In the developing nervous system, axons are guided to their targets by the growth cone. Lamellipodial and filopodial protrusions from the growth cone underlie motility and guidance. Many molecules that control lamellipodia and filopodia formation, actin organization, and axon guidance have been identified, but it remains unclear how these molecules act together to control these events. Experiments are described here that indicate that, in Caenorhabditis elegans, two WH2-domain-containing activators of the Arp2/3 complex, WVE-1/WAVE and WSP-1/WASP, act redundantly in axon guidance and that GEX-2/Sra-1 and GEX-3/Kette, molecules that control WAVE activity, might act in both pathways. WAVE activity is controlled by Rac GTPases, and data are presented here that suggest WVE-1/WAVE and CED-10/Rac act in parallel to a pathway containing WSP-1/WASP and MIG-2/RhoG. Furthermore, results here show that the CED-10/WVE-1 and MIG-2/WSP-1 pathways act in parallel to two other molecules known to control lamellipodia and filopodia and actin organization, UNC-115/abLIM and UNC-34/Enabled. These results indicate that at least three actin-modulating pathways act in parallel to control actin dynamics and lamellipodia and filopodia formation during axon guidance (WASP–WAVE, UNC-115/abLIM, and UNC-34/Enabled).
THE growth cone of an extending axon senses and responds to extracellular guidance cues that direct axon pathfinding in the developing nervous system (Mortimer et al. 2008). Growth cone motility and guidance are mediated by the dynamic extension and retraction of lamellipodia and filopodia that are themselves the result of actin cytoskeleton-plasma membrane dynamics and interactions (Gallo and Letourneau 2004; Zhou and Cohan 2004; Pak et al. 2008). The peripheral region of the growth cone is rich in dynamic, actin-based structures including bundled microfilaments in filopodia and a lamellipodial-like meshwork of actin filaments (Dent and Gertler 2003). Multiple actin regulatory molecules are required to drive the formation of these distinct domains of actin cytoskeletal architecture in the growth cone, and the control of actin dynamics in the growth cone in response to guidance signals is an area of much interest and active study.
Rac GTPases are key regulators of cytoskeletal dynamics and cellular protrusion and have been shown to control axon guidance as well as other morphogenetic events (Lundquist 2003; Watabe-Uchida et al. 2006). In Caenorhabditis elegans, the Rac GTPase CED-10 has overlapping function with the Rac-like GTPase MIG-2 in axon guidance (Lundquist et al. 2001). While MIG-2-like molecules (Mtl in Drosophila) (Hakeda-Suzuki et al. 2002) are not found in vertebrates, MIG-2 might be the C. elegans functional equivalent of vertebrate RhoG (Debakker et al. 2004). Previous work also showed that the UNC-115/abLIM actin binding protein might act downstream of Rac signaling to control lamellipodia and filopodia formation (Lundquist et al. 1998; Struckhoff and Lundquist 2003; Yang and Lundquist 2005). The Enabled actin-regulatory molecule is involved in filopodia formation and acts downstream of axon guidance receptors to mediate filopodia formation in the growth cone (Krause et al. 2003; Lebrand et al. 2004). In C. elegans, UNC-34/Enabled acts in parallel to UNC-115/abLIM and Rac signaling in axon guidance (Withee et al. 2004; Shakir et al. 2006). Other actin-modulating molecules in addition to UNC-115 are likely to act downstream of Racs, as unc-115 null mutations have very little effect on axon guidance alone but synergize with mig-2, ced-10, and unc-34 mutations.
Molecules containing a WH2 domain, alternately called a VCA domain, are activators of Arp2/3, a seven-polypeptide complex that stimulates actin nucleation and branching (Beltzner and Pollard 2007). Arp2/3 activity has been implicated in formation of the growth cone peripheral actin meshwork (Goldberg et al. 2000). However, interference with Arp2/3 activity using a dominant-negative N-WASP WH2 domain in cultured neurons caused more rapid growth cone advance but had little apparent effect on growth cone morphology (Strasser et al. 2004). In C. elegans, wve-1 encodes the sole WAVE/SCAR-like molecule and wsp-1 encodes the sole WASP-like molecule (Sawa et al. 2003; Withee et al. 2004). Both WSP-1 and WVE-1 have C-terminal WH2 domains and are likely to act as Arp2/3 activators. WSP-1 has been shown to act redundantly with WVE-1 and UNC-34 Ena in embryonic gastrulation and CAN cell migration, and WVE-1 acts redundantly with UNC-34 Ena in motor axon guidance (Withee et al. 2004; Sheffield et al. 2007).
The activity of WAVE/SCAR as an Arp2/3 activator is regulated by a plasma membrane-associated complex of proteins including Sra-1/PIR121, HEM2/NAP1/Kette, Abi1, and HSPC300 (Eden et al. 2002; Innocenti et al. 2004; Steffen et al. 2004; Stradal et al. 2004; Vartiainen and Machesky 2004). Activated Rac-GTP interacts physically with Sra-1, leading to activation of WAVE/SCAR and thus Arp2/3. Rac-GTP might relieve an inhibition of WAVE/SCAR by this complex (Eden et al. 2002) or it might activate the entire complex including WAVE/SCAR (Innocenti et al. 2004). This complex is involved in lamellipodia formation in cultured cells (Steffen et al. 2004), and WAVE/SCAR, the Arp2/3 complex, and Kette have been shown to affect neuronal development and axon pathfinding in Drosophila (Zallen et al. 2002; Bogdan and Klambt 2003; Schenck et al. 2004). In C. elegans, gex-2 encodes a Sra-1-like molecule and gex-3 encodes a Kette-like molecule. GEX-2, GEX-3, and CED-10/Rac are involved in cell movements in gastrulation, and GEX-2 and GEX-3 interact physically with one another (Soto et al. 2002). Furthermore, GEX-2 and GEX-3 are present at cell margins as observed in other systems. Perturbation of gex-2 and gex-3 caused defects in ventral closure of the embryo by hypodermal cells, resulting in failure of endodermal cells to be internalized and a gut on the exterior (Gex) phenotype. gex-2 and gex-3 also control the morphogenesis of other tissues including body wall muscle, pharynx, and vulva (Soto et al. 2002).
