The Ubiquitin Ligase RPM-1 and the p38 MAPK PMK-3 Regulate AMPA Receptor Trafficking

Ubiquitination occurs at synapses, yet its role remains unclear. Previous studies demonstrated that the RPM-1 ubiquitin ligase organizes presynaptic boutons at neuromuscular junctions in C. elegans motorneurons. Here we find that RPM-1 has a novel postsynaptic role in interneurons, where it regulates the trafficking of the AMPA-type glutamate receptor GLR-1 from synapses into endosomes. Mutations in rpm-1 cause the aberrant accumulation of GLR-1 in neurites. Moreover, rpm-1 mutations enhance the endosomal accumulation of GLR-1 observed in mutants for lin-10, a Mint2 ortholog that promotes GLR-1 recycling from Syntaxin-13 containing endosomes. As in motorneurons, RPM-1 negatively regulates the pmk-3/p38 MAPK pathway in interneurons by repressing the protein levels of the MAPKKK DLK-1. This regulation of PMK-3 signaling is critical for RPM-1 function with respect to GLR-1 trafficking, as pmk-3 mutations suppress both lin-10 and rpm-1 mutations. Positive or negative changes in endocytosis mimic the effects of rpm-1 or pmk-3 mutations, respectively, on GLR-1 trafficking. Specifically, RAB-5(GDP), an inactive mutant of RAB-5 that reduces endocytosis, mimics the effect of pmk-3 mutations when introduced into wild-type animals, and occludes the effect of pmk-3 mutations when introduced into pmk-3 mutants. By contrast, RAB-5(GTP), which increases endocytosis, suppresses the effect of pmk-3 mutations, mimics the effect of rpm-1 mutations, and occludes the effect of rpm-1 mutations. Our findings indicate a novel specialized role for RPM-1 and PMK-3/p38 MAPK in regulating the endosomal trafficking of AMPARs at central synapses.


Introduction
Synapses are the sites of cellular communication between presynaptic neurons and their postsynaptic partners. Presynaptic terminals contain multiple synaptic vesicles, which release neurotransmitter [1]. Receptors on the postsynaptic side of the synapse receive the neurotransmitter signals from the presynaptic cell. Changes in the localization and regulation of these receptors in turn mediate the changes in synaptic efficacy that occur during learning and memory [2]. The formation of presynaptic terminals and postsynaptic specializations is coordinated, but requires distinct sets of proteins.
A particularly well-studied PHR protein is RPM-1, which regulates the formation of NMJs by two parallel mechanisms.
First, RPM-1 forms an SCF-like complex with the F-box protein FSN-1, the SKP1 ortholog SKR-1, and the Cullin CUL-1; the resulting ubiquitin ligase ubiquitinates DLK-1, an upstream component of a C. elegans p38 MAPK pathway [14,15]. Second, RPM-1 binds to GLO-4, an RCC1-like GEF that regulates GLO-1, a Rab GTPase [16]. RPM-1 is thought to positively regulate a Rab GTPase pathway to promote vesicular trafficking via late endosomes, which is critical for the organization of presynaptic terminals. Little is known about the function of PHR proteins like RPM-1 outside of the presynaptic terminal of NMJs, although they are abundantly expressed in the CNS.
The formation of postsynaptic specializations at excitatory central synapses has also been well studied. Ionotropic glutamate receptors (GluRs) form tetrameric channels on the postsynaptic face of central synapses, where they receive glutamatergic signals from the presynaptic cell [17]. The regulated trafficking of AMPAtype GluRs (AMPARs) into and out of the postsynaptic membrane is thought to underlie several forms of synaptic plasticity [18][19][20][21]. Ubiquitination and endocytosis are key mechanisms that regulate AMPAR postsynaptic accumulation [22][23][24][25][26][27]. However, the specific proteins that mediate the ubiquitin-dependent regulation of AMPARs are not well characterized.
