Identification of a novel RPM-1 binding protein by mass spectrometry
In order to understand how RPM-1 signal transduction is mediated, we previously performed a proteomic screen to identify RPM-1 binding proteins (Grill et al., 2007
). In brief, RPM-1::GFP was transgenically expressed in the neurons of C. elegans
(using the rpm-1
promoter), biochemically purified using an anti-GFP antibody, and RPM-1 binding proteins were identified by LC-MS/MS mass spectrometry. Using this approach we identified the functional RPM-1 binding protein, GLO-4 (Grill et al., 2007
). Among the other candidate RPM-1 binding proteins, we also identified C. elegans
(Ce) RAE-1 (also called NPP-17) based on 13 unique peptide sequences obtained from multiple rounds of purification of RPM-1::GFP and mass spectrometry. CeRAE-1 is composed of 373 amino acids and has a homolog in yeast, flies and mammals (50% identity and 63% conservation between CeRAE-1 and human Rae1). The crystal structure of human Rae1 shows that it is composed of 7 WD repeats that fold into a 7-bladed propeller structure (Ren et al., 2010
). Sequence comparison predicts a similar structure for CeRAE-1 ().
PHR proteins bind to RAE-1. A, a schematic of the CeRAE-1 protein which is composed of an N-terminal domain and 7 WD repeats. Shown below is the segment of the RAE-1 protein that is deleted by rae-1(tm2784).
To confirm our proteomic results, we generated transgenic animals with extrachromosomal arrays that coexpress RPM-1::GFP and FLAG::RAE-1. The rpm-1 promoter was used to drive RPM-1 expression, and RAE-1 expression was driven by Prgef-1, a panneuronal promoter. As shown in , RPM-1::GFP coimmunoprecipitated with FLAG::RAE-1.
The human orthologs of RPM-1 and CeRAE-1 are called Pam and Rae1, respectively. To test if the interaction between RPM-1 and CeRAE-1 is conserved, we analyzed the biochemical relationship between human Pam and rat Rae1. Myc tagged human Pam coprecipitated with FLAG tagged rat Rae1 when expressed in HEK 293 cells (). As a negative control, a protein phosphatase of similar size and expression levels to Rae1 did not bind to Pam ().
human Pam and Rae1 interact biochemically via a conserved protein motif
Previous studies in yeast have shown that Rae1 directly interacts with its known binding proteins, such as Nu
orin (Nup) 98, through a conserved 14–20 amino acid motif (Wang et al., 2001
). To determine if PHR proteins have a similar motif, we used Clustal W to align the sequences of RPM-1, Highwire and Pam with the Rae1 binding motif of Nup98. Our search identified six amino acids that were highly conserved between Nup98 and the PHR proteins. This motif was located in a conserved protein domain in the PHR proteins with no known function, which we now term the Rae1 binding domain (RBD) ().
Identification of a Rae1 binding motif in the PHR proteins.
To test if the putative RBD of PHR proteins are sufficient to mediate binding to Rae1, we performed coimmunoprecipitation from transfected HEK 293 cells. An N-terminal domain, a C-terminal domain, and the RBD of Pam (aa 1984–2512), as well as the RBD of RPM-1 (aa 1766–2253) were cloned (). These constructs were coexpressed as GFP fusion proteins in 293 cells with rat Rae1 tagged with a FLAG epitope (FLAG-Rae1). Only the Pam RBD and RPM-1 RBD coprecipitated with FLAG-Rae1 ().
In yeast, three residues (LXXLR) in Nup98 are essential for binding to Rae1 (, underlined residues) (Pritchard et al., 1999
). To test whether similar residues mediate the interaction between PHR proteins and Rae1, we mutated the corresponding residues in the RBD of RPM-1 (VIR to AAA
), and found that binding to Rae1 was reduced (). Mutation of a single residue in the RPM-1 RBD (R 2088 A
) also abolished binding to Rae1 (). Similar results were observed for the RBD of Pam (data not shown).