In work presented here, the effects of mutations in wve-1/WAVE and wsp-1/WASP on C. elegans axon guidance are described. Genetic interactions suggested that WVE-1 and WSP-1 have overlapping roles in axon guidance and that WVE-1 might act specifically in the CED-10/Rac pathway and that WSP-1 might act with MIG-2/RhoG in parallel to CED-10/Rac in axon guidance. Further studies revealed roles of GEX-2 and GEX-3 in axon guidance and suggested that these molecules might act with both WVE-1/WAVE and WSP-1/WASP. These results are consistent with a model in which GEX-2 and GEX-3 both act in two parallel pathways in axon guidance, one including CED-10 and WVE-1 and the other including MIG-2 and WSP-1.
MATERIALS AND METHODS
C. elegans strains and RNAi:
All experiments were performed at 20° using standard C. elegans techniques unless otherwise noted (Brenner 1974; Sulston and Hodgkin 1988). The mutations and rearrangements used were X: unc-115(ky275 and mn481), mig-2(mu28), lqIs2; I: wve-1(ne350), hIn1; II: juIs73; IV: gex-2(ok1603), gex-3(zu196), ced-10(n1993), wsp-1(gm324), dpy-9(e12), nT1 qIs51 (IV; V), lqIs3; and V: unc-34(e951).
The gex-2(ok1603) deletion, provided by the C. elegans Gene Knockout Consortium (G. Moulder, B. Barstead, M. Edgley, and D. Moerman), was outcrossed to wild-type N2 five times and was tracked by polymerase chain reaction (PCR) to avoid confusion. The gex-2(ok1603) chromosome was then balanced in trans to dpy-9(e12). The gex-2 genomic region (bases 570,843–583,813 of chromosome IV) was amplified by PCR from wild-type genomic DNA in two fragments with 1 kb of overlap. These PCR products were used in micro-injection experiments to generate the gex-2(+) rescuing extrachromosomal array. wve-1 RNAi was performed by injection of double-stranded RNA into the gonad or body cavity of adults (Timmons 2006). A wve-1 genomic fragment encompassing the first two exons of wve-1, including the first intron, was used as a template to generate dsRNA. gex-2 and gex-3 RNAi was performed by feeding using the Geneservice clones representing each locus (Kamath et al. 2003) and standard techniques (Timmons et al. 2001). Axon defects were scored in progeny of injected animals. Construction of transgenic strains was accomplished using DNA micro-injection into the gonad (Mello and Fire 1995). Sequences of all primers used in this work are available upon request.
Double mutant construction and maintenance:
mig-2(mu28) and ced-10(n1993): All double mutants with mig-2(mu28) and ced-10(n1993) were confirmed phenotypically by gonadal distal-tip cell-migration defects and cell-corpse-engulfment defects (for ced-10) (Zipkin et al. 1997; Reddien and Horvitz 2000).
gex-2(ok1603): gex-2(ok1603) and double mutants with gex-2(ok1603) were balanced with the closely linked morphological marker dpy-9(e12). The presence of the ok1603 deletion in each strain was verified by PCR. Before phenotypic analysis, the presence of gex-2(ok1603) and dpy-9(e12) in the strain was verified by the maternal-effect Gex, Egl, pVul phenotype of gex-2 and the Dpy phenotype of dpy-9. PDE axons were scored in adult animals with the pVul and Egl phenotype [e.g., gex-2(ok1603) homozygotes].
gex-3(zu196): gex-3(zu196) (Soto et al. 2002) was balanced with the reciprocal translocation nT1 harboring a pharyngeal GFP marker (qIs51) (Clark and Baillie 1992). gex-3(zu196)/nT1 qIs51 heterozygotes were pharyngeal GFP-positive whereas gex-3(zu196) homozygotes had no pharyngeal GFP expression. Before scoring, the presence of gex-3(zu196) was assessed phenotypically by the presence of maternal effect Gex, Egl, and pVul phenes. Young adult animals with no pharyngeal GFP expression were scored.
wve-1(ne350): wve-1(ne350) was balanced with the inversion chromosome hIn1 (Zetka and Rose 1992), which carries a recessive paralyzed unc mutation. The presence of wve-1(ne350) was assessed by the presence of maternal-effect Gex, Egl, and pVul animals, and PDE axons were scored in adults with this phenotype.
wsp-1(gm324): PCR was used to test for the presence of the gm324 deletion, and, when appropriate, wsp-1 was balanced by the nT1 qIs51 translocation described for gex-3.
unc-115 and unc-34(e951): The presence of these mutations was confirmed by the Unc phenotype displayed by both.
Scoring and analysis of PDE and VD/DD axon guidance defects:
The PDE axons were visualized with an osm-6 promoter∷gfp transgene as previously described (Lundquist 2003). osm-6∷gfp is expressed in all ciliated sensory neurons including PDE, and the PDE axon can be unambiguously identified from other axons (Collet et al. 1998). A PDE axon was considered to have defective guidance if the axon failed to reach the ventral nerve cord (VNC) or if it reached the ventral nerve cord at a position greater than approximately a 45° angle from the PDE cell body. Generally, young adult animals were scored. Percentages of defective PDE axon guidance in each strain were determined, and significance was judged by a two-sided t-test with unequal variance unless otherwise noted.
The VD and DD motor axons were scored using an unc-25∷gfp reporter expressed in these cells (juIs73) (Jin et al. 1999). The cell bodies of the 13 VD and 6 DD neurons lie along the ventral nerve cord. Axons extend anteriorly in the ventral nerve cord, turn dorsally, and extend commissurally to the dorsal nerve cord (White et al. 1986). The axons of two neurons, DD1 and VD2, extend on the left side of the animal, while the remaining 17 extend on the right side. The guidance of commissural axons was considered abnormal if the axons failed to extend directly to the dorsal nerve cord and instead wandered laterally, often intersecting with other axons or failing to reach the dorsal nerve cord. Furthermore, axons extending abnormally on the left side of the animal (DD1 and VD2 excluded) were assayed. The average number and standard deviation of misguided axons and left-side axons per animal were determined, and significance of differences was calculated using a two-sided t-test with unequal variance.
