Post-translational processing of Schizosaccharomyces pombe YPT proteins.

ras proteins are post-translationally processed at their carboxyl-terminal CAAX motif by a triplet of modifications: prenylation of C with farnesyl, proteolytic trimming of AAX, and carboxyl-methylation. These modifications co-operate with palmitoylation of nearby sites or a polybasic region to target plasma membrane localization. The related YPT/rab proteins in contrast are localized to compartments of the endo-membrane system and may be involved in directing membrane traffic. These proteins end in XCC or CXC motifs. We have analyzed the processing of members of this subfamily form the fission yeast Schizosaccharomyces pombe. We find using in vitro translation in reticulocyte lysates that YPT1, -3, and -5 are prenylated with geranylgeranyl and that they incorporate label from [3H]mevalonic acid when expressed in transfected COS cells in vivo. Furthermore, prenylation was necessary for membrane binding in vivo. The CXC protein YPT5, but neither of the two XCC proteins YPT1 and YPT3, was carboxyl-methylated in S. pombe and in COS cells in vivo. However, YPT5 was not carboxyl-methylated in vitro in lysates which were able to methylate ras protein. YPT3 was detectably palmitoylated when expressed in COS cells, though at a much lower level than ras.

tion of the cysteine residue with the 15-carbon steroid precursor farnesol through a thioether linkage, proteolytic removal of the AAX sequence and carboxyl-methylation of the C-terminal a-carboxyl (4)(5)(6)(7). These CAAX modifications, in concert with palmitoylation of adjacent upstream cysteine residues (5) or a polybasic region (8), specify targeting to the cytoplasmic face of the plasma membrane (9). Other members of the ras superfamily which are membrane-bound and terminate in a CAAX motif, but where X = leucine, e.g. the rapla protein, are similarly processed with the exception that the cysteine residue is prenylated with the 20-carbon isoprenoid geranylgeraniol (10).
The YPT/rab family is believed to be involved in the regulation of intracellular trafficking events, and individual members have been localized to specific subcompartments of the endomembrane system (3,11,12). These proteins lack the CAAX sequence but terminate in C-terminal motifs of the type XCC or CXC and are candidates for post-translational modification. Indeed, recent studies have demonstrated, using both in vitro and in vivo approaches, that the XCC and CXC motifs can both signal geranylgeranyl addition (13,14). Furthermore, Farnsworth et al. (14) have reported that purified smgp25A (CXC motif, also called rab3A) from bovine brain membranes contains two thioether-linked geranylgeranyl groups and a methyl group at its C terminus. Since smgp25A contains no cysteine residues upstream of the doubly prenylated CXC motif, the results of this study also indicate that the C-terminal region of smgp25A is not palmitoylated. Whether all CXC proteins and/or any proteins terminating in an XCC motif are prenylated on both cysteine residues remains unclear. Mutational analysis suggests that both cysteine residues within the XCC motif of rablB can act as prenylation sites in uitro (15), but a single report that the Saccharomyces cereuisiae YPTl protein (XCC motif) is palmitoylated on one or both of the C-terminal cysteine residues (16) suggests that the XCC motif might direct only a single prenylation in at least some cases in vivo. Whether the prenylated XCC motif is also carboxyl-methylated is also unknown at present.
We have investigated the post-translational processing of three members of the YPT/rab family cloned from the fission yeast Schizosaccharomyces pombe, namely YPTl (XCC), YPTS (XCC), and YPT5 (CXC) ( Table I). These proteins have been studied following translation in vitro and in S. pombe and transfected COS cells in uiuo. We detect geranylgeranylation of all three proteins in uitro and incorporation of [3H]mevalonic acid in vivo. In other respects, however, there are substantial differences between the processing of YPT1, -3 and -5, which may indicate that each modification has a physiological function that is more specific than simply 11329

Footnote 1
Boldface letters indicate cysteine residues involved in post-translational processing.
increasing the hydrophobicity of the protein and, hence, membrane avidity.

