations in developing Caenorhabditis elegans embryos have identified multiple plasma membrane signals that regulate the Rac GTPase, thus regulating WAVE and Arp2/3 complexes, to promote branched F-actin formation and polarized enrichment. Here, we describe a pathway that acts in parallel to Rac to transduce membrane signals to control epidermal F-actin through the GTPase RHO-1/RhoA. RHO-1 contributes to epidermal migration through effects on underlying neuroblasts. We identify signals to regulate RHO-1-dependent events in the epidermis. HUM-7, the C. elegans homolog of human MYO9A and MYO9B, regulates F-actin dynamics during epidermal migration. Genetics and biochemistry support that HUM-7 behaves as a GTPase-activating protein (GAP) for the RHO-1/RhoA and CDC-42 GTPases. Loss of HUM-7 enhances RHO-1-dependent epidermal cell behaviors. We identify SAX-3/ROBO as an upstream signal that contributes to attenuated RHO-1 activation through its regulation of HUM-7/Myo9. These studies identify a new role for RHO-1 during epidermal cell migration, and suggest that RHO-1 activity is regulated by SAX-3/ROBO acting on the RhoGAP HUM-7.
INTRODUCTION
Morphogenesis requires the migration of adherent sheets of cells. GTPases control the migration and adhesion of cells. We have demonstrated that Caenorhabditis elegans embryos undergo morphogenetic movements that require the GTPase CED-10/Rac1 for epidermal enclosure, the sheet migration that leads to enclosure of the embryo by the six rows of epidermal cells that form at the dorsal surface. To coordinate these migrations, CED-10/Rac1 activates the nucleation-promoting factor WAVE/SCAR, which signals to the branched actin nucleator Arp2/3 to initiate actin polarization. Mutating any component in this pathway prevents the initiation of the sheet migration required for enclosure of the embryo by epidermal cells during morphogenesis in C. elegans (Fig. 1A) (Patel et al., 2008). Our group was the first to show that three axonal guidance receptors, UNC-40/DCC, SAX-3/ROBO and VAB-1/EphB, are required during embryonic morphogenesis to regulate the levels, localization and polarization of this actin nucleation pathway in epidermal cells (Bernadskaya et al., 2012). However, the molecules that connect the axonal receptors to the regulation of GTPase activity during this migration are unknown.
Recent studies by several groups suggest that besides CED-10/Rac1, the GTPases RHO-1 and CDC-42 also contribute to the morphogenetic process of ventral enclosure (Ouellette et al., 2016; Walck-Shannon et al., 2016; Wernike et al., 2016; Zilberman et al., 2017). However, cytoskeletal consequences of altering RHO-1 or CDC-42 function during ventral enclosure are poorly understood. Myosin is required in the underlying neuroblasts to support epidermal cell migration, suggesting important roles for RHO-1 in neuroblasts during embryonic morphogenesis (Fotopoulos et al., 2013; Wernike et al., 2016). It has been proposed that myosin-based mechanical forces in neuroblasts influence constriction of the overlying epidermal cells. An alternative role for the neuroblasts is to promote a chemical cue that guides ventral enclosure, but this cue has not been identified.
In C. elegans, the GTPase RHO-1/Rho1/RhoA is needed throughout embryogenesis to promote cytokinesis, and also contributes to a step in morphogenesis that occurs after the epidermis encloses the embryo, called embryonic elongation. During elongation, the epidermis constricts circumferentially to squeeze the embryo into a tubular worm. Elongation depends on RHO-1 regulators, and targets including the Rho-binding kinase LET-502 and non-muscle myosins (Chisholm and Hardin, 2005; Diogon et al., 2007; Fotopoulos et al., 2013; Gally et al., 2009; Martin et al., 2014; Piekny and Mains, 2003; Piekny et al., 2003; Quintin et al., 2008; Wissmann et al., 1997; Zhang et al., 2010). High levels of tension drive elongation, mainly in the two rows of lateral seam cells, initiated by the Rho-binding kinase LET-502 and its target, MLC-4/myosin regulatory light chain. In contrast, MEL-11/myosin phosphatase opposes this tension. While the lateral cells drive constriction, expression of the RhoGAP RGA-2 in the two rows of dorsal and ventral epidermal cells promotes decreased RHO-1 activity and tension (Diogon et al., 2007). This balance of forces acting on RHO-1 is required for proper elongation. It is possible that changes in the balance of forces acting on RHO-1 also contribute to ventral enclosure.
Rho GTPases are regulated by GTPase-activating proteins (GAPs), which enhance the hydrolysis of GTP to GDP, thus promoting the dynamic turnover of GTPases from active to inactive states. Specific levels of GTPases, and their subcellular enrichment, help promote correct signals that organize the actin cytoskeleton (Clay and Halloran, 2013). The ability of GAPs to turn over activated GTPases makes them attractive candidates to regulate GTPase function by controlling location and levels. However, there are many more GAP proteins than there are target GTPases, so the removal of individual GAP proteins often gives subtle phenotypes that make them harder to characterize. Removal of some GAPs that are known to result in strong changes in the cytoskeleton in tissue culture cells causes mild or minimal effects when removed from developing organisms (Zaidel-Bar et al., 2010).
The RhoGAP Myo9/myosin IX is proposed to regulate multiple cellular functions including adhesion, cell migration and immune function. Mammals have only two of these myosins, Myo9a and Myo9b (reviewed in Omelchenko and Hall, 2012). Loss-of-function mutations or overexpression of these proteins has been linked to various cancers and immune defects. Some studies support a role for MYO9b in human intestinal diseases such as inflammatory bowel disease and Crohn's disease (Li et al., 2016; Prager et al., 2014; Wang et al., 2016). The complex structure of Myo9 proteins includes a C-terminal GAP domain, which studies show attenuates RhoA activity. Myo9 proteins also have an N-terminal single-headed processive myosin motor that moves the protein towards the plus, or barbed, ends of F-actin fibers (Chandhoke and Mooseker, 2012; Hanley et al., 2010; Hegan et al., 2016; Inoue et al., 2002; Liao et al., 2010; Omelchenko and Hall, 2012). Conserved Ras-associated (RA) and C1 domains are less understood. The mouse knockout of Myo9a resulted in central nervous system and kidney defects (Abouhamed et al., 2009; Thelen et al., 2015) that might involve changes in protein trafficking, whereas the mouse Myo9b knockout resulted in altered morphology and motility of immune cells (Hanley et al., 2010) and impaired intestinal barrier function (Hegan et al., 2016). A pathway has been proposed in a human lung cancer model in which SLIT/ROBO acts to inhibit Myo9b, which releases RhoA. However, the reason why this double inhibitory pathway is needed to regulate RhoA, and how misregulated RhoA promotes lung cancer, is unclear (Kong et al., 2015). In metastatic prostate cancer cell lines, MYO9B is elevated, while knockdown of MYO9B results in altered NM2A (non-muscle myosin), and defective migration, further supporting a function in RhoA attenuation (Makowska et al., 2015).
C. elegans has a single Myo9 homolog, HUM-7, which has been characterized extensively at the biochemical level, and was used to analyze how this single-headed myosin motor is able to move processively towards the plus end of actin fibers (Liao et al., 2010). However, the in vivo role of HUM-7 has not been reported. C. elegans is thus an excellent in vivo system to probe the mechanisms behind the complex phenotypes seen for mutations in HUM-7/Myo9.
We sought novel regulators of morphogenesis and identified the HUM-7/Myo9/myosin IX protein. In this study, we characterize the contribution of HUM-7/Myo9 to C. elegans morphogenesis. We investigate which GTPases are affected by this RhoGAP and place hum-7 in a genetic pathway known to regulate actomyosin contractility. Additionally, we address how hum-7 affects F-actin levels, polarization and dynamics, and how it compares to known RHO-1/RhoA pathway mutants, including mel-11. We also investigate the contribution of HUM-7/Myo9 to axonal guidance. Genetic epistasis studies place this GAP downstream of two receptors, SAX-3/ROBO and VAB-1/EphB, known to regulate F-actin during morphogenesis, to modulate RHO-1 signaling.
RESULTS
Three axonal guidance receptors are upstream regulators of F-actin during morphogenesis through their effects on WAVE/SCAR (Fig. 1A) (Bernadskaya et al., 2012). To identify proteins that might connect the axonal guidance receptors to WAVE/SCAR regulation, we performed an RNA interference (RNAi) screen to identify genes that enhance the low levels of embryonic lethality (14%) caused by loss of one of the axonal guidance receptors, UNC-40. We screened through a 2000-clone feeding RNAi library (Sieburth et al., 2005), which represents ∼10% of all C. elegans genes and identified several enhancers of unc-40 lethality using the putative null allele unc-40(n324). RNAi depletion of HUM-7, an unconventional myosin heavy chain protein, increased unc-40 embryonic lethality caused by Gex-like failures in morphogenesis to 26%. hum-7 was thus a candidate gene to act in parallel to unc-40 during morphogenesis.