To determine if similar point mutations in full length RPM-1 affect binding to CeRAE-1, we generated transgenic C. elegans that used a pan-neuronal promoter (Prgef-1) to express FLAG tagged CeRAE-1 and GFP fusion proteins of wild-type or point mutated RPM-1 (VIR to AAA). As shown in , the binding of RPM-1 (VIR to AAA) to CeRAE-1 was greatly reduced compared to wild-type RPM-1. Overall, our data show that the RBD of PHR proteins are necessary and sufficient for binding to Rae1.
rae-1 mutants have defects in axon termination
Previous studies have shown that loss-of-function mutations in rpm-1
or molecules that function downstream of rpm-1
result in defective axon termination in the mechanosensory neurons (Schaefer et al., 2000
; Grill et al., 2007
). To determine if loss of function in rae-1
affects axon termination, we analyzed the allele tm2784
, which deletes amino acids 108–241 of CeRAE-1 (30% of rae-1
coding sequence) (). Based on the crystal structure of human Rae1 (Ren et al.), we predict that tm2784
deletes the C-terminal half of WD repeat 2, all of repeats 3 and 4, and the N-terminal half of repeat 5 of CeRAE-1 (). rae-1(tm2784)
animals are sterile in the F1 generation, and have a transparent vulval region (), which is consistent with vertebrate studies showing that Rae1 is essential for mitosis and mitotic spindle assembly (Babu et al., 2003
; Blower et al., 2005
). Presumably, maternal expression of RAE-1 may rescue mitosis in the F1 generation allowing development of rae-1(tm2784)
animals to adulthood.
Transgenic animals expressing GFP with a cell specific promoter, Pmec-7, were used to visualize the morphology of the mechanosensory neurons. In wild-type animals, each ALM neuron sends an axon toward the anterior of the animal where the axon stops short of the nose (). rpm-1−/− animals had a highly penetrant phenotype in which the ALM axon fails to terminate extension properly, and overextends and hooks posteriorly (). rae-1−/− animals also had defective axon termination in the ALM neurons, although the length of the overextension was generally less, and the penetrance of overextension defects was milder than in rpm-1−/− mutants ().
Figure 3 rae-1−/− mutants have defects in axon termination and synapse formation. A and C, mechanosensory neurons (ALM and PLM) were visualized using muIs32. A, ALM axon termination defects in rpm-1−/− and rae-1−/− (more ...)
In wild-type animals, each PLM neuron extends a single axon that terminates extension prior to the ALM cell body (). In rpm-1−/− animals two PLM phenotypes were observed: a severe, highly penetrant phenotype in which the PLM axon extends beyond the ALM cell body and hooks towards the ventral cord, and a milder, less penetrant phenotype in which the PLM axon only overextends beyond the ALM cell body (). rae-1−/− animals displayed primarily overextension defects, although infrequent hooking defects were also observed ().
In summary, axon termination in the ALM and PLM neurons is defective in rae-1 −/− animals, although the defects are generally less severe and less penetrant than in rpm-1−/− mutants.
rae-1 enhances axon termination defects caused by fsn-1 and glo-4
Our previous studies have shown that two RPM-1 binding proteins, GLO-4 and FSN-1, function in parallel genetic pathways to regulate axon termination (Grill et al., 2007
). To test how rae-1
function relates to glo-4
, we generated rae-1−/−;glo-4−/−
double mutants. With regard to the ALM neurons, the penetrance of defects in axon termination were enhanced in rae-1−/−;fsn-1−/−
double mutants compared to either single mutant (). Likewise, the penetrance of defects in rae-1−/−;glo-4−/−
double mutants was increased compared to either single mutant (). With regard to the PLM neuron, the penetrance of hooking defects was strongly enhanced in rae-1−/−;fsn-1−/−
double mutants ().
