The nerve-cell cytoskeleton is essential for the regulation of intrinsic neuronal activity. For example, neuronal migration defects are associated with microtubule regulators, such as LIS1 and dynein, as well as with actin regulators, including Rac GTPases and integrins, and have been thought to underlie epileptic seizures in patients with cortical malformations. However, it is plausible that post-developmental functions of specific cytoskeletal regulators contribute to the more transient nature of aberrant neuronal activity and could be masked by developmental anomalies. Accordingly, our previous results have illuminated functional roles, distinct from developmental contributions, for Caenorhabditis elegans orthologs of LIS1 and dynein in GABAergic synaptic vesicle transport. Here, we report that C. elegans with function-altering mutations in canonical Rac GTPase-signaling-pathway members demonstrated a robust behavioral response to a GABAA receptor antagonist, pentylenetetrazole. Rac mutants also exhibited hypersensitivity to an acetylcholinesterase inhibitor, aldicarb, uncovering deficiencies in inhibitory neurotransmission. RNA interference targeting Rac hypomorphs revealed synergistic interactions between the dynein motor complex and some, but not all, members of Rac-signaling pathways. These genetic interactions are consistent with putative Rac-dependent regulation of actin and microtubule networks and suggest that some cytoskeletal regulators cooperate to uniquely govern neuronal synchrony through dynein-mediated GABAergic vesicle transport in C. elegans.
EPILEPSY affects 1–2% of the world population and is associated with imbalances between excitatory and inhibitory neurotransmission in the brain (Locke et al. 2009). In particular, interneurons expressing gamma-aminobutyric acid (GABA), the principal inhibitory neurotransmitter in the human brain, are essential for normal neuronal synchronization and maintenance of a seizure threshold in humans (Cossette et al. 2002), rodents (Delorey et al. 1998), and zebrafish (Baraban et al. 2005). A failure of the brain to properly regulate neuronal synchrony can result from ion channel defects (Xu and Clancy 2008), neuropeptide depletion (Brill et al. 2006), brain malformations (Patel et al. 2004), interneuron loss (Cobos et al. 2005), and/or synaptic vesicle recycling failure (Di Paolo et al. 2002), all of which may be caused by disrupting the nerve-cell cytoskeleton. Therefore, further exploration of putative links between cytoskeletal components and neurotransmission may accelerate development of novel therapeutics for epilepsy.
Epilepsy associated with cytoskeletal dysfunction often has a developmental basis (Di Cunto et al. 2000; Wenzel et al. 2001; Keays et al. 2007). For example, mutations in LIS1, a dynein motor complex regulator, lead to classical lissencephaly, which is characterized by neuronal migration defects, a lack of convolutions in the brain, mental retardation, and epileptic seizures (Lo Nigro et al. 1997). Yet, observations that lissencephaly-associated seizures worsen after neuronal migration ceases, while LIS1 expression persists, imply that LIS1 also acts in the adult brain (Cardoso et al. 2002).
We previously reported that C. elegans with a predicted null mutation (t1550) in lis-1, the worm ortholog of human LIS1, exhibited synaptic vesicle misaccumulations, but not neuronal migration or axon-pathfinding defects, in GABAergic motor neurons. We also observed anterior “epileptic-like” convulsions, which were intense, frequent, and repetitive, with lis-1(t1550) homozygotes in the presence of pentylenetetrazole (PTZ; Williams et al. 2004), an epileptogenic GABAA receptor antagonist (Huang et al. 2001; Fernandez et al. 2007). PTZ sensitivity was also increased in heterozygous lis-1(t1550) mutants following RNA interference (RNAi) against worm orthologs of associated cortical malformation genes, such as cdk-5 and nud-2, which are known to interact with LIS1 and the dynein motor complex. Depletion of these gene products was coincident with dynein-mediated synaptic vesicle transport defects, not with architectural defects, in GABAergic motor neurons (Locke et al. 2006).
Plausible functional interactions among LIS-1, dynein, and Rac GTPases (Rehberg et al. 2005; Kholmanskikh et al. 2006) have not been explored in an intact adult nervous system. C. elegans is ideal for characterizing these interactions due to the availability of weak and strong Rac pathway mutants (Lundquist et al. 2001; Poinat et al. 2002; Lucanic et al. 2006), a comprehensive RNAi library (Kamath et al. 2003), and GFP-based neuronal markers. Here, we combine these tools with pharmacological modifiers of neuronal activity and establish an experimental paradigm that reveals a novel regulatory pathway. This pathway is composed of integrins at the plasma membrane that signal through Racs to dynein-associated proteins, which function to coordinate synaptic vesicle transport in larval and adult GABAergic motor neurons.
MATERIALS AND METHODS
Worm strains and maintenance:
C. elegans were maintained via standard procedures (Brenner 1974). The following strains were used: Bristol N2, avr-14(ad1302) avr-15(ad1051) glc-1(pk54), cat-2(e1112), ced-2(n1994), ced-5(n1812), ced-10(n1993), ced-10(n3246), dgk-1(nu62), dhc-1(js121)/hT2[bli-4(e937) let-?(q782) qIs48]; jsIs37, eat-4(ky5), egl-8(md1971), egl-10(md176), goa-1(sa734), ina-1(gm39), ina-1(gm144), lis-1(t1550) unc-32(e189)/qC1 dpy-19(e1259) glp-1(q339); him-3(e1147), mig-2(gm103), mig-2(mu28), mig-15(rh80), mig-15(rh148), mig-15(rh326), rab-3(y251), swan-1(ok267), tom-1(ok188), tph-1(mg280), unc-5(e53), unc-17(e245), unc-25(e156), unc-26(e1196), unc-30(e191), unc-32(e189), unc-34(e315), unc-34(e566), unc-40(n324), unc-49(e407), unc-51(e396), unc-73(e936), unc-73(ev802), unc-73(rh40), unc-115(ky275), Punc-115∷mig-2(G16V), Punc-115∷rac-2(G12V), juIs1 (Punc-25∷SNB-1∷GFP), and oxIs12 (Punc-47∷GFP).