To investigate these processes in a genetic system, we and others previously examined the trafficking of the GLR-1 AMPAR subunit in C. elegans. GLR-1 is expressed in the command interneurons, where it mediates nose-touch mechanosensation and regulates the frequency of spontaneous reversals in locomotion [28][29][30]. Functional GLR-1 receptors, fused with green fluorescent protein (GLR-1::GFP), are localized to punctate clusters at central synapses in the nerve ring (proximal neurites that encircle the pharynx) and along the ventral cord (the fascicle of distal neurites that run along the ventral midline) [31]. The synaptic abundance of GLR-1 is regulated by ubiquitination and endocytosis [24,27,32,33]. GLR-1 synaptic abundance is also controlled by LIN-10, a PDZ-domain protein of the Mint family, which is thought to stimulate the membrane recycling of GLR-1 [31,[34][35][36][37]. Mutants that lack LIN-10 activity have decreased levels of punctate, synaptic GLR-1, and instead accumulate GLR-1 receptors in large, aberrant compartments at non-synaptic sites throughout their neurites [31,37,38]. Mutations that block endocytosis suppress the aberrant accumulation of GLR-1 in lin-10 mutants, suggesting that GLR-1 is accumulating in an internal, post-endocytosis compartment within the neurites of these mutants [37]. Consistent with this model, lin-10 mutants also have deficits in GLR-1-mediated behaviors, which can be suppressed by blocking endocytosis [37].
Here we report findings from our screen for genetic modifiers of lin-10 mutations. We identify rpm-1 as an enhancer of lin-10, as mutations in rpm-1 enhance the aberrant accumulation of GLR-1 in neurites and the cell body that is observed in lin-10 mutants. Mutants for rpm-1 alone also accumulate GLR-1 in large compartments, although to a lesser extent than lin-10 mutants. Whereas rpm-1 mutants have disorganized presynaptic terminals at motorneuron NMJs, we find that presynaptic terminals at interneuron central synapses of rpm-1 mutants are normal at a gross level. Restoration of rpm-1 function in presynaptic neurons does not rescue the GLR-1 trafficking defects of rpm-1 mutants. Instead, we find that RPM-1 functions in the postsynaptic interneurons that express GLR-1. As in the motorneurons, we find that RPM-1 regulates signaling via the PMK-3/p38 MAPK pathway in interneurons.
In addition, our findings point to the regulation of GLR-1 endocytosis as the mechanism for PMK-3 and RPM-1 function. Mutations that positively or negatively alter endocytosis both mimic and occlude the effects of rpm-1 or pmk-3 mutations, respectively, on GLR-1 trafficking to Syntaxin-13-containing endosomes. We propose that p38 MAPK stimulates GLR-1 endocytosis, and that RPM-1 inhibits p38 MAPK signaling, thereby acting to reduce GLR-1 endocytosis and to stabilize GLR-1 at the synapse. Our findings demonstrate a novel function for PHR proteins: the regulation of postsynaptic elements at central synapses via the regulation of endocytosis.

RPM-1 Regulates GLR-1 Trafficking in Neurites
To better understand how GLR-1 is trafficked to synapses, we performed an EMS screen for suppressors and enhancers of a lin-10 loss of function mutation as this mutation results in the aberrant accumulation of GLR-1 [37]. We identified two mutants, od14 and od22, with an enhanced lin-10 phenotype. Whereas wild-type animals contain punctate GLR-1::GFP at synapses along neurites (Fig. 1A), mutants for lin-10 ( Fig. 1B) or rpm-1 (Fig. 1C) accumulate GLR-1 in large accretions found within the neurites. The shift in GLR-1 localization from small puncta to large accretions can be quantified by measuring mean puncta size in the ventral cord (Fig. 1I). Double mutants using molecular null alleles for lin-10 and rpm-1 have significantly larger accretions of GLR-1 than either single mutant alone (Fig. 1D,I), and have reduced numbers of small GLR-1 puncta compared to wild type (Fig. 1J), suggesting that these two genes might act in separate pathways to regulate GLR-1 trafficking.
We identified od14 and od22 as alleles of rpm-1 by genetic mapping and by their failure to complement known rpm-1 alleles ( Fig. 2A; see Supplementary Materials and Methods for details). We examined three other previously identified alleles of rpm-1 (ok364, ju41, js317), and found that all three behaved like od14 and od22 with regard to GLR-1 localization (data not shown). To determine the molecular nature of our alleles, we sequenced genomic DNA from both mutants. The od14 mutation alters glycine 1092 to a glutamate ( Fig. 2A,B). This residue falls within the first PHR domain, and is conserved in all PHR protein family members, suggesting that glycine 1092 is critical for PHR domain function, and that the PHR domain is required to regulate GLR-1 trafficking. The od22 mutation alters the conceptual translation of RPM-1 protein from arginine to an Opal stop codon at amino acid 30, resulting in a protein lacking all known functional domains ( Fig. 2A).
