A reduction‐of‐function mutation in ect‐2 was isolated as a suppressor of the Multivulva phenotype of a lin‐31 mutation. Analysis using markers indicates that this mutation causes defects in both the cytokinesis and migration of epidermal P cells, phenotypes similar to those caused by expressing a rho‐1 dominant‐negative construct. ect‐2 encodes the Caenorhabditis elegans orthologue of the mouse Ect2 and Drosophila Pebble that function as guanine nucleotide exchange factors (GEFs) for Rho GTPases. The ect‐2∷GFP reporter is expressed in embryonic cells and P cells. RNA interference of ect‐2 causes sterility and embryonic lethality, in addition to the P‐cell defects. We have determined the lesions of two ect‐2 alleles, and described their differences in phenotypes in specific tissues. We propose a model in which ECT‐2GEF not only activates RHO‐1 for P‐cell cytokinesis, but also collaborates with UNC‐73GEF and at least two Rac proteins to regulate P‐cell migration.
The small GTPases of the Rho family have roles in fundamental cellular functions such as cytokinesis, migration and polarization (Schmidt & Hall, 2002). Molecular and genetic studies have shown that the Drosophila RhoGEF Pebble (PBL) and its mammalian orthologue, the proto‐oncogene product ECT2, are necessary for cytokinesis (Miki et al, 1993; Prokopenko et al, 1999; Tatsumoto et al, 1999). This function is mediated by PBL binding to Rho1, but not to Rac1 or Cdc42. More recently, PBL has been shown to genetically and physically interact with the Rho family GTPase‐activating protein (GAP), RacGAP50C (Somers & Saint, 2003). In Caenorhabditis elegans, LET‐502/Rho‐binding kinase has been shown to regulate embryonic elongation (Wissmann et al, 1997). A weak loss‐of‐function (lf) allele of ect‐2 (also known as let‐21), which encodes the likely PBL/ECT2 orthologue, has been shown to enhance the cytokinesis defect during the first embryonic cell division caused by RNA interference (RNAi) of the par‐2 gene (Dechant & Glotzer, 2003), suggesting a potential role for ect‐2 in mitosis.
Rho and Rac GTPases have crucial roles during the development of epidermal P cells. At hatching, 12 P cells are present on the sublateral sides of the larvae. In the mid‐first larval (L1) stage, these cells migrate to the ventral midline. After migration, each P cell divides along the anterior–posterior axis to produce two daughter cells, Pn.a and Pn.p. The Pn.a cells continue to divide for up to three rounds to generate motor neurons, whereas the Pn.p cells of hermaphrodites will later either be induced to adopt a vulval cell fate or fuse to the surrounding epidermal syncytium (Sulston, 1976). The RHO‐1 GTPase appears to be required for both P‐cell migration and cytokinesis, as these processes are disrupted in animals that carry a dominant‐negative (dn) Rho mutant transgene (Spencer et al, 2001). Consequently, the Pn.a and Pn.p cells are missing in these animals and they are therefore defective in vulval formation and movement. Two Rac homologues, MIG‐2 and CED‐10, have been shown to act redundantly to regulate P‐cell migration, but not cytokinesis (Spencer et al, 2001; Wu et al, 2002). These Rac proteins, along with a third Rac homologue, RAC‐2, also act redundantly in the migration, axon guidance and morphogenesis of several other cells (Zipkin et al, 1997; Lundquist et al, 2001; Kishore & Sundaram, 2002; Wu et al, 2002). These studies indicate that Rho and Rac small GTPases collaborate to regulate P‐cell migration, whereas Rho also functions in P‐cell cytokinesis. Additionally, unc‐73, which encodes a Trio‐like guanine nucleotide exchange factor (GEF) protein, also has a role in axon guidance and cell migration (Zipkin et al, 1997; Steven et al, 1998; Lundquist et al, 2001; Spencer et al, 2001; Kishore & Sundaram, 2002; Wu et al, 2002). Although UNC‐73 has two GEF domains that have been shown to catalyse the exchange reaction for both Rho and Rac in vitro, unc‐73(lf) mutations do not disrupt P‐cell cytokinesis (Spencer et al, 2001). Therefore, another GEF protein may interact with RHO‐1 for the cytokinesis function. In this report, we describe our analysis of the ect‐2/RHO GEF functions in P‐cell development.