EXPERIMENTAL PROCEDURES
cDNA Clones-Isolation of the YPTl and YPTB (formerly gene 8) clones has been previously described (17)(18)(19). A further member of the YPT gene family, YPT5, was isolated from a S. pombe genomic library in A-DASH (kindly provided by Alistair Craig, University of Oxford, United Kingdom) using oligonucleotides encoding the conserved sequence DTAGQE of ras-like proteins (20). ' The cDNAs for YPT1, YPT3, and YPTB were obtained by polymerase chain reaction as follows. Total RNA was isolated from S.
pombe (21). First-strand cDNA was prepared from 10 pg of RNA using 20 units of avian myeloblastosis virus reverse transcriptase (GIBCO-Bethesda Research Laboratories) and oligo(dT) (Pharmacia LKB Biotechnology Inc.). Oligonucleotides corresponding to 20 nucleotides of the 5'-and 3'-complementary coding regions of each gene, with additional BamHI or BglII sites at the 5' end were prepared, and polymerase chain reaction was performed using 100 pmol of each oligonucleotide, one quarter of the cDNA preparation, and 2.5 units of Tag polymerase (Perkin-Elmer/Cetus) according to the manufacturer's instructions, for 25 cycles of 92 "C (1 min), 50 "C (1 min), and 70 "C (2 rnin). The products were digested with BarnHI or BglII, cloned into plasmids, and sequenced entirely to confirm the absence of mutations. Production of Antibodies to YPTl, YPT3, and YPT5"Antibodies against YPTl were generated in rabbits by immunization with a peptide from residues 2 to 21 from the carboxyl terminus coupled to keyhole limpet haemocyanin (Calbiochem) using m-maleimidobenzyl-N-hydroxysulphosuccinimide (Pierce) (22). After two boosts with this peptide conjugate, the same animals were boosted three times with a second keyhole limpet haemocyanin conjugate made with a peptide from residues 12 to 30 from the carboxyl terminus, plus an additional cysteine for coupling, which overlapped with the original peptide by 10 residues.
DNA fragments containing the entire YPTB and YPT5 coding sequences (214 and 211 amino acids, respectively) were subcloned in frame into the Escherichia coli expression vector pATH3 (23). The trpE fusion proteins were expressed in E. coli JMlOl or TG1 cells, purified by preparative SDS-PAGE: and immunized into rabbits.
The antisera resulting from these immunizations specifically recognized the relevant protein by immunoblotting and immunoprecipitation in each case, though the efficiency of the YPTl peptide antiserum was substantially lower than that of either the trpE-YPT3 or trpE-YPT5 antisera (data not shown).
In Vitro Translation-The cDNAs for all three YPT clones were subcloned into pSP64T (24). K-ras(B) cDNAs were subcloned into the expression vector pGEM-9Z(f-) (Promega). RNA was translated in vitro using T7 or SP6 DNA-dependent RNA polymerase and linearized plasmid as template.
Premid Analysis of in Vitro Translated YPT Proteins-50 pCi of R-[5-3H]mevalonic acid was dried under vacuum at -60 "C and taken up into 50 p1 of reticulocyte lysate. Following translation the whole lysate was partitioned in Triton X-114, and the proteins in the detergent phase were precipitated with 10% trichloroacetic acid as described above. The pellet was washed three times with acetone, dried, and boiled for 4 min in SDS-PAGE loading buffer. Following SDS-PAGE the gel was soaked in Enlightening (Du Pont-New England Nuclear) and fluorographed overnight. The labeled bands were excised, washed three times with 50 mM Tris-C1 (pH 7.5), and digested out of the gel with Pronase. 10-30,000 cpm of 3H-labeled peptides were treated with methyl iodide essentially by the method described by Casey et al. (7) and extracted with chloroform-methanol (9:l). The dried products of the cleavage were dissolved in 100 pl of HPLC solvent B (100% acetonitrile containing 25 mM phosphoric acid). Immediately before injection an aliquot was diluted with an equal volume of water. Analysis was on a Brownlee RP300 cartridge (0.46 X 10 cm) with a 1-cm guard cartridge using a 30-min linear gradient from 100% solvent A (50% aqueous acetonitrile -25 mM phosphoric acid) to 100% solvent B, followed by 10 min at 100% solvent B, at 0.5 ml/min. One-min fractions were collected and counted in 6 ml of Ready-Safe scintillation mixture (Beckman). A Beckman System Gold HPLC was used, and standard isoprenoids were monitored at 254 nm. Geraniol and all-trans-farnesol were from Sigma. All-transgeranylgeraniol was a kind gift of Pat Casey (Duke Comprehensive Cancer Center) and John Glomset (Howard Hughes Medical Institute, University of Washington).
