MOLECULAR AND CELLULAR BIOLOGY, Dec. 1994, p. 7839-7854 Vol. 14, No. 120270-7306/94/$04.00 + 0Copyright ( 1994, American Society for Microbiology

The Srp54 GTPase Is Essential for Protein Export in theFission Yeast Schizosaccharomyces pombeSTEVEN M. ALTHOFF, SCOTTI W. STEVENSt AND JO ANN WISE*Department of Biochemistry, University of Illinois, Urbana, Illinois 61801

Received 18 February 1994/Returned for modification 11 April 1994/Accepted 29 August 1994

Signal recognition particle (SRP) is a cytoplasmic ribonucleoprotein required for targeting a subset ofpresecretory proteins to the endoplasmic reticulum (ER) membrane. Here we report the results of a series ofexperiments to define the function of the Schizosaccharomyces pombe homolog of the 54-kDa subunit ofmammalian SRP. One-step gene disruption reveals that the Srp54 protein, like SRP RNA, is essential forviability in S. pombe. Precursor to the secretary protein acid phosphatase accumulates in cells in which Srp54synthesis has been repressed under the control of a regulated promoter, indicating that S. pombe SRP functionsin protein targeting. In common with other Srp54 homologs, the S. pombe protein has a modular structureconsisting of an amino-terminal G (GTPase) domain and a carboxyl-terminal M (methionine-rich) domain. Wehave analyzed the effects of 17 site-specific mutations designed to alter the function of each of the four GTPaseconsensus motifs individually. Several alleles, including some with relatively conservative amino acidsubstitutions, confer lethal or conditional phenotypes, indicating that GTP binding and hydrolysis are criticalto the in vivo role of the protein. Two mutations (R to L at position 194 [R194L] and R194H) which weredesigned, by analogy to oncogenic mutations in ras, to dramatically decrease the catalytic rate and one (T248N)predicted to alter nucleotide binding specificity produce proteins that are unable to support growth at 18°C.Consistent with its design, the R194L mutant hydrolyzes GTP at a reduced rate relative to wild-type Srp54 inenzymatic assays on immunoprecipitated proteins. In strains that also contain wild-type srpS4, this mutantprotein, as well as others designed to be locked in a GTP-bound conformation, exhibits temperature-dependentdominant inhibitory elects on growth, while a mutant predicted to be GDP locked does not interfere with thefunction of the wild-type protein. These results form the basis of a simple model for the role of GTP hydrolysisby Srp54 during the SRP cycle.

The signal recognition particle (SRP) serves as a biologicaladaptor between the cytoplasmic machinery for protein syn-thesis and the translocation apparatus of the endoplasmicreticulum (ER) (for reviews, see references 31, 59, 70, and 82).SRP recognizes ribosomes translating proteins destined forexport from the cytoplasm via the ER by virtue of theirhydrophobic signal sequences, usually located at the aminoterminus. Association of SRP with the ribosome transientlyhalts elongation of the nascent polypeptide at natural transla-tional pause sites (107), preventing synthesis of the completedprotein in the cytoplasm. Upon reaching the cytoplasmic faceof the ER, the SRP-ribosome-nascent chain complex firstinteracts with a heterodimeric membrane protein known as theSRP receptor (32, 42, 95) or docking protein (51). Followingrelease of the signal sequence by SRP, the particle is liberatedfrom its receptor and recycled to participate in another roundof targeting. Finally, a tight junction between the ribosome andprotein components of the ER is formed (23, 90), and trans-lation resumes concurrently with translocation of the presecre-tory protein across the membrane. Despite intensive biochem-ical investigation of SRP-mediated protein targeting for over adecade, the mechanistic details of the individual steps in theSRP cycle remain to be elucidated.Canine SRP, the first homolog identified, is a stable 11S

ribonucleoprotein composed of one ca. 300-nucleotide RNA

* Corresponding author. Present address: Case Western ReserveUniversity School of Medicine, Department of Molecular Biology andMicrobiology, 10900 Euclid Ave., Cleveland, OH 44106-4960. Phone:(216) 368-1876. Fax: (216) 368-1877.

t Present address: Department of Microbiology and Immunology,University of North Carolina, Chapel Hill, NC 27599.

molecule (103) and six polypeptides, designated Srp72, Srp68,Srp54, Srpl9, Srpl4, and Srp9 on the basis of their apparentmolecular weights (102). The sequences of cloned cDNAsencoding each protein component of mammalian SRP havenow been reported (11, 37, 46, 48, 74, 92, 93). Of these, by farthe most extensively characterized is the Srp54 protein, forwhich homologs in a variety of other organisms have beenidentified. In addition to the two mammalian cDNAs, SRP54genes in the bacteria Escherichia coli (references 11 and 74 andreferences therein), Bacillus subtilis (39), and Mycoplasmamycoides (80) and the yeasts Saccharomyces cerevisiae andSchizosaccharomyces pombe (3, 34) have been identified, andcDNAs for both the cytoplasmic and the chloroplast formsfrom a plant, Arabidopsis thaliana (30, 45), have been isolated.SRP RNA homologs in an even more diverse spectrum oforganisms, including representatives of the archaea and bac-teria (formerly designated archaebacteria and eubacteria, re-spectively [106]), as well as a variety of eukaryotes (reviewed inreferences 2 and 41), have been identified. Several recentreports demonstrate that secretary protein precursors accumu-late in vivo upon depleting components of the E. coli, B.subtilis, and S. cerevisiae particles (35, 39, 63, 66, 73).

Comparative sequence analysis, together with the results ofpartial proteolysis of the mammalian homolog, indicates thatthe Srp54 protein consists of two structurally distinct domains(75, 109). The carboxyl-terminal M domain, so named becauseit is rich in methionine (11), binds directly to SRP RNA (75,109) and can be photochemically cross-linked to a signalsequence (38, 47, 109). The amino-terminal region is desig-nated the G domain on the basis of the presence of the fourconsensus motifs that define the GTPase superfamily (11, 74),which includes small ras-related proteins, a variety of transla-

7839

7840 ALTHOFF ET AL.

tion factors, and the ot subunits of heterotrimeric G proteins(reviewed in references 13, 14, and 97). A vast body ofbiochemical and genetic evidence supports the idea that theseproteins employ a common mechanism to regulate a variety ofbiological processes (reviewed in reference 13). High-resolu-tion crystal structures reveal that although these motifs areshort and the amino acid sequences between them are gener-ally not conserved, they are embedded within very similartertiary structures even in highly divergent proteins. For exam-ple, the GTP-binding domains of ras, EF-Tu, and transducin-ax,which are only 16 to 18% identical in their primary sequences,are virtually superimposable (57, 98). The four conservedmotifs cradle the bound GTP molecule and serve to positioncritical residues for guanine ring and phosphate binding and1--y phosphate bond cleavage (57, 62, 69). The pattern ofamino acid sequence conservation in SrpS4 places it in abranch of the GTPase superfamily with few as yet discoveredmembers (reviewed in references 2 and 14), including suchdiverse proteins as dynamin, the mammalian homolog ofDrosophila shibire, which participates in receptor-mediatedendocytosis (36, 99); the yeast vacuolar protein sorting factorVpsl (77); and a B. subtilis protein required for flagellarfunction (17). Notably, the G domain of Srp54 shares extensivehomology with the G domain of the ot subunit of the SRPreceptor (SRot) even outside the GTPase consensus motifs,suggesting that they interact with one or more commonpartners during the SRP cycle (discussed further in reference2). The second subunit of the SRP receptor, SRO3, also containsGTPase consensus motifs, but these are more closely related tothose of classical G proteins involved in signaling pathways(unpublished data cited in reference 61).

Since it has been known for some time that GTP is requiredfor ER import (19) and GTPase consensus motifs are presentin at least three proteins that function in the SRP pathway, theevents that set the stage for protein translocation are likely tobe controlled by a cascade of GTP hydrolysis reactions. Datafrom in vitro reconstitution assays indicate that nonhydrolyz-able analogs of GTP stabilize the association of SRP with themembrane (20, 22), implying that GTP hydrolysis is requiredfor dissociation of SRP from its receptor or an event thatprecedes this step (see Discussion). Using an in vitro mem-brane repopulation assay, it has been shown that site-specificmutations in the GTPase consensus motifs of mammalian SRadisrupt its function (71). Although photoaffinity labeling invitro demonstrates that Srp54 and both subunits of the SRPreceptor are GTP-binding proteins (20), GTP hydrolysis bymammalian Srp54 could be detected only in the presence ofthe SRP receptor and SRP RNA (21, 53, 54, 66). In contrast,the M. mycoides Srp54 protein displays readily detectableGTPase activity in the absence of other factors (81). Theapparent discrepancy between kinetic parameters for the bac-terial and mammalian proteins remains unexplained.

Earlier work demonstrated that S. pombe Srp54 is containedwithin a ribonucleoprotein with a sedimentation coefficient,chromatographic properties, and subunit composition similarto those of the canine particle (15, 67, 86). In this report, wefirst show by disruption of the srp54 gene that the protein itencodes is essential for growth in S. pombe, in contrast to itsdispensability in S. cerevisiae (35). Second, the consequences ofrepressing Srp54 synthesis imply that the indispensability ofSrp54 can be attributed to its role in secretion. Finally, theearly suggestion that Srp54 is not an active GTPase (22) andsubsequent difficulties arising from the undetectable GTPaseactivity of the mammalian homolog in the absence of anotherprotein composed of two GTPases (54) prompted us to studythe role of this protein by using genetics. To this end, we have

analyzed the effects of a variety of amino acid substitutions,designed by analogy to Ras and other extensively studiedGTPases, to disrupt the function of each of the consensusmotifs individually. The severity of the phenotypic conse-quences of these mutations implies that GTP hydrolysis,probably by a mechanism similar to that of Ras, is required forSrp54 function in vivo. The pattern of dominance displayed byour mutants leads us to propose a specific model for the role ofGTP hydrolysis by Srp54 in the SRP cycle.

MATERIALS AND METHODS

Materials. Restriction enzymes were purchased from eitherBethesda Research Laboratories or New England Biolabs. T4DNA ligase and the Klenow fragment of DNA polymerase Iwere purchased from Bethesda Research Laboratories. TaqDNA polymerase was obtained from Perkin Elmer CetusCorporation. [oa-35S]dATP for DNA sequencing as well asmost mutagenesis reagents was supplied by Amersham Corp.[oL-32P]dCTP for uniformly labeling DNA was obtained fromICN Radiochemicals, Inc., and labeling was performed withthe Multiprime kit from Amersham Corp. Tran-35S-Label forin vivo labeling of proteins was purchased from ICN Radio-chemicals. Oligonucleotides used for sequencing and mutagen-esis were synthesized either at the University of Illinois Ge-netic Engineering Facility or by Operon Technologies, Inc.

Cloning and disruption of the srp54 gene. In general,recombinant DNA manipulations were performed by methodsdescribed by Sambrook et al. (79). The srp54 gene wasoriginally isolated on a plasmid carrying a 3-kb Sacl genomicfragment in which little S. pombe DNA flanked the codingsequence, particularly at its 3' end (34). Since extensive tractsof homology (preferably >1 kb) are generally required forefficient chromosomal replacements in S. pombe (78), it wasnecessary to isolate the gene on a larger fragment prior todisrupting it. To this end, 36 genome equivalents (36,000plaques) of a X-DASH library constructed in our laboratory(88) were screened with a DNA fragment corresponding to theM domain of the protein generated by PCR amplification withthe SP54M1 and SP54M2 primers (86) and labeled by usingthe Multiprime system. The probe hybridized to 39 plaques,suggesting that the gene is present in a single copy per genome.All three isolates chosen for further characterization hadoverlapping restriction maps. Positively hybridizing plaqueswere analyzed by restriction digestion followed by Southernblotting to locate the srp54 gene. Finally, a 5.8-kb EcoRIfragment containing the coding sequence was subcloned intopTZ18U to create pTZSP54, which was used for subsequentmanipulations.To create a null allele for disrupting the srp54 gene, the 3'

ca. 60% of the coding sequence and 193 bp of 3'-flankingsequence were deleted by excising a 1.1-kb NheI fragment andligating together the complementary ends. This plasmid, des-ignated pTZSP54-A, was linearized with NheI, the overhangingends were filled in with Klenow fragment, and the plasmid wasrecircularized with an XhoI linker. An XhoI-SalI fragmentderived from the plasmid YEp13 and containing the S. cerevi-siae LEU2 selectable marker, which complements fission yeastleul mutations (78), was then inserted at the XhoI site. Theresulting plasmid, designated pTZSP54::LEU2, was digestedwith BglII, and the 5.2-kb fragment containing the disruptedgene was gel purified. The fission yeast strain Sp629 (ade6-M210/ade6-M216 ura4-/ura4- leul-32/leul-32), which can bemaintained as a stable diploid through interallelic complemen-tation at the ade6 locus (1), was transformed with the linearfragment by the lithium acetate method. Screening for stable

MOL. CELL. BIOL.