Generation of WSP-1 and WVE-1 WH2 transgenes:
The WH2 domains of WVE-1 (3755–4234 of R06C1.3) and WSP-1 [9108–9583 of the wsp-1 gene (C07G1.4a in Wormbase)] were amplified from genomic DNA by PCR and placed downstream of the osm-6 promoter in vector pPD49.26 (kindly provided by A. Fire). The wve-1 or wsp-1 initiator methionine and six upstream nucleotides were included in the PCR primer upstream of the WH2 region.
wve-1/WAVE controls PDE axon guidance:
wve-1 encodes a C. elegans WAVE-family molecule, with an N-terminal WAVE homology domain, proline-rich domain, and C-terminal WH2 domain (Figure 1A). The sequence of the wve-1 gene was determined from wve-1(ne350) mutants, and a C-to-T nucleotide transition in the wve-1 coding region resulted in a premature stop codon in the reading frame (Figure 1, A and B). wve-1(ne350) is likely a strong loss-of-function allele. wve-1(ne350) homozygous mutants from a heterozygous mother (M+) survived to adulthood but were egg-laying defective (Egl), had protruding vulvae (pVul), and gave rise to embryos that arrested due to previously described defects in embryonic gastrulation (gut on the exterior; Gex phenotype) (Soto et al. 2002).
Neuronal development was assayed in wve-1(ne350M+) larvae and young adults. The PDE neuron is born in the L2 larval stage and sends an axon directly ventrally to the VNC, where the axon bifurcates and extends anteriorly and posteriorly in the VNC (Figure 2A). A ciliated dendrite extends dorsally from the PDE cell body. wve-1(ne350M+) animals alone displayed defects in PDE axon guidance (14%; Figure 2G). Instead of extending directly to the ventral nerve cord as in wild type, some axons of wve-1(ne350M+) animals made lateral turns before reaching the VNC (Figure 2B). Some axons failed to reach the VNC and emanated from the sides of the PDE cell body instead of from the ventral surface and wandered laterally along the body wall (Figure 2C). This could be a defect in polarity of axon initiation or could be a secondary consequence of subsequent aberrant growth cone migration. Some axons bifurcated before reaching the VNC, with anterior and posterior extensions along the lateral body wall (Figure 2D). Alone and in the double-mutant combinations below, occasional ectopic neurites were observed on misrouted axons but rarely on axons that extended normally to the ventral nerve cord (Table 1). This is in contrast to other mutations that affect PDE axon pathfinding (e.g., mig-2; ced-10 double mutants and unc-115 mutants), which also displayed ectopic axon branches or neurites on axons that otherwise extended correctly to the ventral nerve cord (Lundquist et al. 2001; Struckhoff and Lundquist 2003). Thus, wve-1 might affect guidance of the PDE axon but not ectopic axon formation as do ced-10, mig-2, and unc-115.
wve-1/WAVE mutation enhances mig-2/RhoG but not ced-10/Rac in PDE axon guidance:
Biochemical studies indicate that WAVE acts downstream of Rac GTPases (Eden et al. 2002; Innocenti et al. 2004; Steffen et al. 2004; Stradal et al. 2004; Vartiainen and Machesky 2004). We assayed PDE axon pathfinding in double mutants of wve-1(ne350) with mig-2/RhoG and ced-10/Rac. It was shown previously that mig-2 and ced-10 mutations have little effect on PDE axon guidance alone but display strong phenotypic synergy indicative of overlapping function in axon guidance (Lundquist et al. 2001; Shakir et al. 2006). mig-2(mu28) is a putative null mutation (a premature stop codon) (Zipkin et al. 1997) whereas ced-10(n1993) is an incomplete loss-of-function mutation (a missense mutation in the C-terminal CaaX prenylation domain) (Reddien and Horvitz 2000). ced-10(tm597) and ced-10(n3417) are deletion alleles that are likely to be null, but have few PDE axon defects on their own (Shakir et al. 2006). Thus, the null phenotypes of mig-2 and ced-10 are not likely to include significant PDE axon guidance defects.
wve-1(ne350M+); mig-2(mu28) double mutants displayed enhanced PDE axon guidance defects [39% compared to 14% for wve-1(ne350M+) alone (P < 0.0001)] (Figure 2G). In contrast, wve-1(ne350M+); ced-10(n1993) double mutants did not exhibit enhanced PDE axon guidance defects [10% compared to 14% for wve-1(ne350M+) alone (P = 0.26)] (Figure 2G). Thus, mig-2(mu28) but not ced-10(n1993) enhanced wve-1(ne350). Few ectopic axon branches were observed in double mutants (Table 1).
This interaction was confirmed using wve-1 RNAi by injection of dsRNA into the gonad. wve-1 RNAi resulted in ∼60% embryonic lethality with a Gex phenotype. wve-1 RNAi into mig-2(mu28) and ced-10(n1993) resulted in an increase in embryonic lethality to 85 and 90%, respectively, indicating that wve-1, mig-2, and ced-10 might redundantly control embryonic development. While wve-1 RNAi resulted in embryonic lethality, PDE axons were scored in surviving viable animals. wve-1 RNAi in wild-type animals resulted in 5% PDE axon guidance defects in survivors (Figure 2G). This was increased to 50% in mig-2(mu28); wve-1(RNAi) animals (P < 0.0001). In contrast, ced-10(n1993); wve-1(RNAi) was not significantly more severe (11%) than wve-1(RNAi) alone (5%) (P = 0.12). Together, these results suggest that wve-1 might act in parallel to mig-2 in axon guidance, possibly in the ced-10 pathway, as wve-1 did not synergize with ced-10.