HUM-7/Myo9 interacts genetically with axonal guidance receptors and is required during embryonic morphogenesis. (A) Axonal guidance receptors regulate the CED-10/Rac1 GTPase. Active CED-10 recruits the WAVE/SCAR complex to turn on Arp2/3, which promotes branched actin nucleation. Knockdown of any WAVE/SCAR pathway component leads to full embryonic lethality, while null mutations in upstream regulators cause partial embryonic lethality. To identify new morphogenesis regulators, we screened for enhancers of unc-40 embryonic lethality. (B) The effects of loss of HUM-7 on embryonic lethality in axonal guidance receptor mutants. Genetic null mutants of three axonal guidance receptors (UNC-40, SAX-3 and VAB-1) were crossed into dlg-1::gfp to make scoring easier and treated with hum-7 RNAi. (C) Postembryonic effects of HUM-7 on regulators of the WAVE/SCAR pathway. The mec-4::gfp(zdIs5) transgene is expressed in six mechanosensory neurons. Wild-type AVM axon migrates ventrally then anteriorly. Mutants show initial migrations in other directions, including a complete loss of ventral migration [sax-3(ky123)]. Micrographs show representative ventral migration patterns of AVM axon, reflected in the cartoons below (green cells). Black line, ventral nerve cord. The second neuron in images is the ALM, shown to orient the worm positioned anterior to the left. (D) Table summarizing ventral migration AVM defects. More than 200 neurons were analyzed for each genotype. (E) C. elegans HUM-7 shares similar domains and sequence homology with human MYO9A and MYO9B. All three proteins possess a Ras-associated (RA) domain, which is embedded in the ubitiquin (UBQ) domain of human myosin IXA, a myosin 9/IX domain and a RhoGAP domain. Both human MYO9 proteins have one phorbol ester/diacylglycerol-binding domain (C1), while HUM-7 has two. The percentage sequence identities between C. elegans HUM-7 and the human myosin IX proteins are listed between domains (within labeled domains) and overall (on the right). Error bars show 95% confidence intervals. *P<0.05, **P<0.001, ***P<0.0001. For B and D, statistical significance was determined by Welch's, or unequal variance, t-test. n>600 embryos. See also Table 1 and the Materials and Methods.
HUM-7/Myo9 interacts genetically with axonal guidance receptors and is required during embryonic morphogenesis. (A) Axonal guidance receptors regulate the CED-10/Rac1 GTPase. Active CED-10 recruits the WAVE/SCAR complex to turn on Arp2/3, which promotes branched actin nucleation. Knockdown of any WAVE/SCAR pathway component leads to full embryonic lethality, while null mutations in upstream regulators cause partial embryonic lethality. To identify new morphogenesis regulators, we screened for enhancers of unc-40 embryonic lethality. (B) The effects of loss of HUM-7 on embryonic lethality in axonal guidance receptor mutants. Genetic null mutants of three axonal guidance receptors (UNC-40, SAX-3 and VAB-1) were crossed into dlg-1::gfp to make scoring easier and treated with hum-7 RNAi. (C) Postembryonic effects of HUM-7 on regulators of the WAVE/SCAR pathway. The mec-4::gfp(zdIs5) transgene is expressed in six mechanosensory neurons. Wild-type AVM axon migrates ventrally then anteriorly. Mutants show initial migrations in other directions, including a complete loss of ventral migration [sax-3(ky123)]. Micrographs show representative ventral migration patterns of AVM axon, reflected in the cartoons below (green cells). Black line, ventral nerve cord. The second neuron in images is the ALM, shown to orient the worm positioned anterior to the left. (D) Table summarizing ventral migration AVM defects. More than 200 neurons were analyzed for each genotype. (E) C. elegans HUM-7 shares similar domains and sequence homology with human MYO9A and MYO9B. All three proteins possess a Ras-associated (RA) domain, which is embedded in the ubitiquin (UBQ) domain of human myosin IXA, a myosin 9/IX domain and a RhoGAP domain. Both human MYO9 proteins have one phorbol ester/diacylglycerol-binding domain (C1), while HUM-7 has two. The percentage sequence identities between C. elegans HUM-7 and the human myosin IX proteins are listed between domains (within labeled domains) and overall (on the right). Error bars show 95% confidence intervals. *P<0.05, **P<0.001, ***P<0.0001. For B and D, statistical significance was determined by Welch's, or unequal variance, t-test. n>600 embryos. See also Table 1 and the Materials and Methods.
HUM-7 functions in postembryonic axonal guidance and genetically interacts with WAVE/SCAR axonal guidance regulators in embryos and larvae
To test whether hum-7 affects other signals that regulate the WAVE-1 complex, we fed hum-7 RNAi to animals missing other axonal guidance receptors. Although loss of HUM-7 significantly increased the embryonic lethality caused by loss of UNC-40, it had no significant effect on the embryonic lethality caused by the vab-1(dx31) null mutation, and suppressed the embryonic lethality caused by the sax-3(ky123) null mutation from 44% to 29% (Fig. 1B). Therefore, loss of hum-7 had distinct effects on the three upstream signals that organize the cytoskeleton in developing embryos.
One human homolog of HUM-7, MYO9A, is mutated in patients with defects in the neuromuscular junction, leading to myasthenic syndrome (O'Connor et al., 2016). To test for HUM-7 postembryonic roles, particularly in neurons, we characterized the loss of HUM-7 in six mechanosensory neurons, including the AVM neuron, that are easily visualized using the neuronal transgene mec-4::gfp (Yu et al., 2002). In a wild-type background, the AVM neuron has a single round cell body and sends out one axonal projection that travels ventrally to the nerve cord and then anteriorly (Fig. 1C). RNAi depletion of HUM-7 in the mec-4::gfp transgenic strain caused significantly increased AVM ventral migration defects.
To determine whether hum-7 and the axonal guidance receptor mutants interact genetically during postembryonic development, we analyzed the effects of hum-7 RNAi on strains containing null alleles of the axonal guidance mutants and the mec-4::gfp transgene. The ventral migration defects caused by unc-40(n324) were enhanced from 50% to 62% by hum-7 RNAi. In contrast, sax-3(ky123) ventral migration defects were suppressed from 51% to 39% by hum-7 RNAi. There were no significant differences in ventral migration defects in vab-1(dx31) mutants fed hum-7 RNAi (Fig. 1C,D). Therefore, the postembryonic genetic interactions paralleled the embryonic genetic interactions, suggesting a conserved function for HUM-7 throughout development.
HUM-7 functions in embryonic morphogenesis
If HUM-7 is an important regulator of morphogenesis, it should have a phenotype on its own. Depleting hum-7 via RNAi in a wild-type (N2) background resulted in 6.5% embryonic lethality. The hum-7(ok3054) mutation generated by the C. elegans Gene Knockout Consortium is a 653 bp deletion spanning the 3′ end of the myosin domain and two IQ domains (binding sites for EF-hand proteins, including regulatory myosin light chains and calmodulin) (Bähler and Rhoads, 2002) (Fig. 2A). We outcrossed this strain three times, and detected 5.8% embryonic lethality. Because RNAi and the small deletion could mask the true loss-of-function phenotype, we generated an almost full-length deletion using CRISPR, hum-7(pj62), and observed similar phenotypes and slightly higher embryonic lethality of 9.5%. Approximately half of the dead embryos displayed a phenotype known as ‘gut on the exterior’ (Gex), which occurs 100% of the time when any gene in the WAVE complex is removed. In Gex embryos, epidermal ventral migration fails, resulting in exposed internal organs, such as the pharynx and intestine, and the embryo dies (Soto et al., 2002). Other dead embryos displayed 2-fold arrest, indicating a defect at the next morphogenesis step, elongation. All of these hum-7 phenotypes were observed in both hum-7 deletion mutants and in hum-7 RNAi embryos, in similar percentages (Fig. 2A-D, Table 1). This is similar to putative null mutations in upstream regulators of F-actin, like unc-40, which are only 14% embryonic lethal. Therefore, loss of hum-7 alone results in partially penetrant and strong morphogenesis phenotypes.