Behavioral and pharmacological assays:
Convulsion assays were performed, as previously described (Williams et al. 2004; Locke et al. 2008). Concentrations of PTZ (Sigma) employed are indicated in the text. Aldicarb-induced paralysis assays also were performed, as described (Nonet et al. 1998), by transferring young adult hermaphrodites to NGM plates with 0.5 mm aldicarb (Supelco). Worms were observed in 30-min intervals for a period of 3 hr.
For thrashing assays, 1-day-old adult worms were washed clean of bacteria with M9, transferred to 7.5-ml NGM plates with 2 ml M9, and allowed to recover for 2 min. Movies of worms thrashing were captured in real-time with a Q Imaging Retiga Exi digital video camera at 25 frames/sec. Movies were saved onto an Intel Pentium computer using ImageJ (National Institutes of Health, Bethesda, MD) and scored at a reduced frame rate for accuracy. A thrash was defined as a change in direction at the worm mid-body.
Other methods used:
Details of the rescue of ina-1 are available in the supporting information, File S1. Descriptions of the genetic crosses and fluorescence microscopy analyses used to observe possible alterations in GABAergic D-type motor neuron architecture and/or synaptic vesicle distribution are provided in File S1. RNAi by bacterial feeding was performed, as described (Locke et al. 2006), with slight modifications (see File S1).
Statistical analyses of all data sets, except for those obtained from thrashing assays, were performed using Fisher's exact test (http://www.langsrud.com/fisher.htm). Results given are two-tail P-values, which were found by comparing two appropriate data sets for specified comparisons. Statistical analyses of data sets from thrashing assays were performed using Student's t-test. Data are shown as mean ± SD and were deemed significant, if P < 0.05, for both statistical tests.
PTZ induces anterior convulsions in Rac GTPase-signaling-pathway mutants:
We previously demonstrated that worms lacking LIS-1 exhibit PTZ-induced anterior convulsions. We also found these behavioral responses to be associated with loss of GABA because systemic GABA mutants (i.e., unc-25 and unc-47) demonstrated similar responses to PTZ (Williams et al. 2004). Moreover, we previously used PTZ to reveal genetic interactions among lis-1 and other microtubule-dependent cortical malformation genes (Locke et al. 2006). Intriguingly, LIS-1 orthologs have also been shown to interact with Rac1 in Dictyostelium and mammals to shape the actin cytoskeleton (Kholmanskikh et al. 2003; Rehberg et al. 2005). These findings suggest an evolutionarily conserved role for LIS-1 and actin regulators, whose impact on intrinsic neuronal activity is undefined.
To determine if C. elegans Racs modulate neuronal synchrony, we exposed loss- and gain-of-function Rac mutants to PTZ. The products of worm ced-10, mig-2, and rac-2 genes act redundantly in multiple developmental processes, including axon pathfinding in GABAergic D-type motor neurons (Lundquist et al. 2001). The Rac gain-of-function (gf) mutant allele, ced-10(n3246), confers axon-pathfinding defects similar to those resulting from multiple Rac loss-of-function (lf) mutations and is predicted to interfere with a common guanine nucleotide exchange factor (GEF) of redundant Racs (Shakir et al. 2006). A second Rac gf mutant allele, mig-2(gm103), has been shown to perturb axon pathfinding (Zipkin et al. 1997), vulval cell migration (Kishore and Sundaram 2002), and male tail development (Dalpé et al. 2004) in a manner consistent with cross-inhibition of redundant Rac pathways (Zipkin et al. 1997; Lundquist et al. 2001). We found that, like a strong systemic GABA mutant (Figure 1A; Figure S1; File S2), both ced-10(n3246) (Figure 2; File S3) and mig-2(gm103) (Figure S1; File S4) mutants had anterior PTZ-induced convulsions (Figure 1A). The frequency and intensity of convulsions positively correlated with PTZ concentration, similar to lis-1(t1550) and systemic GABA mutant responses to PTZ (Williams et al. 2004). Neither of the Rac lf muta nts, ced-10(n1993) and mig-2(mu28), had altered PTZ sensitivity, compared to C. elegans N2, wild-type, worms (Figure 1A). A putative rac-2 lf allele, ok326, did not alter PTZ sensitivity (Table 1). These data suggest that CED-10, MIG-2, and perhaps RAC-2 play redundant roles in controlling C. elegans inhibitory GABA transmission.
Rac GTPases are known to function in a variety of neurobiological processes, including neurite branching and extension, axon pathfinding, and synapse formation. Furthermore, Rac GTPases are regulated by a number of other proteins, which are not simply limited to GEFs and GTPase-activating proteins (GAPs) (de Curtis 2008). To determine which, if any, known C. elegans Rac regulators modulate neuronal synchrony, we exposed an array of Rac regulator mutants to PTZ. A Rac GTPase-signaling pathway, which is governed by interactions between C. elegans orthologs of integrins and Nck-interacting kinase (NIK), mediates GABAergic D-type motor neuron axon pathfinding (Poinat et al. 2002). Consistent with a role for these Rac regulators in GABAergic neurotransmission, multiple ina-1 (α-integrin) and mig-15 (NIK) hypomorphs demonstrated anterior epileptic-like convulsions in response to PTZ (Figure 1B; Figure 2; Figure S1; File S5; File S6). The percentages of mig-15 mutants with anterior convulsions (Figure 1B) were roughly coincident with the strengths of mutant alleles, where rh326 > rh80 > rh148 (Shakir et al. 2006). Yet, the frequencies of rh326 and rh80 mutant convulsions were not significantly different from each other (Figure 1B). Surprisingly, ina-1(gm39) mutants had higher percentages of PTZ-induced convulsions at all concentrations than ina-1(gm144) mutants, which are predicted from cell migration defects to be weaker (Baum and Garriga 1997) (Figure 1B). Perhaps gm39 disrupts the binding of a different number and/or class of ligands than gm144 disrupts (Baum and Garriga 1997). Future studies with candidate INA-1 ligands should assess this hypothesis.