wve-1/WAVE mutation partially suppresses overactive ced-10/Rac:
To test the idea that wve-1 acts in the ced-10 pathway in parallel to mig-2, the ability of wve-1 to suppress activated ced-10 was determined. From biochemical studies, WAVE is thought to act in a complex of molecules downstream of Rac (Eden et al. 2002; Innocenti et al. 2004). Previous studies showed that constitutive activation of CED-10 in the PDE neuron resulted in the ectopic formation of dynamic lamellipodia and filopodia structures from the axons and cell bodies of PDE neurons (Struckhoff and Lundquist 2003). If WVE-1 acts downstream of CED-10, loss of WVE-1 activity might suppress the effects of activated CED-10. A constitutively-active ced-10(G12V) transgene, consisting of ced-10 with the activating glycine 12 to valine mutation expressed in the PDE, caused 29% of PDE neurons to exhibit ectopic lamellipodia and filopodia (Figure 3, A and B). Alone, wve-1(ne350M+) mutants displayed no ectopic lamellipodia and filopodia. wve-1(ne350M+); ced-10(G12V) animals displayed significantly fewer lamellipodia and filopodia (11%) than ced-10(G12V) alone (29%) (P < 0.0001). ced-10(G12V) also caused extensive ectopic neurite formation in the PDE neuron. This defect was not significantly suppressed by wve-1(ne350M+) (P = 0.8). The mig-2(rh17) mutation is the equivalent activating mutation in mig-2 (Zipkin et al. 1997), but wve-1(ne350); mig-2(rh17) double mutants could not be analyzed due to embryonic lethality that wild-type maternal product did not rescue. That wve-1(ne350) lessened ectopic lamellipodia and filopodia induced by ced-10(G12V) is consistent with the idea that WVE-1 acts downstream of CED-10 in this process.
wsp-1/WASP and wve-1/WAVE act redundantly in PDE axon guidance:
The results above suggest that wve-1 might act in the ced-10 pathway in parallel to mig-2. wsp-1 encodes the C. elegans WASP molecule (Figure 1A). WASP also activates the Arp2/3 complex via a C-terminal WH2 domain (Beltzner and Pollard 2007). wsp-1(gm324) was previously shown to affect CAN cell migration in parallel to wve-1 and unc-34/Enabled (Withee et al. 2004). wsp-1(gm324), a deletion that removes exons 2 and 3 and prevents WSP-1 protein accumulation (Withee et al. 2004), is likely to be a strong loss-of-function allele. wsp-1(gm324) animals are viable as homozygotes.
In this study, wsp-1(gm324) alone was found to have little effect on PDE axon guidance (Figure 2G). However, wsp-1(gm324); wve-1(ne350M+) animals exhibited robust PDE axon guidance defects; 52% of PDE axons were misguided and wandered along the body wall compared to 14% for wve-1(ne350M+) alone (P < 0.0001) (Figure 2, E and G). Only guidance and not ectopic axon formation was strongly affected (Table 1). These data indicate that wsp-1 and wve-1 have overlapping roles in PDE axon guidance.
wsp-1/WASP enhances ced-10/Rac more strongly than mig-2/RhoG in PDE axon guidance:
The above data indicate that WSP-1 and WVE-1 act redundantly in axon guidance and that WVE-1 might act in the CED-10/Rac pathway in parallel to MIG-2/RhoG. To determine how WSP-1, CED-10, and MIG-2 interact, double mutants of wsp-1(gm324) with ced-10(n1993) and mig-2(mu28) were analyzed. Both mig-2(mu28); wsp-1(gm324) and ced-10(n1993) wsp-1(gm324) were homozygous viable although ced-10(n1993) wsp-1(gm324) had uncoordinated movement (Unc) and slightly dumpy body morphology (Dpy). ced-10(n1993) wsp-1(gm324) displayed 28% PDE axon guidance defects (P < 0.0001 compared to each single) (Figure 2, F and G), suggesting that ced-10 and wsp-1 have overlapping roles in PDE axon guidance. mig-2(mu28); wsp-1(gm324) also showed increased PDE axon guidance defects (16%; P < 0.0001 compared to single mutants). However, this was a significantly lower frequency than ced-10(n1993) wsp-1(gm324) (P = 0.004). Ectopic axon formation was not strongly affected (Table 1). Both ced-10(n1993) and mig-2(mu28) enhanced wsp-1(gm324), but the enhancement by ced-10(n1993) was significantly stronger.
That ced-10 enhanced wsp-1 more robustly than did mig-2 is consistent with the idea that mig-2 and wsp-1 might act together in a pathway in parallel to ced-10/wve-1. The activated mig-2(rh17) allele (the equivalent to the G12 activating mutation in ced-10) displayed robust ectopic neurite extension (45%) and some ectopic lamellipodia and filopodia (5%) (Figure 4). wsp-1(gm324); mig-2(rh17) double mutants were viable and fertile, in contrast to the lethality of wve-1(ne350); mig-2(rh17). In wsp-1(gm324); mig-2(rh17) animals, no ectopic lamellipodia and filopodia were evident, and the percentage of ectopic neurites was slightly reduced (from 45 to 34%; t-test P = 0.03, Fisher exact P = 0.007) (Figure 4C). Thus wsp-1(gm324) slightly suppressed the neuronal defects of mig-2(rh17). wsp-1(gm324) slightly but not significantly suppressed the ectopic lamellipodia and filopodia of ced-10(G12V) (29 to 22%, P = 0.21) and had no effect on neurite formation (100% in both instances). While suppression of mig-2(rh17) by wsp-1(gm324) was weak, wsp-1(gm324); mig-2(rh17) animals were viable whereas wve-1(ne350); mig-2(rh17) animals were inviable. Furthermore, the loss-of-function studies described above indicate that wsp-1 and mig-2 might have some parallel function, as mig-2(mu28); wsp-1(gm324) showed weak synthetic axon defects. Possibly, this parallel activity of mig-2 is not suppressed by wsp-1(gm324). That wsp-1 also slightly suppressed ced-10(G12V) hints that wsp-1 might also act in the ced-10 pathway.
wve-1/WAVE and wsp-1/WASP synergize with mig-2/RhoG and ced-10/Rac, respectively, in VD/DD motor axon guidance:
The above data suggest that ced-10 and wve-1 might act together in parallel to mig-2 and wsp-1. This could be an interaction specific to the PDE neurons or it could be a more generalized principle of these molecules' interactions in axon guidance. The interactions of wve-1(ne350), wsp-1(gm324), ced-10(n1993), and mig-2(mu28) were assayed in VD/DD motor axon guidance. The cell bodies of the 13 VD and 6 DD motor neurons lie along the ventral nerve cord (Figure 5A). These neurons extend projections anteriorly in the VNC, which then turn dorsally and extend commissurally to the dorsal nerve cord (DNC) (White et al. 1986). All but DD1 and VD2 extend commissurally on the right side of the animal (DD1 and VD2 extend on the left). The commissural guidance of these axons was assessed, as was left-right guidance of these axons, using an unc-25∷gfp reporter (juIs73) (Jin et al. 1999) (Figure 5, B–D; see materials and methods for scoring).
