Functional Overlap between Retinitis Pigmentosa 2 Protein and the Tubulin-specific Chaperone Cofactor C*

Mutations in the X-linked retinitis pigmentosa 2 gene cause progressive degeneration of photoreceptor cells. The retinitis pigmentosa 2 protein (RP2) is similar in sequence to the tubulin-specific chaperone cofactor C. Together with cofactors D and E, cofactor C stimulates the GTPase activity of native tubulin, a reaction regulated by ADP-ribosylation factor-like 2 protein. Here we show that in the presence of cofactor D, RP2 protein also stimulates the GTPase activity of tubulin. We find that this function is abolished by mutation in an arginine residue that is conserved in both cofactor C and RP2. Notably, mutations that alter this arginine codon cause familial retinitis pigmentosa. Our data imply that this residue acts as an “arginine finger” to trigger the tubulin GTPase activity and suggest that loss of this function in RP2 contributes to retinal degeneration. We also show that in Saccharomyces cerevisiae, both cofactor C and RP2 partially complement the microtubule phenotype resulting from deletion of the cofactor C homolog, demonstrating their functional overlap in vivo. Finally, we find that RP2 interacts with GTP-bound ADP ribosylation factor-like 3 protein, providing a link between RP2 and several retinal-specific proteins, mutations in which also cause retinitis pigmentosa.

Retinitis pigmentosa is a degenerative disease and the major cause of heritable blindness (1). It is caused by mutations in a dizzying variety of genes; some are structural proteins of photoreceptors, many are involved in photoreception and transduction, and a few are seemingly unrelated ubiquitously expressed proteins (see RetNet, www.sph.uth.tmc.edu/retnet/). Most (80 -90%) cases of X-linked retinitis pigmentosa are caused by mutations in two genes, XRP2 and XRP3 (2). These were isolated by positional cloning and shown to encode, respectively, the RP2 1 protein (3) and the retinitis pigmentosa GTPase regulator (RP3) (4,5). The function of RP2 is unknown, and the gene is expressed at a low level in all tissues examined; it is targeted to the plasma membrane by myristoylation and palmitoylation of its amino terminus (6).
RP2 has amino acid sequence similarity over half of its length to the tubulin-specific chaperone protein cofactor C (3). It has been shown that cofactor C acts in a pathway together with four other tubulin-specific chaperone proteins (cofactors A, B, D, and E) to chaperone quasi-native ␣and ␤-tubulin subunits released from the cytosolic chaperonin CCT and assemble the ␣/␤-tubulin heterodimer. Each step in this pathway has been deduced from in vitro reconstitution experiments using purified components (7)(8)(9). The pathway hinges on the formation of a supercomplex containing ␣Ϫ and ␤-tubulin and cofactors C, D, and E. The hydrolysis of GTP by tubulin, stimulated by cofactors C and D, is part of the heterodimer assembly reaction; the stimulated hydrolysis of GTP by ␤-tubulin acts as a switch for the release from the supercomplex of native, newly made tubulin heterodimers (10). Cofactor C and D in combination have also been shown to be a GTPase activator (GAP) for native tubulin; cofactor E enhances this tubulin-GAP activity (10). These biochemical data are consistent with the wealth of genetic evidence on cofactor homologs in the yeasts Saccharomyces cerevisiae and Schizosaccharomyces pombe (11)(12)(13)(14). In S. cerevisiae, these genes were first identified in screens for chromosomal loss and supersensitivity to microtubule poisons. CIN1, CIN2, CIN4, and PAC2 are the putative homologs of cofactor D, cofactor C, Arl2 (see below), and cofactor E, respectively.
The mammalian Arl (ADP ribosylation factor (ARF)-like) proteins constitute a family of Ras-related small GTP-binding proteins with at least eight members (15)(16)(17)(18). They share 40 -60% amino acid sequence identity with ARF proteins but do not act as cofactors in the cholera toxin-induced ADP-ribosylation of G␣ S nor do they possess the other biochemical activities that characterize ARF proteins (19 -21). Arl2 modulates the tubulin-GAP activity of cofactors C and D (22), and mutations in its putative yeast homologs affect microtubule stability (11,12,23). Mutations in Arl2 in Arabidopsis thaliana also result in the loss of microtubules (24,25). Given the sequence similarity between cofactor C and RP2 and the interaction of Arl2 with tubulin-folding cofactors, we decided to compare the functional and biochemical properties of RP2 and cofactor C and to investigate the possibility that RP2 might interact with one or more Arl proteins. Our data show that RP2 acts as a GAP for tubulin in concert with cofactor D, but unlike cofactor C, it does not catalyze tubulin heterodimerization. We also show that both cofactor C and RP2 partially complement CIN2 deletion in S. cerevisiae and that RP2 binds to Arl3 in a nucleotide and myristoylation-dependent manner.