As RPM-1 forms an E3 ubiquitin ligase complex with the F-box protein FSN-1 and the Cullin CUL-1 [15], we tested whether mutations in fsn-1 result in a similar GLR-1 localization phenotype as that in rpm-1. The fsn-1(hp1) mutation introduces a stop codon before the SPRY domain, resulting in a protein null allele. Like rpm-1 mutants, fsn-1 mutants accumulate GLR-1 in large accretions (Fig. 1E,K). Moreover, mutations in fsn-1 enhance the accumulation of GLR-1 observed in lin-10 mutants (Fig. 1F,L) and depress the number of small GLR-1 puncta (data not shown).

RPM-1 Has Different Roles in Different Neuron Types
RPM-1 is required for proper presynaptic bouton formation at motorneuron NMJs [5,6]. To determine whether our newly identified rpm-1 mutations also impair presynaptic bouton formation, we examined the subcellular localization of SNB-1 (synaptobrevin) using juIs1, a transgene that expresses a SNB-1::GFP protein fusion in motorneurons. In wild-type animals, SNB-1::GFP is localized to large (1-2 micron) NMJ boutons along both the dorsal and ventral cords [14,15,39] (Fig. 3A). In rpm-1(od14) mutants, we observed a decreased number of SNB-1::GFP NMJ boutons and an irregularity in inter-bouton spacing, as was observed for other alleles of rpm-1 mutants (Fig. 3C,I). Given the genetic interaction between rpm-1 and lin-10 with regard to GLR-1 localization, we examined SNB-1::GFP at the NMJ boutons of lin-10 single mutants (Fig. 3E) and rpm-1 lin-10 double mutants (Fig. 3G). NMJ boutons from lin-10 mutants were indistinguishable from those in wild type, and rpm-1 lin-10 double mutants were indistinguishable from those in rpm-1 single mutants. We quantified the total number of the dorsal cord SNB-1::GFP boutons in rpm-1(od14) mutants, and found it to be similar to that observed in known alleles of rpm-1 (Fig. 3I). As the od14 mutation alters the first PHR domain, our findings suggest that the PHR domain is critical for RPM-1 function in motorneurons.
Given the changes in presynaptic boutons observed in rpm-1 mutant motorneurons, we reasoned that the changes in GLR-1 ventral cord accumulation in rpm-1 mutants could reflect general defects in synapse formation on interneurons. To test this possibility, we examined the localization of SNB-1 using odIs1, a transgene that expresses a SNB-1::GFP protein fusion in the GLR-1-expressing interneurons of the central nervous system. In wild-type animals, SNB-1::GFP is localized to small (,0.5 micron) boutons along the ventral cord (Fig. 3B). In rpm-1 mutants, there is no observable difference in SNB-1-labeled bouton number or size at central synapses (Fig. 3D). We also examined SNB-1-labeled central synapse boutons in lin-10 single mutants (Fig. 3F) and rpm-1 lin-10 double mutants (Fig. 3H), and found them to be indistinguishable from wild type (Fig. 3J). Finally, we examined the localization of a second presynaptic marker, RFP::SNN-1 (synapsin, [40]), at wild-type and rpm-1 mutant interneuron synapses. RFP::SNN-1 is properly localized in both (Fig. 3K,L).
As mutations in rpm-1 behaved similarly to lin-10 mutations with regard to their effect on GLR-1 trafficking, we also tested whether RPM-1 regulates LIN-10 localization. A LIN-10::GFP chimeric protein is colocalized with GLR-1 in the ventral cord in wild-type animals [31,38]. We found that LIN-10::GFP is localized to similar punctate structures in both wild type and rpm-1 mutants (Fig. 3M,N). Taken together, our findings indicate that the defects in GLR-1 localization observed in rpm-1 mutants are unlikely to be due to gross defects in the formation of central synapses on the interneurons. In addition, our findings indicate that RPM-1 has a distinct role in organizing the presynaptic face of NMJ synapses in motorneurons but not central synapses in interneurons.