For Labeled cells were harvested by centrifugation a t 5,000 X g for 5 min, washed once with spheroplasting medium (1.2 M sorbitol, 50 mM sodium citrate, 50 mM NaZHP04, and 40 mM EDTA, pH 5.6) and resuspended in 500 p1 of spheroplasting medium. Zymolyase-20.T (0.4 mg/ml; Seikagaku, Japan) was added and samples were heated a t 37 "C for 30 min. Spheroplasts were pelleted and then lysed in 40 pl of 1% SDS, 50 mM Tris-HCI, pH 6.8, followed by heating a t 100 "C for 5 min. The cell extract was clarified by centrifugation a t 10,000 X g for 5 min, diluted into 360 pl of RIPA buffer (without SDS) containing protease inhibitors. and then immunoprecipitated with 10 pl of the relevant antiserum.
Expression in COS Cells-The cDNAs for all three YPT proteins were subcloned into the mammalian expression vector pcEXV-3 (29). COS cells were transfected by electroporation using a method modified from Chu et al. (30). In brief, cells were grown to 75% confluence, harvested by trypsinization, washed twice in HeBS (20 mM Hepes, 137 mM NaCI, 5 mM KCI, 0.7 mM NaZHP04, 6 mM dextrose; pH 7.05) and counted. 3 X 10" cells were resuspended in 230 pl of HeRS and 10 pg of plasmid DNA plus 100 pg of sonicated herring sperm DNA (Promega) in a total additional volume of 20 pl. The suspension was dispensed into a 0.4-cm Rio-Rad electroporation cuvette and pulsed at 250 V/125 pF (giving a time constant of approximately 6 ms). Cells were allowed to rest a t room temperature for 10 min before seeding into tissue culture flasks. Cells were metabolically labeled or prepared for immunoblotting 48-72 h post-transfection.
General Methods-Methods for metabolic labeling in COS cells, Triton X-114 partitioning, determination of alkali-labile carboxylmethyl groups, Western blotting, immunoprecipitation, and SDS-PAGE have been reported previously (4,s). Quantitation of radioactivity in immunoprecipitated YPT proteins was performed by scanning fluorograms using a Joyce-Loebl densitometer.
Sl00 and PI00 fractions were prepared from ["Sjmethioninelabeled COS cell lysates by a method modified from Chu et al. (30). Approximately 4 X 10" transfected COS cells in tissue culture dishes were washed twice with ice-cold phosphate-buffered saline (140 mM NaCI, 4 mM KCI, 1.7 mM KHZP04, 8 mM Na,HPO,; pH 7.4), once with ice-cold low ionic strength lysis buffer ( 5 mM Tris-CI, 1 mM MgCI?, 1 mM EGTA, 0.1 mM EDTA; pH 7.5), and then scraped into 0.5-ml lysis buffer containing protease inhibitors and left to swell on ice for 20 min. The cells were broken by 20 passages through a 21gauge hypodermic needle, and insoluble debris was removed by centrifugation a t 1000 X g for 5 min a t 4 "C. The supernatant was centrifuged a t 55,000 rpm a t 4 "C for 60 min in a Reckman TLl00 bench-top ultracentrifuge. The pellet (P100 fraction) was resuspended in 1 ml of RIPA huffer, and the supernatant (S100 fraction) was diluted with an equal volume of double strength RJPA buffer prior to immunoprecipitation.