EXPERIMENTAL PROCEDURES
Plasmid Construction-DNA fragments encoding human RP2, cofactor C, BART (binder of Arl2) and carboxyl-terminally His 6 -tagged Arl2, Arl3 and Arl4 were generated by PCR using reverse transcribed human testis RNA (CLONTECH, Inc., Palo Alto, CA) as template. These were inserted into the pET23b vector (Novagen, Inc., Madison, WI). From these plasmids, mutated cDNAs (encoding RP2-R118H, RP2-⌬S6, cofactor C-R262A, Arl3-T30N, and Arl3-Q70L) were generated by PCR, using the recombinant pET23b plasmids as template, with specific mismatched primers and the QuikChange site-directed mutagenesis kit (Stratagene Inc., La Jolla, CA). For the expression of glutathione S-transferase-RP2 (GST-RP2), a fragment encoding human RP2 was generated and subcloned into the pGEX-4T-1 vector (Stratagene Inc., La Jolla, CA) in frame with the GST-coding region. For expression in yeast, human RP2 and RP2-R118H, human cofactor C wild type, and R262A were subcloned from pET23b into pRS413-GPD carrying the HIS3 gene and the 2 ori (26). CIN2 was cloned into the same vector by PCR from a yeast genomic library. All constructs were verified by DNA sequencing.
Expression and Purification of Proteins-Bovine brain tubulin (27) was obtained free from associated proteins by passage through phosphocellulose (28). The cytosolic chaperonin CCT and cofactors B, D, and E were purified as described previously (8,9,29). Full-length, untagged wild type and mutant cofactor C were expressed in Escherichia coli BL21 (DE3) cells using the expression plasmids described above. Harvested cells were lysed using a French press and clarified by centrifugation, and the recombinant proteins were purified through two dimensions as follows: 1) a cation exchange column of SP-Sepharose (Amersham Biosciences), developed with a linear gradient of 10 -250 mM sodium phosphate buffer (pH 6.8) containing 1 mM each DTT, MgCl 2 , and EGTA and 2) an anion exchange column (Q15; Amersham Biosciences), also developed using a phosphate buffer gradient (10 -400 mM sodium phosphate buffer, pH 7.3). The recombinant proteins were monitored by SDS-PAGE analysis of fractions emerging from the columns. Wild type and mutant RP2 proteins were expressed from pET23b recombinant plasmids harbored in E. coli BL21(DE3). Cell lysates were cleared by centrifugation at 100,000 ϫ g and applied to a Q15 anion exchange column developed with a linear gradient of NaCl (0.01-0.7 M NaCl in 10 mM phosphate buffer, pH 7.4, 0.5 mM MgCl 2 , 1 mM EGTA, and 1 mM DTT). The purified material from this dimension was loaded on a column of hydroxyapatite (Pentax, American International Chemical Co., Inc., Natick, MA) in 0.5 mM MgCl 2 , 50 M CaCl 2 , and 1 mM DTT. Elution was performed in the same buffer using a linear gradient of 0 -0.25 M sodium phosphate buffer, pH 6.7. Fractions enriched in RP2 were further purified by application to a MonoQ anion exchange column (Amersham Biosciences) and eluted in 20 mM Tris-HCl, pH 8, and 1 mM DTT via a 0.0 -0.5 M gradient of MgCl 2 . Following each dimension of purification, aliquots of fractions emerging from the column were analyzed by Western blotting using a polyclonal anti-RP2 antibody for the detection of immunoreactive material. His 6 -tagged Arl2, Arl3, Arl4, and GST-tagged RP2 were affinity-purified on cobalt resin (CLONTECH, Inc., Palo Alto, CA) and agarose-coupled glutathione (Amersham Biosciences), respectively, according to the manufacturer's recommended protocols.