PMK-3 and DLK-1 Act Downstream of RPM-1
One known target of RPM-1 regulation is the p38 MAPK cascade. To test for a role of p38 MAPK in GLR-1 trafficking, we examined pmk-3(ok169) mutants. The pmk-3(ok169) mutation is a deletion spanning most of the pmk-3 coding sequence. Whereas the mutation in pmk-3 did not affect the number or size of GLR-1 puncta in otherwise wild-type neurites (Fig. 4A), it did suppress the accumulation of GLR-1 observed in lin-10 mutants, rpm-1 mutants, and rpm-1 lin-10 double mutants ( Fig. 4B,C,D,G,H). These results indicate that PMK-3/p38 MAPK function is required for GLR-1 accumulation in lin-10 and rpm-1 mutants.
Given the nuclear localization of PMK-3, we hypothesized that PMK-3 might regulate the transcription of glr-1 mRNA. We isolated total mRNA from wild type, pmk-3 mutants, rpm-1 mutants, lin-10 mutants, pmk-3 lin-10 double mutants, rpm-1 lin-10 double mutants, and pmk-3 rpm-1 lin-10 triple mutants. We measured glr-1 mRNA levels by real time PCR (using snb-1 mRNA levels as a control), and detected no significant differences among these different genotypes (data not shown). Removing one of the two copies of the nuIs25[glr-1::gfp] transgene, which results in twofold mRNA level differences that are detectable by real time PCR, does not suppress the GLR-1 accumulation observed in lin-10 mutants [31]. Thus, it is unlikely that PMK-3 and RPM-1 regulate GLR-1 trafficking by affecting glr-1 mRNA levels.
As RPM-1, an ubiquitin ligase, is a negative regulator of the p38 MAPK cascade, the abundance of one or more components of the cascade is likely to be limiting within neurons. To determine if the levels of PMK-3 are limiting, we introduced the Pglr-1::pmk-3(+) transgene into animals that also contained both wild-type alleles of the endogenous pmk-3 locus. The elevation of PMK-3 levels in wild-type animals did not affect GLR-1 localization (data not shown). Also, lin-10 mutants that contained elevated levels of PMK-3 did not demonstrate the dramatic enhancement observed in rpm-1 lin-10 double mutants (Fig. 4E,J).
In C. elegans, GLR-1 signaling positively regulates spontaneous reversals during forward locomotion as animals forage for food [41,42]. Mutants with either reduced GLR-1 activity (e.g., glr-1 mutants) or reduced levels of GLR-1 receptor at the synaptic membrane surface (e.g., lin-10 mutants) have a lower frequency of spontaneous reversals [37]. We examined GLR-1-mediated behaviors in wild-type animals and in animals with mutations in lin-10, glr-1, rpm-1, or pmk-3. We found that wild-type animals spontaneously reversed about 3.7 times per minute (20 animals, 5 minute trial per animal), whereas lin-10 and glr-1 mutants only spontaneously reversed direction about 1.5 and 1.4 times per minute, respectively (Fig. 4K). Mutants for either rpm-1 or pmk-3 spontaneously reversed direction with a frequency similar to that of wild-type animals. Interestingly, double mutants for lin-10 and pmk-3 reversed direction 2.5 times per minute (Fig. 4L), suggesting that mutations in pmk-3 can partially restore GLR-1-mediated reversal behavior to lin-10 mutants. Thus, the effect of pmk-3 mutations on GLR-1 subcellular localization correlates with behaviors that reflect the synaptic strength of the interneuron reversal circuit.

RPM-1 Regulates DLK-1 Abundance in Interneurons
If DLK-1 levels are limiting with regard to the regulation of GLR-1 trafficking, then RPM-1 might regulate DLK-1 levels. To observe changes in DLK-1 protein levels using a more sensitive reporter than our mRFP::DLK-1 chimera, we generated a transgene, P glr-1 ::gfp::dlk-1(+), containing GFP sequences fused in frame to dlk-1 cDNA sequences at the DLK-1 N-terminus, under the control of the glr-1 promoter. We introduced the transgene into wild-type animals and found that GFP::DLK-1 protein was present both in neuron cell bodies (near the membrane and excluded from the nucleus; Fig. 5G) and in ventral cord neurites in a punctate pattern (Fig. 5H). We crossed the transgene into rpm-1 mutants and observed a significant increase in GFP::DLK-1 fluorescence (the same transgenic line is shown in Fig. 5G,H,I,J). To quantify GFP::DLK-1 fluorescence, we removed background fluorescence from our images, then defined either whole cell bodies or whole ventral cords as single objects. We quantified the mean fluorescence for these objects and found a several fold increase in mean GFP::DLK-1 fluorescence in rpm-1 mutants compared to wild type (Fig. 5K,L; same transgenic line is quantified). These differences were observed in both neuron cell bodies and ventral cord neurites. Mutations in lin-10 had little effect on GFP::DLK-1 (Fig. 5K,L). Our findings suggest that RPM-1 negatively regulates DLK-1 protein levels in the command interneurons.