RESULTS
Post-translational Processing of YPTl, YPT3, and YPT5 in Vitro-We have shown previously that a rabbit reticulocyte lysate, supplemented with mevalonic acid and intracellular membranes, can fully process the C A M motif of in uitro translated p2lK'"('" (31). In vitro isoprenylation of nuclear lamins (32) and mammalian rab proteins (13,15) in rabbit reticulocyte lysates has also been demonstrated. Therefore, to investigate the post-translational processing of YPT1, YPT3, and YPT5 the genes were transcribed in oitro, and the mRNA produced was used to direct in uitro translation in a nucleasetreated (message-dependent) rabbit reticulocyte lysate.
T o determine whether the YPT proteins could be prenylated in uitro, translations were performed in lysates supplemented with ['H]mevalonic acid and then immunoprecipitated with the appropriate antiserum prior to SDS-PAGE. Fig. 1 shows that label was incorporated into all three YPT proteins, though the efficiency of incorporation into YPT5 was reproducibly 4-8-fold higher than into YPTl or YPT3 when normalized for methionine incorporation and content. The ability to incorporate large amounts of labeled mevalonic acid into these proteins in oitro was exploited to allow a n analysis of the chain length of the incorporated label. In uitro translations were supplemented with ['H]mevalonic  are shown, as is the void volume of the column ( V,,). An unidentified peak ( X ) was seen in several analyses.
acid, Triton X-114-partitioned, trichloroacetic acid-precipitated, and resolved by SDS-PAGE. After fluorographic detection of the labeled bands, they were excised, digested out of the gel pieces, and subjected to methyl iodide cleavage ( 7 ) and HPLC analysis (Fig. 2). Material not retained on the column probably represents uncleaved peptides, and a similar peak was obtained in mock cleavage reactions without methyl iodide. Of the counts which were retained on the column, the majority coeluted with a geranylgeraniol standard. No label was detected in the position of farnesol (CIS isoprenoid). In each case some label eluted later than geranylgeraniol. This probably corresponds to an isomer of geranylgeraniol (33) and never represented more than 30% of the counts (usually less than 10%). Following farnesylation, p21" undergoes proteolysis to remove the AAX amino acids (4) and expose the a-carboxyl group of the now C-terminal cysteine residue which undergoes methyl esterification (4,34). Since the a-carboxyl group is already exposed on the C-terminal cysteine of the YPT proteins, no proteolysis would be required prior to methylation, and, indeed, methylation at the C terminus of the CXC protein rab3A purified from bovine brain membranes has been reported (14). For proteolysis and methylation of ~2 1~-" '~' t o occur in vitro the reticulocyte lysate must be supplemented with intracellular membranes (31). We therefore investigated whether the S. pombe YPT proteins could be methyl-esterified in vitro by translating the proteins in mevalonic acid-supplemented lysates containing canine pancreatic microsomes and S-adenosyl-[methyl-"Hlmethionine as methyl donor. Fig. 3 shows that no incorporation of label into any of the YPT proteins was evident under these conditions. However, methylation occurred both on wild type, farnesylated p21K-"'"' (Fig. 3, lane a) and also on a CAAX box mutant p21K-r"R' protein which is geranylgeranylated (Fig. 3, lane b) (35).
Post-translational Processing of YPTl, YPT3, and YPT5 in COS Cells in Vivo-Having demonstrated that YPT1, -3, and -5 all incorporated geranylgeraniol in vitro, but were apparently unmethylated, we extended the analysis to in vivo processing. COS cells have been used extensively to assay the post-translational processing of mammalian ras proteins after transfection (4, 5), and we have used similar techniques to study the processing of S. pombe YPT1, -3, and -5.