Generation of Antisera to Cofactor C, RP2, and Arl3-Purified recombinant proteins were used as antigens for the production of antisera in rabbits (Arl3, cofactor C) or guinea pigs (RP2) (Cocalico Biologicals Inc., Reamstown, PA). For this purpose, Arl3 was first coupled to keyhole limpet hemocyanin. The resulting sera were tested for their titer and specificity by Western blotting against both their respective antigens and whole cell lysates.
Tubulin-GAP Assays-Purified bovine brain tubulin (1 M) was incubated at 37°C in the presence of 25 M [␥-32 P]GTP (10 Ci) in folding buffer in the presence of various combinations of tubulin-folding cofactors and RP2 (see Fig. 1 legend). Release of radiolabeled inorganic phosphate was measured at 1 min intervals as described (10).
Complementation Assay in S. cerevisiae-A CIN2 deletion strain (413.4B:MAT␣ ade2 his3 leu2 lys2 ura3 cin2::LEU2) (11) was transformed according to the lithium acetate method with either pRS413 alone or pRS413 carrying the wild type or mutant versions of RP2 or cofactor C. CIN2-pRS413 was used as a positive control. Transformants were selected on synthetic medium. Single colonies were picked and . Reaction products were resolved by native gel electrophoresis followed by autoradiography to visualize those products that non-exchangeably bound radiolabeled GTP. Only in lane 1 is native heterodimeric tubulin regenerated. B, RP2 has tubulin-GAP (GTPase-activator protein) activity. Tubulin (1.0 M) was incubated with cofactors C (80 g/ml), D, and E (30 g/ml each) (open squares); cofactors D and E (30 g/ml each) and RP2 (250 g/ml) (asterisks); cofactor D (30 g/ml) and RP2 (250 g/ml) (crosses); cofactors D and E (30 g/ml each) and RP2 (80 g/ml) (open triangles); and RP2 (250 g/ml) (closed diamonds). The experiment marked with closed circles represents incubation of RP2 without added tubulin. The release of inorganic phosphate from [␥-32 P]GTP by hydrolysis was measured at 1-min intervals. grown at 30°C overnight in liquid medium lacking the appropriate amino acids. Transformants were resuspended in sterile water at the same optical density, and growth on plates was tested at 26°C by spotting serial dilutions of cell suspensions onto yeast extract/peptone/ dextrose plates containing 0, 1, or 5 g/ml of benomyl.
Translation in Vitro and Binding Assays-cDNAs encoding RP2, cofactor C, and BART were transcribed and translated in vitro using TnT rabbit reticulocyte lysate (Promega, Inc., Madison, WI) in the presence of [ 35 S]methionine (0.8 mCi/ml) at 30°C for 1 h. Samples were cleared of ribosomes by centrifugation at 200,000 ϫ g for 20 min and preincubated with uncoupled Talon cobalt affinity beads (CLONTECH, Inc.) to remove nonspecifically bound translated protein. Transcription/ translation reactions were diluted 20-fold in reaction buffer (0.3 M NaCl, 50 mM phosphate, pH. 7.2, 1 mM MgCl 2 ) and incubated for 1 h at room temperature with cobalt affinity beads preloaded with about 25 g of bound His 6 -Arl2, 3, or 4 or (in a control reaction) without bound protein.
The resin-bound complexes were extensively washed with reaction buffer containing 0.05% Nonidet P-40 and eluted in SDS gel loading buffer. Some of the in vitro transcription/translation reactions were done by addition of ethanol alone or 0.6 mM DL-␣-hydroxymyristic acid (HMA, Sigma) dissolved in ethanol (the maximum that the system could tolerate). GST-RP2 binding assays were carried out by incubating 0.5 mg of GST-RP2 bound to glutathione-conjugated-Sepharose 4B beads (Amersham Biosciences). For binding to immobilized RP2-GST, unfractionated soluble protein from a bovine testis lysate (9) was transferred into PBS by passage over Sephadex G25. 0.6 ml of this material (containing 20 mg/ml total protein) was used for each binding reaction. Incubations were performed at room temperature in the presence of various nucleotides each at a concentration of 1 mM. The beads were washed with PBS containing that nucleotide and 0.05% Tween 20. Protein was eluted with 15 mM glutathione in 50 mM Tris-HCl, pH 8, fractionated on SDS-PAGE and analyzed by Western blotting using rabbit anti-Arl3 antisera or a mouse monoclonal anti-␤-actin antibody (Sigma).