RPM-1 Is Required for the Ubiquitin-Mediated Turnover of GLR-1 GLR-1 trafficking and synaptic abundance are regulated by ubiquitination. Because of a limiting cellular concentration of monoubiquitin, overexpression of Myc epitope-tagged ubiquitin (MUb) by a nuIs89 transgene has been shown to negatively regulate GLR-1 abundance in neurites [24]. While multiple E3 ligases have been shown to be partially required for the turnover of GLR-1 after an overexpression of ubiquitin, mutations in no single E3 ligase have been shown to completely block GLR-1 turnover, suggesting that multiple ligases are involved. To determine if RPM-1 is required for the ubiquitin-mediated turnover of GLR-1, we introduced nuIs89 into lin-10 mutants, rpm-1 mutants, and rpm-1 lin-10 double mutants. Elevated ubiquitin depresses the number of GLR-1 puncta in wildtype animals (Fig. 6A,F). Mutations in lin-10 and rpm-1 combined block the effect of overexpressed ubiquitin on GLR-1 puncta number (Fig. 6D,F), indicating that these genes are required for some of the ubiquitin-mediated removal of GLR-1. However, in lin-10 and rpm-1 single mutants, as well as in lin-10 rpm-1 double mutants, elevated ubiquitin can nevertheless still depress the size of GLR-1 accretions (Fig. 6B,C,D,E), suggesting that the GLR-1 receptors that accumulate internally in these mutants can still be removed in part by other ubiquitin-dependent mechanisms. Thus, RPM-1 is required for part, but not all, of the ubiquitin-mediated degradation of GLR-1.

RPM-1 Regulates GLR-1 Endocytosis
Previous studies have shown that p38 MAPK can modulate endocytosis in other systems, raising the possibility that PMK-3 regulates GLR-1 endocytosis [43][44][45]. Indeed, the accumulation of GLR-1 receptors into large accretions in lin-10 mutants is due to defects in GLR-1 recycling from endosomes to synapses, suggesting that the large accretions are endosomes swollen with trapped receptors [37]. Direct colocalization with endosomal markers in the ventral cord has been difficult: fluorescently tagged endosomal residents have not been bright enough to visualize endosomes in the neurites ( [37,46] and our own unpublished observations). Recently, Chun et al demonstrated that an mRFPtagged Syntaxin-13, a resident of early endosomes, colocalizes with GLR-1::GFP in neuron cell bodies [46]. In this study, unc-108/rab-2 trafficking mutants, which contain mislocalized GLR-1 in large accretions in the ventral cord (similar to lin-10), also accumulated GLR-1 in Synxtaxin-13-labeled endosomes in the cell body. Using the glr-1 promoter, we coexpressed RFP::Syntaxin-13 with GLR-1::GFP in wild-type animals, as well as in lin-10 and pmk-3 mutants, then observed subcellular localization using confocal microscopy. GLR-1::GFP is localized in small tubule-like structures inside neuron cell bodies, and a significant fraction of these structures colocalize with RFP::Syntaxin-13 ( Fig. 7A-D,Q). In lin-10 mutants, GLR-1::GFP accumulates in large accretions in the cell bodies, and these show more colocalization with RFP::Syntaxin-13 ( Fig. 7E-H,Q). By contrast, pmk-3 mutant neuron cell bodies accumulate GLR-1 at or near the outer membrane, with significantly less colocalization with RFP::Syntaxin-13 ( Fig. 7I-L,Q). Moreover, mutations in pmk-3 suppress the accumulation and colocalization of GLR-1 with Syntaxin-13 in lin-10 mutants ( Fig. 7M-P,Q). These results indicate that PMK-3 and LIN-10 have opposite effects on the accumulation of GLR-1 in endosomes.