HUM-7/Myo9 interacts genetically and molecularly with Rho GTPases. (A) Molecular model of the C. elegans myosin IX protein, HUM-7, including N-terminal myosin, two IQ, two C1 and RhoGAP domains. Genetic mutations from the C. elegans Gene Knockout Consortium (ok3054 deletion) or from our CRISPR studies are indicated: pj62 deletion, pj63 GAP mutant and pj72 endogenous CRISPR N-terminal GFP tag, OX681 hum-7(pj72)[gfp::hum-7] (see Fig. 3). (B) Embryonic morphogenesis defects in hum-7 mutants observed with differential interference contrast (DIC) optics. Embryos are positioned with anterior to the left and dorsal up here and in other figures unless otherwise noted. Filled-head arrows indicate anterior intestine; line arrows indicate anterior pharynx. Some hum-7 mutants arrest during ventral enclosure (top row), similar to embryos missing ced-10/Rac1 or WAVE complex components like gex-3. Other hum-7 embryos arrest at the 2-fold stage (bottom row), similar to mutants that regulate RHO-1. Line arrows point to the swollen anterior region, a RHO-1 pathway phenotype. All embryos shown were grown at 23°C and mounted in egg salts (see Materials and Methods). Scale bar: 10 μm. (C) Embryonic lethality percentages in genetic doubles of hum-7 and axonal guidance mutants. At least 900 embryos were counted for each genotype. (D) Embryonic lethality percentages in genetic and RNAi doubles of hum-7 and GTPase mutants. We analyzed the three main C. elegans GTPases (RHO-1, CED-10 and CDC-42). We combined a hypomorphic allele of CED-10 [ced-10(n1993)] with hum-7 RNAi, RNAi of cdc-42 with the hum-7(ok3054) mutant allele, and a hypomorphic allele of a RHO-1 guanine exchange factor [ect-2(ax751)] with hum-7 RNAi. n>600 per genotype. See also Table 1. (E) The GAP domain of HUM-7 was tagged with GST and used in a GST pulldown assay to test binding between three His-tagged GTPases (RHO-1, CDC-42 and CED-10) in both the active GTP-bound (Q63L, Q61L and Q61L) and inactive GDP-bound (T19N, T17N and T17N) states. The pulldown assays were performed with double the concentrations of the loading controls shown. The GST binding assay was performed at least twice. Error bars show 95% confidence intervals. **P<0.001, ***P<0.0001, ****P<0.00001.
HUM-7/Myo9 interacts genetically and molecularly with Rho GTPases. (A) Molecular model of the C. elegans myosin IX protein, HUM-7, including N-terminal myosin, two IQ, two C1 and RhoGAP domains. Genetic mutations from the C. elegans Gene Knockout Consortium (ok3054 deletion) or from our CRISPR studies are indicated: pj62 deletion, pj63 GAP mutant and pj72 endogenous CRISPR N-terminal GFP tag, OX681 hum-7(pj72)[gfp::hum-7] (see Fig. 3). (B) Embryonic morphogenesis defects in hum-7 mutants observed with differential interference contrast (DIC) optics. Embryos are positioned with anterior to the left and dorsal up here and in other figures unless otherwise noted. Filled-head arrows indicate anterior intestine; line arrows indicate anterior pharynx. Some hum-7 mutants arrest during ventral enclosure (top row), similar to embryos missing ced-10/Rac1 or WAVE complex components like gex-3. Other hum-7 embryos arrest at the 2-fold stage (bottom row), similar to mutants that regulate RHO-1. Line arrows point to the swollen anterior region, a RHO-1 pathway phenotype. All embryos shown were grown at 23°C and mounted in egg salts (see Materials and Methods). Scale bar: 10 μm. (C) Embryonic lethality percentages in genetic doubles of hum-7 and axonal guidance mutants. At least 900 embryos were counted for each genotype. (D) Embryonic lethality percentages in genetic and RNAi doubles of hum-7 and GTPase mutants. We analyzed the three main C. elegans GTPases (RHO-1, CED-10 and CDC-42). We combined a hypomorphic allele of CED-10 [ced-10(n1993)] with hum-7 RNAi, RNAi of cdc-42 with the hum-7(ok3054) mutant allele, and a hypomorphic allele of a RHO-1 guanine exchange factor [ect-2(ax751)] with hum-7 RNAi. n>600 per genotype. See also Table 1. (E) The GAP domain of HUM-7 was tagged with GST and used in a GST pulldown assay to test binding between three His-tagged GTPases (RHO-1, CDC-42 and CED-10) in both the active GTP-bound (Q63L, Q61L and Q61L) and inactive GDP-bound (T19N, T17N and T17N) states. The pulldown assays were performed with double the concentrations of the loading controls shown. The GST binding assay was performed at least twice. Error bars show 95% confidence intervals. **P<0.001, ***P<0.0001, ****P<0.00001.
HUM-7 is expressed in muscles and regulated by SAX-3/ROBO. (A-D) GFP::HUM-7 expression in 3-fold embryos (A, top row), young adults (A, middle and bottom rows; B,C) and L4 larvae (D). (A) Beginning in late embryos, GFP::HUM-7 expression is enriched in the body wall muscles, as shown by overlap with Pmyo-3::RFP. Strong signal in the posterior bulb of the pharynx begins in late embryos. (B) HUM-7 expression in the striated muscle fibers was verified in the larvae by double labeling with anti-muscle antibody 5.6 (Miller et al., 1983) in young adults. (C) Comparison of expression to HMR-1::mKATE2. HMR-1/E-Cadherin is enriched at apical epithelia (top row, epidermal seam cells; bottom row, apical pharynx) and in the nerve ring (boxed region, anterior arrow). (D) Effects of mutations in axonal guidance proteins on levels of GFP::HUM-7. White arrows point to the posterior bulb of the pharynx; yellow dashed lines cross the body wall muscle. (E,F) Maximum intensity at the pharynx (E) and body wall muscles (F) was compared. Error bars show 95% confidence intervals. **P<0.001, ****P<0.00001.
HUM-7 is expressed in muscles and regulated by SAX-3/ROBO. (A-D) GFP::HUM-7 expression in 3-fold embryos (A, top row), young adults (A, middle and bottom rows; B,C) and L4 larvae (D). (A) Beginning in late embryos, GFP::HUM-7 expression is enriched in the body wall muscles, as shown by overlap with Pmyo-3::RFP. Strong signal in the posterior bulb of the pharynx begins in late embryos. (B) HUM-7 expression in the striated muscle fibers was verified in the larvae by double labeling with anti-muscle antibody 5.6 (Miller et al., 1983) in young adults. (C) Comparison of expression to HMR-1::mKATE2. HMR-1/E-Cadherin is enriched at apical epithelia (top row, epidermal seam cells; bottom row, apical pharynx) and in the nerve ring (boxed region, anterior arrow). (D) Effects of mutations in axonal guidance proteins on levels of GFP::HUM-7. White arrows point to the posterior bulb of the pharynx; yellow dashed lines cross the body wall muscle. (E,F) Maximum intensity at the pharynx (E) and body wall muscles (F) was compared. Error bars show 95% confidence intervals. **P<0.001, ****P<0.00001.
The hum-7 CRISPR alleles and Consortium deletion were used to further probe hum-7 genetic interactions with the axonal guidance genes. Testing the genetic interactions of hum-7 using the two deletions, ok3054 or pj62, gave similar results in combination with null mutations in unc-40 (increased lethality), sax-3 (suppressed lethality) or vab-1 (no change) as we saw with hum-7 RNAi (Fig. 2C). This suggests that RNAi causes strong loss of hum-7 function. We used CRISPR to generate a point mutation in the GAP of hum-7, pj63, and found different effects. This arginine to alanine (R to A) mutation, predicted to perturb the GTPase-activating function of RhoGAPs, resulted in increased lethality compared with the deletions (30%), and more than additive lethality in combination with unc-40 and vab-1. Unlike deletion alleles, the pj63 GAP mutant could not rescue sax-3 embryonic lethality (Fig. 2C). This result further supports an interaction between hum-7 and sax-3, and suggests that sax-3 might regulate HUM-7 activity.
HUM-7 functions as a GAP for RHO-1 and CDC-42, but not for CED-10/Rac1
HUM-7 has strong sequence homology with Myo9/myosin IX proteins from a variety of organisms, and includes the typical N-terminal myosin and C-terminal RhoGAP domains (Abouhamed et al., 2009; Chandhoke and Mooseker, 2012; Hanley et al., 2010; Omelchenko and Hall, 2012). HUM-7 has greater than 30% overall protein identity with human MYO9A and MYO9B (Fig. 1E). Although a C. elegans GAP that regulates the GTPase CED-10 during cell corpse engulfment has been identified (Neukomm et al., 2011, 2014), this GAP, SRGP-1, is dispensable for embryonic viability, and has limited morphogenesis phenotypes (Zaidel-Bar et al., 2010). We therefore tested whether HUM-7 functions as a GAP for one or more C. elegans GTPases.