Alone, ced-10(n1993), mig-2(mu28), wsp-1(gm324), and wve-1(ne350M+) had mild VD/DD axon guidance defects (Table 2). For example, wsp-1(gm324) averaged 1.30 misguided VD/DD axons per animal and 0.02 axons on the left side per animal. wsp-1(gm324); wve-1(ne350M+) double mutants displayed severe misguidance and left-right defects in VD/DD axon pathfinding (4.36 misguided axons per animal and 2.07 axons on the left side per animal; Table 2 and Figure 5C), indicating redundancy of function of WSP-1 and WVE-1 in VD/DD axon guidance.
The wve-1(ne350M+); ced-10(n1993) double was slightly but not significantly more severe than wve-1(ne350M+) alone (e.g., 1.29 misguided axons vs. 0.83 misguided axons and 0.6 vs. 0.51 left-side axons; P = 0.08 and 0.65) (Table 2). In contrast, wve-1(ne350M+); mig-2(mu28) double mutants were much more severe than wve-1(ne350M+) alone (2.7 vs. 0.83 misguided axons and 1.73 vs. 1.07 axons on the left side; all P values < 0.0001) (Table 2 and Figure 5B). Thus, mig-2(mu28) enhanced wve-1 more strongly than ced-10(n1993), similar to the situation in the PDE neuron.
The wsp-1(gm324); mig-2(mu28) double was no more severe than wsp-1(gm324) alone (P = 0.28 and 0.31 for guidance and left-side defects, respectively) (Table 2). In contrast, the wsp-1(gm324) ced-10(n1993) double mutant showed significantly more defects than wsp-1(gm324) alone and the wsp-1(gm324); mig-2(mu28) double (Table 2 and Figure 5D). wsp-1(gm324) ced-10(n1993) averaged 3.16 guidance defects and 1.76 left-side defects per animal compared to 1.3 and 0.02 for wsp-1(gm324) alone (P < 0.0001 for all differences). In agreement with PDE neuron data, ced-10 enhanced wsp-1 whereas mig-2 did not. Together, these data are consistent with the idea that mig-2 acts with wsp-1 in a pathway in parallel to ced-10 and wve-1 in VD/DD axon guidance.
Transgenic expression of the WH2 domains of WVE-1 and WSP-1 causes neuronal defects:
In cultured mammalian cells, overexpression of the WH2 Arp2/3 activation domains of WASP family members leads to excess cellular protrusions (Yamaguchi et al. 2000). To determine if WVE-1 and WSP-1 WH2 domains were similarly active in neurons, the WH2 domains of WVE-1 and WSP-1 were placed under the control of the osm-6 promoter for expression in the PDE neurons. PDE defects were observed in transgenic animals, including axon guidance defects, ectopic axon branches, and a low proportion of cell-body and axon morphological defects that resembled the lamellipodia and filopodia caused by activated MIG-2 and CED-10 (Figure 6). These defects were never observed in animals expressing GFP from the osm-6 promoter.
gex-2/Sra-1 and gex-3/Kette affect axon guidance:
Mutation of gex-3 was previously shown to affect cell movements in gastrulation resulting in failure of the endodermal cells to be enclosed by the ectoderm (the Gex phenotype) (Soto et al. 2002). gex-3(zu196) is a Tc1 transposon insertion in the tenth exon (out of 12) (Soto et al. 2002) and might not cause complete loss of gex-3 activity. RNAi of gex-2 gave a similar gastrulation phenotype to that of gex-3(zu196).
To determine if GEX-2 and GEX-3 affect neuronal development, the PDE axons were analyzed in gex-2 and gex-3 mutations. A new deletion mutation in gex-2, called ok1603, was isolated and provided by the C. elegans Gene Knockout Consortium). The gex-2(ok1603) deletion removed part of exon 2, all of exons 3 and 4, and part of intron 4 (Figure 1C). gex-2(ok1603) displayed a similar gross morphological phenotype as gex-3(zu196M+). Homozygotes from heterozygous mothers grew to adulthood and were slightly Unc, Egl, and pVul.
Both gex-2(ok1603M+) and gex-3(zu196M+) animals displayed defects in PDE axon guidance (21% in gex-2(ok1603M+) and 15% in gex-3(zu196M+) (Figure 7A). For gex-2(ok1603), the PDE axon guidance defects as well as the Unc, Egl, and pVul defects were rescued by transgenic expression of a wild-type gex-2(+) gene amplified by PCR (see materials and methods; Figure 7A; 21% PDE guidance defects compared to 3% in transgenic gex-2(+) animals, P < 0.0001). RNAi by feeding animals gex-2 dsRNA caused embryonic and larval lethality, and viable animals were Unc, pVul, and Dpy. PDE axon defects were detected in viable gex-2(RNAi) animals (11%) (Figure 7A). These data indicate that the axon defects in gex-2(ok1603M) were likely due to loss of gex-2 gene function.
In gex-2(ok1603M+) and gex-3(zu196M+), a similar spectrum of PDE guidance defects was observed as seen in wve-1(ne350M+): PDE axons wandered laterally and sometimes emanated from the sides of the PDE cell body instead of from the ventral surface as in wild type (Figure 7, B and C). Occasionally, ectopic axon branches or neurites were observed in gex-2(ok1603M+) and gex-3(zu196M+) PDE axons, but these were generally in axons that were already misguided as described for wve-1(ne350). Thus, gex-2 and gex-3 might affect axon guidance but not ectopic axon formation, similar to wve-1 and wsp-1 (Table 1).