Functional Comparison of RP2 and Cofactor C-Human RP2
and human cofactor C share 53% similarity and 29% identity over a domain of ϳ200 amino acids. The untagged, bacterially expressed proteins were purified to homogeneity through several dimensions of ion exchange chromatography. Using these purified recombinant proteins, together with the cytosolic chaperonin CCT and the other tubulin-specific chaperone proteins required for tubulin heterodimer formation (i.e. cofactors B, D, and E) (9), we performed in vitro ␣-tubulin refolding reactions in the presence of radiolabeled GTP. In these reactions, tubulin is cycled through the folding (CCT-driven) and heterodimerization (tubulin-folding cofactors) machinery, in the process incorporating non-exchangeable GTP bound to ␣-tubulin in the ␣/␤ heterodimer (10,31). The products were analyzed by non-denaturing gel electrophoresis (Fig. 1A). We found that despite its sequence similarity, RP2 was unable to substitute for cofactor C in these in vitro folding reactions and hence cannot participate in the heterodimerization of tubulin under these experimental conditions. Increasing the concentration of RP2 in these reactions had no effect (data not shown). However, reactions containing RP2, cofactors D, or D and E, and native tubulin showed tubulin-GAP activity (Fig. 1B). No GTP hydrolysis was observed in control reactions containing either RP2 alone or RP2 plus tubulin. We conclude that RP2, like cofactor C, acts in concert with cofactor D as a tubulin-GAP. As is the case with cofactor C, cofactor E enhances this tubulin-GAP activity but is not essential to it (Fig. 1B). A higher concentration of RP2 relative to cofactor C is needed to elicit comparable tubulin-GAP activity; possible reasons for this difference are discussed below.
Mutation of the Arginine Conserved in RP2 and Cofactor C-It has been proposed that GTPase activators share a common active site consisting of an arginine finger (32). Arg-118 in RP2 and Arg-262 in cofactor C constitute the only pair of conserved arginines found in these two proteins ( Fig. 2A). Therefore, using site-directed mutagenesis, we engineered plasmid constructs expressing either cofactor C or RP2 mutated at this position, i.e. cofactor C-R262A and RP2-R118H. In the latter case, mutation of arginine to histidine was chosen because this mutation has been shown to cause X-linked retinitis pigmentosa in some families (3,33). We also made a plasmid expressing another disease mutant RP2 protein, RP2-⌬S6. This mutation is known to prevent myristoylation of the protein, a defect leading to intracellular misrouting (2,6). The three mutant proteins were expressed in E. coli and purified using precisely the same chromatographic dimensions used to purify the wild type protein. The mutant proteins behaved indistinguishably from their wild type counterpart on all dimensions (see "Materials and Methods"). We conclude that these mutations do not result in any global changes in the structures of the respective proteins. We tested each mutant protein in tubulin-GAP reactions; the results are shown in Fig.  2. Mutation of the invariant arginine common to both RP2 and cofactor C completely abolished their tubulin-GAP activity (Fig. 2, B and C). However, deletion of serine 6 in RP2 had no effect on its GAP activity (Fig. 2C). Given the structural integrity of the mutant proteins, it seems likely that the invariant arginine is critical for tubulin-GAP activity and suggests that it is indeed part of an arginine finger.
Complementation Assay in S. cerevisiae-Cofactor C and RP2 are distantly related in sequence to the yeast protein CIN2 (11,12,40). CIN1, CIN2, CIN4, and PAC2 (putative homologs of cofactor D, cofactor C, Arl2, and cofactor E, respectively) act together in a pathway affecting microtubule stability. Yeast harboring mutations in these genes show supersensitvity to cold and to benomyl, both of which destabilize microtubules. To test whether human cofactor C or RP2 can compensate for the loss of CIN2 in yeast, we transformed plasmids expressing wild type and mutant versions of these proteins into a CIN2 deletion strain. We assayed the resulting strains for their benomyl resistance in the cold (26°C) where benomyl supersensitivity of the parental strain is enhanced. The positive control was CIN2cloned into the same vector, and the negative control was the vector itself. As shown in Fig. 3, only wild type RP2 or cofactor C but not RP2-R118H, cofactor C-R262A, or vector alone could complement for the loss of CIN2. Neither RP2 nor cofactor C complemented as well as CIN2 itself, and cofactor C was more effective than RP2 in restoring benomyl resistance.