All three yeast proteins were well expressed in COS cells transfected with the relevant cDNA in the mammalian expression vector pcEXV-3. All three proteins incorporated label from ['H]mevalonic acid (Fig. 4), though the extent of labeling was insufficient to allow HPLC analysis of the isoprenoid moiety. The proteins were found to be distributed between the aqueous and detergent phase of Triton X-114 after an overnight label with ["%]methionine, with the detergent forms of YPTl and YPT5, but not YPT3, migrating slightly faster than the form found in the aqueous phase (Fig.  5A). When 50 p~ mevinolin was included during the labeling period to block isoprenoid synthesis, all three proteins were found exclusively in the aqueous phase, suggesting that the acquisition of detergent binding properties and, in the cases  and PI00 ( P ) fractions as described under "Experimental Procedures" prior to immunoprecipitation and SDS-PAGE.
Gels were fluorographed with PPO/Me,SO and exposed a t -70 "C overnight.
of YPTl and YPT5, an increased electrophoretic mobility correlated with prenylation.
An identical pattern was observed in parallel experiments in which ["S]methionine-labeled transfected COS cells were subjected to hypotonic lysis and separated into SI00 and PI00 fractions (Fig. 5 R ) . These data indicate that prenylation is necessary for membrane binding.
The CXC protein YPT5 incorporated alkali-labile counL9 from [methyl-'Hlmethionine (Table II), indicating that this protein is carboxyl-methylated in COS cells in viuo. The inclusion of mevinolin in the labeling medium prevented the incorporation of alkali-labile counts (Table II), which indi-TARLE I1 Incorporation of alkali-labile methyl groups into Y P T protein9 in S. pornbe and COS celLv The primary sequence of YPT5 contains 2 methionine residues. The initiator methionine may be removed in the processed form (38). Processing of Y P T proteins in COS cells is saturated due to overexpression; this results in the accumulation of unprocessed protein (Fig. 5). This unlabeled protein can subsequently incorporate alkalilabile counts from [methyl-'Hlmethionine, giving an artefactually high MeOH/Met ratio. For this reason, MeOH/ Met ratios for COS cell experiments are not presented. Background counts (-60 cpm) have been subtracted. cpm

Mevinolin pretreatment was for 60 min a t 50 p M .
cates that carboxyl-methylation of YPT5 in COS cells is prenylation-dependent. In contrast, neither of the two XCC proteins YPTl and YPT3 incorporated alkali-labile counts from [methyl-'Hlmethionine.
The mammalian Hand N-ras proteins are palmitoylated in uiuo and are efficiently labeled by ['HHJpalmitic acid when expressed in COS cells, partly due to the rapid turnover of the thioester-linked palmitate moiety on these proteins (36). The S. cereuisiae YPTl protein, which terminates in a XCC motif, has also been reported to be palmitoylated (16). COS cells expressing the S. pombe YPT proteins 1, 3, or 5, or Hras protein as a positive control, were incubated overnight in the presence of [:'H]palmitic acid. Fig. 4 shows that the Hras-positive control protein was strongly labeled under these conditions. The XCC protein YPT3 also incorporated counts from ['H]palmitic acid, though a t only -10-15% of the efficiency of the H-ras-positive control when normalized for protein expression. Incorporation of counts into YPTl and YPT5 could only be detected after prolonged exposure, though at levels 13-15 times lower than incorporated into YPT3. In order to investigate whether the label was incorporated as intact palmitate through a thioester linkage, as is the case for palmitoylated ras proteins (36), the immunoprecipitates from a second experiment were split into two and run on two separate SDS gels; one of these was then soaked in 1 M hydroxylamine, pH 8, which cleaves thioester bonds, prior to fluorography with PPO/Me2S0. Densitometric analysis of the "H-labeled bands showed that more than 85% of the counts incorporated into both ras and YPT3 were sensitive to hydroxylamine treatment (Fig. 6). In contrast, a "-labeled band nonspecifically immunoprecipitated by the trpE-YPT3 fusion protein antiserum was unaffected by hydroxylamine treatment (Fig. 6).