FUNCTION OF THE FISSION YEAST Srp54 GTPase 7841

integrants, random spore analysis, and Southern blotting toconfirm that the correct gene replacement had occurred wereperformed essentially as described earlier (68). The diploidstrain heterozygous for gene disruption at the srp54 locus wasdesignated SpAS1.Plasmid construction and mutagenesis. To create the

pSP54-U plasmid to be used in complementation assays, theLEU2 marker in the S. pombe-E. coli shuttle vector pIRT2 (78)was first replaced with ura4 (7) to create pURT2, which isequivalent to pIRT5 (55). A 4.2-kb BglII fragment carrying thewild-type fission yeast srp54 gene from pTZSP54 was theninserted at the BamHI site to yield pSP54-U. Mutagenesis ofthe srp54 gene was carried out by using the Amershammutagenesis system essentially as previously described (44, 87).Single-stranded phagemid pSP54-U template DNA was pre-pared from E. coli NM522 by using M13K07 helper phage(Promega). Since pSP54-U is very large (11.7 kb) and the Ncilsite used for nicking the wild-type strand is distant from theposition of the mutations being made, the time for exonucleasedigestion of the parental strand was increased to 60 min. Thesequences of the oligonucleotides used for mutagenesis are asfollows: region G-1, 54Q11OX, 5' GTAGGT'l'A(G/A)(G/A)AGGCAGTGGTAAAACG 3' (creates Q11OG/E/IR/K); BoxI,5' CAGTGGT(A/G)GAACGACAACATG 3' (creates K114R/G); and T15N/S, 5' CAGTGGTAAAA(A/G)TACAACATGCTC 3'; region G-2, D138N, 5' GGTTGCTGCCAATAC'l'l'CCG 3'; region G-3, R194HIL, 5' GATACATCTGGTC(A/T)CATCAGCAGGAGC 3'; R194Q/E, 5' GATACATCTGGT(G/C)A(A/G)CATCAGCAGGAGC 3'; and BoxII, 5' CATCGTTAATACATCT7GG 3' (creates D19ON); and region G-4,7248X, 5' ATAATCA(A/G)(C/G)AAATTGGGC 3' (createsT248N/S/K/R). Transformants were screened initially by re-striction mapping and then by double-stranded sequencingwith a Sequenase version 2.0 kit from United States Biochemi-cals. The degenerate oligonucleotides used to make mutationsin each region were used to sequence the mutations in thenearest downstream region, except in the case of the G-1 motif,for which an upstream sequencing primer (54Seql, 5' GACCCTAAAGTTGATGC 3') was required.A plasmid in which the srpS4 coding sequence was inserted

between the nmtl promoter and the polyadenylation signal wasconstructed as follows. ABamHI site was first created at the 3'end of the gene in pSP54-U by using the oligonucleotide SrpS43' BamHI (5' GAAGACGTTAGGATCCT[CCCTCGTTA3') by site-specific mutagenesis followed by creation of an NdeIsite precisely at the ATG initiation codon by site-specificmutagenesis with SrpS4 S'-NdeI (5' TTAATCTGACCATATGGY'l'''G 3'). The NdeI-BamHI fragment was then excisedand subcloned into the same sites of pREP2, pREP42, andpREP82 (plasmid series marked with the ura4 gene [9]). Theresulting plasmids, designated pREP254, pREP4254, andpREP8254, were then transformed into a haploid strain carry-ing the chromosomal disruption allele (leu2::srpS4) covered bya wild-type allele on an ade6 plasmid (pSP54-A [see below]).After ca. five generations of growth in EMM2 liquid mediumcontaining adenine to allow the wild-type gene to be lost, red(Ade- Ura+) colonies were isolated on EMM2 plus minimaladenine (10 mg/liter) plates. We refer to this technique asreverse plasmid shuffle. The strains carrying the null allelemarked with LEU2 covered by ura4-bearing pREP(X)54 plas-mids were then maintained on minimal media lacking thia-mine.A plasmid for expressing the S. cerevisiae SRPS4 gene

(provided by B. Hann and P. Walter) in S. pombe wasconstructed by inserting the 2.3-kb SacI-BamHI fragment from

the plasmid pRS316 (35) into the pURT2 fission yeast shuttlevector described above.

In vivo 35S labeling and translocation assays. In vivolabeling of proteins was begun by growing 50-ml cultures ofpREP4254 and pREP8254 to an optical density at 600 nm(OD600) of 1.0 in EMM2 plus adenine at the normal growthtemperature (30'C). The cells were then subcultured into 500ml of EMM2 plus adenine with or without 5 mM thiamine(vitamin B1) at an OD600 of 0.03, and growth was monitored bymeasuring the OD600. At three time points, 5 OD units wereremoved to a sterile 50-ml disposable Corning tube andcollected by centrifugation at 5,000 X g for 3 min at theambient temperature. The cells were resuspended in 5 ml offresh media and pulse-labeled with 250 puCi of Tran-35S-Labelper ml (22.5 jud) for 10 min at 30'C with shaking. To terminatethe pulse, 5 ml of ice-cold 20 mM NaN3 was added to eachtube, which was then incubated on ice for 5 min. The labeledcells were collected by centrifugation as described above, andthe pellets were resuspended in 500 ,ul of trichloroacetic acid(TCA) lysis buffer (35) containing a cocktail of proteaseinhibitors (1 mM phenylmethylsulfonyl fluoride [PMSF], 1 pRgof pepstatin per ml, 300 pRg of leupeptin per ml, 1 ,ug ofaprotinin per ml), transferred to 1.5-ml conical bottom screw-cap tubes, and lysed by two 50-s pulses separated by 1 min onice in a Mini Bead-Beater (Biospec Products) with 1 ml ofprechilled 0.3-mm-diameter zirconium oxide beads (BiospecProducts) and 500 ,u of 20% TCA. The supernatant wascollected and combined with one 550-,I wash with a 1:1 mix ofice-cold 20% TCA and TCA lysis buffer (20 mM Tris HCl [pH7.9]), 50 mM ammonium acetate, 2 mM EDTA) containingprotease inhibitors. Proteins were precipitated by centrifuga-tion at 12,000 X g for 5 min at 4°C. The white pellet wasresuspended in 100 RI ofTCA resuspension buffer (3% sodiumdodecyl sulfate [SDS], 100 mM Tris base, 3 mM dithiothreitol)and boiled for 5 min.To assay for accumulation of secretary protein precursors,

denaturing immunoprecipitations were performed by adding50 ,u of the resuspended TCA pellet to 1 ml of NET-2 (50 mMTris HCl [pH 7.4], 150 mM NaCl, 0.05% Nonidet P-40)containing protease inhibitors on ice. The particulate materialwas removed by centrifugation at 12,000 X g for 5 min at 4°Cand then the supernatant was transferred to a clean microcen-trifuge tube. Next, either 5 ,u of rabbit polyclonal anti-BiP (64)or 300 [lI of mouse monoclonal antibody 7B4 directed againstS. pombe acid phosphatase (85) was added and allowed to bindon ice. After 30 min, 5 ,u of rabbit anti-mouse immunoglobulinG was added to the tubes containing the mouse monoclonalsera to enhance binding to protein A-Sepharose beads. Thereaction mixtures were incubated on ice for a total of 1 h, andthen 60 ,u of a 1:1 suspension of protein A-Sepharose beads(Pharmacia) in NET-2 containing protease inhibitors wasadded. After 2 h of gentle rocking at 4°C, the beads werewashed twice with ice-cold NET-2 containing protease inhibi-tors, twice with ice-cold NET-2 plus 2 M urea, and once withice-cold 1% P-mercaptoethanol, using brief, gentle vortexingto resuspend the pellet at each wash. The immune complexeswere resuspended in 50 pI of 2x SDS-polyacrylamide gelelectrophoresis (PAGE) sample buffer, boiled for 5 min, andseparated from the beads by brief centrifugation. A total of 15,u of each acid phosphatase immunoprecipitation was loadedonto a Mini-Protean SDS-10% polyacrylamide gel (Bio-Rad).A total of 25 pA of each sample immunoprecipitated withanti-BiP was loaded onto a 10% standard-size gel (Hoefer).Following electrophoresis at 100 V, gels were fixed and treatedwith En3Hance (Dupont), dried, and subjected to autoradiog-raphy for 3 weeks at -70°C.

VOL. 14, 1994

7842 ALTHOFF ET AL.

Phenotypic analysis. Media and transformation proceduresemployed in the following experiments are described byMoreno et al. (55) and Alfa et al. (1).

(i) Testing for complementation. To determine whetherplasmid-borne alleles could complement the gene disruption,we initially used random spore analysis as described previously(43, 68). After transforming the strain SpAS1, heterozygous forthe srp54::LEU2 disruption allele, with either pSP54-U orvector (pURT2) alone, the diploid was sporulated and theresulting asci were treated with glusulase to release the sporesprior to plating. Haploid colonies were gridded and scored forthe presence of nutritional markers by replica plating. Subse-quently, a simpler complementation test that yields compara-ble results was developed, taking advantage of the red-whiteassay to distinguish haploids and diploids that is available forstrains derived from Sp629 (1). In this assay, Ura' diploidtransformants are restreaked, and single colonies are pickedinto 3 ml of minimal medium (EMM2 [55]) and allowed togrow to saturation. S. pombe diploids form asci in minimalmedia at a fairly high frequency. The sporulating cultures arediluted 1:1,000 into sterile water, and 10 or 100 ,ul is platedonto EMM2 plus minimal adenine (10 mg/liter; sufficient forAde- colonies to survive yet continue to accumulate redpigment), with or without 200 mg of leucine per liter. If theura4-containing plasmid bears an allele that complements thedisruption, then haploids that are Ade- (red) Ura+ Leu+ willbe abundantly present on both plates. If the allele does notcomplement, then the plate with leucine will have many redLeu- Ura+ colonies, indicating that sporulation was efficient,while the plate lacking leucine will contain only a few coloniesarising from diploids that are Ade+ (white) Leu+ Ura+.Complementing haploids isolated in this way are then re-streaked before being gridded for phenotypic analysis toensure that they are not derived from a mixed colony.

(ii) Testing for dominance. To determine whether theproteins encoded by mutant alleles exhibit dominant interfer-ence with wild-type function in the diploid state, we examinedthe SpAS1 diploids transformed with mutagenized pSP54-Uplasmids. These cells, which carry a single copy of the wild-typesrp54 sequence in the chromosome and a mutant allele on themulticopy plasmid, were tested for growth at 37, 30, and 18°Cby streaking a single colony with a sterile toothpick onto threeplates of selective media, always using the last plate streaked asthe control (incubated at 30°C). In each case, multiple isolateswere analyzed to eliminate interpretation errors due to differ-ences in the number of cells streaked. The plates wereremoved from the incubator when large colonies were visibleon the plate streaked with an isogenic wild-type control strain.As a further control, SpAS1 cells transformed with the vectorlacking the srp54 gene were also examined for growth at eachtemperature.To test for dominance in haploid strains, it was first neces-

sary to make a version of the complementation plasmid inwhich the ura4 marker was replaced by a gene conferringadenine prototrophy. The first step in this procedure was tomutagenize the cloned ade6 gene (94) (kindly provided by P.Szankasi) with the following oligonucleotides: AdeAJ-ind, 5'CAAGCCTCAATGGTG 3' (eliminates the HindIII sitewithin the coding sequence);Ade3'Hind, 5' TTGAAGCTTAATTATCGG 3' (introduces a HindIII site in the 3' flankingsequence); and Ade5'Hind, 5' AACAAGCTlllTGGGC 3'(introduces a HindIII site in the 5' flanking sequence). The2.6-kb HindIII fragment carrying the ade6 gene was then usedto replace the LEU2 gene in the plasmid pIRT2 (78) toproduce pAD2. To create a more versatile plasmid, the BamHIsite in the ade6 coding sequence was eliminated by mutagen-

esis with the oligonucleotide AdeABam (5' CCTCCAAGGATACCTACAACCTG 3'), producing pAD3. The BglII fragmentcontaining the srp54 gene from pTZSP54 was then subclonedinto the polylinker BamHI site of pAD3 to create pSP54-A. Totest for dominant negative effects of mutants in the presence ofa wild-type gene at an equal copy number, haploid strainscontaining the srp54::LEU2 gene disruption in the chromo-some and the pSP54-A (wild-type) plasmid and each of themutant alleles on a multicopy plasmid derived from pSP54-Uor the pURT2 vector alone as a control were constructed.Phenotypic analysis was then carried out as described fordominance testing in diploids, selecting continuously for boththe Ade and Ura markers. To test for dominant negativeeffects on growth in a haploid containing an excess of themutant alleles, the strain CHP428 (ade6-M210 leul-32 his7-366ura4-d18 srp54 [5]) was transformed with pSP54-U derivativescontaining each altered gene. These were analyzed as for thestrains containing the two alleles at an equal gene dosage.