We also observed PTZ-induced convulsions with a pair of unc-73 mutants, which are deficient in UNC-73 GEF-dependent activation of the triply redundant Racs (Struckhoff and Lundquist 2003). The weaker unc-73(e936) mutants, which carry a splice donor mutation that disrupts all Rac GEF-containing isoforms of UNC-73, had higher percentages of PTZ-induced convulsions than stronger unc-73(rh40) mutants, which lack UNC-73 guanosine-5′-diphosphate to guanosine-5′-triphosphate exchange activity (Steven et al. 1998) (Figure 1B; Figure S1; File S7). This result could be due to disruption of GEF2 Rho GEF activity by e936 (Steven et al. 1998). GEF2 Rho GEF activity may contribute to PTZ sensitivity, as it has been previously linked to normal synaptic transmission in C. elegans (Steven et al. 2005; Williams et al. 2007). Yet, GEF2 Rho GEF-deficient unc-73(ev802) mutants (Steven et al. 2005) did not exhibit altered PTZ sensitivities (Table 1), indicating that loss of GEF2 Rho GEF activity alone is not sufficient to allow for PTZ-induced convulsions.
We also did not observe altered PTZ sensitivities in ced-2(n1994) or ced-5(n1812) lf mutants (Table 1), which disrupt Rac-dependent cell corpse engulfment (Kinchen et al. 2005), cell migration, and axon pathfinding (Wu et al. 2002), suggesting that a different Rac-dependent mechanism underlies PTZ-induced convulsions. Likewise, we were unable to detect PTZ-induced convulsions with unc-40(n324) netrin receptor null mutants (Table 1), which are deficient in Rac-dependent GABAergic D-type motor neuron axon pathfinding (Lucanic et al. 2006) and non-GABAergic neuron axon pathfinding (Gitai et al. 2003). We also did not observe PTZ-induced convulsions with multiple lf mutants of unc-34, which has been shown to function in parallel to Rac mutants, downstream of UNC-40, in C. elegans axon pathfinding (Gitai et al. 2003; Lucanic et al. 2006; Shakir et al. 2006) or a second netrin receptor null mutant, unc-5(e53) (Lucanic et al. 2006; Table 1). Mutants lacking SWAN-1, which negatively regulates Rac mutants in neuronal and non-neuronal cells (Yang et al. 2006), or UNC-115, which positively regulates RAC-2, but not CED-10 or MIG-2, in GABAergic D-type motor neuron axon pathfinding (Struckhoff and Lundquist 2003), also did not have PTZ-induced convulsions (Table 1). These data suggest that UNC-73 activation of redundant Rac mutants, not Rho, lowers PTZ sensitivity in a manner that depends on integrins, not netrins. We have found that Rac-signaling mutants phenocopy lis-1(t1550) and systemic GABA mutants by exhibiting anterior convulsions on PTZ (Table 2). Yet, generalized synaptic transmission defects lower a threshold for PTZ-induced convulsions, as another general synaptic transmission mutant, rab-3(y251), convulses on PTZ (Figure 1, A and B; Figure 2; File S8). Thus, we cannot resolve if this Rac pathway plays a GABA-specific role in worm neurons by PTZ.
The C. elegans inhibitory GABAergic nervous system is composed of 19 D-type motor neurons that innervate body-wall muscles and four nerve ring RME (ring motor) neurons that innervate head muscles (McIntire et al. 1993). Deficits in a single class of inhibitory GABAergic neurons could be sufficient for PTZ-induced convulsions. To address this possibility, we examined unc-30(e191) null mutants, which express wild-type levels of GABA in RME neurons but fail to express GABA in D-type motor neurons (Jin et al. 1994). We did not observe convulsions in these mutants with any PTZ concentration tested (Table 1), implying that RMEs may be chiefly responsible for this behavioral phenotype. Yet, these data do not rule out a contributory role for D-type motor neurons in convulsions.
GABA may not be the only source of inhibitory transmission at head muscles because unc-49(e407)-predicted null GABAA receptor mutants respond to PTZ in a dose-dependent manner (Williams et al. 2004). The C. elegans genome contains sequences for many other GABAA receptor homologs, including glutamate-gated chloride (GluCl) channels, which could function redundantly with unc-49 (Schuske et al. 2004). In support of this argument, predicted null unc-49 mutants have been shown to respond to a GABA agonist, muscimol, albeit less strongly than wild-type worms (McIntire et al. 1993). Likewise, predicted null unc-49 GABA mutants also do not display RME neuron-mediated “loopy” foraging defects, unlike other GABA-deficient worms, further suggesting redundancy among GABAA receptor homologs (McIntire et al. 1993). To elucidate roles of other neurotransmitters in PTZ-induced convulsions, we assayed mutants systemically lacking acetylcholine [unc-17(e245)] (Miller et al. 1996), dopamine [cat-2(e1112)] (Sulston et al. 1975), glutamate [eat-4(ky5)] (Rankin and Wicks 2000), or serotonin [tph-1(mg280)] (Keane and Avery 2003) on PTZ. None of these mutants had PTZ-induced convulsions (Table 1). Yet, 28.9% of triple GluCl channel mutants [avr-14(ad1302); avr-15(ad1051) glc-1(pk54)], which lack inhibitory glutamate transmission at the nerve ring (Dent et al. 2000), displayed anterior convulsions in the presence of 10 mg/ml PTZ (File S9), whereas 100.0% of these mutants had anterior convulsions in the presence of 20 mg/ml PTZ. We did not observe convulsions in N2 worms at these concentrations of PTZ. These results may explain the failure of strong GABA mutants, such as unc-25(e156), to spontaneously convulse (Williams et al. 2004). Indeed, C. elegans GluCl channels may also be perturbed by PTZ. Picrotoxin (PTX), a second epileptogenic GABAA receptor antagonist, has been shown to block invertebrate GluCl's (Etter et al. 1999). Since PTX and PTZ interact with overlapping domains of GABAA receptors (Huang et al. 2001), they may similarly block worm GluCl's. Yet, because the triple GluCl mutant convulsions are substantially less frequent and intense than convulsions of Rac-signaling mutants or systemic GABA mutants, it is likely that Rac-signaling mutants are GABA deficient. We hypothesize that loss of UNC-49 significantly lowers an intrinsic threshold, which PTZ overcomes by antagonizing other partially redundant GABAA receptor homologs on RME-innervated muscles. Future experiments, including electrophysiology, will be needed to better describe the roles of GABAA receptor homologs in C. elegans neurons.