Discussion
We identified a novel role for RPM-1 and PMK-3 as regulators of AMPAR endocytosis. Previously, LIN-10 was shown to mediate the recycling of GLR-1 from endosomes back to the synapse [37]. Mutations in lin-10 result in the accumulation of GLR-1 in endosomes; this accumulation is suppressed by mutations that decrease endocytosis and enhanced by mutations that increase endocytosis. Several lines of evidence suggest that RPM-1 and PMK-3 regulate the endocytosis of GLR-1 receptors at central synapses ( Fig. 9), independent of their role in presynaptic differentiation at NMJs. First, animals with mutations in rpm-1 accumulate GLR-1 receptors in large accretions, and mutations in rpm-1 enhance the GLR-1 accumulation observed in lin-10 mutants. Second, the GLR-1 accretions observed in these mutants colocalize with the endosomal marker Syntaxin-13. Third, while mutations in rpm-1 result in defective presynaptic boutons at motorneuron NMJs, these same mutations do not appear to alter presynaptic boutons at interneuron central synapses. Fourth, mutations in PMK-3, a p38 MAPK, suppress the accumulation of GLR-1 in endosomal accretions in both lin-10 and rpm-1 mutants. Moreover, mutations in PMK-3 suppress the behavioral defects of lin-10 mutants. Fifth, elevated levels of DLK-1, a MAPKKK upstream of PMK-3, result in a GLR-1 accumulation phenotype similar to that of rpm-1 mutants. Moreover, DLK-1 levels are negatively regulated by RPM-1 in interneurons. Sixth, mutations resulting in decreased GLR-1 endocytosis occlude the effect of pmk-3 mutations, whereas transgenic manipulations resulting in increased GLR-1 endocytosis bypass the requirement for PMK-3 and occlude the effect of mutations in rpm-1 (both on GLR-1 accumulation and GLR-1-mediated behaviors). Based on these results, we propose that RPM-1 regulates GLR-1 trafficking by promoting the ubiquitin-mediated degradation of DLK-1, thereby maintaining reduced levels of PMK-3/p38 MAPK signaling and hence GLR-1 endocytosis.
Mutations in rpm-1 were originally identified in screens for mutants with aberrant NMJ presynaptic boutons [5,6]. RPM-1 is expressed in the central as well as the peripheral nervous system, as are its mammalian orthologs Pam and Phr1, but their function in the CNS is unknown. Surprisingly, we found that RPM-1 is not required to organize presynaptic boutons at central synapses; thus, instead of a presynaptic role at central synapses, RPM-1 regulates the endocytosis of AMPARs from postsynaptic elements. This is the first postsynaptic function to be identified for a PHR protein.
While we did not observe obvious presynaptic defects at the central synapses of rpm-1 mutants, we cannot rule out the possibility that there might be subtle presynaptic defects. Nevertheless, the GLR-1 defects observed in rpm-1 mutants cannot simply be explained based on a presynaptic role for rpm-1. First, the GLR-1 trafficking defect observed in rpm-1 mutants can be rescued by cell-autonomous, postsynaptic (but not presynaptic) expression of an rpm-1 minigene. Second, mutants for unc-104 (KIF1A), which lack most synaptic vesicles from central synapse terminals, do not have the gross GLR-1 localization defects observed in rpm-1 mutants. In addition, mutants for unc-18 (Munc18/Sec1), which are deficient in synaptic vesicle release, also are not similar to rpm-1 mutants [31,50].
How does RPM-1 function presynaptically in one neuron type and postsynaptically in another? Interestingly, p38 MAPK signaling is the target of RPM-1 regulation at both central synapses and NMJs, suggesting that these genes function as a signaling cassette. One difference for RPM-1 function is the requirement for the GEF GLO-4 and the small GTPase GLO-1 [16]. In motorneurons, RPM-1 acts as an E3 ubiquitin ligase via FSN-1, and as an activator of the GLO-1 Rab via the GLO-4 GEF. Mutations in either fsn-1 or glo-4 alone yield mild presynaptic defects, whereas mutations in both are similar to mutations in rpm-1, suggesting that RPM-1 relies on both of these factors in parallel to organize NMJ presynaptic boutons. By contrast, mutations in fsn-1 alone yield the same postsynaptic defects with regard to the trafficking of GLR-1 as do rpm-1 mutations. Mutations in glo-1 and glo-4 appear to have no major postsynaptic defects, suggesting that, in interneurons, RPM-1 acts primarily as an E3 ubiquitin ligase to conduct its postsynaptic function. Thus, the nature of the effecter molecules that associate with RPM-1 in each neuron type might dictate the role of RPM-1 in that neuron.