Previous studies have shown that mutations in dlk-1
(a target of RPM-1's ubiquitin ligase activity), and pmk-3
(a p38 MAPK that functions downstream of dlk-1
) suppress the axon termination defects caused by rpm-1
(lf) (Nakata et al., 2005
; Grill et al., 2007
). In order to determine if loss of function in rae-1
phenotypes, we constructed rae-1−/−;rpm-1−/−
double mutants. rae-1−/−;rpm-1−/−
animals showed a similar severity and penetrance of phenotypes to those seen in rpm-1−/−
single mutants (). This observation is consistent with rae-1
functioning in the same genetic pathway. Because rae-1−/−;rpm-1−/−
animals were not suppressed, it is unlikely that RAE-1 is a target of RPM-1's ubiquitin ligase activity. Consistent with this interpretation, axon termination defects caused by rae-1(tm2784)
were not suppressed by dlk-1
rae-1 functions cell autonomously, downstream of rpm-1 to regulate axon termination
Given that rpm-
1 functions cell autonomously in the mechanosensory neurons to regulate axon termination, we sought to test if rae-1
also functions cell autonomously (Schaefer et al., 2000
; Grill et al., 2007
). To do so, we generated transgenic animals that were rae-1−/−;fsn-1−/−
double mutants and used cell specific promoters to express rae-1
. We performed our rescue analysis on rae-1−/−;fsn-1−/−
double mutants in order to show that the enhanced penetrance of defects in these animals was specifically caused by loss of function in rae-1
. As shown in , expression of rae-1
using a promoter that drives expression in the mechanosensory neurons (mec-7
promoter) rescued the defects in rae-1−/−;fsn-1−/−
animals, but expression of rae-1
in the surrounding muscle cells (myo-3
promoter) showed no rescue. Importantly, expression of rat Rae1 was also sufficient to rescue the defects in rae-1−/−;fsn-1−/−
double mutants (). These results are consistent with rae-1
functioning cell autonomously through an evolutionarily conserved mechanism to regulate axon termination.
Our genetic observations suggested that rae-1 and rpm-1 function in the same pathway. The next step in our analysis was to determine if rae-1 functions downstream of rpm-1 by transgenic expression of rae-1 in rpm-1−/− animals. As shown in , overexpression of rae-1 using the mec-7 promoter partially reduced the penetrance of defects in rpm-1−/− animals. In contrast, overexpression of the mec-7 promoter alone had no effect on the defects in rpm-1−/− mutants. These results are consistent with rae-1 functioning downstream of rpm-1.
rae-1 functions in the rpm-1 pathway to regulate synapse formation in motor neurons
Previous studies have shown that rpm-1
not only functions in axon termination, but also functions in synapse formation (Zhen et al., 2000
). Presynaptic terminals in the GABAergic DD motor neurons can be visualized using a transgene that expresses a fusion protein of S
revin (SNB)-1 and GFP (juIs1
). In wild-type animals SNB-1::GFP was localized to discrete puncta that are evenly distributed along the dorsal nerve cord (). In rpm-1
−/− mutants, the organization of SNB-1::GFP puncta was disrupted (), and the number of puncta were reduced (). In fsn-1
−/− mutants, mild defects in SNB-1::GFP puncta distribution and number were observed compared to rpm-1−/−
animals. Both glo-4−/−
−/− single mutants had wild-type distribution and numbers of puncta (). rae-1−/−;fsn-1−/−
double mutants displayed enhanced defects in the organization and number of SNB-1::GFP puncta, although the defects were not as strong as those observed in rpm-1
−/− mutants, or glo-4−/−;fsn-1−/−
double mutants (). rae-1−/−;rpm-1−/−
double mutants were not enhanced compared to rpm-1
−/− single mutants. Overall, our data are consistent with rae-1
playing an important role in synapse formation by functioning in the same genetic pathway as rpm-1
RAE-1 is localized to the perisynaptic zone of presynaptic terminals
Given our observations that rae-1
functions in the same pathway as rpm-1
to regulate axon termination and synapse formation, it was important to determine if RAE-1 was localized to the same subcellular compartment as RPM-1. Previous studies have shown that RPM-1 localizes to a poorly understood region of the presynaptic terminal called the perisynaptic zone (Abrams et al., 2008
). When RPM-1::GFP is transgenically overexpressed in the GABAergic motor neurons using the unc-25
promoter, it is present in both the cell bodies and at the perisynaptic zone of presynaptic terminals (Zhen et al., 2000
) ( and data not shown). When a fusion protein of mCherry and RAE-1 (mCherry::RAE-1) was coexpressed in GABAergic neurons with RPM-1::GFP, mCherry::RAE-1 was present in the nucleus and cell body, ventral cord, axon commissures and in the dorsal nerve cord. This pattern of distribution is much broader than RPM-1, which is consistent with RAE-1 having multiple functions in neurons. Importantly, mCherry::RAE-1 was concentrated in puncta along the dorsal nerve cord and colocalized with RPM-1::GFP (). This observation is consistent with RAE-1 binding to RPM-1, and rae-1
functioning in the same genetic pathway as rpm-1
to regulate synapse formation.