Post-translational Processing of YPTl, YPT3, and YPT5 i n S. pombe in Viuo-Some studies of the processing of YPT1, -3, and -5 in their natural environment were made possible by their relatively high level of endogenous expression in S. pombe (Fig. 7). We have recently shown that S. pombe can incorporate ['H]mevalonic acid into proteins in the form of both farnesyl and geranylgeranyl isoprenoids (37). However, we could not detect any effects of mevinolin or other hydroxymethylglutaryl-CoA reductase inhibitors on either the growth of S. pombe or on any cellular proteins, including YPT1, YPT3, and YPT5 (data not shown). Nevertheless it was possible to label YPTS (but not YPTl or YPT3) with [:'H]mevalonic acid (Fig. 7). Insufficient counkq were incorporated to confirm that geranylgeranyl was the isoprenoid added.
Endogenously expressed YPT5, but not YPTl or YPT3, incorporated alkali-labile counts from [methyl-'H]methionine, confirming the results obtained in COS cells (Table 11). Assuming equally efficient incorporation into methyl groups and methionine, and given that there are 2 methionines in YPT5, the stoichiometry of the methylation varied hetween 1.2 and 1.7 mol of methyl groups/mol of protein in different analyses (Table 11). Since the second residue of YPT.5 is Ala, the N-terminal methionine may be removed (38). If so, this would change the stoichiometry to 0.6-0.85 mol of methyl groups/mol of protein. We could not detect incorporation of ["Hlpalmitic acid into any of the YPT proteins, although a . . large number of cellular polypeptides were labeled (data not shown).

DISCUSSION
Members of the YPT/rab family of G T P binding proteins are thought to function in the different stages of eukaryotic membrane trafficking (12), a role which requires the proteins t o associate with cellular membranes. The association of ras proteins with membranes is dependent upon a series of posttranslational modifications centered around the C-terminal CAAX motif, namely farnesylation of the cysteine residue, AAX proteolysis, and methyl esterification of the now Cterminal cysteine (4-7). These modifications then combine with second signals to target plasma membrane localization of p21" (5,8,35). The YPT/rab family also have conserved C-terminal cysteine residues, arranged in CXC or XCC motifs, which are essential for the biological activity of the proteins (16,39). Recent studies have demonstrated that both XCC and CXC motifs can direct geranylgeranyl addition in uitro and in uiuo (13)(14)(15), and there is evidence that prenylation is necessary for membrane attachment of an XCC protein, rablB (13), and the CXC protein rab3A (40).
We have shown here that three members of the YPT family from the fission yeast S. pornbe, of which two have XCC carboxyl termini and one has a CXC carboxyl terminus, are all isoprenylated in vitro with C2o geranylgeranyl. The reticulocyte lysate supplemented with mevalonic acid, therefore, has all the activities necessary for synthesis and addition of geranylgeranyl to YPT proteins, as well as farnesyl to wildtype ras and geranylgeranyl to mutant ras proteins terminating in a leucine residue (31,35). Recent evidence suggests that separate enzyme activities may geranylgeranylate members of the ras superfamily ending in a C A M sequence where X = Leu (41, 42) and those which terminate in XCC/CXC motifs (13,14,43,44). We found that the CXC protein YPT5 was prenylated a t a much higher efficiency than either of the XCC proteins YPTl and YPTB under identical conditions in uitro. While this could indicate that separate enzymes were responsible for prenylation of XCC and CXC proteins in the reticulocyte lysate, a substantial difference between the efficiency of in uitro prenylation of two XCC proteins, rablB and rab2, has also been reported (15). We also detected isoprenylation of YPT1, -3, and -5 overexpressed heterologously in monkey kidney (COS) cells, but only the CXC protein YPT5 was detectably labeled by ['HI mevalonic acid in S. pornbe in uiuo. It is likely that the isoprenoid attached is C2o geranylgeranyl although the level of expression of the proteins has not allowed us to demonstrate this directly. The reason why neither of the two XCC proteins were labeled by ["Hlmevalonic acid in S. pornbe is not clear. However, YPTl and YPTB pulse-labeled with ['"'SI methionine chased rapidly from the aqueous to detergent phase of Triton X-114 following translation in S. pornbe (data not shown), compatible with the addition of a hydrophobic moiety. One explanation of this apparent discrepancy could be that the XCC proteins YPTl and YPT3, in contrast to the CXC protein YPT5, are prenylated in S. pornbe from a pool of geranylgeranyl diphosphate which is either inaccessible to exogenous ['H]mevalonic acid or has a very slow turnover rate, with the result that very little ['H]geranylgeranyl diphosphate was generated in this pool during the labeling period.