The p38 MAPK proteins phosphorylate both nuclear and cytosolic targets. In mammals, p38 MAPK regulates endocytosis by phosphorylating components of the endocytosis machinery [43][44][45]. The net effect is increased endocytosis, although how such increased endocytosis is applied to specific cargo is unclear. Similarly, our findings suggest that PMK-3 stimulates GLR-1 endocytosis, perhaps via the activation of RAB-5. It should be noted that we do not have a more direct measure of surface GLR-1, and thus cannot directly address the level of surface GLR-1 in pmk-3 and rpm-1 mutants. Indeed, measuring surface receptors directly in a living, intact organism remains an elusive goal for most systems. Nevertheless, we have assayed GLR-1 endocytosis (1) by measuring endogenous GLR-1 via its behavioral function, and (2) by measuring its colocalization with the endosomal marker Syntaxin-13. The findings from these independent approaches reinforce one another, supporting our argument that GLR-1 intracellular trafficking is regulated by pmk-3 and rpm-1.
It should also be noted that, while p38 MAPKs have been shown to regulate endocytosis in the cytosol [43][44][45], we currently cannot rule out a role for PMK-3 in the nucleus of these interneurons. One way to test for a nuclear function of PMK-3 would be to introduce a PMK-3 protein mutated at its nuclear localization signal, then test for the ability of this ''cytoplasmicallyrestricted'' PMK-3 mutant protein to rescue GLR-1 trafficking. Unfortunately, the mechanism of p38 MAPK nuclear localization is unknown (p38 MAPKs, including PMK-3, lack a consensus nuclear localization signal), precluding us from conducting such a test [51].
Why does the p38 MAPK cascade require negative regulation by RPM-1? One possibility is that p38 MAPK signaling has to be continually kept in check, with RPM-1 constitutively maintaining low levels of DLK-1 protein. In support of this idea, we found that the levels of DLK-1, but not PMK-3, are limiting with respect to GLR-1 trafficking, suggesting that the activity of this p38 MAPK pathway requires regulation at the MAPKKK step. An alternatively role for RPM-1 might be to act as a feedback to shut off p38 MAPK signaling after the initial activation event. A connection between p38 MAPK activity and RPM-1 levels or ubiquitin-ligase activity would support this model. Finally, in a de-repression mechanism, external stimuli might activate p38 by deactivating RPM-1. To address these questions, we will need a better understanding of both the extrinsic and intrinsic factors that signal via this pathway.
The activation of the PMK-3 pathway by specific extrinsic cues could explain why mutations in pmk-3 and rpm-1 in an otherwise wild-type animal yield milder effects on GLR-1 trafficking or GLR-1-mediated behaviors than those observed in lin-10 mutants: PMK-3 signaling might only be required under certain circumstances. Indeed, an environment-dependent requirement for p38 MAPK in the brain has been observed in knockout mouse strains reared under different environmental conditions [52]. Alternatively, PMK-3 might function redundantly with other pathways; indeed, there are two other p38 MAPK paralogs in the genome [53].
The p38 pathway clearly mediates extrinsic signals in the mammalian brain, including ones that induce both metabotropic glutamate receptor (mGluR)-dependent and NMDAR-dependent long-term depression [44,54,55]. In both cases, p38 MAPK is activated by Rap1, which is activated by either G bc (triggered by mGluRs) or calcium (from NR2B-containing NMDARs) [44,56]. The p38 MAPK pathway is also activated by reactive oxygen species (ROS) and helps mediate the oxidative stress response in many different tissues [57]. Activated AMPARs can contribute to excitotoxic neuronal death by excessive calcium influx and ROS production [58,59]. Thus, in addition to the role of p38 MAPK in LTD, it is possible that p38 MAPK is activated by excess ROS, and in turn triggers AMPAR endocytosis. Such negative feedback could protect neurons from excitotoxicity by minimizing oxidative stress. An exploration of AMPAR trafficking in C. elegans under conditions of stress or in different environments should help to fully elucidate p38 MAPK signaling in neurons.