Mutations in ced-10/Rac and mig-2/RhoG enhance gex-2/Sra-1 and gex-3/Kette:
PDE axon guidance defects of gex-2(ok1603M+) and gex-3(zu196M+) were strongly enhanced by both mig-2(mu28) and ced-10(n1993) (Figure 7A). For example, ced-10(n1993); gex-2(ok1603M+) displayed 50% PDE guidance defects compared to 19% for gex-2(ok1603M+) alone (P < 0.0001), and mig-2(mu28); gex-3(zu196M+) displayed 62% compared to 15% for gex-3(zu196M+) alone (P < 0.0001). While some ectopic neurites were observed, these were generally on axons that were already misguided as described above (Table 1). These data suggest that gex-2 and gex-3 have overlapping function with ced-10 and mig-2 in axon guidance. GEX-2 and GEX-3 might act in both the CED-10 and MIG-2 pathways and therefore display overlapping function with both pathways.
gex-3(zu196M+); mig-2(mu28) and gex-3(zu196M+); ced-10(n1993) displayed defects not observed in the double mutants with gex-2(ok1603M+), including a high-penetrance PDE cell-body defect: cell bodies were larger in size and multiple sheet-like and finger-like protrusions emanated from the cell bodies (Figure 7D and Table 1). The nature of this defect is unclear, but it resembled the ectopic lamellipodia and filopodia formation induced by MIG-2 and CED-10 overactivation in the PDE (Struckhoff and Lundquist 2003). Another defect seen in gex-3(zu196M+) double mutants with mig-2(mu28) and ced-10(n1993) and not observed in other genotypes, including mig-2(mu28); ced-10(n1993) double mutants, was apparent PDE axon termination (Figure 7E and Table 1). Instead of wandering along the body wall, the PDE axon terminated prematurely and was often thicker than the normal axon. While the nature of this defect is not understood, it could represent a failure in the axon to extend or elongate. Cell-body and axon-termination defects were generally more severe in mig-2(mu28); gex-3(zu196M+) than ced-10(n1993); gex-3(zu196M+) animals. These phenotypic differences between gex-2 and gex-3 could be because the gex-2(ok1603) allele caused a less severe reduction of function than did gex-3(zu196). This is unlikely, as gex-2(ok1603) is a deletion whereas gex-3(zu196) is a transposon insertion which may cause incomplete loss of function. Alternatively, the phenotypic differences between gex-2 and gex-3 could represent distinct roles of gex-2 and gex-3 in neuronal development (i.e., both gex-2 and gex-3 affect axon guidance, but only gex-3 is involved in cell shape and axon extension).
Together, these data suggest that gex-2 and gex-3 might act redundantly with mig-2 and ced-10 in PDE axon guidance and that gex-3 might act in parallel to mig-2 and ced-10 in multiple other aspects of neuronal development including regulation of cell-body shape and axon extension.
gex-2 and gex-3 RNAi do not enhance guidance defects of the wsp-1; wve-1(M+) double mutant:
Enhancement of gex-2 and gex-3 by mig-2 and ced-10 could mean that the gex genes act in both the wve-1/ced-10 and mig-2/wsp-1 pathways, or that the gex genes act in a pathway parallel to these. gex-2 and gex-3 were silenced using feeding RNAi in the wsp-1(gm324); wve-1(ne350M+) background. As expected, gex-2 and gex-3 RNAi resulted in embryonic lethality. However, viable surviving animals displayed no enhancement of PDE axon guidance defects (Figure 2G): wsp-1(gm324); wve-1(ne350M+) displayed 52% PDE guidance defects compared to 44% after treatment with gex-2 RNAi (P = 0.31) and 54% after treatment with gex-3 RNAi (P = 0.80). A caveat of this experiment is that viable, surviving animals might be those least affected by RNAi. However, enhancement of axon defects of mig-2 were observed with wve-1 RNAi in surviving animals (Figure 2G). The wsp-1(gm324); wve-1(ne350M+) guidance defect was more severe than either gex-2(ok1603M+)or gex-3(zu196M+) alone, a result not expected if the molecules act largely in the same pathway. This could be explained by the wild-type maternal product supplied in gex-2(ok1603M+) and gex-3(zu196M+) animals [i.e., true loss of all gex-2 and gex-3 activity in the PDE might resemble wsp-1; wve-1(M+)].
unc-115/abLIM enhances gex-2, gex-3, wve-1, and wsp-1 in PDE axon guidance:
Previous studies implicated unc-115/abLIM, which encodes an actin-binding protein, as a downstream effector of Rac signaling during neuronal development (Struckhoff and Lundquist 2003). unc-115 mutations synergized with mig-2 and ced-10 and partially suppressed the ectopic lamellipodia and filopodia induced by activated Rac. unc-115 double mutants with gex-2(ok1603M+), gex-3(zu196M+), wve-1(ne350M+), and wsp-1(gm324) all displayed enhanced PDE axon guidance defects compared to each single alone (Figure 8, A–C, and F). For example, unc-115(ky275); gex-2(ok1603M+) displayed 62% PDE guidance defects (all differences of double mutants from singles P < 0.0001). That unc-115 enhances each of these mutations could mean that unc-115 acts in parallel to each pathway. Alternatively, it could mean that unc-115 acts in each pathway (i.e., is controlled by both ced-10 and mig-2). The latter explanation is supported by other studies that show that unc-115 might act downstream of Rac signaling in PDE development and that the SWAN-1 7-WD repeat protein interacts physically with UNC-115 as well as with MIG-2 and CED-10 and might connect UNC-115 to Rac signaling (Yang et al. 2006). unc-115, mig-2 and ced-10 mutants also affected formation of ectopic neurites (Lundquist et al. 2001; Struckhoff and Lundquist 2003), but double mutants of unc-115 with gex-2, gex-3, wve-1, and wsp-1 did not display increased ectopic neurite formation.