Binding of RP2 to ADP Ribosylation Factor-like Protein 3 (Arl3)-The tubulin-GAP activity of cofactors C and D is modulated by the small G protein Arl2 (ARF-like protein 2) (22). We therefore examined the possibility that one or more members of the Arl family of proteins might bind to RP2. To do this, we generated constructs for the expression in E. coli of Arl2, Arl3, and Arl4, each His 6 -tagged at their carboxyl termini. The Arl2and Arl3-binding protein BART (34, 35) was used as a positive control. Agarose-bound nickel beads loaded with each of the His 6 -tagged proteins were incubated with either 35 S-labeled RP2, BART, or cofactor C generated by translation in vitro. Analysis of the bound products by SDS-PAGE showed that Arl3 bound to in vitro translated RP2 but only in the presence of the slowly hydrolyzable GTP analogue, GTP-␥-S (Fig. 4A). This suggested a specific binding between RP2 and GTP-Arl3. As a positive control, both Arl2 and Arl3 bound to BART, while none of the three Arl proteins tested bound to cofactor C (Fig. 4A).
We also made two mutant forms of Arl3, Arl3-Q71L and Arl3-T30N, based on mutant proteins used for the study of the well characterized and quite similar Ras family of G-proteins (36,37). These mutations correspond to Q61L and T17N in Ras. The former mutation results in a defect in GTP hydrolysis, leading to GTP-bound mutant protein, whereas the latter is defective in GTP binding, resulting in nucleotide-free or GDPbound protein. We found that the putative GTPase-defective mutant Arl3 binds most strongly to RP2, reinforcing our conclusion that it is GTP-Arl3 that binds to RP2 (Fig. 4B). The interaction between RP2 and Arl3 was also found using purified tagged proteins (data not shown), demonstrating that the interaction is direct and not mediated by some third protein.
To confirm the binding of RP2 to GTP-Arl3 in tissue extracts, we expressed RP2 as a GST fusion protein, immobilized it on agarose-bound glutathione beads, and examined the ability of these beads to specifically bind proteins contained in an unfractionated tissue extract. These experiments were done in the presence of GTP, GDP, GTP-␥-S, or no nucleotide. In each case, the bound proteins were eluted, resolved by SDS-PAGE and detected by Western blotting with an anti-Arl3 antiserum. These experiments showed that binding of Arl3 is strongest in the presence of GTP-␥-S (Fig. 4C), supporting a specific interaction between RP2 and GTP-Arl3 in the context of whole cells.
We have shown that Arl2 inhibits the tubulin-GAP activity of the tubulin-specific chaperones known as cofactors C and D (22). However, we found that neither wild type nor mutant   FIG. 3. Cofactor C and RP2, but neither cofactor C-R262A or  RP2-R118H, complement a CIN2 deletion phenotype in yeast. CIN2, cofactor C (wild type and R262A), RP2 (wild type and R118H), and vector alone were each transformed into a CIN2 deletion strain. Individual transformants were grown in selective medium overnight and spotted in serial dilutions onto yeast extract/peptone/dextrose plates containing 0, 1.0, or 5.0 g/ml benomyl to assess growth under these conditions. Incubation was at 26°C to enhance benomyl sensitivity. Cc, cofactor C. A 3 ), or Arl4 (lane A 4 ) bound to cobalt affinity beads or beads alone (lane C) were incubated in the presence of 35 S-radiolabeled BART, RP2, or cofactor C either in the presence or absence of GTP-␥-S. Eluted products were analyzed by SDS-PAGE. I, input translated BART, RP2, or Cofactor C corresponding to 40% of the amount used in the other lanes. B, interaction between RP2 and either wild type or mutant forms of His 6 -tagged Arl3. Wild type and His 6 -tagged mutant (Q71L and T30N) forms of purified recombinant Arl3 were incubated with 35 S-radiolabeled RP2 and analyzed as described in A. C, GST-RP2 binds to Arl3 in a whole tissue extract. Immobilized GST-RP2 or GST alone was incubated with a total soluble protein extract from bovine testis. Bound and eluted material was analyzed by Western blotting with either an anti-Arl3 or anti-␤-actin antibody. Left-hand panel, Western blot of input bovine testis soluble extract (I) corresponding to the amount used in the lanes shown in the center panel. Center panel, Western blot of material binding to beads (lane C) or to GST alone. ϪL and ϩL, material binding to GST-RP2 in the absence or presence, respectively, of bovine testis lysate. Right-hand panel, Western blot of (I) tissue extract corresponding to the amount used in the adjacent binding experiments; GTP, GDP, GTP-␥-S, Ϫ, material bound to GST-RP2 in the presence of GTP, GDP, GTP-␥-S (all at 1 mM) or no nucleotide. His 6 -tagged Arl3 modulated the tubulin-GAP activity of RP2 and cofactor D, nor did RP2 behave as a GTPase activator of Arl3 (data not shown). However, these negative results could result from using bacterially synthesized Arl3 protein that is both tagged and unmodified, as well as recombinant RP2, which is not acylated.