Results and Discussion
Isolation of ku427 as a suppressor of a lin‐31 mutant
We isolated the mutation ku427 as a suppressor of the Multivulva (Muv) phenotype of a lin‐31 allele (Fig 1; see the supplementary information online for details). The suppression of the Muv phenotype was probably owing to a defect in generating epidermal P cells, as, under Nomarski optics, we observed that Pn.p cells were often missing from the ventral midline in ku427 animals. Quantitative analysis indicated that the number of Pn.p cells is decreased from 11 to 7.1 in ku427 (Table 1). Consistent with this result, the ku427 mutant had an egg‐laying‐defective (Egl) phenotype (80%, n=276). In addition, these animals were uncoordinated (Unc), which suggested that Pn.a cells were also missing or defective.
ku427 is an allele of ect‐2 RhoGEF
By single nucleotide polymorphism mapping and microinjection transformation, we determined that the ect‐2 (T19E10.1) gene could rescue the ku427 mutation (see the supplementary information online for details). RNAi of ect‐2 phenocopied ku427 in suppressing the Muv phenotype of lin‐31(lf) as well as causing P‐cell defects (Table 1; data not shown). We then identified a lesion in the promoter region of ect‐2 (−155G to A) in ku427 DNA (Fig 1A,B). ku427 also failed to complement the existing ect‐2 allele e1778, which we determined as having an in‐frame deletion in ect‐2 (Fig 1A,C). These results indicate that ku427 is an lf allele of ect‐2, which is defined by the sterile mutations e1778 and oz204 (Riddle et al 1997; Dechant & Glotzer, 2003).
ect‐2 encodes a likely orthologue of the mouse proto‐oncogene product ECT2 and Drosophila Pebble (Fig 1C), both of which are GEF proteins for Rho family GTPases (Miki et al, 1993; Prokopenko et al, 1999). Similarly to ECT2 and Pebble, ECT‐2 has BRCT1 and BRCT2 domains in its amino‐terminal half and a GEF domain in its carboxy‐terminal half. The GEF domain contains a Dbl homology (DH) domain and a pleckstrin homology (PH) domain. BRCT domains are thought to be important for cell‐cycle regulation, whereas the DH/PH domain interacts with Rho family proteins and catalyses the GDP to GTP exchange reaction (Prokopenko et al, 1999). In addition, ECT‐2 has two putative nuclear localization signals (amino acids 9–16 and 361–367) in the N‐terminal and middle regions.
Because ku427 has a lesion in the promoter region, we compared the sequence of the promoter region of ect‐2 with that in another Caenorhabditis species, C. briggsae. This region is highly conserved between the two species and contains the FOXO‐binding consensus site (Fig 1B) (Pierrou et al, 1994). We then used quantitative reverse transcription–PCR to determine whether the mutation results in a decrease in the transcription level of ect‐2. We observed that the level of ect‐2 in the ku427 mutant was sevenfold lower than that in wild type (Fig 1D). We also failed to detect green fluorescent protein (GFP) fluorescence from an ect‐2∷GFP reporter transgene that contained this lesion (data not shown).
ect‐2∷GFP is strongly expressed in P cells
To investigate the expression patterns of ect‐2 in C. elegans, we made transgenic worms containing a transcriptional ect‐2∷GFP fusion construct (pKM37 in Fig 1A). In these worms, GFP expression was detected from embryogenesis to adulthood. In embryogenesis, almost all cells, other than those in the gut lineage, showed GFP expression (Fig 2A). Before P‐cell migration in the L1 stage, expression was detected only in Q cells (Fig 2B). After P‐cell migration and division, expression was seen in the P‐cell derivatives (Fig 2C) and distal tip cells (data not shown). In L3, GFP expression was present only in the vulval precursor cells and their derivatives (Fig 2D). In L4 and adulthood, GFP fluorescence was detected only in some neuronal cells (data not shown).