(iii) Microscopic examination. The change in cellular mor-phology observed upon temperature shift of logarithmic-phasecells was recorded by growing a 50-ml culture of either of thehaploid mutant strains T248N and R194L, as well as anisogenic wild-type strain, to an OD600 of ca. 0.5 at 30°C,collecting the cells by centrifugation, and resuspending them infresh medium preequilibrated at either 30 or 18°C. Thecultures were incubated for an additional 4 h at the permissiveor restrictive temperature, and an aliquot was placed on amicroscope slide and photographed on Kodak Ektachrome100 color slide film at a X400 magnification, using an Olympuslight microscope with a 35-mm camera attachment.

(iv) Determination of growth rates. Growth rates for haploidstrains harboring a mutant allele as their sole source of Srp54were obtained by monitoring the increase in OD600 in minimalmedium (EMM2 [55]) with the appropriate supplements, usinga Beckman DU-64 spectrophotometer. For experiments re-quiring a temperature shift, an overnight culture in EMM2 wassubcultured into 200 ml of fresh media and grown at 30 or37°C, and its OD600 was monitored until growth resumed. Cellsin mid-log phase were collected by centrifugation; resuspendedin fresh, precooled EMM2; and placed in a shaking water bathat 18°C, and growth was again monitored by measuring OD600.

(v) GTPase assays on immunoprecipitated proteins. Wild-type and mutant Srp54 proteins were assayed for GTPaseactivity by first preparing an SRP-enriched cell fraction asfollows. Cells were grown to an OD600 of ca. 1.0, and the samenumber of cell equivalents were harvested from each cultureby low-speed centrifugation. Each cell pellet was resuspendedin 500 ,u of high-salt lysis buffer A (50 mM Tris [pH 8.0], 5 mMmagnesium acetate, 500 mM potassium acetate, 20 mg ofaprotinin per ml, to which PMSF was freshly added to 0.5 mM)and transferred to a conical bottom 1.5-ml screw-cap cryogenictube ca. one-third filled with 0.3-,um-diameter acid-washedglass beads (Sigma) on ice. Lysis was performed at 4°C by four50-s pulses at medium speed in a chilled Mini Bead-Beater,with chilling on ice for 1 min between pulses. The supernatantwas removed and combined with 500 RI of buffer A used towash the beads. The 1 ml of cloudy supernatant was clarified bylayering over 500 pI of cold high-salt cushion buffer B (bufferA with 750 mM sucrose), loaded into a Beckman TLS55 rotor,and centrifuged at 12,000 rpm for 10 min. The top layer wasthen removed, layered over 500 pA of cold buffer B, andcentrifuged in a TLA100.2 rotor at 64,000 rpm in a BeckmanTL100 tabletop ultracentrifuge for 1 h at 4°C. The resultingsupernatant (SEC [87]), which is moderately enriched for SRP,was then used for native immunoprecipitation. A total of 100RI of swollen, equilibrated protein A-Sepharose beads was

MOL. CELL. BIOL.

FUNCTION OF THE FISSION YEAST Srp54 GTPase 7843

IS) N cn CO) LOCD CD 0 co q' C) CO

CD - a: - - - -qOllc = LO CC O ,

= 0L u

Ini z 1. ) zg)

ISenlIngZY zc

000DC) C)00') CD) cm (D CO LO C,

FC = cc- 0

Q LI) 0) UILUl cc LULLnco L jL

0 0 0

C)CO "It

= cc a0 0-a 0 C)

M aU LLJ

CO CO C Co

4 Cl C") -)

0 - 0=Q CL ;C ilk11 2 XU)

EUI2 a

= c:- 0cnc

B.

5300- A -. -588004200-

- 3957

To a TO

0D C' 0LO uL (D

= Cl C- 0 0-a Qin aL LU

FIG. 1. Disruption of the srp54 gene. (A) Restriction maps of the wild-type and disrupted alleles of srp54. Numbering of restriction sites isrelative to the ATG start codon. The positions complementary to the SP54M1 and SP54M2 primers used for PCR amplification to isolate the gene(86) are indicated on the map of the wild-type allele at the top. (B) Genomic Southern blot of the diploid SpAS1, which was transformed with thesrp54::LEU2 allele. The DNA was digested with the indicated enzyme, transferred to GeneScreen Plus, and probed with a gel-purified, 32P-labeled4.2-kb BglII fragment containing the coding sequence. The wild-type gene is carried on a 4.2-kb BglII band, while the modified allele yields a 5.3-kbBg!II band, the net result of replacing a 1.1-kb fragment with a 2.2-kb fragment. Digestion with EcoRI, which cuts within the LEU2 fragment,produces a 5.8-kb fragment corresponding to the wild-type gene and two fragments of 4.0 and 2.9 kb arising from the disrupted allele (seepanel A).

preloaded with either polyclonal antisera directed againstSrp54 or with preimmune sera as previously described (86).The washed beads were pelleted in a microcentrifuge, and thesupernatant was removed. A total of 1 ml of NET-2 containing20 mg of aprotinin per ml and 0.5 mM PMSF was added to 350RA of the SEC preparation from above and combined with theantibody-bead complexes in a microcentrifuge tube on ice.Precipitation was performed with gentle rocking at 40C for 1 h.The complexes of antigen, antibody, and protein A-Sepharosebeads were then carefully washed five times with ice-coldNET-2 and collected by centrifugation. The immune com-plexes were then evenly resuspended in 50 AI of GTPasereaction buffer (1 mM CHAPSO {3-[(3-cholamidopropyl)-dimethylammonio]-2-hydroxy-1-propanesulfonate}, 5 mg ofbovine serum albumin [BSA] per ml, 2 mM MgCl2, 2 mMdithiothreitol, 20 mM Tris HCl [pH 8.0]), and 25 AL wasremoved for each GTPase assay, which was carried out exactlyas previously described for the related GTPase Vpsl (100).Thin-layer chromatograms on which the GTP and GDP wereresolved were analyzed and the data were quantitated on aMolecular Dynamics Phosphorlmager, using Image Quantsoftware for Microsoft Windows version 3.1.

RESULTSThe Srp54 protein is essential for viability in S. pombe. A

gene encoding a polypeptide homologous to both mammalianSrp54 and the E. cohffl gene product was previously isolatedvia PCR amplification from S. pombe genomic DNA of a

fragment bounded by short stretches of sequence conservedbetween the human and bacterial proteins (34). The predictedfission yeast protein is a component of an SRP-like ribonucleo-protein particle (86), as is the product of an S. cerevisiae geneisolated by the same strategy (34, 35). Although SRP RNA wasshown several years ago to be required for viability in S. pombe(15, 72), budding yeast cells lacking components of the SRP

cycle, including Srp54, the scR1 RNA with which it is associ-ated, and the a-subunit of the SRP receptor, are viable (27, 35,61). To assess the generality of this apparent discrepancy, wedisrupted the S. pombe srp54 gene to determine whether theprotein it encodes is essential for growth. A cloned copycontaining extensive flanking DNA was isolated, and a nullallele was constructed and introduced into the chromosome ina diploid strain as described in Materials and Methods (Fig.1A). Of 37 transformants selected for further examination, 8had a stable Leu+ phenotype, indicating that the marker hadbeen integrated. To confirm that integration occurred at thesrp54 locus, we performed Southern blot analysis on genomicDNA from several of these diploids. Results for the straindesignated SpAS1, shown in Fig. 1B, indicate that it is het-erozygous for gene replacement at the srp54 locus. We con-sider it unlikely that a protein with even partial function isproduced from the disrupted allele, since the deleted sequence

begins 20 amino acids preceding the fourth GTPase consensusmotif, removes a putative GDS-binding site within the Gdomain (2), and eliminates the entire M domain.To assess the phenotypes of cells lacking a functional copy of

srpS4, random spore analysis was performed on SpAS1. Analy-sis of over 400 haploids derived from SpAS1 yielded no Leu+colonies; data from a representative experiment are shown inTable 1. Microscopic examination of the sporulation platesafter a 5-day incubation at 30'C revealed the presence ofmicrocolonies containing two to three cells. Since this is theterminal phenotype observed after tetrad dissection of spores

containing a disruption of the srp7 gene encoding fission yeastSRP RNA (15), we conclude that cells harboring a null alleleof srp54 are likewise able to divide only once or twice.Additional evidence that Srp54 is required for vegetativegrowth is provided by the observation that no colonies arosewhen haploid cells carrying the null allele complemented bypSP54-U were plated on media containing 5-fluoroorotic acid,

- 2943

VOL. 14, 1994

7844 ALTHOFF ET AL.

TABLE 1. Random spore and plasmid complementation analysis ofa strain heterozygous for the srp54::LEU2 allelea

Diploid strain No. of haploids No. of coloniessporulated analyzedb EMM+ALC EMM+Ad

SpAS1 38 0 0SpAS1/pSP54-U 76 72 38SpAsl/pURT2 43 43 0

a The diploid strain SpAS1 (srp54::LEU2Isrp54 ade6-M210/ade6-M216 ura4-iura4- leul-32/leul-32) was transformed with a plasmid containing the wild-typesrpS4 gene (pSP54-U) or the empty vector (pURT2). All three strains weresporulated and tested for growth on selective medium as described in Materialsand Methods and in the other footnotes.

b Spores were plated first on rich medium (YEA [55]) to allow growth of allviable spores, which are distinguished from residual diploids by their red color;these were tested for growth on minimal medium.

c Minimal medium (55) supplemented with adenine plus leucine, each at 100mg/liter. All cells harboring the plasmid (which carries a functional ura4 gene)will grow on these plates.

d Minimal medium supplemented with adenine. Cells must carry both thedisrupted allele and the plasmid to grow.

which is lethal to cells with a functional copy of the ura4 gene(89). Most definitively, we took advantage of the red-whiteassay (see Materials and Methods) for discriminating betweenhaploids and diploids in the Sp629 strain background to ruleout the possibility that Leu+ haploids are viable but grow veryslowly. When a sporulating culture of SpAS1 was spread at 104to 105 cells per plate on EMM2 plus uracil plus minimaladenine, no haploid (red) Leu+ colonies were observed evenafter more than 3 weeks of incubation at 30°C. The presence ofthousands of Leu- haploid colonies on a control plate con-taining leucine indicates that the strain sporulated efficiently.These data demonstrate that the Srp54 protein is essential forgrowth in S. pombe. Finally, we used genetic complementationto show that the inviability of haploids carrying the null alleleis solely a consequence of deleting a portion of the srp54 gene.The wild-type gene on the plasmid pSP54-U rescues sporescontaining the gene disruption (Table 1), while the vectoralone does not (Table 1). As expected, an approximately equalnumber of colonies carrying the wild-type and disrupted alleleswere observed upon sporulating the diploid harboring a plas-mid-borne copy of the srp54 gene.

Given that the Srp54 sequence is highly conserved throughevolution (reviewed in reference 2), we decided to test whetherthe gene encoding the S. cerevisiae homolog of this protein(provided by B. Hann and P. Walter) is able to complementour srp54::LEU2 gene disruption. A fragment harboring bud-ding yeast SRP54 was subcloned into a fission yeast shuttlevector (see Materials and Methods), and this construct wasintroduced into the SpAS1 diploid. Surprisingly, no viableLeu+ Ura+ cells were recovered upon sporulation. A Northern(RNA) blot confirms that the mRNA encoding S. cerevisiaeSrp54 is expressed in the diploid (data not shown), and otherdata (see below) indicate that the heterologous protein stablyaccumulates in S. pombe. Thus, even though the domainboundaries of the budding yeast and fission yeast proteinsclosely correspond and they exhibit extensive similarity inamino acid sequence (2, 4, 34), they are not functionallyinterchangeable.