LIS-1, dynein, and Rac-signaling-pathway mutants are hypersensitive to aldicarb:
Excitatory acetylcholine (Ach) transmission and inhibitory GABA transmission antagonize each other at C. elegans body-wall neuromuscular junctions (NMJs) (Richmond and Jorgensen 1999). Accordingly, C. elegans mutants with deficits in general synaptic transmission, such as rab-3(y251), are resistant to paralysis from an acetylcholinesterase inhibitor, aldicarb (Figure 3, A–C; Nguyen et al. 1995; Miller et al. 1996). Conversely, C. elegans mutants with an inability to negatively regulate ACh transmission (Robatzek and Thomas 2000) or positively regulate inhibitory GABA transmission, such as unc-25(e156) (Figure 3, A–C; Jiang et al. 2005; Vashlishan et al. 2008), are hypersensitive to aldicarb. To better characterize synaptic transmission defects associated with lis-1 and dynein mutations, we exposed a collection of mutants to aldicarb and monitored the time course of paralysis. The lis-1(t1550) lf mutant, which is genetically linked to unc-32(e189) and convulses on PTZ (Williams et al. 2004), was hypersensitive to aldicarb (Figure 3A). The lis-1-associated aldicarb hypersensitivity was substantial, considering that unc-32(e189) mutants were resistant to aldicarb (Figure 3A), as previously reported (Nguyen et al. 1995). Consistent with a role for the dynein motor complex in lis-1-associated aldicarb sensitivity, a hypomorphic dhc-1 mutant allele, js121, also conferred hypersensitivity to aldicarb (Figure 3A). In support of our findings, another study has shown that RNAi against other members of the C. elegans dynein motor complex results in hypersensitivity to aldicarb (Vashlishan et al. 2008). These data suggest that LIS-1 and the dynein motor complex govern neuronal synchrony at worm NMJs by controlling ACh transmission.
To determine if a Rac GTPase-signaling pathway, such as LIS-1 and the dynein motor complex, participates in synaptic transmission at C. elegans NMJs, we exposed Rac and Rac regulator mutants to aldicarb. Consistent with functionally redundant roles for Racs in synaptic transmission, both ced-10(n1993) and mig-2(mu28) lf mutants demonstrated wild-type sensitivities to aldicarb (Figure 3B). Conversely, transgenic worms, expressing constitutively active mig-2(G16V) or rac-2(G12V) under the control of the neuron-specific unc-115 promoter (Punc-115) (Lucanic et al. 2006), were hypersensitive to aldicarb (Figure 3B). Similarly, the Rac gf mutants ced-10(n3246) and mig-2(gm103) were hypersensitive to aldicarb (Figure 3B), suggesting cross-inhibition of redundant Racs.
Hypomorphic ina-1 and mig-15 mutants, as well as Rac-deficient unc-73 mutants, were also hypersensitive to aldicarb (Figure 3C). INA-1 is known to function upstream of Racs during axon pathfinding of GABAergic motor neurons (Poinat et al. 2002) and may also function upstream of Racs post-developmentally in synaptic transmission on the basis of our pharmacological results. To test if INA-1 contributes to aldicarb sensitivity through integrin and Rac-signaling pathways in GABAergic motor neurons, we used a GABA-specific unc-47 promoter (Punc-47) to drive the expression of ina-1 in various genotypic backgrounds. Overexpression of ina-1 in GABAergic neurons significantly decreased the aldicarb sensitivity of wild-type worms (Figure 3C). Overexpression of ina-1 also partially rescued the aldicarb hypersensitivity of ina-1(gm144), mig-15(rh148), and unc-73(e936) hypomorphs (Figure 3D). Conversely, ina-1 overexpression did not rescue the aldicarb hypersensitivity of the dominant Rac gf mutant or the strongest available Rac-deficient GEF mutant, unc-73(rh40) (Figure 3D), which is consistent with INA-1 functioning upstream of Racs in GABAergic motor neurons. Overexpression of ina-1 in GABAergic neurons of the Rac gf mutants n3246 and gm103 may trend toward significance (Figure 3D) due to integrin-dependent modulation of parallel Rac pathways. These results suggest that a Rac-signaling pathway, dependent upon Trio, integrin, and NIK, could mediate sensitivity to aldicarb via the same GABA-based mechanism by which it mediates sensitivity to PTZ.