ect‐2 is required for cytokinesis in P cells
ect‐2(ku427) and ect‐2(RNAi) caused the absence of Pn.p cells (Table 1). To determine the cause of this phenotype, we used unc‐47∷GFP to mark GABAnergic neurons (Fig 3C), as 13 of the 26 GABAnergic neurons are derived from P cells (Pn.a cells) (McIntire et al, 1997), and followed the P‐cell lineage in ku427 mutants. We observed that in ect‐2(ku427) and ect‐2(RNAi) animals, although nuclear divisions of P cells seemed to be completed at the late larval stage, cytokinesis seemed to be disrupted (Fig 3; Table 1). Owing to the likely failed cytokinesis of some P cells (and also Pn.a derivatives) in ku427 mutants, multinuclear cells were formed, as visualized by UNC‐47∷GFP fluorescence (Fig 3E). The Pn.p nuclei were observed in these cells, but the stereotypical Pn.p nuclear morphology was no longer seen. We previously showed that, when expressed under an epidermis‐specific promoter, a rho‐1(T19N dn) transgene caused the same cytokinesis defect in P cells (Spencer et al., 2001) (Table 1). To confirm the role of ect‐2 in cytokinesis, we also examined the effect of ect‐2(RNAi) on divisions of early embryos. Examination of tubulin∷GFP‐marked live embryos showed a strong cytokinesis defect in these embryos, whereas mitosis seemed to be normal (supplementary Fig 1 online). Our data are consistent with the evidence that Pebble in Drosophila and ECT2 in the mouse activate Rho GTPases to regulate cytokinesis (Prokopenko et al, 1999; Tatsumoto et al, 1999). Interestingly, as reported for the Drosophila rho gene (Schumacher et al, 2004), we failed to rescue either the cytokinesis or the migration defects in P cells in the ku427 mutant by expressing a hyperactive rho‐1 mutant gene (G14V gf) (data not shown). This may indicate that cycling between the GDP‐bound and the GTP‐bound forms is important for the proper functioning of Rho.
As mentioned above, ect‐2∷GFP is expressed in a limited number of cell lineages during postembryonic development. This raises the possibility that ECT‐2‐mediated cytokinesis is not required for all cell divisions. Such an idea is consistent with a recent report that states that activation of Rho A for cytokinesis is tissue specific in mammals (Yoshizaki, 2004).
The P‐cell cytokinesis defect was not observed in the mutants of two rac genes (ced‐10 and mig‐2) or in unc‐73 (Table 1). Similarly, in Drosophila, mutations in Rac genes do not result in any obvious cytokinesis defects (Hakeda‐Suzuki et al, 2002).
ect‐2 is also involved in P‐cell migration
We have previously shown that rho‐1(dn) causes a strong defect in P‐cell migration from the sublateral position to the ventral midline, shortly after hatching (Spencer et al, 2001). The worms show P‐cell‐derived neurons on the lateral surface instead of the normal ventral midline positions (Table 1). We also observed misplaced neurons in the ect‐2(ku427) and ect‐2(RNAi) animals, although this phenotype is significantly weaker than that of rho‐1(dn) (Fig 3F,G; Table 1). In addition, we have also observed multinucleated P cells in the lateral positions in late‐stage larvae (Fig 3H,I), which suggests that the cytokinesis and migration defects can be associated with the same P cells. An unc‐73 mutation also displaced neurons in the lateral position in 58% of animals (Table 1) (Spencer et al, 2001). These data indicate that ect‐2 is also involved in regulating P‐cell migration, but the contribution by this gene to P‐cell migration is minor compared with that of rho‐1 or unc‐73. Consistent with a proposal that both ECT‐2 and UNC‐73 contribute to the exchange reaction of RHO‐1 during P‐cell migration (Fig 4), ect‐2(ku427); unc‐73(RNAi) showed a stronger cell migration defect than either ku427 or unc‐73(RNAi) alone (Table 1).
A role for ect‐2 in cell migration is consistent with the recent finding in Drosophila that Pebble is involved in the cell migration of the mesoderm in embryogenesis (Schumacher et al, 2004; Smallhorn et al, 2004). However, it was suggested that Rho is not involved in this Pebble‐mediated cell migration, as a cell migration defect was not observed when a dominant‐negative (dn) form of Rho was expressed in the cells (Schumacher et al, 2004).
Phenotypic differences among the two ect‐2 alleles and RNAi
The ect‐2(ku427) mutant showed a strong Egl phenotype and a relatively weak Unc phenotype owing to the cytokinesis and migration defects of P cells. The average brood size was 68 (n=14) and no obvious germline defects were observed. In addition, this strain showed no obvious embryonic lethal phenotype (1%, n=222). In contrast, 47% (n=299) of the ect‐2 (RNAi) animals were embryonic lethal. Escapers showed strong Unc and Ste (sterile) phenotypes (100%, n=50) as well as stronger cytokinesis defects in P cells (Table 1). These animals sometimes showed a ball‐like gonad (data not shown). The differences in phenotypes may be due to the difference in the extent of loss of ect‐2 gene activity under the two conditions. As described earlier, the ku427 mutant has a lesion in the promoter region that causes a decrease in transcription of the gene. The residual low expression level of ect‐2 in ku427 animals appears to be sufficient to cover the gene functions in the embryo as well as in the germ line, and P cells may be more sensitive to the reduction in the gene dosage. Alternatively, it is also possible that the reduction of the transcription level by the ku427 lesion is more drastic in the P‐cell lineage, as we failed to detect any expression from the GFP reporter driven by the mutant promoter in P cells.