Repressing Srp54 synthesis results in the accumulation ofsecretary protein precursors. To determine the effect ofgradual depletion of the Srp54 protein on growth and secre-tion, we took advantage of a recently developed regulatedexpression system for S. pombe. We were initially surprised toobserve that cells harboring srp54 cloned behind the nmtl (no

message on thiamine [50]) promoter continued to grow at arate indistinguishable from that of a wild-type strain even afterlong periods in medium supplemented with thiamine. How-ever, pulse-labeling experiments revealed that Srp54 synthesispersists in these cells (data not shown), and data from otherlaboratories confirm that this promoter produces residualmessage even under repressing conditions (9, 29). We there-fore adopted two weakened versions of the nmtl promoter, themost impaired of which produces no detectable mRNA underrepressing conditions (9). Figure 2A shows growth data in thepresence and absence of thiamine for these cells, in which thesole copy of the srp54 gene is under thiamine regulation. Thestrain containing the less active variant, pREP8254, shows asignificant growth defect, while the strain harboringpREP4254, which produces some mRNA, shows apparentlywild-type growth. After an extended period of growth onmedium containing thiamine, the pREP8254 strain resumesand sustains a wild-type rate of growth. The rapidity with whichthis occurs (data not shown) suggests that it is due to aphysiological adaptation to diminished SRP function, althoughwe cannot completely eliminate the possibility that it arisesthrough outgrowth of a subpopulation of cells in which thia-mine regulation has been bypassed.A critical issue is whether the reduced growth rate of the

pREP8254 cells is due to a secretion defect, as would beexpected on the basis of recent data for other organisms(discussed in detail in reference 2). To determine whether S.pombe cells depleted of Srp54 display a translocation defect,we used pulse-labeling followed by immunoprecipitation toexamine the fates of two ER-targeted proteins, acid phos-phatase, which is secreted, and BiP, which remains in thelumen of this organelle. Acid phosphatase exists in threepreviously characterized forms: a 50.5-kDa cytoplasmic precur-sor, a 70-kDa core-glycosylated form produced upon translo-cation into the ER, and a diffuse high-molecular-weight bandcorresponding to the mature, secreted form (85). Figure 2Bshows the results of immunoprecipitating extracts of the cellsfor which growth data are shown in Fig. 2A with antibodiesdirected against acid phosphatase. In the culture lackingthiamine, both core-glycosylated and mature acid phosphataseare produced throughout the time course. In contrast, accu-mulation of a band corresponding in size to the primarytranslation product of this protein (16) and a correspondingdiminution of the ER translocated forms occurs at Ti, 13.5 hafter repression of Srp54 sythesis. At T2, 17.5 h postshift, noneof the three forms of acid phosphatase is detectable. In analiquot of the culture removed at T3, after a wild-type rate ofgrowth had resumed, a faint band corresponding to core-glycosylated acid phosphatase is observed. Thus, the ability tosustain a rapid rate of growth correlates with restoration of thecell's ability to translocate pre-acid phosphatase across the ERmembrane.The other half of the pulse-labeled lysate from each time

point was immunoprecipitated with antibodies directed againstBiP, an ER-resident protein of the Hsp7O family, with theresults shown in Fig. 2C. S. pombe BiP exists in three previouslycharacterized forms: a major 70-kDa unglycosylated ER formlacking the signal sequence, a minor ca. 72-kDa ER formcarrying an asparagine-linked carbohydrate chain, and a ca.73-kDa precursor form that is difficult to resolve from theglycosylated ER species (64). The two higher-molecular-weight forms of BiP comigrated in the experiment for whichresults are shown in Fig. 2C. Thus, to distinguish between themature, glycosylated form of the protein and the cytoplasmicprecursor, we divided the immunoprecipitates and treated onehalf with endoglycosidase F (Endo F) which cleaves N-linked

MOL. CELL. BIOL.

FUNCTION OF THE FISSION YEAST Srp54 GTPase 7845

B.A.

Time (hr after thiamine addition)

C.

pREP8254 -B ITi T2

pREP8254 +B1T7S3 T I T2 T3

Mature [

Core -pGlvcosvlated

UnglycosylatedPrecursor

MW (kDa1--I

-200

-97.4

-68

-49.5

-29

1 2 3 4 5 6

- Endo F + Endo FpREP8254-B pREPX8254+B pREPI 54-B I pREP8254+B

'TI T2 T3 ' TI T2 T3 " TI T2 T3T--TI T2 T3 'MW (kDa)

-97.4PrecursorMature N-glyc

Mur-e_Mature -68

2 3 4 5 6 7 8 9 10 I I 12

FIG. 2. Assays of growth and protein secretion in strains expressing Srp54 under thiamine regulation. (A) Growth curves in the presence andabsence of thiamine (vitamin B1) in strains harboring plasmids in which the srp54 coding sequence is cloned behind two nmtl promoter variants(see Materials and Methods for details). 0, pREP4254; A, pREP4254 plus B1; O, pREP8254; O, pREP8254 plus B1. Thiamine was added to theindicated cultures at time zero, and growth was monitored over time by measuring OD600. T1 and T2 indicate the time points at which aliquotsof the cultures were removed for pulse-labeling. (B) Denaturing immunoprecipitation with monoclonal antibody directed against S. pombe acidphosphatase. At three time points after the addition of thiamine, cells were pulse-labeled with Tran-35S-Label as described in Materials andMethods. In addition to T1 and T2 indicated in panel A, a third aliquot of the culture was labeled at T3, after a wild-type rate of growth had beenreestablished in the pREP8254-containing cells propagated in medium containing thiamine (see text). Cells were harvested and disrupted, and halfof each lysate was subjected to denaturing immunoprecipitation with anti-acid phosphatase (see Materials and Methods for details). The labeledproteins were resolved by electrophoresis and visualized by autoradiography. (C) Denaturing immunoprecipitation with antibodies directed againstS. pombe BiP. The second aliquot of the pulse-labeled lysate from each time point was immunoprecipitated with antibodies directed against thisER-resident protein. The BiP immunoprecipitates were divided in two, and half of the material was treated with Endo F to remove N-linked sugars.

carbohydrate chains at the junction with the protein; the EndoF-resistant high-molecular-weight band represents the residualuntranslocated BiP precursor. At time points T1 and T2, aslightly higher fraction of the total radioactivity corresponds tothis species in the samples taken from cells grown in thepresence rather than in the absence of thiamine (Fig. 2C,compare lanes 11 and 8). The notion that the BiP precursoraccumulates to a minor degree at these time points is sup-ported by the results of another experiment (not shown) inwhich the primary translation product was resolved from thecore-glycosylated form. Nonetheless, the effect of Srp54 deple-tion on the translocation of BiP across the ER membraneappears to be minimal compared with the dramatic effectobserved for acid phosphatase (see Discussion). Importantly,radioactively labeled products were precipitable with anti-BiPat all time points, in contrast to acid phosphatase, which isvirtually undetectable a short time after the growth defect ismanifest. The lack of apparent BiP precursor accumulation atT3 (Fig. 2C, compare lanes 9 and 12), after wild-type growthhas resumed, correlates with the reappearance of core-glyco-sylated acid phosphatase at this time point.

Point mutations in the G domain are deleterious to thegrowth of haploid cells. (i) Design and analysis of mutations.As noted in the introduction, members of the GTPase super-

family contain four signature sequence motifs, designated G-1through G-4 (14), which position critical residues for nucle-otide binding and hydrolysis (57, 62, 69). The conservation oftopological features surrounding the bound GTP even inhighly divergent family members makes it reasonable to as-sume that the corresponding regions of proteins for whichX-ray crystallographic data are not yet available, such as Srp54,adopt similar tertiary structures. Consistent with this notion, asecondary structure prediction on the S. pombe Srp54 Gdomain sequence, using the PHD program (76), places theG-1, G-3, and G-4 motifs in the same secondary structurecontexts as in Ras and EF-Tu (49). Thus, to design mutationsin fission yeast Srp54 whose effects would be interpretable, weused other members of the GTPase superfamily in whichanalogous substitutions have been studied as a guide. Residueswere targeted within each of the four consensus motifs in orderto determine whether any or all of the known GTPase func-tions of guanine ring binding, A- and -y-phosphate binding, andGTP hydrolysis are functions of Srp54 in vivo. The 17 mutantsconstructed are shown in Fig. 2, together with an alignment ofboxes G-1 through G-4 from the S. pombe protein with thecorresponding regions of a representative set of diverse familymembers. Each allele was tested for the ability to function asthe sole copy in a haploid, using a complementation assay

VOL. 14, 1994

7846 ALTHOFF ET AL.

TABLE 2. Phenotypic analysis of Srp54 mutants in haploid strains'

Plasmid Complementation Phenotype

ControlspURT2 (negative) - LethalpSp54-U (positive) + Wild type

G-domain mutationsG-4 (guanine binding)pSp54-U T248S + Wild typepSp54-U T248K - LethalpSp54-U T248R - LethalpSp54-U T248N + Cold sensitive

G-3 (catalytic site)pSp54-U R194U - LethalpSp54-U R194L + Cold sensitivepSp54-U R194H + Cold sensitivepSp54-U R19ON - Lethal

G-2 (effector domain), - LethalpSp54-U D138N

G-1 (phosphate binding)pSp54-U T115N - LethalpSp54-U Tl15S + Wild typepSp54-U K114R - LethalpSp54-U K114G - LethalpSp54-U Ql1OG + Wild typepSp54-U QllOE + Wild typepSp54-U Q11OR - LethalpSp54-U QllOK - Lethal

Heterologous, pUSc54 - Lethala Mutants were tested for the ability to function in the absence of a wild-type

copy, using a complementation assay (see Materials and Methods for details).Briefly, a plasmid bearing each allele was transformed into the diploid strainSpAS1, which is heterozygous for gene disruption at the srp54 locus; this wasfollowed by sporulation and analysis of haploids for the relevant nutritionalmarkers.

G-1

30c

T248N

T248S

R194-l

) I(C; T248N

1 941. T248S

U'l 1 80

I I(G

1941,

R19411FIG. 4. Plate assay for cold sensitivity of Srp54 G domain mutants.

This photograph shows the growth of haploid S. pombe cells thatharbor a disrupted chromosomal copy of the srp54 gene and anextrachromosomal plasmid carrying either a wild-type (wt) or theindicated mutant allele. The cells were streaked onto selective medium(EMM2 [55]), and the plates were incubated at either 30 or 18°C.

followed by screening of viable strains for conditional growthdefects, with the results shown in Table 2. In the followingsections, the Srp54 GTPase mutants are grouped according totheir functional implications.

(ii) Point mutations designed to affect guanine ring binding.A hallmark of the subfamily of GTPases that includes Srp54and a few other proteins, represented in Fig. 3 by dynamin, isthe presence of threonine in place of asparagine at a conservedposition (248 in S. pombe Srp54) in the G-4 motif (reviewed inreferences 2 and 14). The amide functional group of E-116, thecorresponding residue in Ras, stabilizes the guanine bindingpocket through hydrogen bonds to the main chain at-aminogroup of V-14 in G-1 (98), and mutations at this positionweaken the affinity of the enzyme for nucleotide (reviewed inreference 65). Restoration of the GTPase superfamily consen-sus in Srp54 produces cold sensitivity (Table 2; Fig. 4) and, incommon with other conditional mutants described below, cellswith a T-to-N change at position 248 (T248N) display anaberrant morphology at the nonpermissive temperature. Theconservative substitution T248S supports wild-type growth ofhaploid cells, while T248K and T248R, which introduce bulky,

G-2 G-3 G-4GTPase consensusHs H/K/N rasHs Gza subunitE. coli EF-TuRat DynaminCanine SRaE. coli FtsYpHuman SRP54pE. coli FfhpS. pombe Srp54p

XOOOOGXXGXG KS5k LVVVGa gGv GKSal t 20k- LLLGtsnsGKStivnVGt I GhvdhGKtt I t

q I AVVGgq s aGKSsvIvVt FcGvnGvGKStnlv I LMVGvnGvGKt t tv I MFVG I qGsGKt t tc

D- (,"9n- T33D pTiedsy40DilrsrdmTt giveDnapeekargiTintDr vTgt nkgDTf ragaDTfraaaDTf raga

vVLMAG I qGaGKt t sv Dvyrpaa103v I MMVG I qGsGKt t tc118 138DTfraga144

G RN NE GSRK

OJOODXAGJ X53LDl LDtAGQe62FKMVDv g GQr

ts YahVDcpGHaLTLVDI pGmtVv LVDt AGRmVI IADtAGR II IVDtsGRhVI LVDtAGRI

186Vi IVDt sGRh195

N QHL

OOOON KXD112VLVGNKcD11

I LFLNKkDI VFLNKcDI GV 1 KI DGIVLU KfDGI t LR KIDsVMIV KI DGVVLI KvD

244AV I t2KID251

NSRK

FIG. 3. Comparison of GTPase consensus motifs and mutagenesis strategy. The consensus sequences of the G-1 through G-4 motifs (14) areshown at the top. X, any amino acid; J, a hydrophilic amino acid; 0, a hydrophobic amino acid. The next four lines show the corresponding regionsof proteins from diverse branches of the GTPase superfamily (taken from reference 14), followed by two SRa sequences and three Srp54sequences. Uppercase letters, amino acids that fit the consensus sequence; lowercase letters, amino acids that do not fit the consensus sequence;boldface letters, highly conserved residues; open type letters, residues characteristic of the SRP branch of the GTPase superfamily. Amino acidsubstitutions generated by oligonucleotide-directed mutagenesis (see Materials and Methods) are shown below the S. pombe Srp54 sequence.

MOL. CELL. BIOL.

FUNCTION OF THE FISSION YEAST Srp54 GTPase 7847

A. B.

FIG. 5. Effects of incubation at the nonpermissive temperature onthe morphology of an Srp54 G domain mutant. (A) Photograph of S.pombe cells harboring the T248N mutant allele, taken from a culturein the logarithmic phase of growth at 30'C immediately prior toshifting to the nonpermissive temperature. (B) Photograph of R194Lcells taken from a culture 4 h after shifting to 18'C. See Materials andMethods for details.

positively charged amino acids, fail to complement (Table 2).These observations indicate that correct recognition of theguanine ring is critical for Srp54 function in vivo, since grosschanges at position 248 are lethal and even a subtle structuralperturbation has deleterious consequences.