Genetic tests of GAP function
We hypothesized that if HUM-7 acts like a GAP for CED-10/Rac1, then loss of hum-7 would rescue partial loss of CED-10/Rac1. However, although the ced-10(n1993) hypomorphic mutation resulted in 14% embryonic lethality, depletion of hum-7 via RNAi in ced-10(n1993) increased embryonic lethality to 34% (Table 1, Fig. 2D). An increase was seen with a second hypomorphic allele, ced-10(n3246) (Table 1). Thus, hum-7 and ced-10 appear to function in parallel pathways.
To test whether HUM-7 can function as a GAP for RHO-1/RhoA, we used a temperature-sensitive allele of the known RhoGEF, ect-2(ax751). This allele has been well characterized to partially knock down RHO-1 signaling (Zonies et al., 2010). At 23°C, ect-2(ax751) resulted in 61% embryonic lethality on control RNA, which dropped to 43% lethality in animals also depleted of hum-7 via RNAi. Similarly, ect-2 RNAi leads to almost 100% dead embryos, but this drops to 84% dead embryos if ect-2 RNAi is fed to hum-7 mutants (Fig. 2D, Table 1). Further, all of the ect-2 phenotypes, including poor differentiation that suggests ect-2 defects in the early embryo, were suppressed by the loss of hum-7 (Table 2).
To test whether HUM-7 can function as a GAP for CDC-42, we used RNAi depletion of cdc-42, which in wild-type animals resulted in embryonic lethality above 60%. In contrast, depletion of cdc-42 via RNAi in the putative null strain hum-7(ok3054) resulted in embryonic lethality of only 27% (Table 1, Fig. 2D). The fact that loss of hum-7 was able to suppress embryonic defects caused by partial loss of cdc-42 or rho-1 [using ect-2(ax751)] suggested that hum-7 acts in a pathway with rho-1 and cdc-42. Because there are no other reports of a myosin IX protein regulating cdc-42, we also tested whether loss of hum-7 could suppress the lethality of a known cdc-42 effector, wsp-1. Deletion of mutant wsp-1(gm324) results in 18.4% embryonic lethality, which drops to 11.4% when these animals are depleted of hum-7 via RNAi (Table 1).
Molecular tests of GAP function
To test whether the HUM-7 GAP domain can bind to the activated form of specific GTPases we performed GST pulldown assays. Purified His-tagged GTPases in active (GTP-loaded) and inactive (GDP-loaded) form were individually tested for binding to the GAP domain of HUM-7. The GAP domain of HUM-7 bound strongly to GTP-loaded RHO-1 and CDC-42 and failed to bind GTP-loaded CED-10. No binding was observed between HUM-7 GAP and any of the GDP-loaded constructs tested for all three GTPases (Fig. 2E). In short, HUM-7 behaves functionally, as well as molecularly, as a GAP for RHO-1 and CDC-42, but not CED-10.
HUM-7 is expressed in muscles
If HUM-7 is important for embryonic development, we would expect it to be expressed in embryos. We therefore created an N-terminally endogenously tagged allele of hum-7, pj72, via CRISPR. CRISPR-tagged gfp::hum-7 is faintly expressed in embryonic muscle, including the developing body wall muscles and pharynx (Fig. 3A). Expression in other embryonic tissues, like neuroblasts and epidermis, could not be determined with the CRISPR allele. Staining gfp::hum-7 adults with monoclonal antibody 5.6 (Miller et al., 1983), which recognizes body wall muscles, confirmed the muscle enrichment (Fig. 3B, enlarged boxed region).
The gfp::hum-7 CRISPR strain was crossed to other strains to analyze expression in other tissues in larvae and adults. A strain carrying gfp::hum-7 and hmr-1::mKate2 (Marston et al., 2016) allowed us to compare expression in the central nervous system, pharynx and epidermis (Fig. 3C). In the posterior pharyngeal bulb, gfp::hum-7 is expressed throughout, including the pm6 and pm7 cells that help the grinder contract during feeding (Wormatlas, www.wormatlas.org), whereas hmr-1::mKate2 is enriched only in apical regions, as expected. hmr-1::mKate2 is expressed throughout the nerve ring axons, while gfp::hum-7 is expressed in adjacent cells, perhaps glia. Viewing this same strain on the surface shows that although hmr-1::mKate2 is highly expressed in the seam cells, gfp::hum-7 shows no obvious overlap in the seam cells (Fig. 3C). A strain expressing RFP under control of the myosin heavy chain promoter, pmyo-3::rfp (Viveiros et al., 2011), shares expression with gfp::hum-7 in body wall muscles (Fig. 3A). Therefore, gfp::hum-7 appears to have broad expression, with enrichment in several types of muscle tissue, including regions of the pharynx, and in body wall muscles. Expression is not overall enhanced in neurons, but might include support cells for neurons. Expression does not appear to be enhanced in epidermal cells, although there is precedent for the epidermal signal to appear striped due to impingement from muscle or the pharynx (Wormatlas). Antibody staining with antibodies specific to body wall muscle support that at least some of the striped signals are in muscle (Fig. 3B).
sax-3 regulates levels of gfp::hum-7 in muscle
hum-7 genetic interactions with the axonal guidance receptors led us to test whether any of them might be upstream regulators of gfp::hum-7. Loss of unc-40 or vab-1 had limited effects on gfp::hum-7, but loss of sax-3 resulted in significantly elevated levels of gfp::hum-7 in expressing tissues. Measuring the effect on the posterior bulb, or body wall muscles, showed an increase of almost 100% (Fig. 3E). The genetic and molecular interactions both suggest that sax-3 is upstream of hum-7 and that SAX-3 signaling, directly or indirectly, results in lower HUM-7 expression.
HUM-7 affects F-actin dynamics during the initial cell migrations required for epiboly
Some embryonic phenotypes resulting from loss of HUM-7 resemble the Gex phenotype seen when members of the WAVE/SCAR complex are mutated (Figs 1A and 2B). Loss of WAVE/SCAR components during epidermal cell migration leads to defects in F-actin levels, ventral F-actin enrichment and actin dynamics in the migrating epidermis (Bernadskaya et al., 2012; Patel et al., 2008). We therefore tested HUM-7 effects on the actin cytoskeleton using two epidermal F-actin transgenic strains, plin-26::vab-10(ABD)::gfp(mcIs51) (Gally et al., 2009) and plin26::Lifeact::mCherry (Havrylenko et al., 2015). These strains, which express an actin-binding domain (ABD) (vab-10 ABD or Lifeact) under the lin-26 epidermal promoter, enabled us to perform live imaging of actin levels, ventral enrichment and dynamics in the epidermis of embryos during morphogenesis. In wild-type embryos, early in morphogenesis (250 min after the first cell division), F-actin becomes enriched at the ventral edge of the leading cells of the epidermis. When WAVE components are depleted via genetic mutation or RNAi, overall F-actin levels drop (Patel et al., 2008). In hum-7 mutants or hum-7(RNAi) embryos the overall levels of F-actin are higher than in wild type using the plin-26::vab-10 ABD strain, or plin26::Lifeact::mCherry strain (Fig. 4A-E). We conclude that the loss of hum-7 increases epidermal F-actin levels. As HUM-7 is not enriched in epidermis, these effects might be indirect.
HUM-7 affects F-actin levels and dynamics in migrating epidermis during embryonic morphogenesis. (A-H) F-actin levels in embryos during epidermal enclosure visualized by Plin-26::VAB-10-ABD::GFP (Gally et al., 2009) (A-C) and Plin-26::LIFEACT::mCherry (Havrylenko et al., 2015) (D-H). (A,B) Epidermal F-actin levels in hum-7(ok3054); plin-26::vab-10(ABD)::gfp. Embryos were imaged at 2 min intervals beginning at 240 min after first cleavage. Images shown here represent embryos at the early (250 min), middle (270 min) and late (285 min) stages of epidermal enclosure. In the yellow outline boxes, highest signal was measured in the leading cells. Embryos were pseudocolored using the ‘Fire’ function in ImageJ, and intensity is shown from low (blue) to high (yellow). See also Movie 1. (C) Ventral enrichment of F-actin in migrating epidermal cells in hum-7(ok3054) embryos. Close-ups of representative epidermal leading cells with the ventral (V) edge at the bottom and the dorsal (D) region at top. Intensity of F-actin in the leading cells during protrusion initiation was analyzed with a ventral to dorsal line through the leading cells (ImageJ ‘plot profile tool’). Two graphs use different y-axis scales due to higher F-actin levels in hum-7(ok3054). n=10 embryos per genotype. See also Movie 2. (D) plin-26::Lifeact::mCherry transgenic line (Havrylenko et al., 2015) in control animals and embryos depleted of hum-7, sax-3 or both. (E) The maximum F-actin intensity measured in ventral leading-edge cells 270 min after first cleavage using ImageJ ‘line tool’. (F-H) Effects of hum-7 and sax-3 on the rate of dynamic actin protrusions. Using movies shown in D, average F-actin protrusions or retractions per time point in two leading cells. (F) Representative images. Arrowheads mark protrusions and asterisks mark retractions. (G,H) Beginning at leading-edge migration, average protrusions (G) and retractions (H) during an 8 min period were measured. In F only, the exposure of individual images is enhanced to better show protrusions and retractions. See also Movie 2. Error bars show 95% confidence intervals. ****P<0.00001.