The od14 mutation was mapped between dpy-11 and vab-8 on the right arm of LGV. A dpy-11 od14 vab-8 recombinant chromosome was constructed and placed in trans over the CB4856 LGV chromosome to facilitate three-factor SNP mapping. Multiple recombinants placed od14 between SNP Figure 9. A model for the regulation of GLR-1 AMPAR trafficking by PMK-3 and RPM-1. GLR-1 (gray channels) endocytosis can occur via clathrin (pit indicated to left of the synapse). Gray arrows indicate major trafficking steps, positively regulated by the factor(s) indicated next to the arrows. Black arrows indicate positive (stimulatory) genetic regulatory interactions between two factors. GLR-1 endocytosis is mediated by UNC-11/AP180 and RAB-5. Once endocytosed, receptors can either be recycled to the synapse in a step requiring LIN-10, or degraded. PMK-3/p38 MAPK stimulates GLR-1 endocytosis via RAB-5 activation. PMK-3 is activated by a MAP kinase cascade, which includes MKK-4 and DLK-1. RPM-1 and FSN-1, working as an E3 ligase, negatively regulate DLK-1 (and hence p38 MAPK signaling) by ubiquitin-mediated turnover. doi:10.1371/journal.pone.0004284.g009 pkP5063 (map position +1.68) and SNP pkP5116 (map position +1.51). This region contains rpm-1, and od14 failed to complement rpm-1 mutations. A second enhancer, od22, was linked to and failed to complement od14. We sequenced rpm-1 genomic DNA from od14 and od22 mutants. For od14, we identified a missense mutation resulting in a glycine to glutamate change at amino acid 1092, a conserved residue within a PHR domain. For od22, we identified a premature stop codon mutation at nucleotide 88 of the predicted rpm-1 cDNA sequence.

Fluorescence Microscopy
GFP-and RFP-tagged fluorescent proteins were visualized in nematodes by mounting larvae on 2% agarose pads with 10 mM levamisole. Fluorescent images were observed using a Zeiss Axioplan II. A 1006 (N.A. = 1.4) PlanApo objective was used to detect GFP and RFP signal. Imaging was done with an ORCA charge-coupled device (CCD) camera (Hamamatsu, Bridgewater, NJ) using IPLab software (Scanalytics, Inc, Fairfax, VA). Exposure times were chosen to fill the 12-bit dynamic range without saturation. Maximum intensity projections of z-series stacks were obtained and out-of-focus light was removed with a constrained iterative deconvolution algorithm. We used a macro written for IPLabs to automatically calculate the outlines of fluorescent objects in ventral cord neurites when they were two standard deviations above the unlocalized baseline fluorescence. This algorithm allowed both the small GLR-1::GFP puncta in wildtype animals and the large, aberrant accretions in lin-10 and rpm-1 mutants to be defined as single objects. Object size was measured as the maximum diameter for each outlined cluster. Object number was calculated by counting the average number of clusters per 10 microns of dendrite length.
For the quantification of GFP::DLK-1, mean intensities of transgenic animals were measured from images of cell bodies (captured by a 206 objective) and ventral cords (captured by a 636 objective) using IPLabs software. For both, background signals were subtracted.
For the quantification of GLR-1::GFP and RFP::Syntaxin-13, animals were immobilized on ice for 10 minutes, then fixed with ice cold 1% paraformaldehyde in PBS for 10 min. Images for neuronal cell bodies were taken using a Carl Zeiss confocal microscope equipped with the BD CARV II TM Confocal Imager and a Carl Zeiss 1006 Plan-Apochroma objective (N.A. = 1.4). For quantitative colocalization analysis, all image manipulations were performed with iVision v4.0.11 (Biovision Technologies, Exton, PA) software using the FCV colocalization function. We applied an empirically derived threshold to all images for both the GLR-1::GFP channel and the RFP::Syntaxin-13 channel to eliminate background coverslip fluorescence and other noise (typically ,5% of pixels for each channel). The fluorescent intensity values for both the GLR-1::GFP and RFP::Syntaxin-13 channels were then scatter plotted for each pixel. Pixels with similar intensity values for both channels (within an empiricallyestablished tolerance factor) were counted as colocalized. To acquire the fraction of GLR-1::GFP colocalized with RFP::Syntaxin-13, the number of colocalized pixels was normalized to the number of GLR-1::GFP pixels under threshold. To maximize our resolving power while observing the relatively small C. elegans neuron cell bodies, we restricted our analysis to a single focal plane taken through the middle of the cell body.

Behavioral Assays
The reversal frequency of individual animals was assayed as previously described, but with some modifications [30]. Single young adult hermaphrodites were placed on NGM plates in the absence of food. The animals were allowed to adjust to the plates for 5 minutes, and the number of spontaneous reversals for each animal was counted over a 5-minute period. Twenty animals were tested for each genotype, and the reported scores reflect the mean number of reversals per minute.