Interestingly, the ect‐2(e1778) allele caused only a weak Unc phenotype, but a complete Ste phenotype. This allele also resulted in P‐cell defects that were weaker than those of ku427 (Table 1). The difference between the two alleles may indicate that there is some tissue specificity in the effects of these mutations. ect‐2(e1778) has an in‐frame deletion that removes the BRCT1 domain (111–207 aa; Fig 1E). Thus, it is possible that the BRCT1 domain has a more crucial role in the germ line. Alternatively, the e1778 allele could have eliminated the gene activity. However, a strong maternal effect masks the requirements of the gene during embryogenesis and in other lineages, but not during the germline development that occurs in the later stages. This leads to the strong Ste phenotype associated with the homozygous e1778 animals from the heterozygous mother. A similar phenomenon has been observed in several essential C. elegans genes (Fay & Han, 2000; Antoshechkin & Han, 2002).
zen‐4 and cyk‐4 function in cytokinesis of P cells
ect‐2 and rho‐1 mutants have defects in both P‐cell migration and cytokinesis. To determine whether the migration defect is related to the role of ect‐2 and rho‐1 in the cytokinesis mechanism, we examined the effects of RNAi of two other genes, zen‐4 and cyk‐4. zen‐4 encodes a kinesin‐like protein and cyk‐4 encodes a RhoGAP, both of which are part of a central spindlin complex that is involved in the formation of the central spindles and cleavage furrow during cytokinesis of early embryos (Raich et al, 1998; Jantsch‐Plunger et al, 2000; Severson et al, 2000; Mishima et al, 2002; Portereiko et al, 2004). As indicated in Table 1, worms treated with zen‐4 RNAi or cyk‐4 RNAi showed the cytokinesis defect in P cells, but not the migration defect, suggesting that ect‐2/RhoGEF and rho‐1 interact with different factors for the migration function.
In Drosophila, Pebble, Racgap50C (homologue of cyk‐4) and Pavarotti (homologue of zen‐4) form a complex that is involved in the positioning of the contractile ring and coordinates F‐actin and microtubule remodelling during cytokinesis (Somers & Saint, 2003). Our results indicate that the functions of Rho, its regulators and associated proteins in cytokinesis are conserved.
A model for ect‐2 function, in relationship with other genes, is shown in Fig 4. The mechanism by which the Rho GEF proteins are regulated for the cell migration function is at present not clear in C. elegans or other systems (Raftopoulou & Hall, 2004). The mechanism acting downstream of Rho for the P‐cell migration function also remains to be understood. We have shown that an lf mutation in let‐502, which encodes a Rho‐activated kinase (Wissmann et al, 1997), shows a partially penetrant defect in P‐cell migration but not a defect in cytokinesis (Spencer et al, 2001) (Table 1), suggesting that let‐502 is involved in the migration process. However, its weak phenotype suggests that there are other Rho effectors acting in the process.
Methods for scoring phenotypes. For scoring P‐cell numbers, we counted the Pn.p cells of L2 larvae under Nomarski optics. For P‐cell migration and cytokinesis defects, we examined L4 larvae containing the unc‐47∷GFP transgene for abnormal lateral positions of GFP‐positive cells and for the appearance of multinuclear GFP‐positive cells, respectively.
Mutant isolation and position cloning. ect‐2(ku427) was isolated as a suppressor of the Muv phenotype of the lin‐31(n301); eff‐1(hy21) double mutants. Details regarding the strains and the methods for the screen, mapping as well as cloning, are given in the supplementary information online.
Quantitative RT–PCR, RNAi and construction of plasmids for transgenic animals. See the supplementary information online for details. Supplementary information is available at EMBO reports online (http://www.nature.com/embor/journal/vaop/ncurrent/extref/7400533‐s1.pdf).
Supplementary Figure 1
We thank the C. elegans Genetics Center for strains, R. Stevens for help in constructing the rho‐1(gf) transgene, Y. Kohara for clones, T. Schedl and A. Hajnal for communicating unpublished results and A. Sewell, Y. Suzuki and D. Eastburn for comments on the manuscript. This work was supported by a grant from the National Institutes of Health (GM47869) and by the Howard Hughes Medical Institute, of which K.M. is an associate and M.H. is an investigator.
- Copyright © 2005 European Molecular Biology Organization