Small GTPases^*

RNAi-mediated knockdown of rho-1 results in early embryonic arrest, often with a failure of cytokinesis (Jantsch-Plunger et al., 2000; Spencer et al., 2001). Embryos that arrested later in development displayed severe defects in tissue morphogenesis. Other mutations that affect embryonic morphogenesis and cytokinesis identified other genes that interact genetically with rho-1 in this process. let-502 encodes a Rho-binding kinase and mel-11 encodes a myosin phosphatase (Piekny and Mains, 2002; Wissmann et al., 1999; Wissmann et al., 1997). These molecules likely affect actin-myosin cytoskeletal dynamics of the contractile ring during cytokinesis in response to Rho signaling. The cyk-4 gene, identifed in a screen for mutations affecting embryonic cytokinesis, encodes a Rho GAP that stimulates GTPase activity of Rho, Rac, and Cdc42 and that likely modulates RHO-1 activity in cytokinesis (Jantsch-Plunger et al., 2000).

LET-502 and MEL-11 also control embryonic elongation. The hypodermal cells which epibolize the embryo contain circumferential rings of actin that contract, resulting in a "squeezing" of the embryo and a subsequent increase in length. LET-502 and MEL-11 have opposing roles in this process (LET-502 normally enhances contraction and MEL-11 normally inhibits contraction; Wissmann et al., 1999; Wissmann et al., 1997; Figure 2). Interestingly, the Mtl Rac GTPase MIG-2 and the UNC-73 Trio Rac/Rho GEF act in the MEL-11 pathway in this process. UNC-73 Trio acts as a GEF for RHO-1, CED-10, and MIG-2 (Spencer et al., 2001; Steven et al., 1998; Wu et al., 2002). These results indicate that MIG-2 Mtl Rac and RHO-1 might have opposing roles in embryonic elongation (Piekny et al., 2000; Figure 2).

Later in development, rho-1 controls the ventral migrations of the P cells (Spencer et al., 2001). The P cell nuclei are born in sublateral locations and migrate ventrally to align at the ventral midline. RNAi of rho-1 and dominant-negative rho-1 transgenic expression perturb this migration of the P cells. The UNC-73 Trio Rac/Rho GEF acts upstream of RHO-1 in P cell migration (UNC-73 acts as a GEF on Rho and activated RHO-1 partially rescues an unc-73 mutant), and the LET-502 Rho-binding kinase acts downstream of RHO-1 in this process (a let-502 mutation is not rescued by activated rho-1; Spencer et al., 2001; Steven et al., 1998; Figure 3). Interestingly, the Rac GTPases CED-10 and MIG-2 control P-cell migration in a parallel redundant manner with RHO-1 (e.g., RHO-1 activity can partially compensate for loss of Rac activity in P cell migration; Spencer et al., 2001).

RHO-1 also controls neuronal cell shape after the establishment of the normal axon and dendrite morphology of the neuron. Expression of dominant-negative RHO-1(T19N) in the ASE sensory neuron resulted in an expanded cell body morphology and ectopic neurite initiation (Zallen et al., 2000). This phenotype is similar to that caused by loss of function of the SAX-1 Ndr kinase (Zallen et al., 2000), a known downstream target of Rho activity in other systems. RHO-1 and SAX-1 might act together to regulate neuronal morphogenesis.

RNAi of cdc-42 causes defects in embryonic cytokinesis similar to rho-1(RNAi) (Kay and Hunter, 2001). cdc-42(RNAi) also perturbs the polarity of the single-celled zygote and results in defects in anterior-posterior axis formation and mitotic spindle orientation (Gotta et al., 2001; Kay and Hunter, 2001). Upon fertilization, sperm entry results in a rearrangement of the zygote cytoplasm such that anterior cortex accumulates the PAR-3 and PAR-6 proteins and the posterior cortex accumulates the PAR-2 protein, which together define the anterior-posterior axis of the organism (Etemad-Moghadam et al., 1995; Hung and Kemphues, 1999; Levitan et al., 1994; Watts et al., 1996). The PAR proteins associate with the actin cytoskeleton at the cell cortex, and the actin cytoskeleton is required for their localization. As a result of PAR gene localization, the first cell division generates a larger anterior cell and a smaller posterior cell, whose spindle rotates 90° in relation to the anterior cell (Kemphues et al., 1988). Mutations in the par genes disrupt zygotic polarity and result in the symmetric division of zygote and defects in spindle orientation. RNAi of cdc-42 results in a similar loss of zygotic polarity and an associated failure to localize the PAR proteins to their proper domains (Gotta et al., 2001; Kay and Hunter, 2001). Thus, cdc-42 might control the localization of the PAR proteins or their interaction with the actin cytoskeleton. CDC-42 physically associates with PAR-6, a CRIB- and PDZ-domain-containing component of a conserved complex including PAR-3, and PKC-1, a protein kinase C ortholog, that defines the anterior cortical domain in the zygote (Gotta et al., 2001).