HUM-7 affects F-actin levels and dynamics in migrating epidermis during embryonic morphogenesis. (A-H) F-actin levels in embryos during epidermal enclosure visualized by Plin-26::VAB-10-ABD::GFP (Gally et al., 2009) (A-C) and Plin-26::LIFEACT::mCherry (Havrylenko et al., 2015) (D-H). (A,B) Epidermal F-actin levels in hum-7(ok3054); plin-26::vab-10(ABD)::gfp. Embryos were imaged at 2 min intervals beginning at 240 min after first cleavage. Images shown here represent embryos at the early (250 min), middle (270 min) and late (285 min) stages of epidermal enclosure. In the yellow outline boxes, highest signal was measured in the leading cells. Embryos were pseudocolored using the ‘Fire’ function in ImageJ, and intensity is shown from low (blue) to high (yellow). See also Movie 1. (C) Ventral enrichment of F-actin in migrating epidermal cells in hum-7(ok3054) embryos. Close-ups of representative epidermal leading cells with the ventral (V) edge at the bottom and the dorsal (D) region at top. Intensity of F-actin in the leading cells during protrusion initiation was analyzed with a ventral to dorsal line through the leading cells (ImageJ ‘plot profile tool’). Two graphs use different y-axis scales due to higher F-actin levels in hum-7(ok3054). n=10 embryos per genotype. See also Movie 2. (D) plin-26::Lifeact::mCherry transgenic line (Havrylenko et al., 2015) in control animals and embryos depleted of hum-7, sax-3 or both. (E) The maximum F-actin intensity measured in ventral leading-edge cells 270 min after first cleavage using ImageJ ‘line tool’. (F-H) Effects of hum-7 and sax-3 on the rate of dynamic actin protrusions. Using movies shown in D, average F-actin protrusions or retractions per time point in two leading cells. (F) Representative images. Arrowheads mark protrusions and asterisks mark retractions. (G,H) Beginning at leading-edge migration, average protrusions (G) and retractions (H) during an 8 min period were measured. In F only, the exposure of individual images is enhanced to better show protrusions and retractions. See also Movie 2. Error bars show 95% confidence intervals. ****P<0.00001.
In WAVE mutants, F-actin fails to enrich ventrally in the migrating cells and therefore fails to drive ventral enclosure (Bernadskaya et al., 2012; Patel et al., 2008). However, in hum-7 mutants, F-actin is correctly enriched ventrally in the leading cells (Fig. 4C). Therefore, hum-7 does not appear to regulate ventral enrichment.
We next compared actin dynamics in hum-7 mutants, using F-actin reporters. Previous work using the plin-26::vab-10 ABD transgene showed that wild-type ventral epidermal cells undergo frequent protrusions and retractions at the edge of the leading cells, whereas WAVE/SCAR mutants undergo protrusions and retractions less frequently (Bernadskaya et al., 2012; Patel et al., 2008). To assess the effects of hum-7 on actin protrusion dynamics, we imaged F-actin using plin26::Lifeact::mCherry, every 2 min from 240 min to 400 min after first cleavage. Wild-type embryos exhibited dynamic protrusions at the edge of the leading cells, averaging 1.0 protrusion from five time points sampled during 8 min of enclosure. hum-7 mutants averaged 1.6 protrusions per time point (Fig. 4F,G, Movies 1 and 2). Similar results were seen in movies made at 15 s intervals during the first 5 min of enclosure, which showed an increase from 1.1 to 1.6 protrusions per time point. Therefore, the rate of protrusion formation is greater in hum-7 mutants than in wild type, the opposite of the defect seen in WAVE mutants.
F-actin enrichment and dynamics were affected in sax-3 mutants, and this depended, at least partially, on hum-7. Previous work using plin-26::vab-10 ABD showed that sax-3(ky123) embryos have relatively wild-type levels of F-actin at the beginning of ventral enclosure, which becomes more disrupted as enclosure proceeds. The enclosure process was also delayed (Bernadskaya et al., 2012). sax-3(ky123) crossed into plin26::Lifeact::mCherry resulted in elevated F-actin during enclosure, whereas the double mutant with hum-7 resulted in lower levels (Fig. 4D,E). Average protrusions in the leading cells during enclosure were elevated in sax-3 mutants, and this number also dropped in the double mutant with hum-7 (Fig. 4F,G, from 2.2 to 1.6). Movies made at 15 s intervals showed the same pattern of average protrusions during enclosure, with elevated protrusions in hum-7 and sax-3 single mutants, and closer to wild type in the double mutant (1.1 in controls, compared with 1.6 and 1.4 in hum-7 and sax-3 single mutants, respectively, and 1.2 in the double mutant. The retraction rate in the 2 min and 15 s movies showed no significant change in hum-7 or sax-3 single mutants, or in double mutants (Fig. 4H). Therefore, the increased F-actin levels seen in the single mutants correlate with increased numbers of protrusions at the leading edge, while the drop in F-actin in the double mutants correlates with protrusions and retractions closer to wild type.
HUM-7 regulates the timing of epidermal morphogenesis
Mutations in WAVE/SCAR, or its upstream regulators, including UNC-40/DCC, SAX-3/ROBO and VAB-1/EphB, lead to delayed migrations (Bernadskaya et al., 2012). We compared the timing of enclosure in hum-7 mutants and found that, although a low percentage of embryos arrest partway through enclosure, embryos on average are not delayed, and some can enclose faster than wild type. For example, in one set of experiments, 1/7 embryos arrested, while 2/7 embryos enclosed significantly faster than wild type. In Movies 1 and 2, the hum-7(ok3054) embryo in the center achieves ventral enclosure faster than wild type, whereas the hum-7(ok3054) embryo on the right arrests during enclosure. We saw similar results in the other hum-7 mutants (pj62, 20% faster; pj63, 10% faster), when compared with wild-type embryos, timed from 240 min after first cleavage to ventral enclosure, by 300 min. Together, these results demonstrate a role for HUM-7 in the regulation of F-actin dynamics that contributes to the correct initiation of epidermal cell migration during epidermal enclosure and correct timing of morphogenetic events.
HUM-7 functions as a GAP upstream of RHO-1 to regulate embryonic morphogenesis
The role of the GTPases RHO-1/RhoA and CDC-42/Cdc42 in the ventral enclosure step of morphogenesis is only beginning to be described (Fotopoulos et al., 2013). In contrast, the role of RHO-1 in a later morphogenetic step, epidermal elongation, is exceedingly well studied (Diogon et al., 2007; Gally et al., 2009; Martin et al., 2014; Piekny et al., 2003, 2000). Because hum-7 behaved genetically and molecularly like a candidate GAP for RHO-1 (and CDC-42), we investigated whether hum-7 mutants have defects in RHO-1-dependent processes, like elongation. As shown in Fig. 2B, half of the dying hum-7 embryos show defects in elongation, including swelling of the anterior region. This defect is shown by known RHO-1 pathway mutants, including rga-2, a RHO-1 GAP that acts during elongation but not during ventral enclosure (Diogon et al., 2007) (Fig. 2B, thin arrows; Table 1).
If hum-7 is a GAP for RHO-1, loss of HUM-7 might suppress mutations that decrease Rho-dependent contractility, such as let-502/RHO-1, and enhance mutations that increase Rho-dependent contractility, such as mel-11/myosin phosphatase. These genetic interactions have been demonstrated for the Rho GAP rga-2 (Diogon et al., 2007). The let-502 ts allele, sb118ts, resulted in 75% or 98% lethality at 25°C or 25.5°C, respectively. However, this lethality dropped to 60% or 72%, respectively, when we crossed in hum-7(ok3054) (Table 1, Fig. 5A). In a wild-type background, mel-11 RNAi resulted in 36% lethality, which was enhanced to 60% in a hum-7 background (Table 1, Fig. 5A). Interestingly, our initial RNAi screen for enhancers of unc-40 embryonic lethality identified both hum-7 and mel-11. These results support that hum-7 might be a GAP for RHO-1.