FIG. 4. RP2 binds to GTP-Arl3 (ADP-ribosylation factor-like protein 3). A, His 6 -tagged Arl2 (lane A 2 ) Arl3 (lane
Because our data established that RP2 binds to GTP-Arl3, we decided to see whether there was differential binding among the RP2 mutant proteins that we used in the GAP experiment shown in Fig. 2. We found that in vitro translated RP2-R118H binds to the GTPase-defective form of Arl3 more weakly than wild type RP2, whereas RP2-⌬S6 bound much more strongly to this form of Arl3 and to wild type Arl3 (Fig.  5A). Since deletion of serine 6 in RP2 is known to prevent its myristoylation (6), we tested whether the enhanced binding of RP2-⌬S6 might be a result of its inability to acquire this posttranslational modification. To do this, we included an inhibitor of myristoylation (DL-␣-hydroxymyristic acid) (38) in in vitro translation reactions and showed that a greater proportion of wild type RP2 is bound to Arl3 when myristoylation is inhibited (Fig. 5B). We conclude that myristoylation of RP2 indeed weakens its binding to Arl3 while increasing the nucleotide dependence of that interaction. DISCUSSION Retinitis pigmentosa 2 protein, RP2, is similar in sequence over half its length to the tubulin-specific chaperone cofactor C. We investigated the functional similarities and differences between the two proteins. We demonstrate here that the two proteins have overlapping but not identical functions. Both stimulate the GTPase activity of native tubulin, in both cases only with the cooperation of cofactor D. However, only cofactor C participates in the heterodimerization of newly folded tubulin subunits. Whereas the ADP-ribosylation factor-like 2 pro-tein (Arl2) regulates the tubulin-GAP activity of cofactors C and D (22), the related protein Arl3, to which RP2 binds specifically (see below), does not affect the tubulin-GAP activity of RP2 and cofactor D. The fact that RP2 and cofactor C are not functionally identical is not surprising given that they only share one of two putative domains and that the former is largely a membrane-associated protein, whereas the latter is cytosolic.
Remarkably, we found that mutation of the only conserved arginine residue present in both RP2 and cofactor C (R118H in RP2 and R262A in cofactor C) causes a total loss of tubulin-GAP activity in each protein. The mutation R118H in RP2 is one that has been shown to cause familial retinitis pigmentosa (3). This mutation does not, however, lead to the mislocalization or destabilization of the RP2 protein (2,6). When expressed in cultured cells (either tagged or untagged), RP2-R118H is expressed at high levels and is targeted to the plasma membrane in a manner indistinguishable from wild type RP2. Combining these genetic studies with the biochemical data presented here, we conclude that the loss of tubulin-GAP activity itself may cause retinitis pigmentosa in families harboring this mutation.
The active site of many different GAPs is hypothesized to contain an arginine finger (32). It is likely that the arginine residue we have mutated plays this role in cofactor C and RP2 for two reasons. First, it is the only arginine that is invariant in all sequenced RP2 and cofactor C proteins (2). Second, mutation of this arginine leads to a total loss of GAP activity, while leaving the stability and chromatographic properties of both proteins unchanged. From the latter fact we conclude that mutation of the conserved arginine does not lead to a global rearrangement or misfolding of the proteins. If structural analysis confirms our inference that cofactor C and RP2 act as FIG. 5. Differential binding of RP2 and RP2 mutants to Arl3 and the effect of myristoylation on binding of wild type RP2 to Arl3. A, RP2-⌬S6 binds more strongly than wild type RP2 or RP2-R118H to both wild type Arl3 and Arl3-Q71L. Left-hand panel, input RP2 wild type and mutant translation products. Center and right-hand panels, binding of wild type Arl3 (center panel) or Arl3-Q71L (right panel) to radiolabeled wild type or mutant RP2. B, effect of an inhibitor (HMA) of myristoyl-CoA:protein N-myristoyltransferase on binding of wild type RP2 to Arl3-Q71L. Wild type RP2 was translated in the absence (control) or in the presence of HMA, and the reaction products were incubated in the presence of His 6 -tagged Arl3-Q71L conjugated to cobalt beads. Recovered complexes were eluted and analyzed by SDS-PAGE along with an equivalent amount of inputtranslated RP2. The bands corresponding to bound and input RP2 were quantitated by phosphorimaging. The ratio of these two numbers, expressed as a percentage and averaged from three independent experiments, is shown in the figure. tubulin-GAPs through an arginine finger, this would be very striking given the fact that tubulin has a very different GTP binding pocket than the small GTP-binding proteins (39).