(iii) Point mutations designed to afect catalysis. WhileArg-194 in the G-3 motif of S. pombe Srp54, like Thr-248, isconserved within the SRP protein subfamily (reviewed inreference 2), Gln is found at the corresponding position of ras

and other small GTPases, and His is conserved among elon-gation factors (14). Structural studies imply a direct role forthis residue in catalysis (69), and the mutation Q61L in ras

produces a 90% drop in the hydrolysis rate without affectingGTP binding (96). The R194L and R194H substitutions infission yeast SrpS4 give rise to cold sensitivity (Table 2; Fig. 4),as did a Q-to-L change at the equivalent position of thedistantly related S. cerevisiae Sec4 protein, a small GTPase ofthe ras-related rab family required for the final stage ofsecretion (104). As illustrated in Fig. SB, cells harboring theR194L allele as their sole source of SrpS4 protein display a

highly aberrant morphology after prolonged incubation at thenonpermissive temperature: at 4 h postshift, the culture con-tains a significant fraction of enlarged, highly vacuolated cellsthat are frequently rounded and exhibit a marked tendency toaggregate. Even at the standard growth temperature, 30°C,some R194L cells appear rounded (data not shown). Remark-ably, despite the grossly altered morphology of this mutant andT248N at low temperature, their growth arrest is reversibleeven after a week at 18°C (data not shown). In contrast, whenthese strains are propagated at 30°C, the cultures containexclusively cells with an elongated, cylindrical appearance anda medially located septum (illustrated for T248N in Fig. SA);this is the typical morphology observed in a rapidly dividingfission yeast culture.A surprising result from this series of mutants is that

restoration of the ras consensus in the R194Q SrpS4 mutant islethal (Table 2), which may imply that the structure of theactive site or the catalytic mechanism differs slightly in the SRPsubfamily or, more likely, that SrpS4 has an intrinsically lowerhydrolysis rate than Ras. The lethality of mutations designed toreduce catalytic efficiency provides the first evidence that SrpS4hydrolyzes GTP during its functional cycle in vivo.

(iv) Other point mutations designed to produce GTP-lockedSrp54. X-ray crystallographic studies of Ras indicate that the

£-amino group of Lys-16 hydrogen bonds to the 3- and-y-phosphates of GTP, which has been proposed to stabilize thetransition state (69, 105). The effects of mutating several otherpositions in the G-1 and G-3 motifs of fission yeast Srp54 arecompatible with the idea that they also participate directly inGTP hydrolysis, since mutating the homologous position (K-114) in Srp54 to G or R is lethal (Table 2). Also located in G-1is the infamous Gly-12, the residue most frequently mutated inras-related human tumors (8). Although not strictly conserved(14) (Fig. 3), it is evident from the crystal structures of Ras andEF-Tu that only amino acids with small side chains can beaccommodated at this position. In S. pombe Srp54, restoringthe Ras consensus (QlOG) or substituting a small but nega-tively charged side chain (QilOE) confers no obvious growthdefect in haploid cells (Table 2; Fig. 4), while Ql1OR andQ1 10K, which are analogous to the most deleterious rasmutants, do not support growth.

(v) A point mutation designed to produce GDP-lockedSrp54. The final goal of our mutagenesis experiments was totest the in vivo effects of locking the protein in the GDP-boundstate. Ser-17 in Ras binds a Mg2" ion that coordinatesnonbridging oxygens of the f- and y-phosphates, and mutatingthis residue to Asn produces a protein with a much higheraffinity for GDP than GTP (26). A mutation equivalent toS17N in S. cerevisiae Cdc42 also produces functional defectsand, like the analogous ras allele, interferes with the functionof the wild-type protein (see below). In Srp54, the conservativesubstitution T115S supports wild-type growth, while T115N isinviable (Table 2), indicating that the ability to correctlyrecognize the terminal phosphates of GTP is critical for thefunction of Srp54 in vivo. The mutation D19ON, which affectsthe position in Srp54 homologous to a second Mg2" ligand inRas (14), is also lethal (Table 2).The R194L mutant is a less proficient GTPase than wild-

type Srp54. To confirm that a mutant predicted to be catalyt-ically defective on the basis of analogy to other GTPases doesindeed display impaired enzymatic activity, we assayed GTPhydrolysis by wild-type Srp54 and the R194L mutant immuno-precipitated under native conditions from an SRP-enrichedcell fraction. The results of a typical experiment with thewild-type protein, shown in Fig. 6A, reveal that significantGDP is produced by enzymatic hydrolysis compared with thecontrol containing buffer only even after 1 min. Figure 6Bshows a graph of data derived from quantitating the assaysshown in Fig. 6A, together with analogous data for the R194Lmutant. As a control for hydrolysis due to other GTPases,assays were performed on proteins precipitated from bothwild-type and R194L cells by preimmune sera. Both thewild-type and the mutant protein display activity which isparticularly evident over the relatively high background at the1-min time point. However, the R194L mutant does nothydrolyze nearly as much GTP in this interval as the wild-typeprotein, and its activity appears to level off by the 10-min timepoint. These differences cannot be attributed to the efficiencyof the immunoprecipitations, since nearly equal quantities ofprotein were recovered in each case (data not shown). In adifferent experiment (not shown), a mutant designed to affectguanine ring binding, T248N, was found to hydrolyze anamount of GTP equivalent to the wild-type protein throughoutthe interval measured.

Several alleles produce dominant inhibition of growth. Acommon observation in studies of a diverse array of GTPasesin vivo is that a subset of mutant alleles exhibit dominantinhibition of growth when expressed in conjunction with thewild-type protein (91, 108). We were therefore somewhatsurprised to obtain abundant, healthy diploid transformants

VOL. 14, 1994

7848 ALTHOFF ET AL.

A. B.60

*m %.

GDP --'

GTP --t-

A'.,-v"IV I

5()

H0

0}

Time (min)

it 10' 20' 30' 45' 1' 45'I1

*U- - anLi-Srp54p, wt

- - --- - anLi-Srp54p. R 1941L

- - 0 - pre-immune. At

O... p [[res- mrnunle.R1941.

anti-Srp54p Buffer .- ......H.suffer OrilsFIG. 6. Enzymatic assays of GTP hydrolysis by wild-type and mutant Srp54. (A) Thin-layer chromatogram illustrating GTP hydrolysis by

wild-type Srp54. Proteins were immunoprecipitated from an SRP-enriched cell fraction and incubated with oa-32P-labeled GTP as described inMaterials and Methods. Nucleotides were resolved by thin-layer chromatography and are displayed as a 600-dot-per-inch digitized print-out froma Molecular Dynamics Phosphorlmager. At the far right is a control in which buffer was added in place of immunoprecipitated protein. (B) Graphshowing the results of GTPase assays and control reactions for the wild-type protein and for the R194L mutant. The amount of each nucleotideresolved by thin-layer chromatography was quantitated by Phosphorlmager analysis. Two different exposures were taken when necessary tooptimize the quantitation of both unreacted GTP and the GDP product.

with all of our mutant alleles. However, the observation thathalf the normal dosage of wild-type srp54 in a diploid strain isdeleterious to growth at high or low temperatures (data notshown) prompted us to test whether the GTPase mutants inthe heterozygous state are growth impaired at 18 and 370C. Asdocumented in Table 3 (middle columns) and Fig. 7, incuba-tion at extreme temperatures of either haploid or diploidstrains containing the mutant allele on a high-copy-numberplasmid and a single copy of the wild-type gene in thechromosome resulted in several cases in either severely inhib-ited or fully wild-type growth. The vast majority of the mutantsshowed colonies of intermediate size, comparable to the vectorcontrol, and were not examined further. To determine whetheralleles that produced growth inhibition when present in excessover the wild type are truly dominant, we examined them inhaploid strains carrying either a disrupted allele (Table 3) or awild-type copy (Fig. 8) in the chromosome and both a wild-typeand a mutant gene on a plasmid. The same subset of mutantsdisplayed dominance when the wild-type and altered geneswere present at equal dosages, although the effects were lessdramatic in some cases than with the mutant in excess.Importantly, a pattern in which the alleles that displayeddominant inhibition of growth were those designed to produceSrpS4 locked in the GTP-bound state was readily discerned,while the sole lethal mutant that exerted no detectable inter-ference with wild-type function in the heterozygous state was

designed to exist predominantly in the GDP-bound form.Unexpectedly, budding yeast Srp54 also produced dominantinhibition of growth in S. pombe (Table 3), suggesting that itinteracts with one or more components of the fission yeast SRPcycle (see Discussion).

DISCUSSION

In the work reported here, we used a battery of genetic teststo demonstrate that the product of the srp54 gene is indispens-able for growth in S. pombe. Moreover, experiments in whichthis protein was expressed under the control of a regulatedpromoter indicate that this protein is involved in the targetingof proteins to the ER in this organism, as in others (reviewedin reference 2). Strikingly, one of the ER-targeted proteinsthat we examined, pre-acid phosphatase, appeared to bealmost completely dependent on SRP for entry into thesecretary pathway, since nearly all of the material synthesizedduring a brief pulse accumulated in the precursor band. Thispattern contrasts with the behavior of the ER resident proteinBiP, in which the majority of the radioactivity was found intranslocated forms of the protein, with only a minor accumu-lation of precursor. In both E. coli and S. cerevisiae, depletingSrp54 also affects different proteins to different degrees (3, 35,63); it is therefore likely that all three organisms possess amechanism for SRP-independent protein targeting. The essen-

MOL. CELL. BiOL.

.r_

a.

tCL.r)I -,

FUNCTION OF THE FISSION YEAST Srp54 GTPase 7849

TABLE 3. Phenotypic analysis of Srp54 mutants in combinationwith the wild-type protein'

DiploidPlasmid

180C 300C 37'C

ControlspURT2 (negative) + ++ +pSp54-U (positive) ++ + + + +

G-domain mutationsG-4 (guanine binding)pSp54-U T248S + + + + +pSp54-U T248K - + + +pSp54-U T248R - + + +pSp54-U T248N + + + +

G-3 (catalytic site)pSp54-U R194Q - + + -pSp54-U R194L - + + -pSp54-U R194H + + + -pSp54-U D19ON + ++ ++

G-2 (effector domain), + + + -pSp54-U D138N

G-1 (phosphate binding)pSp54-U T115N ++ ++ ++pSp54-U T115S ++ ++ ++pSp54-U K114R - ++ -pSp54-U K114G + ++ -pSp54-U Q11OG - ++ +pSp54-U Q11OE + + + + -pSp54-U Ql1OR + ++ +pSp54-U QllOK + ++ +

Heterologous, pUSc54 - + + -

a To test for dominant interference when mutant srpjover the wild-type allele, diploids that are heterozygous fisrp54 locus and harbor a mutant allele on a multicopy 1growth at various temperatures, using a plate assay (Fig.interference at equal gene dosages, a subset of the mutaassay on haploid strains harboring the altered allele on aand a wild-type allele on an ade6 plasmid with the same c8). As a control (not shown), each mutant allele was alsovector lacking the wild-type srp54 gene, with results idein haploid complementation assays (Table 2).

tial nature of both Srp54 and SRP RNA in Sdispensability of these components in S. certhat the SRP pathway is the most heavilyprotein export in this organism. However, ti

wt 300

Kl KI

Sc54 Sc54

KI 4L

FIG. 7. Plate assay of Srp54 G domain mutaithe presence of a lower gene dosage of the wild-the growth of haploid S. pombe cells that harbor v

the chromosome and the indicated G domain mplasmid. The cells were streaked onto selective mand the plates were incubated at either 30 or 18'

Haploid

180C 300C 370C

++ ++ ++

++ ++ ++

+ ++ ++

+ + _

TI 15N 114R T115N

,c54 R194LR194L

LJ I 9UN IJ1MA1

FIG. 8. Plate assay of Srp54 G domain mutants for dominantphenotypes in the presence of an equal dosage of the wild-type gene.Shown is the growth of haploid S. pombe cells that harbor a disruptedchromosomal srpS4 allele and two multicopy plasmids, one carryingwild-type (wt) srp54 and one carrying the indicated G domain mutant.The cells were streaked onto selective medium (EMM2 [55]), and theplates were incubated at either 30 or 18°C.

the S. cerevisiae homolog of BiP is profoundly affected by lossof the SRP pathway, while depleting Srp54 in S. pombe hadonly a minor affect on this protein, suggests that this view isoversimplified. For example, the indispensability of SRP in S.pombe (and E. coli) may be due to an absolute requirement for