NMY-2::GFP colocalizes with F-actin in epidermal cells during enclosure. (A) Schematic of embryos, ventral side up, during ventral enclosure. Magenta, epidermis; gray, neuroblasts. (B-E) To measure myosin in the epidermal cell focal plane, endogenously tagged nmy-2::gfp (myosin) was crossed into plin-26::Lifeact::mCherry (F-actin), which is expressed only in epidermis. Wild-type embryos incubated at 23°C are shown at three different z-planes, 1 µm apart, three time points after first cleavage. (B) 270 min: four leading-edge cells have almost met at the ventral midline. Scale bar: 10 μm. (B′-B‴) In the leading-edge cells, fluorescence intensity was measured: from ventral to dorsal in the middle of the cell, to compare ventral enrichment of NMY-2::GFP and F-actin (B′); from ventral to dorsal along the lateral cell boundary (B″); and across two cell-cell membranes of the same cell (B‴). The color of the surrounding box indicates the z-plane used. (C) 290 min: leading-edge cells have met at the ventral midline, pocket cells have not. (C′) Intensity measured across two pocket cells that have not met. (D) 300 min: pocket cells meet at the ventral midline. (E) Orthogonal view of the middle of the z-plane from D shows overlapping signals of myosin and epidermal F-actin.
NMY-2::GFP colocalizes with F-actin in epidermal cells during enclosure. (A) Schematic of embryos, ventral side up, during ventral enclosure. Magenta, epidermis; gray, neuroblasts. (B-E) To measure myosin in the epidermal cell focal plane, endogenously tagged nmy-2::gfp (myosin) was crossed into plin-26::Lifeact::mCherry (F-actin), which is expressed only in epidermis. Wild-type embryos incubated at 23°C are shown at three different z-planes, 1 µm apart, three time points after first cleavage. (B) 270 min: four leading-edge cells have almost met at the ventral midline. Scale bar: 10 μm. (B′-B‴) In the leading-edge cells, fluorescence intensity was measured: from ventral to dorsal in the middle of the cell, to compare ventral enrichment of NMY-2::GFP and F-actin (B′); from ventral to dorsal along the lateral cell boundary (B″); and across two cell-cell membranes of the same cell (B‴). The color of the surrounding box indicates the z-plane used. (C) 290 min: leading-edge cells have met at the ventral midline, pocket cells have not. (C′) Intensity measured across two pocket cells that have not met. (D) 300 min: pocket cells meet at the ventral midline. (E) Orthogonal view of the middle of the z-plane from D shows overlapping signals of myosin and epidermal F-actin.
The effects of RhoA signaling on cell migration might occur through the activation of actomyosin contractility controlled by non-muscle myosin, NMY-2, in the epidermis. Crossing nmy-2::gfp with plin-26::Lifeact::mCherry to visualize the epidermis revealed nmy-2::gfp puncta in the epidermal focal plane (Fig. 5), as previously shown (Wernike et al., 2016). During ventral enclosure, myosin puncta are enriched towards the leading edge of migrating cells, and at the midline as the cells meet (Fig. 5A,B). We depleted hum-7 via mutation or RNAi, and found brighter nmy-2::gfp puncta in all tissues of the embryo. To measure effects on ventral enclosure, we measured the robust nmy-2::gfp epidermal signal as the pocket cells met at the midline. Plotting the maximum intensity showed an increase of 20% in hum-7 mutants relative to controls (Fig. 6C).
SAX-3/HUM-7/RHO-1 regulate non-muscle myosin, NMY-2, in epidermal cells. (A) Genetic interactions of hum-7, sax-3, vab-1 and known RHO-1 pathway mutants let-502 and mel-11. Experiments shown were performed at 23°C, except for the let-502(sb118ts) experiments, which were performed at 25.5°C. (B) Quantitation of NMY-2::GFP levels in the ventral pocket cells at 300 min, when pocket cells of control embryos meet at the midline. Graph shows mean maximal fluorescent intensity at pocket cells, relative to wild type or relative to control RNAi, normalized to 1. (C) Representative images of embryos used in B. The yellow outline boxes indicate the region measured. Scale bars: 10 μm. (D) SAX-3 is proposed to negatively regulate HUM-7, which functions as a GAP to attenuate RHO-1 activity and thus reduce actomyosin contractility through NMY-2, non-muscle myosin heavy chain 2. We propose that tight regulation of NMY-2-dependent actomyosin contractility promotes timely ventral enclosure.
SAX-3/HUM-7/RHO-1 regulate non-muscle myosin, NMY-2, in epidermal cells. (A) Genetic interactions of hum-7, sax-3, vab-1 and known RHO-1 pathway mutants let-502 and mel-11. Experiments shown were performed at 23°C, except for the let-502(sb118ts) experiments, which were performed at 25.5°C. (B) Quantitation of NMY-2::GFP levels in the ventral pocket cells at 300 min, when pocket cells of control embryos meet at the midline. Graph shows mean maximal fluorescent intensity at pocket cells, relative to wild type or relative to control RNAi, normalized to 1. (C) Representative images of embryos used in B. The yellow outline boxes indicate the region measured. Scale bars: 10 μm. (D) SAX-3 is proposed to negatively regulate HUM-7, which functions as a GAP to attenuate RHO-1 activity and thus reduce actomyosin contractility through NMY-2, non-muscle myosin heavy chain 2. We propose that tight regulation of NMY-2-dependent actomyosin contractility promotes timely ventral enclosure.
We tested how two proposed upstream regulators of hum-7 affected nmy-2::gfp levels during ventral pocket cell meeting. A null mutation in sax-3, ky123, led to a 20% decrease in the pocket cell signal, whereas a null mutation in vab-1, dx31, led to 20% increase (Fig. 6B,C). These results suggest that both SAX-3 and VAB-1 regulate actomyosin contractility in the migrating epidermal cells. To test whether hum-7 is required for these changes, we removed hum-7 via RNAi in each strain. In the vab-1(dx31); hum-7 RNAi strain, the myosin levels are somewhere in the middle, closer to wild type, which suggests that the effects of VAB-1 on HUM-7 might be indirect. In contrast, the levels of nmy-2::gfp in the sax-3(ky123); hum-7(RNAi) strain resembled the levels of hum-7(RNAi) (elevated), further supporting that SAX-3 acts through HUM-7.
DISCUSSION
HUM-7, a new component in the RHO-1 pathway that appears to attenuate RHO-1 signaling
The analysis of Rho GAPs has been complicated by their sheer number. Even in C. elegans, there are 23 proteins with GAP domains. Our analysis of one of these 23 GAP proteins has revealed exciting connections between axonal guidance receptors, the known players in RHO-1 signaling, and the movements of the epidermis during ventral enclosure. Our findings support a requirement for RHO-1 and CDC-42 attenuation by the previously uncharacterized RhoGAP, HUM-7/Myo9. hum-7 mutants have low penetrance ventral enclosure defects, coupled with highly penetrant increased actin dynamics, suggesting that increased actin dynamics in epidermal protrusions interferes with the normal processes that accompany cellular migration. In single cell migration systems, downregulation of actomyosin contractility is required to allow persistent migration in one direction (Theisen et al., 2012). A similar event might be occurring in the migration of single axons, as we show here (Fig. 1) that loss of HUM-7 alters neuronal migration. One interpretation of our results is that in a sheet migration of the embryonic epidermis, increased actomyosin contractility interferes with persistent migration. Evidence in several systems suggests that increased F-actin structures of one type can interfere with F-actin structures of another type (reviewed in Suarez and Kovar, 2016). Because HUM-7 is a candidate GAP for both RHO-1 and CDC-42, loss of HUM-7 might lead to increased formin assembly that would decrease the G-actin available for branched actin assembly by Arp2/3. Our movies of hum-7 mutants show higher F-actin formation and increased myosin, but they do not distinguish which form of F-actin is forming (linear or branched). Therefore, HUM-7 might attenuate activity levels of RHO-1 and CDC-42 so that RAC-1-based F-actin structures form correctly.