Considering that rab-3(y251) mutants (Figure 1, A and B; Figure 3, A–C; Nonet et al. 1997) and other PTZ-responsive general synaptic transmission mutants (Nguyen et al. 1995; Williams et al. 2004) are resistant to aldicarb, whereas systemic GABA mutants are PTZ responsive (Figure 1, A and B; Williams et al. 2004) and hypersensitive to aldicarb (Figure 3, A–C; Jiang et al. 2005; Vashlishan et al. 2008), these results are consistent with GABA-specific roles for our Rac-signaling pathway. As corroboration, heterotrimeric G-protein-signaling mutants, which are hypersensitive to aldicarb (Robatzek and Thomas 2000; Figure S2A) from excess secretion of ACh at NMJs, did not convulse on PTZ (Figure S2B). In fact, lf mutations in two antagonists of DGK-1 and GOA-1 activity, EGL-8 and EGL-10 (Robatzek and Thomas 2000), increased susceptibility to PTZ-induced anterior convulsions (Figure S1; Figure S2B; File S10), implying that GOA-1 inhibits RME activity.
To increase confidence in the interpretation of our pharmacological assays, we also placed young adult hermaphrodites in liquid and analyzed a high-frequency, drug-independent locomotory behavior known as “thrashing.” Mutants lacking egl-10 or other synaptic transmission genes have previously been shown to exhibit reduced thrashing rates (Miller et al. 1996). Despite hypersensitivity to aldicarb, goa-1(lf) mutants did not demonstrate thrashing rates significantly different from wild-type worms. Furthermore, goa-1(lf) mutants thrashed at significantly higher rates than the general synaptic transmission mutant [rab-3(y251)], a systemic GABA mutant [unc-25(e156)], Rac gf mutants, and Rac regulator mutants (Figure S2C). Surprisingly, thrashing assays have revealed another plausible explanation for the failure of unc-25(e156) mutants to spontaneously convulse. Although unc-25(e156) mutants had significantly lower thrashing rates (127.2 ± 17.5 thrashes/min) than wild-type worms (188.2 ± 14.8 thrashes/min), these systemic GABA mutants thrashed at significantly higher rates than unc-49(e407) mutants (96.5 ± 14.6 thrashes/min; Figure S2C), which have been shown to lack detectable inhibitory GABA transmission at body-wall muscles (Richmond and Jorgensen 1999). Perhaps unc-25(e156) mutants have sufficient GABA production, which is undetectable by immunocytochemistry, in RMEs to prevent them from spontaneously convulsing. A different mutant with excess ACh secretion and associated hypersensitivity to aldicarb, tom-1(ok188) (McEwen et al. 2006), also failed to convulse in response to PTZ (Table 1) and thrashed at wild-type rates (185.3 ± 28.8 thrashes/min; n = 30 worms). These data imply that the dynein motor complex and Rac-signaling pathway control neuronal synchrony at C. elegans NMJs by promoting GABA transmission, not by negative regulation of ACh.
GABAergic synaptic vesicles misaccumulate in Rac-signaling mutants:
Despite evidence that disruption of RMEs is most responsible for PTZ-induced convulsions, GABAergic synaptic vesicle misaccumulations in D-type motor neurons correlate well with PTZ sensitivity and are reliably scored (Williams et al. 2004; Locke et al. 2006). Our previous work demonstrates that misaccumulations of synaptic vesicles in GABAergic D-type motor neurons are associated with depletion of LIS-1 and dynein motor complex activity (Williams et al. 2004; Locke et al. 2006). Although Rac GTPase-signaling-pathway mutants mimic lis-1 and dynein motor complex mutants in the presence of PTZ or aldicarb, we cannot assume that their sensitivities to these neural stimulants arise from a common mechanism. A Rac pathway functions in inhibitory GABAergic motor neuron development (Lundquist et al. 2001; Poinat et al. 2002; Lucanic et al. 2006). Thus, Rac pathway mutants may have increased susceptibilities to aldicarb and/or PTZ as a result of architectural defects in the GABAergic nervous system.
To determine if a Rac pathway contributes to post-developmental GABAergic synaptic vesicle transport, we assayed for axonal gaps with SNB-1∷GFP or soluble GFP expressed in GABAergic D-type motor neurons (Figure 4; Figure S3). To correlate axonal gaps with PTZ or aldicarb sensitivity, we first examined ventral nerve cords (VNCs; Figure 4A) and dorsal nerve cords (DNCs; Figure S3) of young adult hermaphrodites (Williams et al. 2004; Locke et al. 2006). However, axons from the two sets of GABAergic D-type motor neurons, known as VD and DD neurons, overlap extensively in the VNCs and DNCs of young adults. To buttress our analysis, we assayed VNCs of L1 larvae hermaphrodites for axonal GFP gaps whose VNCs contain only DD axons (Sakamoto et al. 2005; Figure 4C). Consistent with a role for triply redundant Racs and an integrin–NIK complex in mediating GABAergic D-type commissural navigation (Poinat et al. 2002), we observed significant soluble GFP gaps in DNCs of mig-2(gm103) young adult Rac gf mutants, in ina-1(gm144) and mig-15(rh148) young adult hypomorphs, and in Rac GEF-deficient (Steven et al. 1998), unc-73(e936) and unc-73(rh40) mutant young adults (Figure S3). Conversely, young adult mutants for either ced-10(n3246) or ina-1(gm39) did not exhibit significant soluble GFP gaps in their DNCs (Figure S3). Moreover, none of these Rac-signaling mutants had significant soluble GFP gaps in their VNCs, either as L1 larvae or as young adults (Figure 4).