++ ++ ++ this factor in the export of one or a few essential proteins.++ + + + + These observations underscore the emerging notion that there+ ++ - is a complex interplay between SRP-dependent translocation

and other mechanisms for maintaining proteins in an unfoldedstate for delivery to the ER, at least one of which is likely toinvolve members of the cytoplasmic Hsp7O family (18, 24).Moreover, a recent study demonstrates that depletion of Srp54in S. cerevisiae results in induction of both cytoplasmic heat

+ ++ shock proteins and BiP (6), consistent with our observationthat BiP continues to be synthesized at high levels in S. pombe

54 was present in excess even when acid phosphatase translation products are undetect-:or gene disruption at the able (Fig 2)plasmid were scored for (Fig7). To test for dominant This report also presents evidence derived from both in vivo

Lnts was tested in a plate and in vitro experiments that the fission yeast Srp54 protein ismulticopy ura4 plasmid an active GTPase that most likely hydrolyzes GTP by a

rigin of replication (Fig. mechanism analogous to that of Ras and other extensivelycotransformed with thentical to those obtained studied members of this enzyme superfamily. In cycling be-

tween a GTP-bound on state and a GDP-bound off state, theseproteins adopt dramatically different conformations (10, 52)that have unequal affinities for upstream and downstream

S. pombe versus the effectors. These properties allow GTPases to serve as a versa-evisiae might imply tile binary switch that comes with a built-in timer, since in mosttraversed route for cases their hydrolysis rates are extraordinarily low in compar-he observation that ison with those of other NTPases, except in the presence of a

GTPase-activating protein (GAP) (reviewed in references 12,14, and 65). Because the affinity of GTPases for nucleotides is

18' extremely high, a second accessory factor, most recentlyWt termed a guanine nucleotide dissociation stimulator (GDS), is

D138N required to complete the catalytic cycle by promoting displace-ment of GDP (reviewed in references 12, 14, and 65). Theprotein is then returned to the active state by the immediate

_ l acquisition of GTP, which is present at a higher intracellularconcentration than GDP (14). Well-characterized members of

1T 15N the GTPase superfamily provide two general paradigms for therole of the Srp54 GTPase (reviewed in reference 13). First, the

R194L lag preceding GTP hydrolysis following binding of the EF-Tu-nts for dominance in tRNA complex to the ribosome provides sufficient time for-type allele. Shown is incorrectly paired tRNAs to dissociate from the mRNA,vild-type (wt) srp54 in thereby serving as a kinetic proofreading mechanism. Byutant on a multicopy analogy, GTP hydrolysis by the Srp54 protein has been sug-edium (EMM2 [55]), gested to serve as a timer to improve the fidelity of signal°C. sequence recognition upon binding of the particle to the

300IO.'

180

VOL. 14, 1994

rlnI fx-

7850 ALTHOFF ET AL.

A. 44

GDP

GTP

Srp54pGDS

ER Lumen

FIG. 9. Models to account for the behavior of srp54 mutants. (A) The GTPase cycle of Srp54 superimposed on the known events of the SRPcycle. A ribosome from which the signal sequence of a nascent presecretory protein has just emerged is depicted at the top. For reasons describedin the text, we propose that SRP containing the GTP-bound form of Srp54 recognizes this complex and targets it to the membrane. Upon bindingof Srp54 to SRot on the cytoplasmic face of the ER, an Srp54 GAP (most likely the receptor itself; see the text) stimulates GTP hydrolysis by Srp54,resulting in transfer of the signal sequence to a component of the translocon. SRP containing Srp54 in its GDP-bound form is then released fromthe membrane, followed by a return to the GTP-bound state through the action of a GDS (possibly the Srp68 protein; see the text), thereby settingthe stage for another round of targeting. The dashed arrows at the bottom right of the figure indicate that both the number and the nature of theintervening steps are unknown. (B) Competition between SRP-dependent and SRP-independent pathways for translocation sites. The sequenceof events shown at the left is the same as in panel A. To explain the conditional dominant effects of GTP-locked Srp54 mutants (see the text), wepropose that SRP containing the defective proteins (bottom right, indicated by dark shading) impedes the progress of ribosomes complexed withSRP containing wild-type Srp54 (middle). The inability of these nascent chain-ribosome complexes to disengage from the translocation pore blocksthe access of secretary polypeptides that arrive at the membrane either via SRP-mediated targeting (route 1) or an alternative pathway involvinga heat shock protein or another molecular chaperone (route 2).

MOL. CELL. BIOL.

FUNCTION OF THE FISSION YEAST Srp54 GTPase 7851

ribosome (11, 74, 110). Second, the active, GTP-bound form ofthe a-subunit of a heterotrimeric G protein interacts withdownstream effectors to transduce a signal which is terminatedby hydrolysis of GTP. By analogy to these proteins, SrpS4could function as a component of a switching mechanism toensure unidirectionality of the targeting reaction at the mem-brane (11, 54).Our working model, shown in Fig. 9, is based on the

switching paradigm. Specifically, we believe that GTP hydro-lysis by Srp54 occurs at the ER membrane and serves toregulate transfer of the signal sequence to the translocationmachinery. This model was developed principally from thepatterns of dominance displayed by mutations in the fissionyeast Srp54 protein but also takes into account the results of invitro experiments with mammalian components. First, becausea mutant which was designed to bind preferentially to GDPdoes not interfere with the function of the wild-type protein,we propose that Srp54 GDP does not interact with theribosome-nascent chain complex and, as a consequence, can-not engage and block the function of the membrane-boundtranslocation machinery. On the other hand, we postulate thatthe dominant negative effects of mutations predicted to trapthe protein in the GTP-bound state are due to the formation ofdead-end complexes with the SRP receptor; this is consistentwith the observation that nonhydrolyzable GTP analogs stabi-lize the association of SRP with the ER membrane in vitro(22). At what point does GTP hydrolysis occur, i.e., when doesSrp54 encounter its GAP? We postulate that this function isserved by SRa, since in enzymatic assays with both mammalian(54) and bacterial (53) components, GTP hydrolysis by Srp54was stimulated in the presence of the receptor. Furthermore,our model stipulates that the conformational change in Srp54upon returning to the GDP-bound state results in release ofthe signal sequence to a component of the translocon, a likelycandidate being the Sec6l protein (33, 56, 83). The sequenceof events depicted in Fig. 9A contrasts with a model recentlydescribed by Miller et al. (54), in which Srp54 was proposed toengage the ribosome in an unprecedented nucleotide-freestate, with acquisition of GTP upon binding to the SRPreceptor triggering signal sequence release (see below).Although it is clear from biochemical experiments with

mammalian components that GTP hydrolysis by SRa is criticalfor ER translocation (71), no specific model for which step ofthe SRP cycle is regulated by this protein has been presented.We have used comparative sequence analysis in an effort to fillin this gap. Specifically, based on the conservation of a putativeGDS binding site between SRot and Srp54, together with thepresence in the Srp68 protein of sequence motifs found inguanine dissociation stimulators for small Ras-related GT-Pases, we recently hypothesized that this SRP componentserves as the GDS for SRa (2). To explain the order of eventsat the ER membrane, we propose further that Srp68 can gainaccess to SRa only after the signal sequence is released bySrp54 and that the conformational switch in SRa upon re-placement of GDP with GTP triggers release of SRP, therebyensuring that Srp54 cannot rebind the signal sequence andbegin a futile cycle. While even less information is availableconcerning the role of SRP, it seems reasonable to suggest thatthe G domain of this protein might serve as a switch to relayinformation about the state of the translocation apparatus(71), perhaps via stimulation of GTP hydrolysis by SRa. In Fig.9A, the speculative nature of the steps that occur after GTPhydrolysis by Srp54 but prior to the return of SRP containingGDP-bound Srp54 to the soluble pool is indicated by dashedarrows.To complete its catalytic cycle, Srp54 must exchange GDP

for GTP, presumably through the action of a GDS protein, inpreparation for another round of signal sequence binding. Inaddition to the comparative sequence analysis mentioned inthe last paragraph, our finding that GDP-locked Srp54 doesnot interfere with the function of the wild-type protein isconsistent with the idea that this factor is Srp68. Specifically,Srp54 - GDP displays recessive behavior, while mutations thattrap Ras and other small GTPases in their GDP-bound formsproduce dominant lethality as a result of sequestering therelevant exchange factors (25, 84, 108). This disparity suggeststhat the GDS for Srp54 is a component of signal recognitionparticle itself, since SRP proteins are tightly associated withthe RNA (40), while the small GTPases employ GDS proteinsthat are free to dissociate. In contrast to mutations predicted todiminish the rate of nucleotide exchange, a change in theregion of Srp54 corresponding to the site of GAP stimulationin other GTPases (reviewed in reference 65) produces domi-nant negative growth defects, suggesting by a similar argumentthat the latter factor is not part of SRP. This observation is alsocompatible with our model since, by analogy to Ras (101), themutant protein is expected to compete with wild-type Srp54 forbinding to SRa or another hydrolysis stimulator, yet beunresponsive to its action. The dominance displayed by the S.cerevisiae Srp54 protein may be a consequence of its ability tosequester SrpS4 GAP or another component of the fissionyeast SRP cycle.Why are some mutants detrimental in the presence of

wild-type SrpS4, while others are not? We propose that this isbecause proteins targeted to the ER by SRP compete withthose maintained in a translocation-competent state by othermechanisms for a limited number of protein-conducting chan-nels, as illustrated in Fig. 9B. Dominant negative effects uponmutating proteins that function as dimers or in higher-ordercomplexes are frequently observed, and a polysome translatinga presecretory protein can be thought of as a multimer ofsignal recognition particles joined by a common mRNA. IfSrp54 * GTP forms a dead-end complex with the SRP receptorin which the translation arrest cannot be released, the progressof ribosomes associated with wild-type SRP will also beblocked (Fig. 9B). Stated differently, mutant Srp54 anchorswild-type Srp54 to the ER membrane, where the mixedpolysome sequesters the translocation apparatus and preventsimport of proteins that arrive at the ER through either theSRP (Fig. 9B, route 1) or an alternative (Fig. 9B, route 2)pathway. In the absence of the wild-type protein, the mutantSrp54 sequesters only the SRP receptor, and chaperone-mediated translocation can proceed unimpeded. The net con-sequence is that the alternative pathway(s) for protein exportcan compensate for the effects of srpS4 GTPase mutants whenthey are present in the homozygous but not in the heterozy-gous state.Our data on the effects of srp54 mutants in vivo cannot be

reconciled with models proposed by earlier investigators forthe operation of the SRP cycle. First, Connolly et al. (22)suggested that SrpS4 is not an active GTPase but rather thatnucleotide hydrolysis by SRa. is responsible for release of thesignal sequence. The in vivo consequences of mutations in theGTPase consensus motifs clearly demonstrate their functionalimportance, and moreover, our proposal that a conformationalchange in Srp54 itself causes release of the signal sequenceseems more likely than the alternative view that this event istriggered by another protein. Second, Ogg et al. (60) proposedthat the GDP-bound form of Srp54 binds the signal sequenceand engages the receptor, which is incompatible with therecessive behavior of our SrpS4 mutants locked in this state.Third, Miller et al. (54) proposed that GTP hydrolysis by SrpS4

VOL. 14, 1994

7852 ALTHOFF ET AL.

triggers the release of SRP from its receptor, which is contra-dicted by our observation that catalytically defective proteinsare more detrimental in the heterozygous state; if SRP werereleased from the membrane only after Srp54 had disengagedfrom the signal sequence, then hydrolysis-defective mutantsshould not block access of proteins delivered to the protein-conducting channel by SRP-independent mechanisms (Fig.9B).Two additional considerations make our model even more

compelling. First, the sequence of events that we postulateduring the catalytic cycle of Srp54 (Fig. 9A) incorporates theidea that this protein functions in a manner analogous towell-characterized GTPases, in which the nucleotide-free formoccurs only transiently following GDP release stimulated by aGDS factor (reviewed in references 14 and 65); this contrastswith the proposal of Miller et al. (54) that the nucleotide-binding site is empty at the time of initial contact with thesignal sequence. Second, our model parallels the workingmodel for the rab family of small GTPases involved in vesiculartrafficking (reviewed in references 28 and 58). The commontheme is that a soluble pool of a GTP-bound targeting proteinbinds to its cargo in the cytoplasm and only upon arrival of thecomplex at the membrane surface does it come into contactwith the appropriate GAP. In both cases, interaction of thetargeting molecule with a distinct receptor on the membranesurface endows the targeting reaction with specificity, whileGTP hydrolysis at the membrane ensures efficient and unidi-rectional delivery of the cargo to its destination.

In summary, we have tested the effects of a variety ofsite-specific mutations in the GTPase consensus motifs of thefission yeast Srp54 protein. Most revealing was the pattern ofdominant inhibition of growth displayed by these mutantswhen expressed in combination with the wild-type protein,which led to a new working model for the role of GTPhydrolysis by Srp54 during the SRP cycle. While our data by nomeans provide definitive proof of this hypothesis, they areincompatible with earlier proposals. Because of the complexitypresented by three interacting GTPases, genetic analysis islikely to remain an extremely valuable tool to gain insight intothe function of the SRP cycle.