Axonal guidance receptors regulate the RHO-1 pathway
Our results suggest that HUM-7 plays an important role coordinating the response to external polarity cues by modulating RHO-1 activity to permit the correct levels and dynamics of actin required for proper cellular movements. The role of HUM-7 in embryonic morphogenesis was discovered through a screen for enhancers of the partially penetrant morphogenesis defects caused by loss of an axonal guidance receptor, UNC-40/DCC (Fig. 1). hum-7 showed distinct genetic interactions with each of three axonal guidance pathway receptors. Particularly intriguing was the rescue of embryonic lethality for sax-3(ky123) null animals, from 44% to 29%, when hum-7 is removed via RNAi or genetic mutations (Figs 1B and 2C). Because HUM-7 is proposed to regulate the RHO-1 pathway, we tested whether the axonal guidance mutants affected embryonic non-muscle myosin, using the nmy-2::gfp CRISPR strain. Loss of SAX-3/ROBO alters NMY-2::GFP levels, and this is epistatic to hum-7, suggesting that sax-3 and hum-7 are in a genetic pathway that regulates NMY-2. The connection between SAX-3 and HUM-7 is further supported by the fact that loss of sax-3 results in elevated expression of GFP::HUM-7 levels measured using a CRISPR line (Fig. 3E). Our results support that SAX-3 is a positive regulator of RHO-1 signaling during the migrations of epidermal enclosure, and that SAX-3 acts, at least in part, through negative regulation of the RhoGAP HUM-7 (Fig. 5E). Because loss of hum-7 does not fully rescue loss of SAX-3, it is clear that SAX-3 has other targets during ventral enclosure. Although loss of hum-7 suppressed sax-3 phenotypes (Fig. 4D-H), just as it partially suppressed sax-3 lethality (Fig. 2C), the double mutant phenotype was different from hum-7 alone. This suggested that sax-3 has other targets besides hum-7, and that these targets promote F-actin in the epidermis (Fig. 4D,E).
It is possible that VAB-1/EphB also regulates HUM-7, to regulate the RHO-1 pathway. Overall, loss of vab-1 has opposite effects to loss of sax-3. For example, in vab-1 mutants, levels of nmy-2::gfp increase, whereas in sax-3 mutants they decrease (Fig. 5B). In support of this, loss of vab-1 enhances loss of MEL-11, just like loss of hum-7, and similarly increased levels of nmy-2::gfp. However, the effects of VAB-1 on the RHO-1 pathway might not be through HUM-7. Loss of vab-1 did not significantly affect gfp::hum-7 levels (Fig. 3D,E).
Axonal guidance receptors regulate the RHO-1 pathway in epidermis
During ventral enclosure, rho-1 and nmy-2 are needed in the underlying neuroblasts (Wernike et al., 2016). Analysis of the pattern of nmy-2::gfp in combination with transgenes only expressed in the epidermis confirms that there are important nmy-2::gfp puncta that form during ventral enclosure (Fig. 5). Our findings that these epidermal nmy-2::gfp puncta are altered in embryos missing the axonal guidance receptors, SAX-3/ROBO and VAB-1/EphB, supports an important role for these proteins in guiding the epidermal migrations directly or indirectly (Fig. 6). Although it is accepted that SAX-3 can rescue epidermal migrations when expressed only in the epidermis or the underlying neuroblasts (Ghenea et al., 2005), how VAB-1 influences epidermal migrations is more controversial, because only a subset of epidermal cells is proposed to express VAB-1 (Ikegami et al., 2012). In this context, it is intriguing that the strongest HUM-7 expression is not in the epidermis or neuroblasts, but in the underlying muscle cells (Fig. 3). As we have not tested where SAX-3 and HUM-7 are required for the regulation of epidermal morphogenesis, the effects of SAX-3 on HUM-7, and the effects of both on the epidermis, might be indirect.
The HUM-7 and RHO-1 pathway affects actin dynamics
SAX-3/ROBO regulation of HUM-7/Myo9 to regulate RHO-1/RhoA is conserved from human cells (lung cancer) to C. elegans. The human homologs of HUM-7, MYO9A and MYO9B, are proposed to regulate cellular behaviors as GAPs that act on the RhoA GTPase. Many of the proposed phenotypes can be explained as resulting from excess RhoA signaling. What is not as clear is what are the downstream effects of excess RhoA signaling, and what signals regulate this unusual RhoGAP with a processive myosin motor. We used a complex tissue migration, ventral enclosure, to measure several aspects of RhoA signaling and effects on migration dynamics. As the first analysis of Myo9 function in C. elegans, we have uncovered many features of morphogenesis that depend on Myo9, and have begun to place them in signaling pathways.
Axonal guidance in C. elegans depends on HUM-7/Myo9, which appears to receive signals through SAX-3/ROBO, and possibly VAB-1/EphB. These interactions are conserved in embryos, where they regulate epidermal cell migrations. Signals from axonal guidance receptors alter non-muscle myosin expression in the migrating epidermal cells. The pathway we uncover here, from SAX-3/ROBO to HUM-7/Myo9, to modulate RHO-1/RhoA (Fig. 5E), has been proposed to function in human lung cancer tissue culture cells. The altered timing of the migrations in hum-7 mutant embryos suggests one consequence of misregulated RhoA: overactive RhoA leads to excess protrusions at the leading edge, without matching increased retractions. This change can create various problems, including overly stable protrusions, owing to decreased turnover of RhoA GTPase. Although this can lead to occasional faster migrations, it can also make the cells less responsive to their environment. For migrating epidermal cells, this could reduce detection of tension in the underlying neuroblasts thought to help guide the migrations. The defects in hum-7 mutants might combine failure to properly transmit signals within epidermal cells with failure to detect signals from neighboring tissues. Tissue-specific rescue experiments of hum-7 to resolve this would be challenging, because the overall lethality is low. However, future experiments will need to address the tissue specificity of the signals, and of HUM-7 function in receiving these signals, to address how the epidermis and neuroblasts cooperate during this complex migration.
MATERIALS AND METHODS
Strains
The following strains were used in this study: MT324, unc-40(n324); CX3198, sax-3(ky123); CZ337, vab-1(dx31); FT48, xnIs16[dlg-1::gfp]; VC2436, hum-7(ok3054); OX615, hum-7(ok3054); dlg-1::gfp; WM43, gex-3(zu196)/DnT1; MT5013, ced-10(n1993); MT9958, ced-10(n3246); JH2754, ect-2(ax751); OX646, hum-7(ok3054); ect-2(ax751); JJ1473, zuIs45[nmy-2::gfp]; OX630, hum-7(ok3054); nmy-2::gfp; ML1154, mcIs51[plin26::vab-10 ABD::gfp]; OX595, hum-7(ok3054); plin26::vab-10 ABD::gfp; SK4005, zdIs5[mec-4::gfp]; OX350, unc-40(n324); mec-4::gfp; OX213, sax-3(ky123); mec-4::gfp; IC136, vab-1(dx31); mec-4::gfp; HR1157, let-502(sb118ts); OX644, hum-7(ok3054) let-502(sb118ts); OX635, hmp-1(fe4); hum-7(ok3054); OX681, gfp::hum-7; OX714 hum-7(pj62); OX715, hum-7(pj63); ML773, rga-2(hd102)/hIn1 [unc-54(h1040)]; OX645, hum-7(ok3054) rga-2(hd102)/hIn1 [unc-54(h1040)]; and HR483, mel-11(sb56) unc-4(e120)/mnC1[dpy-10(e128) unc-52(e444)].
In vitro binding assay
The in vitro binding assay was based on Neukomm et al. (2011). GTP-loaded and GDP-loaded GTPase constructs for CDC-42, CED-10 and RHO-1 were gifts from the Hengartner laboratory (Institute of Molecular Life Sciences, University of Zurich, Switzerland). The HUM-7 GAP domain construct (amino acids 1540-1727) was cloned into the pGEX-4T2 GST vector. All the GTPase constructs were His tagged and purified using His-Bind resin (Novagen). The GST-tagged HUM-7 construct was purified using Glutathione Sepharose 4B beads (GE Healthcare) based on the manufacturer’s instructions. For the pulldown assay, 30 μg purified GST-tagged proteins were incubated with 10 μg His-tagged proteins at 4°C for 2 h. Pulldowns were performed with GST beads, proteins were separated on 12% acrylamide gels, and blots were probed with antibody against His (Millipore). The GST binding assay was performed at least twice using the same batch of GST-HUM-7-GAP for all of the pulldowns, and at least two sets of HIS-GTPase lysates.
RNAi
All RNAi bacterial strains used in this study were administered by the feeding protocol as in Bernadskaya et al. (2012). RNAi strains were constructed by cloning cDNA of the genes into the L4440 vector and transforming them into HT115 competent cells. Small overnight bacterial cultures were diluted 1:250 and grown until the optical density at a wavelength of 600 nm was close to 1. The culture was pelleted and resuspended in LB medium containing 100 µg/ml ampicillin and 1 mM isopropyl β-D-1-thiogalactopyranoside (IPTG).