All Rac-signaling mutants that we examined had significantly higher percentages of abnormally distributed synaptic vesicles, as revealed by SNB-1∷GFP axonal gaps, than axon outgrowth defects, as revealed by soluble GFP (Punc-47∷GFP) gaps, in their VNCs at the young adult stage (Figure 4A) and at the L1 larval stage (Figure 4C). Figure 4B is a representative image showing SNB-1∷GFP axonal gaps in the unc-73(e936) mutant background. We never observed SNB-1∷GFP mislocalization in the DNCs of Rac-signaling mutant L1 larvae (data not shown), suggesting that this Rac-signaling pathway does not participate in synaptic vesicle cargo recognition during DD neuron remodeling (Sakamoto et al. 2005). The only Rac pathway mutants that did not exhibit significantly more SNB-1∷GFP axonal gaps than soluble GFP gaps in young adult DNCs were the mig-15 mutant, rh148 (Shakir et al. 2006), and the strongest Rac GEF mutant, rh40 (Steven et al. 1998) (Figure S3). Rac-signaling mutants had similar interpunctal SNB-1∷GFP axonal gap widths and gap numbers at both L1 larval and young adult stages (Table S1). Thus, SNB-1∷GFP axonal gap frequencies, not interpunctal widths or numbers, correlate with Rac-signaling mutant sensitivities to PTZ and aldicarb.
Because significantly reduced SNB-1∷GFP gaps in the majority of the Rac pathway mutants tested could not be explained by axon outgrowth defects, these data suggest that synaptic vesicle misaccumulations in GABAergic motor neurons could be sufficient for enhanced sensitivities of Rac pathway mutants to aldicarb and PTZ. In support of this hypothesis, ina-1(gm39) mutants, unlike ina-1(gm144) mutants, had insignificant architectural defects in GABAergic DD motor neurons of young adults (Figure S3). Yet, ina-1(gm39) mutants were more sensitive to aldicarb (Figure 3C) and PTZ (Figure 1B) than ina-1(gm144) mutants. Defects in non-GABAergic neurons of the stronger mutant might explain its weaker sensitivities to aldicarb and PTZ. However, paralysis from severe defects in synaptic transmission or axon pathfinding does not prevent PTZ-induced convulsions, since 100.0% of unc-26(e1196) and 82.2% of unc-51(e396) immotile mutants exhibited anterior convulsions with 2.5 mg/ml PTZ. As noted above, gm39 may also affect binding of a different number and/or class of ligands than gm144. Regardless, similar synaptic vesicle misaccumulations suggest that LIS-1 and dynein converge with Racs to control GABA transmission at adult C. elegans NMJs.
RNAi confirms role for Rac-signaling pathway in GABAergic vesicle transport:
SNB-1∷GFP misaccumulations implicate Rac signaling in GABAergic vesicle transport. Yet, these data are difficult to interpret in light of subtle axon-pathfinding defects in GABAergic D-type motor neurons. To be confident that GABAergic vesicle transport defects from Rac signaling attenuation are not secondary to axon-pathfinding defects, we used RNAi to produce weaker phenotypes than those of extant mutants.
We have previously shown that lactose-induced RNAi feeding against LIS-1 pathway members is sufficient to cause SNB-1∷GFP misaccumulations in D-type motor neurons (Locke et al. 2006). Accordingly, lactose-induced RNAi feeding uncovered GABAergic vesicle transport abnormalities, which phenocopied lis-1(t1550) and Rac-signaling-pathway mutants, with combinatorial RNAi (Tischler et al. 2006) against both ced-10 and rac-2 (Figure 5; Figure S4) in first generation (F1) progeny of RNAi-treated worms. As with the mutant alleles, SNB-1∷GFP misaccumulated with mig-15(RNAi) in DNCs and with ina-1(RNAi), lis-1(RNAi), or unc-73(RNAi) in both VNCs and DNCs (Figure 5; Figure S4) of F1 progeny. RNAi feeding also revealed SNB-1∷GFP misaccumulations with combinatorial RNAi against the functionally redundant C. elegans PAK orthologs, pak-1 and max-2 (Figure 5; Figure S4), that encode the immediate downstream effectors of activated Racs implicated in GABAergic D-type commissural axon navigation (Lucanic et al. 2006). These results may suggest that PAK-mediated signals between actin and microtubule networks are particularly important for GABAergic vesicle transport, as dynein motor complex subunits, such as dynein light chain (Lu et al. 2005) and dynamitin (Menzel et al. 2007), have also been shown to physically interact with PAK orthologs in humans and Drosophila, respectively.
Additionally, our results suggest novel roles for bicd-1, the worm ortholog of bicaudal-D, a dynein- and NIK-interacting protein (Houalla et al. 2005), and pes-7, the worm ortholog of IQGAP1, a rodent Lis1 and Rac1 interactor (Kholmanskikh et al. 2006), in GABAergic vesicle transport (Figure 5; Figure S4). RNAi against these genes did not result in soluble GFP axonal gaps (Figure 5; Figure S4). In contrast, most of the same RNAi treatments against parent animals, instead of their F1 progeny, resulted in SNB-1∷GFP misaccumulations, but not soluble GFP axonal gaps (Figure S5), such as those observed with young adult Rac-signaling mutants (Table S1). The only treatment that did not uncover a trend toward GABAergic vesicle transport defects was combinatorial RNAi against pak-1 and max-2 in parents (Figure S5). This result may be explained by dilution of phenotypic strength, where RNAi phenotypes are weakened by multiple gene targeting (Tischler et al. 2006).
RNAi reveals synergistic interactions between dynein and Rac-signaling pathways:
To determine if members of a canonical Rac-signaling pathway function together to modulate neuronal synchrony in C. elegans, we performed aldicarb-induced paralysis assays and PTZ-induced convulsion assays with RNAi-treated animals. Consistent with parallel functions of triply redundant Racs in GABAergic motor neurons (Lundquist et al. 2001), RNAi against ced-10 in mig-2(gm103) gf mutants was not sufficient to alter sensitivity to aldicarb (Figure 6A). Likewise, RNAi against rac-2 in either ced-10(n3246) or mig-2(gm103) gf mutant backgrounds was not sufficient to alter aldicarb sensitivity (Figure 6A). However, combinatorial RNAi against ced-10 and rac-2 synergized with a mig-2(gm103) mutation to significantly enhance aldicarb sensitivity and was also sufficient to increase aldicarb sensitivity of N2 wild-type worms (Figure 6A).