Recent work in Drosophila
showed that Highwire levels are reduced in DRae1
mutants (Tian et al., 2011
). To test if this was also the case in C. elegans
, we used an integrated transgene, juIs58
, that uses the rpm-1
promoter to drive expression of RPM-1::GFP panneuronally. In wild-type or rae-1+/−
animals, RPM-1::GFP is localized to perisynaptic puncta in many neurons, and our analysis specifically focused on the nerve ring, the dorsal cord, and the SAB neurons (Abrams et al., 2008
) (). In rae-1
−/− animals, we observed no change in the punctate localization or in the levels of RPM-1::GFP in the nerve ring, dorsal nerve cord, or the SAB neurons (). These results demonstrate that rae-1
is not required for localization of RPM-1, and are consistent with rae-1
functioning downstream of rpm-1
Binding of RAE-1 to RPM-1 is required for RPM-1 function
Our biochemical studies identified point mutations in RPM-1 that reduced binding to CeRAE-1 (). This observation prompted the question of whether RPM-1 needs to bind to RAE-1 to be fully functional. To test this question, we generated transgenic rpm-1−/− animals that used the rpm-1 promoter to express wild-type or point mutated RPM-1::GFP. We found that rpm-1 (lf) phenotypes in the mechanosensory neurons were strongly rescued by expression of wild-type RPM-1::GFP (). In contrast, transgenic expression of RPM-1::GFP (VIR to AAA) that was point mutated to reduce binding to RAE-1 was less efficient at rescuing rpm-1 (lf) phenotypes (). It is unlikely that this effect is due to reduced stability or folding of the RPM-1::GFP (VIR to AAA) point mutant, since it is expressed as a similar size protein to RPM-1::GFP (), and it is localized to perisynaptic puncta similar to RPM-1::GFP (data not shown). Thus, RAE-1 mediates a portion of RPM-1's function. It is important to note that the RPM-1 (VIR to AAA) point mutant only has partially reduced capacity to rescue rpm-1 (lf) defects because this construct presumably still binds to FSN-1 and GLO-4. Our genetic analysis discussed earlier indicates that reduced binding to RAE-1, as well as FSN-1 or GLO-4, would be required to greatly reduce the function of RPM-1 and its ability to rescue rpm-1 (lf) defects. However, we remain unable to test this prediction at present due to a lack of point mutations that specifically abolish binding of GLO-4 or FSN-1 to RPM-1. It is also plausible that we would observe greater differences between RPM-1::GFP (VIR to AAA) and wild-type RPM-1::GFP in their ability to rescue rpm-1 (lf) defects if transgenes were inserted as single copies into the genome, rather than being analyzed as extrachromosomal arrays. Nonetheless, our data are consistent with the conclusion that RPM-1 needs to bind to RAE-1 to be fully functional.
Figure 4 Binding of RAE-1 to RPM-1 is necessary for RPM-1 to be fully functional. Transgenic animals were generated on an rpm-1−/− background in order to quantify rescue of rpm-1 (lf) defects. Rescue was analyzed for transgenic expression of wt (more ...)