unc-34 Ena enhances wve-1 and wsp-1 in PDE axon guidance:
Previous studies showed that unc-34, which encodes C. elegans Enabled, acts in parallel to mig-2 and ced-10 in PDE axon guidance. Furthermore, it was shown that wve-1 and unc-34 act redundantly in VD and DD motor axon guidance and that wsp-1 and unc-34 have overlapping roles in CAN cell migration (Withee et al. 2004). Mutations in unc-34 alone were previously shown to affect PDE axon guidance (Shakir et al. 2006).
wve-1(ne350M+); unc-34(e951M+) animals showed increased PDE axon guidance defects compared to either single alone (P < 0.0001) (Figure 8, D and G). The unc-34(e951); wsp-1(gm324) double mutant was maternal-effect lethal with embryonic arrest as previously described. PDE axon guidance was assessed in unc-34(e951M+); wsp-1(gm324M+) animals with wild-type maternal contribution of both unc-34 and wsp-1. In these animals, 18% PDE axon guidance defects were found compared to 5% for unc-34(e951M+) alone (P < 0.0001) (Figure 8, E and F). This relatively weak effect could be due to the wild-type maternal contributions of each locus. These data indicate that WVE-1, WSP-1, and UNC-34 Ena have overlapping roles in PDE axon guidance. As with UNC-115, these data could mean that UNC-34 Ena acts in parallel to WVE-1 and WSP-1 or that it acts in each pathway. While the data here do not distinguish these possibilities, it is reasonable to think that UNC-34 Ena represents a distinct pathway to the actin cytoskeleton in parallel to WVE-1, WSP-1, Rac GTPases, and UNC-115. This is supported by the finding that UNC-34 Ena acts independently of WASP-1 in cell movements in embryonic gastrulation (Sheffield et al. 2007).
The complex architecture of the actin cytoskeleton of the growth cone is likely regulated by multiple actin regulatory proteins. The mechanisms that control these molecules during axon guidance remain to be described. This is due in part to the overlapping roles that many of these molecules have in axon guidance and other developmental events, resulting in an apparently wild-type phenotype in mutants of genes encoding these molecules. The use of double-mutant analysis has been useful to discover overlapping roles of these molecules and to understand how these molecules work together in signaling pathways and networks.
This work describes genetic interactions in axon guidance of a group of molecules known to regulate actin dynamics downstream of Rac GTPases. Biochemical studies suggest that the Arp2/3 activator WAVE is part of a complex of proteins including Sra-1 and Kette that regulate WAVE function. Activated Rac-GTP activates this complex allowing WAVE to interact with and activate Arp2/3 (Eden et al. 2002; Innocenti et al. 2004; Steffen et al. 2004; Stradal et al. 2004; Vartiainen and Machesky 2004). Here are described experiments that implicate a Rac/Sra-1/Kette/WAVE pathway in axon pathfinding and that suggest that a distinct Arp2/3 activator, WASP, might act with the Rac-like molecule MIG-2 in parallel to the Rac/Sra-1/Kette/WAVE pathway.
WAVE and WASP have overlapping roles in axon pathfinding:
wve-1 mutations and RNAi had relatively mild effects on axon pathfinding, raising the possibility that another molecule acts in parallel to WVE-1 in axon guidance. Previous studies showed that WVE-1 and the C. elegans WASP protein WSP-1 have overlapping roles in cell migration in C. elegans (Withee et al. 2004; Sheffield et al. 2007). Work here shows that they also have overlapping roles in axon guidance: wsp-1 and wve-1 alone had little effect on axon guidance, whereas wve-1; wsp-1 double mutants displayed severe axon guidance defects. Both WVE-1 and WSP-1 have C-terminal WH2 domains, which are activators of the Arp2/3 complex in other systems. Thus, WVE-1 and WSP-1 might act redundantly to control the Arp2/3 complex in axon guidance. Consistent with this idea, overexpression of the WH2 domains of WVE-1 and WSP-1 in the PDE neuron led to neuronal defects including guidance errors, ectopic axon formation, and ectopic lamellipodia and filopodia formation. In cultured mammalian cells, overexpression of these domains also led to the formation of cellular protrusions thought to be mediated by Arp2/3 activation (Yamaguchi et al. 2000).
WSP-1/WASP and MIG-2/RhoG act in a pathway parallel to WVE-1/WAVE and CED-10/Rac:
ced-10/Rac and mig-2/RhoG have overlapping roles in axon pathfinding (Lundquist et al. 2001). Here we present genetic evidence that CED-10/Rac might act in the WVE-1 pathway, consistent with previous biochemical data. Mutation and RNAi of wve-1 caused few axon defects alone. ced-10; wve-1 double mutants also displayed few axon guidance defects whereas mig-2; wve-1 double mutants displayed robust axon guidance defects. This loss-of-function result suggests that CED-10/Rac and WVE-1 might act in the same pathway in parallel to MIG-2/RhoG. Furthermore, the wve-1(ne350) mutation partially suppressed the ectopic lamellipodia and filopodia on PDE neurons driven by activated CED-10 expression in these cells (i.e., the effects of activated CED-10 required functional WVE-1). This is consistent with biochemical studies in which WAVE was shown to be downstream of Rac. This pathway might be conserved in C. elegans axon guidance.
Furthermore, data presented here suggest that WSP-1/WASP might act downstream of the MIG-2/RhoG GTPase in axon guidance. ced-10; wsp-1 double mutants had much more severe axon guidance defects than mig-2; wsp-1. Furthermore, wsp-1(gm324) loss of function weakly suppressed the ectopic neurites and lamellipodia and filopodia formed on the PDE neuron by the activated mig-2(rh17) mutation. That the wsp-1(gm324); mig-2(rh17) double was viable and that the axon defects were certainly no worse than mig-2(rh17) alone are important observations, as wve-1(ne350); mig-2(rh17) double mutants were embryonic lethal, indicative of multiple pathways being perturbed in the double. wsp-1; mig-2 double mutants did have significant axon defects (although not as strong as wsp-1 ced-10), and wsp-1(gm324) also slightly suppressed activated ced-10. Possibly, WSP-1 also acts in the CED-10 pathway in parallel to MIG-2 in axon guidance.