RNAi against bicd-1, ina-1, lis-1, mig-15, pes-7, or unc-73, as well as combinatorial RNAi against the functionally redundant pak-1 and max-2 genes, was sufficient to enhance the aldicarb sensitivity of wild-type worms (Figure 6A). RNAi against these putative Rac-signaling-pathway members, except for pes-7 with ced-10(n3246), resulted in synergistic enhancement of sensitivity to aldicarb in both Rac gf mutant backgrounds (Figure 6A). Despite not being sufficient to induce anterior convulsions in the presence of PTZ, as predicted by our earlier studies (Locke et al. 2006), these same RNAi treatments also synergized with Rac gf mutations to increase sensitivity to PTZ-induced anterior convulsions (Figure 6B). Furthermore, most of these RNAi treatments synergized with hypomorphic ina-1(gm144) and mig-15(rh148) mutations, albeit weakly with the latter, to increase sensitivity to aldicarb (Figure 7A) and PTZ (Figure 7B). Specifically, neither RNAi against lis-1, pes-7, and unc-73 nor combinatorial RNAi against pak-1 and max-2 was sufficient to significantly enhance sensitivity to PTZ with mig-15(rh148) (Figure 7B). Likewise, pes-7(RNAi) was not sufficient to significantly enhance sensitivity to aldicarb (Figure 7A) or PTZ with ina-1(gm144) (Figure 7B). These inconsistencies could result from our use of weaker hypomorphic mutants (Poinat et al. 2002) or subthreshold levels of neural stimulants. Ultimately, however, these results suggest that the canonical Rac GTPase regulators, INA-1, MIG-15, UNC-73, PAK-1, and MAX-2, modulate neuronal synchrony through interactions with triply redundant Racs, integrators of microtubule-actin-signaling networks, such as LIS-1, PES-7, and BICD-1.
We previously reported that disturbance of LIS-1 and the dynein motor complex results in enhanced sensitivities to a GABAA receptor antagonist, PTZ, and synaptic vesicle misaccumulations, which were specific to GABAergic motor neurons (Williams et al. 2004; Locke et al. 2006). In this work, we show that both lis-1 and dhc-1 mutants are also hypersensitive to an acetylcholinesterase inhibitor, aldicarb, which overstimulates body-wall muscles (Nguyen et al. 1995; Miller et al. 1996) in a similar manner to PTZ. These results further suggest that lis-1 and dhc-1 cooperate with GABA to modulate neuronal synchrony. Yet, these data do not suggest a mechanism by which LIS-1 and dynein may function specifically in C. elegans GABA transmission, given their pleiotropy and likely roles in general synaptic function. Here, we present evidence that this mechanism is Rac dependent and active in diverse sets of GABAergic neurons.
GABAergic D-type motor neurons are known to complete their development in the second of four C. elegans larval stages (Knobel et al. 1999). Moreover, previous studies have demonstrated that a tripartite Rac GTPase-signaling cascade mediates commissural axon navigation of these neurons (Lundquist et al. 2001; Poinat et al. 2002; Lucanic et al. 2006). Interestingly, expression of C. elegans Racs persists in the adult GABAergic nervous system (Lundquist et al. 2001). Rac regulators, including ina-1 (Baum and Garriga 1997), mig-15 (Poinat et al. 2002), and unc-73 (Steven et al. 1998; Hunt-Newbury et al. 2007), as well as lis-1 (Dawe et al. 2001; Hunt-Newbury et al. 2007) and dhc-1 (Hunt-Newbury et al. 2007), have also been detected in GABAergic neurons of adult C. elegans hermaphrodites. These findings suggest that a Rac-signaling pathway could govern not only GABAergic neuron development but also GABA transmission, post-developmentally. Consistent with this hypothesis, we show for the first time that a Rac-signaling pathway modulates synaptic transmission at adult C. elegans NMJs. Furthermore, we present evidence that Racs have post-developmental functions, which appear distinct from previously described neuromuscular roles for RHO-1 (McMullan et al. 2006; Williams et al. 2007). More specifically, this Rac cascade is required for dynein-mediated synaptic vesicle transport in GABAergic neurons.
Rho family GTPases, including Racs, may be differentially regulated in space and time to confer cellular identity (de Curtis 2008). Indeed, disparate Rac-signaling pathways have been implicated in several C. elegans developmental processes (Lundquist et al. 2001). Here, we reveal a specific role for an integrin-mediated tripartite Rac cascade in GABAergic motor neurons by combining microscopic, pharmacological, and behavioral approaches. Furthermore, we have shown that Rac regulators, which were previously implicated in GABAergic D-type motor neuron commissural axon navigation (Poinat et al. 2002), genetically interact with Racs and the dynein motor complex to mediate GABAergic vesicle transport. We have also identified roles for C. elegans orthologs of bicaudal-D and IQGAP1 in this dynein–Rac-signaling pathway. It has been speculated that these genes integrate signals between microtubule and actin networks via NIK and Lis1, respectively (Houalla et al. 2005; Kholmanskikh et al. 2006). Although our results support these propositions, a series of bicd-1 and pes-7 hypomorphic and null mutant alleles will be required to fully understand the neuronal functions of these genes.