ACKNOWLEDGMENTS

We acknowledge the invaluable efforts of Pat Brennwald in the earlystages of this project and all members of the Wise laboratory, past andpresent, for helpful discussions. We thank Byron Hann and PeterWalter (University of California) and Philippe Szankasi (University ofBern) for providing plasmids containing the S. cerevisiae SRP54 and S.pombe ade6 genes, respectively; Roger VanHoy for constructingpAD2; and Chris Merli for converting pAD2 to pAD3. Antibodiesdirected against fission yeast BiP were kindly provided by AlisonPidoux and John Armstrong (ICRF), and anti-acid phosphatase wasgenerously supplied by Ernst Schweingruber. We are grateful to ourcolleagues Pat Brennwald, Enrique Martinez-Force, Peter Orlean,David Selinger, and Dane Wittrup for helpful discussions and forcritical reading of the manuscript.

This research was supported by NSF grants DCB 88-16325 and BCS92-13895.

REFERENCES1. Alfa, C., P. Fantes, J. Hyams, M. McLeod, and E. Warbrick. 1993.

Experiments with fission yeast: a laboratory course manual. ColdSpring Harbor Laboratory Press, Cold Spring Harbor, N.Y.

2. Althof, S., D. Selinger, and J. A. Wise. 1994. Molecular evolutionof SRP cycle components: functional implications. Nucleic AcidsRes. 22:1933-1947.

3. Amaya, Y., and A. Nakano. 1991. SRH protein, the yeast homologof the 54 kDa subunit of signal recognition particle, is involved inER translocation of secretary proteins. FEBS Lett. 282:325-328.

4. Amaya, Y., A. Nakano, K. Ito, and M. Mori. 1990. Isolation of ayeast gene, SRHJ, that encodes a homologue of the 54K subunitof mammalian signal recognition particle. J. Biochem. 107:457-463.

5. Apolinaro, E., M. Nocero, M. Jin, and C. S. Hoffman. 1994.Cloning and manipulation of the Schizosaccharomyces pombehis7' gene as a new selectable marker for molecular geneticstudies. Curr. Genet. 24:491-495.

6. Arnold, C., and K. D. Wittrup. The stress response to loss ofsignal recognition particle function in Saccharomyces cerevisiae. J.Biol. Chem., in press.

7. Bach, M.-L. 1987. Cloning and expression of the OMP decarbox-ylase gene ura4 from Schizosaccharomyces pombe. Curr. Genet.12:527-534.

8. Barbacid, M. ras genes. Annu. Rev. Biochem. 56:779-827.9. Basi, G., E. Schmid, and K Maundrell. 1993. TATA box

mutations in the Schizosaccharomyces pombe nmtl promoteraffect transcription efficiency but not the transcription start pointor thiamine repressibility. Gene 123:131-136.

10. Berchtold, H., L. Reshetnikova, C. 0. A. Reiser, N. K. Schirmer,M. Sprinzl, and R. Hilgenfeld. 1993. Crystal structure of activeelongation factor Tu reveals major domain rearrangements.Nature (London) 365:126-132.

11. Bernstein, H. D., M. A. Poritz, K. Strub, P. J. Hoben, S. Brenner,and P. Walter. 1989. Model for signal sequence recognition fromamino-acid sequence of 54k subunit of signal recognition particle.Nature (London) 340:482-486.

12. Boguski, M. S., and F. McCormick. 1993. Proteins regulating Rasand its relatives. Nature (London) 366:643-654.

13. Bourne, H. R., D. A. Sanders, and F. McCormick. 1990. TheGTPase superfamily: a conserved switch for diverse cell func-tions. Nature (London) 348:125-132.

14. Bourne, H. R., D. A. Sanders, and F. McCormick. 1991. TheGTPase superfamily: conserved structure and molecular mecha-nism. Nature (London) 349:117-127.

15. Brennwald, P., X. Liao, K. Holm, G. Porter, and J. A. Wise. 1988.Identification of an essential Schizosaccharomyces pombe RNAhomologous to the 7SL component of signal recognition particle.Mol. Cell. Biol. 8:1580-1590.

16. Brennwald, P., and J. A. Wise. 1994. A homologous cell-freesystem for studying protein translocation across the endoplasmicreticulum membrane in fission yeast. Yeast 10:159-172.

17. Carpenter, P. B., D. W. Hanlon, and G. W. Ordal. 1992. flhF, aBacillus subtilis flagellar gene that encodes a putative GTP-binding protein. Mol. Microbiol. 6:2705-2713.

18. Chirico, W. J., M. G. Waters, and G. Blobel. 1988. 70K heatshock-related proteins stimulate protein translocation into mi-crosomes. Nature (London) 332:805-810.

19. Connolly, T., and R. Gilmore. 1986. Formation of a functionalribosome-membrane junction during translocation requires theparticipation of a GTP-binding protein. J. Cell Biol. 103:2253-2261.

20. Connolly, T., and R. Gilmore. 1989. The signal recognitionparticle receptor mediates the GTP-dependent displacement ofSRP from the signal sequence of the nascent polypeptide. Cell57:599-610.

21. Connolly, T., and R. Gilmore. 1993. GTP hydrolysis by complexesof the signal recognition particle and the signal recognitionparticle receptor. J. Cell Biol. 123:799-807.

22. Connolly, T., P. J. Rapiejko, and R. Gilmore. 1991. Requirementof GTP hydrolysis for dissociation of the signal recognitionparticle from its receptor. Science 252:1171-1173.

23. Crowley, K. S., G. D. Reinhart, and A. E. Johnson. 1993. Thesignal sequence moves through a ribosomal tunnel into a noncy-toplasmic aqueous environment at the ER membrane early intranslocation. Cell 73:1101-1115.

24. Deshaies, R. J., B. D. Koch, M. Werner-Washburne, E. A. Craig,and R. Schekman. 1988. A subfamily of stress proteins facilitatestranslocation of secretary and mitochondrial precursor polypep-tides. Nature (London) 332:800-805.

25. Farnsworth, C. L., and L. A. Feig. 1991. Dominant inhibitorymutations in the Mg2e binding site of rasH prevent its activationby GTP. Mol. Cell. Biol. 11:4822-4829.

26. Feig, L. A., and G. M. Cooper. 1988. Relationship among guaninenucleotide exchange, GTP hydrolysis, and transforming potential

MOL. CELL. BIOL.

FUNCTION OF THE FISSION YEAST Srp54 GTPase 7853

of mutated ras proteins. Mol. Cell. Biol. 8:2472-2478.27. Felici, F., G. Cesareni, and J. M. X. Hughes. 1989. The most

abundant small cytoplasmic RNA of Saccharomyces cerevisiae hasan important function required for normal cell growth. Mol. Cell.Biol. 9:3260-3268.

28. Ferro-Novick, S., and P. Novick. 1993. The role of GTP-bindingproteins in transport along the exocytic pathway. Annu. Rev. CellBiol. 9:575-599.

29. Forsburg, S. 1993. Comparison of Schizosaccharomyces pombeexpression systems. Nucleic Acids Res. 21:2955-2956.

30. Franklin, A. E., and N. E. Hoffman. 1993. Characterization of achloroplast homologue of the 54-kDa subunit of the signalrecognition particle. J. Biol. Chem. 268:22175-22180.

31. Gilmore, R. 1993. Protein translocation across the endoplasmicreticulum: a tunnel with toll booths at entry and exit. Cell75:589-592.

32. Gilmore, R., P. Walter, and G. Blobel. 1982. Protein translocationacross the endoplasmic reticulum.I. Detection in the microsomalmembrane of a receptor for the signal recognition particle. J. CellBiol. 95:463-469.

33. Gdrlich, D*, S. Prehn, E. Hartmann, K.-U. Kalies, and T. A.Rapoport. 1992. A mammalian homolog of Sec6lp and secYp isassociated with ribosomes and nascent polypeptides during trans-location. Cell 71:489-503.

34. Hann, B. C., M. A. Poritz, and P. Walter. 1989. Saccharomycescerevisiae and Schizosaccharomyces pombe contain a homolog ofthe 54-kD subunit of the signal recognition particle that in S.cerevisiae is essential for growth. J. Cell Biol. 109:3223-3230.

35. Hann, B. C., and P. Walter. 1991. The signal recognition particlein S. cerevisiae. Cell 67:131-144.

36. Herskovits, J. S., C. Burgess, R. A. Obar, and R. B. Vallee. 1993.Effects of mutant rat dynamin on endocytosis. J. Cell Biol.122:565-578.

37. Herz, J., N. Flint, K. Stanley,R Frank, and B. Dobberstein. 1990.The 68 kDa protein of signal recognition particle contains aglycine-rich region also found in RNA-binding proteins. FEBSLett. 276:103-107.

38. High, S., and B. Dobberstein. 1991. The signal sequence interactswith the methionine-rich domain of the 54-kD protein of signalrecognition particle. J. Cell Biol. 113:229-233.

39. Honda, K., K. Nakamura, M. Nishiguchi, and K. Yamane. 1993.Cloning and characterization of a Bacillus subtilis gene encodinga homolog of the 54-kilodalton subunit of mammalian signalrecognition particle and Escherichia coli Ffh. J. Bacteriol. 175:4885-4894.

40. Janiak, F., P. Walter, and A. E. Johnson. 1992. Fluorescence-detected assembly of the signal recognition particle: binding ofthe two SRP protein heterodimers to SRP RNA is noncoopera-tive. Biochemistry 31:5830-5840.

41. Larsen, N., and C. Zwieb. 1991. SRP-RNA sequence alignmentand secondary structure. Nucleic Acids Res. 19:209-215.

42. Lauffer, L., P. D. Garcia, R N. Harkins, L. Coussens, A. Ullrich,and P. Walter. 1986. Topology of signal recognition particlereceptor in endoplasmic reticulum membrane. Nature (London)318:334-338.

43. Liao, X., P. Brennwald, and J. A. Wise. 1989. Genetic analysis ofSchizosaccharomyces pombe 7SL RNA: a structural motif thatincludes a conserved tetranucleotide loop is important for func-tion. Proc. Natl. Acad. Sci. USA 86:4137-4142.

44. Liao, X., D. Selinger, S. Althoff, A. Chiang, D. Hamilton, M. Ma,and J. A. Wise. 1992. Random mutagenesis of Schizosaccharomycespombe SRP RNA: lethal and conditional lesions cluster in presump-tive protein binding sites. Nucleic Acids Res. 20:1607-1615.

45. Lindstrom, J. T., B. Chu, and F. C. Belanger. 1994. Isolation andcharacterization of an Arapidopsis thaliana gene for the 54 kDsubunit of the signal recognition particle. Plant Mol. Biol.23:1265-1272.

46. Lingelbach, K., C. Zweib, J. R Webb, C. Marshallsay, P. J.Hoben, P. Walter, and B. Dobberstein. 1988. Isolation andcharacterization of a cDNA clone encoding the 19 kDa protein ofsignal recognition particle (SRP): expression and binding to 7SLRNA. Nucleic Acids Res. 16:9431-9442.

47. Lutcke, H., S. High, K. Romisch, A. J. Ashford, and B. Dobber-

stein. 1992. The methionine-rich domain of the 54 kDa subunit ofsignal recognition particle is sufficient for the interaction withsignal sequences. EMBO J. 11:1543-1551.

48. Lutcke, H., S. Prehn, A. J. Ashford, M. Remus,R Frank, and B.Dobberstein. 1993. Assembly of the 68- and 72-kD proteins ofsignal recognition particle with 7S RNA. J. Cell Biol. 121:977-985.

49. Martinez-Force, E., and J. A. Wise. Unpublished data.50. Maundrell, K. 1990. nmtl of fission yeast: a highly expressed gene

completely repressed by thiamine. J. Biol. Chem. 265:10857-10864.

51. Meyer, D. I., E. Krause, and B. Dobberstein. 1982. Secretoryprotein translocation across the endoplasmic reticulum mem-brane: the role of the 'docking protein'. Nature (London) 297:647-650.

52. Milburn, M. V., L. Tong, A. M. deVos, A. Brunger, Z. Yamaizumi,S. Nishimura, and S.-H. Kim. 1990. Molecular switch for signaltransduction: structural differences between active and inactiveforms of protooncogenic ras proteins. Science 247:939-945.

53. Miller, J. D., H. D. Bernstein, and P. Walter. 1994. Interaction ofE. coli Ffh/4.5S ribonucleoprotein and FtsY mimics that ofmammalian signal recognition particle and its receptor. Nature(London) 367:657-659.

54. Miller, J. D., H. Wilhelm, L. Gierasch,R Gilmore, and P. Walter.1993. GTP binding and hydrolysis by the signal recognitionparticle during initiation of protein translocation. Nature (Lon-don) 366:351-354.