The regulation of WASP by Cdc42-subfamily GTPases is well documented (Rohatgi et al. 1999; Miki and Takenawa 2003; Ho et al. 2004; Tomasevic et al. 2007). Previous studies suggested that WASP was controlled by intramolecular autoinhibitory activity regulated by PIP2 and Cdc42 (Kim et al. 2000; Rohatgi et al. 2000; Ho et al. 2004; Vartiainen and Machesky 2004). Indeed, PIP2 is involved in C. elegans axon guidance (Weinkove et al. 2008). Studies reported here imply that WSP-1/WASP might be controlled by the MIG-2 Rac/RhoG GTPase in C. elegans axon guidance. Furthermore, these studies imply that the WSP-1/WASP molecule might act in a pathway with GEX-2/Sra-1 and GEX-3/Kette. The action of WASP in a regulatory complex of proteins is not unprecedented, as recent studies have shown that Toca-1 and WIP act with Cdc42 in WASP regulation (Ho et al. 2004). Possibly, multiple mechanisms control WASP function in axon guidance, including Cdc42-Toca-1-PIP2 as well as a potential MIG-2/RhoG-Sra-1-Kette pathway.
GEX-2/Sra-1 and GEX-3/Kette control axon pathfinding:
gex-2/Sra-1 and gex-3/Kette mutations had defects in PDE axon guidance that are enhanced by mutations in both ced-10/Rac and mig-2/RhoG. These data could be interpreted to mean that GEX-2 and GEX-3 act in parallel to both GTPases. A more parsimonious explanation that incorporates known biochemical and genetic interactions is that GEX-2/Sra-1 and GEX-3/Kette act in both the CED-10/Rac and MIG-2/RhoG pathways (Figure 9). Enhancement of each by ced-10 and mig-2 could be due to alternate pathways not involving GEX-2 and GEX-3 downstream of each GTPase. Indeed, UNC-115/abLIM might be an alternate pathway to the actin cytoskeleton downstream of Rac signaling (i.e., loss of CED-10 enhances loss of GEX-2 and GEX-3 because CED-10 also utilizes UNC-115/abLIM) (Struckhoff and Lundquist 2003). Consistent with this idea, loss of UNC-115 enhanced loss of GEX-2 and GEX-3 as well as loss of WVE-1 and WSP-1. Loss of UNC-34/Enabled also enhanced loss of WVE-1 and WSP-1, consistent with previous results showing that UNC-34 had overlapping function with CED-10, MIG-2, and UNC-115 (Shakir et al. 2006). UNC-34/Enabled might represent a third distinct pathway to the cytoskeleton in axon guidance.
Some evidence suggests that the WAVE regulatory complex containing Sra-1 and Kette exerts a negative influence on WAVE that is relieved by activated Rac (Eden et al. 2002), whereas other data suggest that the complex is required for WAVE activation of Arp2/3 (Innocenti et al. 2004). In these studies, axon guidance defects of gex-2 Sra-1 and gex-3/Kette mutants resembled loss of function of wve-1/WAVE. This is in agreement with what has been seen in studies of embryonic development (M. C. Soto, unpublished results).
mig-2; gex-3 and ced-10; gex-3 double mutations displayed defects not seen in other mutant combinations, including premature axon termination and ectopic cell-body protrusions that resembled lamellipodia and filopodia produced by activated CED-10 and MIG-2 expression in neurons. These data suggest that GEX-3/Kette might have roles in axon pathfinding independent of GEX-2 Sra-1 and WVE-1/WAVE.
In sum, these results support a model in which GEX-2/Sra-1 and GEX-3/Kette act in both of the parallel CED-10/Rac and MIG-2/RhoG pathways in axon guidance (Figure 9), an idea supported by the result that gex-2 and gex-3 RNAi did not enhance guidance defects of wsp-1(gm324); wve-1(ne350M+). These data also suggest that a distinct Arp2/3 activator might be used in each pathway (WVE-1/WAVE in the CED-10/Rac pathway and WSP-1/WASP in the MIG-2/RhoG pathway). Furthermore, data presented here support previous studies indicating that multiple pathways in addition to WAVE/WASP are employed in signaling to the cytoskeleton during axon guidance, including UNC-115/abLIM, which also acts with Rac signaling, and UNC-34/Enabled (Figure 9). Furthermore, previous studies suggest that Rac signaling also utilizes UNC-115/abLIM (Struckhoff and Lundquist 2003), but it is unknown if the WSP/WAVE/GEX complex is involved in this interaction. How each of these pathways (WAVE/WASP, UNC-115/abLIM, and UNC-34/Enabled) contribute to axon guidance at the cellular and cytoskeletal level will be the subject of further study.
The authors thank the Caenhorhabditis Genetics Center, sponsored by the National Institutes of Health (NIH) National Institute of Research Resources, for C. elegans strains, the C. elegans Gene Knockout Consortium (Bob Barstead, Gary Moulder, Mark Edgley, and Don Moerman) for gex-2(ok1603), Y. Jin and A. Chisholm for juIs73, and Brian Ackley and members of the Lundquist lab for helpful discussions. This work was supported by National Science Foundation (NSF) grant 0641123 to M.C.S., NIH grant P20 RR-016475 from the Kansas Infrastructure Network of Biomedical Research Excellence Program of the National Center for Research Resources (J. Hunt, Principal Investigator), and NIH grant NS-40945 and NSF grant IOS93192 to E.A.L.
↵1 These authors contributed equally to this work.
↵2 Present address: Department of Neurology, Baylor College of Medicine, Neurosensory Center, 6501 Fannin, Suite NB302, Houston, TX 77030.
↵3 Present address: Department of Neurology, Mount Sinai School of Medicine, One Gustave L. Levy Pl., New York, NY 10029.
Communicating editor: K. Kemphues
- Received March 6, 2008.
- Accepted May 14, 2008.
- Copyright © 2008 by the Genetics Society of America