On the basis of the above results, we present a model in which disruption of this dynein–Rac-signaling pathway results in a loss of the axonal transport of synaptic vesicles in GABAergic motor neurons (Figure 8). As suggested in Figure 8, it is plausible that this dynein–Rac-signaling pathway coordinates microtubule plus end-binding of GABAergic synaptic vesicles, thereby allowing them to undergo retrograde transport. Evidence for the interdependency of dynein and kinesin in anterograde transport also exists and likely explains the GABAergic vesicle misaccumulations and associated behavioral phenotypes, which we observed (Williams et al. 2004; Locke et al. 2006). These GABAergic vesicle misaccumulations could also be caused by disruptions in other cellular processes, such as the localization of SNB-1∷GFP to synaptic vesicle membranes, selective degeneration or retraction of axons, or regulated degradation or endocytosis of SNB-1∷GFP (Koushika et al. 2004). We hypothesize that SNB-1∷GFP misaccumulates in these dynein–Rac-signaling mutants due to axonal transport defects. However, these alternative hypotheses could be considered in the future, perhaps best with live imaging.
The identification and characterization of interactions between Racs and the dynein motor complex was accomplished by separating the architectural from the functional defects in GABAergic neurons and knocking down putative interactors in sensitized genotypic backgrounds. To this end, we employed lactose-induced RNAi feeding, which dependably inhibits gene expression in GABAergic motor neurons (Locke et al. 2006). We found that RNAi against members of a Rac cascade was not sufficient to cause architectural defects in the GABAergic nervous system, even though these genes are important for GABAergic D-type motor neuron development (Lundquist et al. 2001; Poinat et al. 2002; Lucanic et al. 2006). Conversely, RNAi against these targets was sufficient to cause SNB-1∷GFP misaccumulations, revealing a link with synaptic vesicle transport. RNAi feeding in hypomorphic genotypic backgrounds was sufficient to uncover genetic interactions without the need to cross mutants with slow-growing hypersensitive RNAi strains. This approach may be used to accelerate the mapping of other signaling cascades and to identify roles for other developmental genes in adult neurons.
Previous studies with C. elegans and other model systems have postulated interactions between the dynein motor complex and Rac GTPases. For example, an integrin-dependent signaling cascade has been shown to mediate trafficking of lipid rafts through Rac1, Arf6, and microtubules (Balasubramanian et al. 2007). Likewise, inhibition of Dictyostelium orthologs of LIS-1 and dynein intermediate chain has been associated with actin depolymerization, possibly through disruption of Rac1 (Rehberg et al. 2005). Consistent with these results, a C. elegans study has shown that lis-1 and dhc-1 mutants are defective in germline cell corpse engulfment (Buttner et al. 2007). Although the molecular basis of these engulfment defects was not investigated, the data suggested that corpse engulfment is dynein dependent and likely required lis-1 and dhc-1 for cargo transport and/or changes in cytoskeletal dynamics (Buttner et al. 2007). These hypotheses are particularly relevant to our own findings, as one of three C. elegans Racs, CED-10, is essential for corpse engulfment (Kinchen et al. 2005). It is possible that dynein and LIS-1 cooperate with CED-10 and other Rac regulators to mediate engulfment. However, it is unlikely that the CED-10-dependent mechanism, which is necessary for engulfment, is the same one responsible for neuronal synchronization and dynein-mediated GABAergic vesicle transport. Indeed, null mutants of ced-2 and ced-5, both of which are important regulators of CED-10 in engulfment (Ellis et al. 1991), exhibited wild-type sensitivities to the GABAA receptor antagonist PTZ (Table 1). Thus, a unique and perhaps uncharacterized set of Rac regulators could govern Rac and dynein activity in adult GABAergic neurons.
Perturbations of inhibitory GABA transmission can promote epileptic activity in a wide range of model systems, including rodents (Delorey et al. 1998), zebrafish (Baraban et al. 2005), Drosophila (Pavlidis et al. 1994; Guan et al. 2005), and C. elegans (Williams et al. 2004). These studies underscore the utility of both mammalian and non-mammalian models to reveal cellular effectors of neuronal activity (Locke et al. 2009). Accordingly, several reports from a variety of animal models further implicate members of a Rac-signaling pathway in seizures and epilepsy. Orthologs of dhc-1, lis-1, the functionally redundant Racs and Paks, ina-1, mig-15, unc-73, bicd-1, and pes-7 are all expressed in GABA-enriched regions of the adult mouse brain (Lein et al. 2007). In addition, limbic seizures have been shown to result in increased expression levels of integrin β 1 (Pinkstaff et al. 1998) and NIK (Arion et al. 2006), while integrins have been shown to modulate long-term potentiation in a mature hippocampus (Chan et al. 2006; Huang et al. 2006). Seizure-like activity in Drosophila has also been shown to induce α-integrin and trio expression and to downregulate a Rac-GAP (Guan et al. 2005). Together with our own findings, these data strongly suggest the plausibility of an evolutionarily conserved network of cytoskeletal regulators that cooperate in the maintenance of a GABA-dependent seizure threshold.
We thank all members of the Caldwell Laboratory, especially Laura Berkowitz, Lindsay Faircloth, Stacey Fox, and David Agee for their collegiality and teamwork. Particular thanks also go to Hwai-Jong Cheng, Gian Garriga, Yishi Jin, Erik Jorgensen, and Erik Lundquist for donating C. elegans strains. All C. elegans mutants came from the Caenorhabditis Genetics Center, which is funded by the National Institutes of Health National Center for Research Resources. An Undergraduate Science Education Program grant from the Howard Hughes Medical Institute supported the undergraduate researchers (K.P.B. and S.K.L.) involved in this study. Additional support came from a National Science Foundation CAREER Award to G.A.C.
Supporting information is available online at http://www.genetics.org/cgi/content/full/genetics.109.106880/DC1.
↵1 These authors contributed equally to this work.
↵2 Present address: Neuroscience Program, University of California, San Francisco, CA 94158.
Communicating editor: K. Kemphues
- Received July 1, 2009.
- Accepted September 17, 2009.
- Copyright © 2009 by the Genetics Society of America