A screen for mutations that affect the development of the excretory cell identified exc-5, which encodes a putative Cdc42 GEF similar to the human faciogenital dysplasia-1 protein (Buechner et al., 1999; Gao et al., 2001; Suzuki et al., 2001). EXC-5 controls the formation of apical versus basolateral cell surfaces and cytoskeleton in the excretory canals: exc-5 loss of function mutants display large varicosities in the canals due to a failure of apical surface formation; and exc-5 overactivity results in foreshortened canals due to overproliferation of apical structures at the expense of basolateral structures. A direct role of CDC-42 in this process has not been demonstrated. However, the similarity of EXC-5 to the Cdc42 GEF FGD1 suggests that CDC-42 might be involved in excretory cell apical-basolateral polarity.

C. elegans WSP-1 is the homolog of the human Wiskcott-Aldrich syndrome protein (WASP). WASP is a downstream effector of Cdc42 that activates the Arp2/3 complex, an actin nucleating and branching complex. During C. elegans gastrulation, the dorsally-located hypodermal cells migrate toward the ventral midline where they meet and completely surround the embryo in a process known as ventral closure (Simske and Hardin, 2001; Sulston et al., 1983; Williams-Masson et al., 1997). C. elegans wsp-1 mutants display a low-penetrance embryonic lethality due to a failure of migration of the hypodermal cells and ventral closure (Withee et al., 2004).

Hypomorphic ced-10 mutations cause failure of phagocytosis of cell corpses after programmed cell death (Ellis et al., 1991; Reddien and Horvitz, 2000). Upon programmed death of a cell, neighboring cells extend actin-based plasma membrane protrusions to engulf and metabolize the cell corpse. ced-10 mutants fail in this process, resulting in the persistence of unengulfed cell corpses. ced-10 might be required to organize the actin cytoskeleton underlying the extension of cellular protrusions of the engulfing cell. Other engulfment mutations identified the ced-1, ced-2, ced-5, ced-6, ced-7, and ced-12 genes (Ellis et al., 1991). These genes can be placed into two pathways based upon redundancy of function in engulfment: mutations in ced-1, ced-6, and ced-7 enhance the engulfment phenotypes of ced-2, ced-5, and ced-10, whereas mutations within the two groups do not enhance each others' phenotypes. This suggests that two parallel pathways exist to control phagocytosis, one of which involves ced-10 Rac.

Molecular and genetic analyses revealed other CED-10 pathway members in phagocytosis (Figure 4), including CED-2, a CrkII-like receptor adapter protein (Reddien and Horvitz, 2000), and CED-5, a CDM-family Rac GEF (Wu and Horvitz, 1998). Furthermore, CED-12 is an ELMO-like protein, which cooperates with CDM proteins in Rac GTP exchange (Gumienny et al., 2001; Lu et al., 2004; Zhou et al., 2001).

Interestingly, MIG-2 Mtl Rac and the UNC-73 Trio Rac/Rho GEF act upstream of the CED-5/ CED-12/CED-10 complex in phagocytosis (deBakker et al., 2004; Figure 4). In response to UNC-73 GTP exchange, active GTP-bound MIG-2 interacts with CED-12 resulting in CED-5-mediated GTP exchange and activation of CED-10 (deBakker et al., 2004). Thus, Rac-like GTPases can act in a heirarchical fashion during phagocytosis.

ced-10 mutations also affect the migration of the gonadal distal tip cells, which results in misshapen gonad arms in ced-10 mutants (Lundquist et al., 2001). ced-5 and ced-12 also affect distal tip cell migration, as does the unc-73 Trio-like Rho/Rac GEF (which also acts with rho-1 in P-cell migration; Gumienny et al., 2001; Lundquist et al., 2001; Zhou et al., 2001).