CIN2, the putative homolog of cofactor C in yeast, acts together with CIN1, CIN4, and PAC2 (the putative homologs of cofactor D, Arl2, and cofactor E, respectively) in a pathway affecting microtubule stability (11,12,40). We have shown here that cofactor C and RP2 can partially complement a deletion in CIN2. However, when mutated at the conserved arginine residue, neither protein can restore benomyl resistance to the CIN2 deletion strain. These findings suggest that tubulin-GAP activity is essential for restoring normal microtubule function in vivo. There was a gradient in the effectiveness of these proteins in conferring benomyl resistance: CIN2 Ͼ cofactor C Ͼ RP2. This result is readily explained by the degree of sequence similarity among the three proteins. We infer that cofactor C is the human homolog of CIN2 and that RP2 (which is similar in sequence over only half its length to cofactor C and CIN2) shares some but not all functions of the latter two proteins.
As tubulin-GAPs, RP2 and cofactor C convert GTP-tubulin to GDP-tubulin. Since it is only the former species that is competent to polymerize into microtubules, these GAP activities could regulate microtubule polymerization in vivo. Microtubules are stabilized against catastrophic depolymerization by a dynamic cap of GTP-tubulin formed because, although hydrolysis of bound GTP by tubulin is coupled to its polymerization, this reaction lags slightly behind the polymerization reaction (41,42). If the GAP activity of either RP2 or cofactor C causes the destruction of this GTP cap, these proteins would be powerful regulators of microtubule dynamics. Since RP2 is plasma membrane-bound, it is possible that it plays a role in maintaining the plasticity of microtubules near the periphery. The fact that RP2 is membrane-bound in vivo means that in the cell, it is effectively concentrated in two dimensions. This may explain the somewhat higher concentrations of RP2 required for GAP activity in the in vitro reaction compared with those of cofactor C, which is a cytosolic protein. In addition, the myristoylation and palmitoylation of RP2 may enhance its GAP activity (we used recombinant, unmodified RP2 in our in vitro GAP assays). There are several reasons to believe that the GAP activity of RP2 is tubulin specific. 1) This activity is totally dependent on the action of a third protein (cofactor D). 2) RP2 displays no GAP activity toward, for example, recombinant ADP ribosylation factor-like proteins. 3) GAP proteins are usually extremely specific (36), sometimes selectively recognizing specific isoforms of G proteins.
Biochemical and genetic data have previously shown that Arl2, a member of the ARF-like (Arl) subfamily of small G proteins, regulates tubulin-specific chaperones (22,23). Here we show that RP2 binds to the related protein Arl3. The interaction of Arl3 and RP2 is very likely to be physiologically significant, because 1) it occurs in whole cell extracts and 2) it is GTP-dependent, as is typically the case for interactions of G proteins with their effectors. Interestingly, we find that while RP2-R118H shows a weak affinity to Arl3, RP2-⌬S6 binds more strongly to it than its wild type counterpart as does unmyristoylated RP2. Our data imply that the absence of myristoylation causes the stronger binding of RP2-⌬S6. It is possible that in vivo Arl3 binds to unmodified RP2 and participates in its targeting.
The interaction of RP2 and Arl3 is particularly interesting since Arl3 has recently been shown to bind to PDE ␦, the ␦ subunit of rod-specific cyclic GMP phosphodiesterase (43), a molecule involved in phototransduction. PDE ␦ also interacts with the retinitis pigmentosa GTPase regulator (44); mutations in the gene (XRP3) encoding this protein are the major cause of X-linked retinitis pigmentosa. Thus, it is tempting to speculate that Arl3 links RP2 with RGPR and PDE ␦ in a common pathway necessary for the maintenance of rod photoreceptor cells.