55. Moreno, S., A. Klar, and P. Nurse. 1991. Molecular geneticanalysis of fission yeast Schizosaccharomyces pombe. MethodsEnzymol. 194:795-823.

56. Musch, A., M. Wiedmann, and T. A. Rapoport. 1992. Yeast secproteins interact with polypeptides traversing the endoplasmicreticulum membrane. Cell 69:343-352.

57. Noel, J. P., H. E. Hamm, and P. B. Sigler. 1993. The 2.2 A crystalstructure of transducin-o complexed with GTPys. Nature (Lon-don) 366:654-663.

58. Novick, P., and P. Brennwald. 1993. Friends and family: the roleof the rab GTPases in vesicular traffic. Cell 75:597-601.

59. Nunnari, J., and P. Walter. 1992. Protein targeting to andtranslocation across the membrane of the endoplasmic reticulum.Curr. Opin. Cell Biol. 4:573-580.

60. Ogg, S. C., J. Nunnari, J. Miller, and P. Walter. 1992. The role ofGTP in protein targeting to the ER, p. 129-136. In W. Neupertand R. Lill (ed.), Membrane biogenesis. Elsevier Science Pub-lishers, New York.

61. Ogg, S. C., M. A. Poritz, and P. Walter. 1992. Signal recognitionparticle receptor is important for cell growth and protein secre-tion in Saccharomyces cerevisiae. Mol. Biol. Cell 3:895-911.

62. Pai, E. F., U. Krengel, G. A. Petsko, R S. Koody, W. Kabsch, andA. Wittinghofer. 1990. Refined crystal structure of the triphos-phate conformation of H-ras p-21 at 1.35 A resolution: implica-tions for the mechanism of GTP hydrolysis. EMBO J. 9:2351-2359.

63. Phillips, G. J., and T. J. Silhavy. 1992. The E. coli ffh gene isnecessary for viability and efficient protein export. Nature (Lon-don) 359:744-746.

64. Pidoux, A., and J. Armstrong. 1992. Analysis of the BiP gene andidentification of an ER retention signal in Schizosaccharomycespombe. EMBO J. 11:1583-1591.

65. Polakis, P., and F. McCormick. 1993. Structural requirements forthe interaction of p2lrnS with GAP, exchange factors, and itsbiological effector target. J. Biol. Chem. 268:9157-9160.

66. Poritz, M. A., H. D. Bernstein, K. Strub, D. Zopf, H. Wilhelm,and P. Walter. 1990. AnE. coli ribonucleoprotein containing 4.5SRNA resembles mammalian signal recognition particle. Science250:1111-1117.

67. Poritz, M. A., V. Siegel, W. B. Hansen, and P. Walter. 1988. Smallribonucleoproteins in Schizosaccharomyces pombe and Yarrowialipolytica homologous to signal recognition particle. Proc. Nati.Acad. Sci. USA 85:4315-4319.

68. Porter, G., P. Brennwald, and J. A. Wise. 1990. Ut small nuclearRNA from Schizosaccharomyces pombe has unique and con-served features and is encoded by an essential single-copy gene.Mol. Cell. Biol. 10:2874-2881.

69. Prive, G. G., M. V. Milburn, L. Tong, A. M. deVos, Z. Yamaizumi,

VOL. 14, 1994

7854 ALTHOFF ET AL.

S. Nishimura, and S.-H. Kim. 1992. X-ray crystal structures oftransforming p2lras mutants suggest a transition-state stabiliza-tion mechanism for GTP hydrolysis. Proc. Natl. Acad. Sci. USA89:3649-3653.

70. Rapaport, T. A. 1992. Transport of proteins across the endoplas-mic reticulum membrane. Science 258:931-936.

71. Rapiejko, P. J., and R. Gilmore. 1992. Protein translocationacross the endoplasmic reticulum requires a functional GTPbinding site in the signal recognition particle receptor. J. CellBiol. 117:493-503.

72. Ribes, V., P. Dehoux, and D. Tollervey. 1988. 7SL RNA fromSchizosaccharomyces pombe is encoded by a single copy essentialgene. EMBO J. 7:231-237.

73. Ribes, V., K. Romisch, A. Giner, B. Dobberstein, and D. Toll-ervey. 1990. E. coli 4.5S RNA is part of a ribonucleoproteinparticle that has properties related to signal recognition particle.Cell 63:591-600.

74. Romisch, K., J. Webb, J. Herz, S. Prehn, R. Frank, M. Vingron,and B. Dobberstein. 1989. Homology of the 54K protein of signalrecognition particle, docking protein, and two E. coli proteinswith putative GTP-binding-domains. Nature (London) 340:478-482.

75. Romisch, K., J. Webb, K. Lingelbach, H. Gausepohl, and B.Dobberstein. 1990. The 54-kD protein of signal recognitionparticle contains a methionine-rich RNA binding domain. J. CellBiol. 111:1793-1802.

76. Rost, B., and C. Sander. 1993. Prediction of protein structure atbetter than 70% accuracy. J. Mol. Biol. 232:584-599.

77. Rothman, J. H., C. K. Raymond, T. Gilbert, P. O'Hara, and T.Stevens. 1990. A putative GTP binding protein homologous tointerferon-inducible Mx proteins performs an essential functionin yeast protein sorting. Cell 61:1063-1074.

78. Russell, P. 1990. Gene cloning and expression in fission yeast, p.244-272. In A. Nasim, P. Young, and B. Johnson (ed.), Molecularbiology of fission yeast. Academic Press, Inc., San Diego, Calif.

79. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecularcloning: a laboratory manual, 2nd ed. Cold Spring HarborLaboratory Press, Cold Spring Harbor, N.Y.

80. Samuelsson, T. 1992. A Mycoplasma protein homologous tomammalian SRP54 recognizes a highly conserved domain of SRPRNA. Nucleic Acids Res. 20:5763-5770.

81. Samuelsson, T., and M. Olsson. 1993. GTPase activity of abacterial SRP-like complex. Nucleic Acids Res. 21:847-853.

82. Sanders, S. L., and R. Schekman. 1992. Polypeptide translocationacross the endoplasmic reticulum membrane. J. Biol. Chem.267:13791-13794.

83. Sanders, S. L., K. M. Whitfield, J. P. Vogel, M. D. Rose, and R. W.Schekman. 1992. Sec6lp and BiP directly facilitate polypeptidetranslocation into the ER. Cell 69:353-365.

84. Schweighoffer, F., H. Cai, M. C. Chevallier-Multon, I. Fath, G.Cooper, and B. Tocque. 1993. The Saccharomyces cerevisiaeSCD25 C-domain gene product overcomes the dominant inhibi-tory activity of Ha-Ras Asn-17. Mol. Cell. Biol. 13:39-43.

85. Schweingruber, A. M., F. Schoenholzer, L. Keller, R. Schwan-inger, H. Trachsel, and M. E. Schweingruber. 1986. Glycosyla-tion and secretion of acid phosphatase in Schizosaccharomycespombe. Eur. J. Biochem. 158:133-140.

86. Selinger, D., P. Brennwald, S. Althoff, C. Reich, B. Hann, P.Walter, and J. A. Wise. 1994. Genetic and biochemical analysis ofthe fission yeast ribonucleoprotein particle containing a homologof Srp54p. Nucleic Acids Res. 22:2557-2567.

87. Selinger, D., P. Brennwald, and J. A. Wise. 1993. Identification ofRNA sequences and structural elements required for assembly offission yeast SRP54 protein with signal recognition particle RNA.Mol. Cell. Biol. 13:1353-1362.

88. Selinger, D. A., G. L. Porter, P. J. Brennwald, and J. A. Wise.1992. The two similarly expressed genes encoding U3 snRNA inSchizosaccharomycespombe lack introns. Mol. Biol. Evol. 9:297-308.

89. Sikorski, R. S., and J. D. Boeke. 1991. In vitro mutagenesis andplasmid shuffling: from cloned genes to mutant yeast. MethodsEnzymol. 194:302-319.

90. Simon, S. M., and G. Blobel. 1991. A protein-conducting channelin the endoplasmic reticulum. Cell 65:371-380.

91. Stacey, D. W., L. A. Feig, and J. B. Gibbs. 1991. Dominant

inhibitory ras mutants selectively inhibit the activity of eithercellular or oncogenic ras. Mol. Cell. Biol. 11:4053-4064.

92. Strub, K., and P. Walter. 1989. Cross-hybridization of differentlyprimed polymerase chain reactions: isolation of a cDNA clone ofthe 14 kDa subunit of the signal recognition particle. Proc. Natl.Acad. Sci. USA 86:9747-9751.

93. Strub, K., and P. Walter. 1990. Assembly of the Alu domain ofthe signal recognition particle (SRP): dimerization of the twoprotein components is required for efficient binding to SRPRNA. Mol. Cell. Biol. 10:777-784.

94. Szankasi, P., W.-D. Heyer, P. Schuchert, and J. Kohli. 1988.DNA sequence analysis of the ade6 gene of Schizosaccharomycespombe. Wild-type and mutant alleles including the recombinationhot-spot allele ade6-M26. J. Mol. Biol. 204:917-925.

95. Tajima, S., L. Lauffer, V. L. Rath, and P. Walter. 1986. The signalrecognition particle receptor is a complex that contains twodistinct polypeptide chains. J. Cell Biol. 103:1167-1178.

96. Temeles, G. L., J. B. Gibbs, J. S. D'Alonzo, I. S. Sigal, and E. M.Scolnick. 1985. Yeast and mammalian ras proteins have con-served biochemical properties. Nature (London) 313:700-703.

97. Valencia, A., P. Chardin, A. Wittinghoffer, and C. Sander. 1991.The ras protein family: evolutionary tree and role of conservedamino acids. Biochemistry 30:4637-4648.

98. Valencia, A., M. Kjeldgaard, E. F. Pai, and C. Sander. 1991.GTPase domains of ras p21 oncogene protein and elongationfactor Tu: analysis of three-dimensional structures, sequencefamilies, and functional sites. Proc. Natl. Acad. Sci. USA 88:5443-5447.

99. van der Blieck, A. M., T. E. Redelmeyer, H. Damke, E. J. Tisdale,E. M. Meyerowitz, and S. L. Schmid. 1993. Mutations in humandynamin block an intermediate stage in coated vesicle formation.J. Cell Biol. 122:553-563.

100. Vater, C. A., C. K. Raymond, K. Ekena, I. Howald-Stevenson, andT. H. Stevens. 1992. The VPS1 protein, a homolog of dynaminrequired for vacuolar protein sorting in Saccharomyces cerevisiae.J. Cell Biol. 119:773-786.

101. Vogel, U. S., R. A. Dixon, M. D. Schaber, R. E. Diehl, M. S.Marshall, E. M. Scolnick, I. S. Sigal, and J. B. Gibbs. 1988.Cloning of bovine GAP and its interaction with oncogenic rasp21. Nature (London) 335:90-93.

102. Walter, P., and G. Blobel. 1980. Purification of a membrane-associated protein complex required for protein translocationacross the endoplasmic reticulum. Proc. Natl. Acad. Sci. USA77:7112-7116.

103. Walter, P., and G. Blobel. 1982. Signal recognition particlecontains a 7S RNA essential for protein translocation across theendoplasmic reticulum. Nature (London) 299:691-698.

104. Walworth, N., P. Brennwald, A. Kabcenell, M. Garrett, and P.Novick. 1992. Hydrolysis of GTP by Sec4 protein plays animportant role in vesicular transport and is stimulated by aGTPase-activating protein in Saccharomyces cerevisiae. Mol. Cell.Biol. 12:2017-2028.

105. Wittinghofer, A., and E. F. Pai. 1991. The structure of Rasprotein: a model for a universal molecular switch. Trends Bio-chem. Sci. 16:382-386.

106. Woese, C. R., 0. Kandler, and M. L. Wheelis. 1990. Towards anatural system of organisms: proposal for the domains Archaea,Bacteria, and Eucarya. Proc. Natl. Acad. Sci. USA 87:4576-4579.

107. Wolin, S., and P. Walter. 1988. Signal recognition particlemediates a transient elongation arrest of preprolactin in reticu-locyte lysate. J. Cell Biol. 109:2617-2622.

108. Ziman, M., J. M. O'Brien, L. A. Ouellette, W. R. Church, andD. L. Johnson. 1991. Mutational analysis of CDC42Sc: a putativeGTP-binding protein involved in the control of cell polarity. Mol.Cell. Biol. 11:3537-3544.

109. Zopf, D., H. D. Bernstein, A. E. Johnson, and P. Walter. 1990.The methionine-rich domain of the 54 kd protein subunit of thesignal recognition particle contains an RNA binding site and canbe crosslinked to a signal sequence. EMBO J. 9:4511-4518.

110. Zopf, D., H. D. Bernstein, and P. Walter. 1993. GTPase domainof the mammalian signal recognition particle is required forprotein translocation but not for signal sequence binding. J. CellBiol. 120:1113-1121.

MOL. CELL. BIOL.