Null alleles of ced-10 cause maternal-effect embryonic lethality with defects in early cell movements and gastrulation resulting in failure of the endoderm to be surrounded by the ectoderm (the gut on the exterior (Gex) phenotype; Lundquist et al., 2001; Soto et al., 2002). ced-10 might control actin rearrangements involved in movement and migration during gastrulation. Mutations in gex-1, gex-2 and gex-3 cause a similar maternal effect lethal phenotype. GEX-2 encodes an Sra-1-like molecule, and GEX-3 encodes a Hem2-like molecule. Sra-1 and Hem2, along with Nck, form a complex with the Arp2/3 complex activator WAVE (Eden et al., 2002). This complex might act downstream of CED-10 Rac in gastrulation (Figure 4).

Null mutations in mig-2 were isolated on the basis of defective migration of the Q cells and their daughters, and activated mig-2 alleles but not null mig-2 alleles affect CAN neuron migration (Zipkin et al., 1997). mig-2 mutations also display defective migration of the gonadal distal tip cells similar to ced-10 (Lundquist et al., 2001). The UNC-73 Trio Rac/Rho GEF acts with MIG-2 in both processes. As described for CED-10, MIG-2 and the UNC-73 Trio Rac/Rho GEF act upstream of the CED-12/CED-5/CED-10 complex to control phagocytosis (Figure 4).

Null mutants of mig-2 and ced-10 and rac-2(RNAi) displayed no detectable defects in axon pathfinding (Lundquist et al., 2001; E.A.L., unpublished results). Each pairwise double mutant revealed synergistic axon pathfinding defects (i.e. the defects were more severe than the additive effects of each mutation alone; Lundquist et al., 2001; Wu et al., 2002; Figure 5). The double mutant phenotype included defects in axon outgrowth, axon guidance, and axon branching, indicating that Racs control multiple aspects of axon pathfinding. Furthermore, Rac double mutants displayed ectopic axon branches, suggesting that Racs also control the pruning of spurious axon branches (Struckhoff and Lundquist, 2003). Consistent with a role of Racs in axon development, constitutively-active CED-10, RAC-2 and MIG-2 perturb axon pathfinding (Struckhoff and Lundquist, 2003; Zipkin et al., 1997; Figure 5). Rac(G12V) molecules induce the formation of ectopic lamellipodia and filopodia on neurons that resemble those found on the growth cone during axon outgrowth (Knobel et al., 1999).

Using the logic that two mutations in the same redundant pathway should not enhance the other, whereas two mutations in distinct redundant pathways should show synergistic effects, other genes that act with rac signaling in axon pathfinding were identified (Figure 6). The UNC-73 Trio Rac/Rho GEF, which acts with RHO-1 in P cell migration, acts with all three Rac genes in axon pathfinding; CED-5, the CDM GEF that acts with in the CED-10 pathway in phagocytosis, acts in the MIG-2 pathway in parallel to CED-10 in axon pathfinding; and the actin-binding protein UNC-115 abLIM acts in the RAC-2 pathway in parallel to CED-10 and MIG-2 in axon pathfinding (Lundquist et al., 1998; Lundquist et al., 2001; Struckhoff and Lundquist, 2003; Wu et al., 2002). UNC-115 abLIM might represent one path to the actin cytoskeleton downstream of RAC-2. Indeed, unc-115 loss of fucntion partially suppresses the ectopic lamellipodia and filopodia induced by activated rac-2, indicating that unc-115 acts downstream of rac-2 in lamellipodia and filopodia formation (Struckhoff and Lundquist, 2003). RNAi of the Arp2/3 complex activator WVE-1 WAVE caused no axon defects alone but synergized with mig-2 mutations but not ced-10 mutations, suggesting that WVE-1 and possibly the Arp2/3 complex act downstream of CED-10 Rac in axon pathfinding (E.A.L., unpublished results). This combination of double mutant analysis and epistasis analysis will be important to identify additional genes that act downstream of Rac signaling in axon pathfinding and will contribute to a more complete understanding of the signaling mechanisms downstream of Rac.