Embryonic lethality counts and imaging
Synchronized L1 worms were plated on AMP/IPTG plates containing the appropriate RNAi bacteria. Plates were cultured at 23°C for 3 days. Temperature-sensitive mutants, such as ect-2(ax751), were cultured at three different temperatures, 15°C, 20°C and 23°C. After the 3-day incubation, embryos were mounted on 3% agarose pads and lethality was counted. Embryos shown in images and graphs were cultured at 23°C unless stated otherwise.
Neuronal migration analysis
For mec-4::gfp neuronal analysis, synchronized L1 worms were plated on AMP/IPTG plates containing the appropriate RNAi bacteria and cultured for 2-3 days at 20°C. L4440 empty vector was used as the control RNAi. The effectiveness of hum-7 RNAi was monitored by counting the percentage embryonic lethality. After incubation, L4 worms were mounted onto 3% agarose pads with 10 mM levamisole to prevent frequent movement and scored for phenotypes within 15 min of mounting. The AVM mechanosensory neuron was checked for ventral migration defects. The table in Fig. 1D is based on more than 200 worms per genotype.
RNAi-feeding screen
To generate liquid cultures of the RNAi genes, 150 μl of LB broth containing the appropriate antibiotic was pipetted into 96-well plates. Using a 96-well pin, frozen glycerol stocks of RNAi genes were transferred to the liquid medium. The culture was inoculated overnight at 37°C with shaking. Cultures were spotted on 24-well NGM plates containing 2% lactose and 25 μg/ml carbenicillin and grown overnight at room temperature. Approximately 20 L1 worms with the unc-40(n324) genotype were seeded onto the 24-well plates containing the RNAi bacteria. Worms were grown at 23°C for ∼3 days and then their progeny screened for embryonic lethality. Clones that produced greater than 15% lethality in combination with unc-40(n324) were selected for secondary screening. We were particularly interested in selecting clones that had good differentiation and phenotypes that resemble Gex.
Live imaging
Some late-stage hum-7 embryos (<2%) showed a subtle phenotype in which they developed correctly, but flattened when mounted on pads for imaging. This flat appearance could be related to the fact that hum-7 embryos behave osmotically sensitively, and have to be mounted in egg salts instead of water to prevent arrest during live-imaging studies. Thus, all embryos shown in this study were mounted in egg salts. For all live imaging shown, embryos at the two- to four-cell stage were dissected from adult hermaphrodites and mounted onto 3% agarose pads, covered with #1.5 cover slips and sealed with Vaseline. Embryos were then incubated at 23°C for 240 min. Imaging was performed in a temperature-controlled room set to 23°C on a laser spinning disk confocal microscope with a Yokogawa scan head, on a Zeiss AxioImager Z1 microscope using the Plan-Apo 63×/1.4 NA or Plan-Apo 40×/1.3 NA oil lenses. Images were captured on a Photometrics Evolve 512 EMCCD Camera using MetaMorph software (BioVision Technologies), and analyzed using ImageJ (National Institutes of Health). Using the ImageJ software, 4D movies were constructed by projecting Z-stacks at maximum intensity. All measurements were performed on raw data. For actin and myosin measurements, background intensity was subtracted using a box or line of the same size and measuring average intensity in the same focal plane, near the embryo.
Actin intensity measurements
To measure actin levels and dynamics, we performed live imaging of strains plin-26::vab-10(ABD)::gfp (Fig. 4A) and plin-26::Lifeact::mCherry (Fig. 4D). Following the 240 min incubation of two- to four-cell embryos, embryos were imaged at 2 min intervals for at least 120 min.
F-actin levels were calculated in the leading cells at three different time points in minutes after first cleavage: 250/early, when migrations begin; 260/middle, when migration is underway; and 280/late, when the ventral cells meet (Fig. 4A). Fig. 4D shows the middle time point to allow comparison of actin levels between wild type, hum-7 mutant and the sax-3 mutant. To compare actin levels in the leading cells, ImageJ ‘box tool’ was used to construct a rectangle (dotted box in figure) around the ventral edge of the leading cells (box size) and the maximum intensity was recorded.
Quantitation of F-actin intensity
Using GraphPad Prism, the maximum intensity for each analyzed embryo was graphed using a scatter plot displaying the mean with 95% confidence intervals (Fig. 4E).
Actin protrusion analysis
We measured actin dynamics by evaluating protrusions and retractions in the two leading cells during early enclosure on one side of the embryo. Time ‘0’ corresponds to the first appearance of epidermal pocket cell protrusions, at ∼250 min after first cleavage in wild type. Because this is delayed in mutants like sax-3, timing based on the early pocket cell protrusion aligns the actin dynamics developmentally (Bernadskaya et al., 2012). For the 2 min interval movies, we examined five time points during the first 8 min of enclosure. For the 15 s interval movies, we examined 20 time points during the first 5 min of enclosure. The average number of protrusions or retractions seen in five time points (2 min movies) or 20 time points (15 s movies) is shown or mentioned in the Results. For the 2 min movies, n=10, n=10, n=8 and n=8 for controls, hum-7, sax-3 and the double mutant, respectively. For the 15 s movies, n=4, n=13, n=6 and n=4 for controls, hum-7, sax-3 and the double mutant, respectively. Protrusion and retraction rates calculated from the 2 min and 15 s movies were graphed using a scatter plot displaying the mean with 95% confidence intervals (4E-H summarizes 2 min data, while 15 ms data is mentioned only in the text).
Imaging for plin-26::LifeAct::mCherry;nmy-2::gfp strains
Live imaging was performed at 40× magnification using a 1.3 NA oil immersion objective lens. Imaging was performed in a temperature-controlled room that was set at 23°C. Embryos were exposed at 30% laser power for 100 ms for GFP filter every 10 min, and at 35% laser power for 100 ms for mCherry filter every 2 min. Z-stacks were made across the entire embryo using a motorized z-stage, ∼27-32 z-stacks total, at 1 µm intervals. Controls and mutants were imaged within 3 days, under the same imaging condition for each dataset to account for technical variability.
Myosin measurements
To compare myosin levels as pocket cells meet, a rectangular box enclosing the pocket cells as they first touch (∼320 min) was drawn (yellow boxes in Fig. 6C), Time points shown and measured are those when the epidermal cells first touch, based on epidermal F-actin signal. To measure myosin puncta on the same focal plane as the epidermis, we co-localized the highest Plin-26::LIFEACT::mCherry intensity with NMY-2::GFP and recorded the highest of three measurements per embryo (Fig. 6B). Maximum intensity values were recorded after subtracting the average background fluorescence. The graph in Fig. 6B records the relative level of NMY-2::GFP, after normalization of either wild type, or wild type on control RNAi, to 1.
Statistical analysis
For grouped data (Figs 3E,F, 4E-H and 6B), statistical significance was established by performing a two-way analysis of variance (ANOVA) followed by the Bonferroni multiple comparison post test. For ungrouped data (Figs 1B,D, 2C,D and 6A) an unpaired t-test, the unequal variance (Welch) t-test, was used. Error bars show 95% confidence intervals. *P<0.05, **P<0.001, ***P<0.0001.
Acknowledgements
We thank the NCRR-funded Caenorhabditis Genetics Center, Jeremy Nance, Paul Mains and Ian Chin-Sang for strains; Lucas Neukomm for GTPase constructs; Rutgers undergraduate Vipin Palukuri for technical assistance; Chris Rongo, Monica Driscoll and Andy Singson for helpful comments on the manuscript; and Alisa Piekny for helpful discussions. We thank the National Institutes of Health Shared Instrumentation Program (S10OD010572) for the Spinning Disk used in this research.
Footnotes
Author contributions
Conceptualization: A.G.W., M.C.S.; Methodology: H.R., M.C.S.; Validation: M.C.S.; Formal analysis: A.G.W., H.R., M.C.S.; Investigation: A.G.W., H.R., J.C., M.C.S.; Resources: A.G.W.; Writing - original draft: A.G.W., M.C.S.; Writing - review & editing: A.G.W., H.R., M.C.S.; Supervision: A.G.W., M.C.S.; Funding acquisition: M.C.S.
Funding
This work was supported by the National Institute of General Medical Sciences [R01GM081670 to M.C.S. and an Institutional Research and Academic Career Development Award (K12GM093854) to A.W.]. Deposited in PMC for release after 12 months.
References
Competing interests
The authors declare no competing or financial interests.