The hydroxymethylglutaryl-CoA reductase inhibitor mevinolin inhibits the synthesis of the isoprenoid precursor mevalonic acid in eukaryotic cells, including S. cereuisiae, and has been used to study the prenylation of a number of proteins, e.g ras (5). We could not demonstrate any effects of this compound in S. pornbe, however, and therefore the effects of mevinolin on YPT1, -3, and -5 were investigated in transfected COS cells. Interestingly, an effect of the related compound compactin has been demonstrated in S. cereuisiae ( 6 ) , suggesting either that mevinolin is not taken up by S. pornbe or that the S. pornbe hydroxymethylglutaryl-CoA reductase is not sensitive to this type of inhibitor. The inclusion of mevinolin in labeling experiments in COS cells confined all three YPT proteins to the aqueous phase of Triton X-114 and to the cytoplasmic (SlOO) fraction, which strongly suggests that prenylation is necessary for membrane attachment in each case. These data are compatible with recent evidence showing that the bet-2 gene is required for membrane attachment of the S. cereuisiae Yptl and Sec4 proteins (45). This gene encodes a protein homologous to the DPRllrarnl gene product, which is a subunit of a protein: farnesyl transferase (46, 47). It is likely that the bet-2 polypeptide is a component of a protein geranylgeranyl transferase activity, and a family of such subunits may exist.
We have also shown that the YPT5 protein is methylated in vivo in S. pornbe and when expressed in COS cells. As for p21", methylation of YPT5 follows, and is dependent on, prenylation of the C-terminal motif since modification did not occur in COS cells treated with mevinolin. Interestingly, the two XCC proteins were not detectably methylated in COS cells or S. pornbe in uiuo. Thus, whereas the CXC motif direcLq both geranylgeranylation and methylation, the XCC motif may direct geranylgeranylation alone.
The CAAX motif of p21K-"'"' can be methylated in uitro in a reticulocyte lysate supplemented with microsomal membranes (30). In contrast, under the same in vitro conditions, none of the YPT proteins we studied were methylated. One interpretation of these data is that YPT/rab proteins with CXC carboxyl termini are methylated by a different enzyme from the ras methyltransferase. However, the methyltransferase activities so far identified exhibit a relatively loose substrate specificity, such that methylation of simple substrates consisting only of a prenylated cysteine residue, e.g N-acetyl farnesyl cysteine or N-acetyl geranylgeranyl cysteine, can be observed with both mammalian and yeast preparations (48).:' A more likely explanation, therefore, may be that YPT5 is only prenylated on the upstream cysteine within the CXC motif in reticulocyte lysates supplemented with mevalonic acid in vitro, and hence did not provide a suitable substrate for methyl esterification in translations performed in the ' R. Rando, personal communication.
presence of mevalonic acid, microsomal membranes, and labeled S-aden~syl-jmethyE-~H]methionine. We are currently investigating these possibilities by studying the prenylation and methylation of a series of C-terminal mutants of YPT5 in which one or both of the CXC cysteine residues is/are replaced by the amino acid serine. The same interpretation could be applied to the lack of methylation of YPTl and YPTB both in vitro and in vivo.