Growth cones detect guidance information using transmembrane guidance receptors. Indeed, CED-10 Rac and the UNC-115 actin-binding protein act downstream of the netrin guidance receptor UNC-40 DCC in ventral AVM axon pathfinding (Gitai et al., 2003). Furthermore, CED-10, MIG-2 and RAC-2 act with the MIG-15 NIK kinase downstream of the integrin complex in VD/DD motor axon circumferential navigation (Poinat et al., 2002).

ced-10 and mig-2 mutations alone affect some cell migrations, including the gonadal distal tip cells and, in the case of mig-2, Q cells and their descendants. The migrations of the CAN neuron cell bodies, which are born in the anterior of the embryo and migrate posteriorly to a position near the vulva, are redundantly controlled by the Rac genes. No single Rac loss-of-function mutant displays CAN migration defects, but each pairwise double mutant combination of ced-10, mig-2 and rac-2 display synergistic failure in the posterior migrations of the CAN neurons (Lundquist et al., 2001).

Rac genes also act redundantly in multiple aspects of vulval development and morphogenesis. mig-2 and ced-10 single mutant vulvae are largely wild-type, whereas mig-2; ced-10 double loss-of-function mutants display defects in the orientiation of asymmetric divisions of the 1° and 2° vulval cells, indicating that mig-2 and ced-10 might redundantly control spindle orientation (Kishore and Sundaram, 2002). mig-2; ced-10 doubles also display a failure in the migrations of 2° vulval cells toward the 1° vulval cells to form a functional vulva. unc-73 mutations cause indistinguishable vulval defects, indicating that the UNC-73 Rho/Rac GEF controls both CED-10 and MIG-2 in vulval cell mitotic spindle orientation and 2° cell migration.

Innate immunity in C. elegans consists of the expression and secretion of anti-microbial peptides, the neuropeptide-like proteins (NLPs) and the caenacins (CNCs; Couillault et al., 2004). Expression of nlp-31 in response to fungal infection requires the tir-1 gene, which encodes a Toll/IL/IR-domain protein similar to vertbrate SARM (Couillault et al., 2004). A yeast two-hybrid screen with TIR-1 identified RAB-1. RNAi of RAB-1 abolished nlp-31 expression in response to fungal infection (Couillault et al., 2004), indicating that RAB-1 mediates fungally-induced nlp-31 expression and might act with TIR-1 in this process. The molecular and cellular mechanism of RAB-1 in this process remains unclear.

The C. elegans RAB-3 protein is associated with acetylcholinergic synapses where it might influence synaptic vesicle trafficking and acetylcholine release (Nonet et al., 1997). Mutations in rab-3 were isolated in a screen for animals that were resistant to the acetylcholinesterase inhibitor aldicarb. Aldicarb causes overaccumulation of acetylcholine in the synaptic cleft which results in paralysis, and mutations in rab-3, which reduce the release of acetylcholine from presynaptic terminals, mitigate this effect. In rab-3 mutants, synaptic vesicles containing acetylcholine fail to be properly localized to acetylcholinergic synapses and accumulate in intersynaptic regions of the axon (Nonet et al., 1997). This phenotype is consistent with RAB-3 controlling the trafficking of synaptic vesicles to the synapse.

aex-3 encodes a Rab3 GEF, and mutations in aex-3 were isolated in a screen for defecation-defective animals (Iwasaki et al., 1997; Iwasaki and Toyonaga, 2000). In aex-3 mutants, RAB-3 protein accumulates in neuronal cell bodies and is not targeted to synapses in axons. Thus, AEX-3 might control the association of RAB-3 with synaptic vesicles.

In addition to affecting synaptic vesicle localization, RAB-3 might also control other aspects of vesicle release. unc-10 encodes a PDZ domain- and C2 domain-containing protein similar to the known Rab3 effector Rim1 (Koushika et al., 2001). Vesicle docking and synaptic organization appear normal in unc-10 mutants, and UNC-10 might interact with RAB-3 and syntaxin to facilitate vesicle fusion.