For correct plasma membrane targeting, ras proteins require a second signal in addition to the modifications of the CAAX motif. In the case of p21H", p21N-", and p2lK."'*' this is palmitoylation of nearby upstream cysteine residues, while p21K-rw'B) requires the presence of a polybasic domain ( 5 , 8,35). In general, the YPT/rab proteins lack either of these features in their amino acid sequences. There is, however, one report which documents palmitoylation of the XCC motif of S. cereuisioe YPTl (16). We detected significant incorporation of counts from [3H]palmitic acid into the XCC protein YPT3. 85% of these counts were hydroxylaminesensitive, suggesting that the majority of the label was attached through a thioester linkage as intact palmitate. The efficiency of incorporation was much lower than that observed on to ras, which may indicate that only a small fraction of the YPT3 protein expressed in COS cells is palmitoylated, or that palmitate does not turn over on this protein, in contrast to ras (36). The fact that another XCC protein, YPT1, was not significantly labeled under the same conditions suggests that the observed palmitoylation of YPT3 does not represent an artefact of overexpression in a heterologous system. Palmitoylation of YPTB was not detected in S. pombe, despite the fact that many other proteins incorporated label from [3H]palmitic acid. We could not investigate whether S. pombe RAS is palmitoylated for use as a comparable positive control for palmitoylation of YPTB because we could not detect the S. pombe RAS protein using the anti-mammalian ras antibody Y13-259 (data not shown). Thus the significance of this negative result cannot be fully interpreted. Treatment of both S. pombe and transfected COS cell lysates with hydroxylamine prior to subcellular fractionation did not redistribute membrane-bound YPTB into the cytoplasmic (SlOO) fraction (data not shown). This result is consistent with the observation that a mutant p21K-"'B' protein that is geranylgeranylated and lacks the polybasic region (i.e. the second signal for membrane localization of wild-type K-ras(B)) nevertheless binds strongly to cellular membranes (35); thus a single geranylgeranyl moiety alone is sufficient to bind the protein stably to membranes.
The proposed function of YPT/rab proteins is to mediate specific targeting and fusion of vesicles involved in membrane traffic. It is thought that these proteins function by binding t o membranes and then being released into the cytosol after their function is complete, to reattach for another cycle (11,12). Isoprenylation is a stable modification at least for ras proteins ( 5 ) and is therefore unlikely to mediate reversible membrane association directly. However, we observed substantial differences in other aspects of the post-translational processing of the three YPT proteins, which may be of functional significance. For example, there may be differences between methylated and nonmethylated Czo modified YPT/ rab proteins, since methylation is required for efficient membrane binding of ~2 1~"~'~' (31). However, we could not detect any turnover of the methyl group on YPT5 which could modulate membrane binding (data not shown). Moreover, a single geranylgeranyl moiety is considerably more hydrophobic than a farnesyl group4 This suggests that any effect of S. Black, personal communication.
methylation on membrane binding is likely to depend upon a mechanism other than its effect on the overall hydrophobicity of the molecule, especially if YPT5 is doubly geranylgeranylated as in the case of another methylated CXC protein, rab3A (14). Prenylation and other potential post-translational modifications, including palmitoylation and phosphorylation (491, may also modulate membrane attachment and functional recycling, either directly or by modulating the interaction of YPT/rab proteins with other proteins. Indeed the rab3A guanine nucleotide dissociation inhibitor interacts only with the post-translationally modified form of the protein and may be involved in the functional recycling of fully modified rab3A between membrane and cytosol (50).
Current evidence would suggest that most, if not all, YPT/ rab proteins are hydrophobically modified by one or two geranylgeranyl groups, and yet individual YPT/rab proteins have been localized to specific subcompartments of the endomembrane system (12). It appears that the hypervariable C-terminal domain contributes to the information that is required for this subcellular targeting (51). Whether variations in the nature of the post-translational modification(s) on these proteins cooperates with such protein sequences to determine the specificity of membrane association remains unclear.
The YPT/rab proteins, which may recycle between membranes and cytoplasm, are post-translationally processed by the attachment of the highly hydrophobic group geranylger-any1 and, in some cases, one or more additional modifications. In contrast, the ras proteins exhibit some biological activity when farnesylated and weakly membrane-bound (5). However, optimal biological activity depends upon efficient membrane-binding promoted by subsequent palmitoylation or a polybasic region (5,8). These and other data suggest that the post-translational modifications on members of the ras superfamily of proteins serve functions that are more subtle than simply increasing their overall hydrophobicity and hence their affinity for cellular membranes.