Molecular Characterization of VACl, a Gene Required for Vacuole Inheritance and Vacuole Protein Sorting*

Cell division requires an accurate partitioning of cytoplasmic organelles. The segregation of vacuoles in the budding yeast Saccharomyces cereuisaie occurs at a specific time in the cell cycle and is spatially targeted to the small bud. Several yeast uac mutants have been isolated which are defective in this process. We have now cloned the VACl gene, corresponding to the first of these mutants, vacl-1. This gene encodes a protein of 616 amino acids, without homolog in the current data bases. It contains neither long hydrophobic stretches nor a classical leader peptide. The most no- table aspect of the sequence is the presence of three zinc fingers. Yeast in which the VACl gene has been entirely deleted are viable. However, they grow more slowly than wild-type cells and only form microcolonies when grown on glycerol at 37 “C. These yeast are defective in vacuole segregation at both the permissive and nonpermissive temperatures. The uacl mutant was previously shown to mislocalize carboxypeptidase Y to the cell surface, suggesting that Vaclp is involved in more than one vesicular traffic pathway.

Cell division requires an accurate partitioning of cytoplasmic organelles. The segregation of vacuoles in the budding yeast Saccharomyces cereuisaie occurs at a specific time in the cell cycle and is spatially targeted to the small bud. Several yeast uac mutants have been isolated which are defective in this process. We have now cloned the VACl gene, corresponding to the first of these mutants, vacl-1. This gene encodes a protein of 616 amino acids, without homolog in the current data bases. It contains neither long hydrophobic stretches nor a classical leader peptide. The most notable aspect of the sequence is the presence of three zinc fingers. Yeast in which the VACl gene has been entirely deleted are viable. However, they grow more slowly than wild-type cells and only form microcolonies when grown on glycerol at 37 "C. These yeast are defective in vacuole segregation at both the permissive and nonpermissive temperatures. The uacl mutant was previously shown to mislocalize carboxypeptidase Y to the cell surface, suggesting that Vaclp is involved in more than one vesicular traffic pathway.
While cytoplasmic organelles with high copy number may segregate by random diffusion at cytokinesis (Birky, 1983), organelles with low copy number must employ specific mechanisms to ensure their accurate partitioning into daughter cells. No single mechanism accounts for the segregation of each type of organelle. For example, mammalian Golgi divide by synchronized vesiculation, segregation of the vesicles, and re-fusion in the daughter cells to form the characteristic stacked cisternae (Warren, 1985). In contrast, the yeast nuclear membrane remains intact during mitosis and divides by septation.
The yeast vacuole partitions by a very different mechanism (Weisman et al., 1987;Weisman and Wickner, 1988). Early in S phase, shortly after bud emergence, the new vacuole is founded in the bud. Studies with zygotes (Weisman and Wickner, 1988) bearing the endogenous ade2 fluorophore (Weisman et al., 1987) in one parental vacuole demonstrated that this new bud vacuole is founded by a highly directed stream of membrane-limited structures. Using a variety of fluorophores, these structures appear as tubules or vesicles (Weisman and Wickner, 1988;Weisman et al., 1990;Raymond et al., 1990;de Mesquita et aZ., 1991). These structures appear * This work was supported by a grant from the National Institute of General Medical Sciences. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) M80596. soon after bud emergence and persist for approximately 20 min (de Mesquita et al., 1991). Thus, vacuole inheritance is both cell cycle-specific and highly spatially directed.
We have isolated several yeast strains which are defective in this process (Weisman et al., 1990;Shaw and Wickner, 1991). In these uac mutants, the bud grows to normal size and at cytokinesis has a normal content of mitochondria and nuclear DNA, yet has little or no vacuole material (Weisman et al., 1990;Shaw and Wickner, 1991). The majority of these mutants are defective in both vacuole segregation as well as in targeting newly made proteins from the Golgi to the vacuole. In this paper, we report the further characterization of the first of these mutants to be identified, uacl-1. We have used this mutant to clone and sequence the wild-type gene. VACl encodes a protein with three zinc fingers which is otherwise without homolog in current data bases. We propose that Vaclp is either a peripheral membrane protein or cytoskeletal protein which functions in several interorganelle traffic pathways.

MATERIALS AND METHODS
Strains, Plasmids, and Cell Culture-Escherichia coli DH5a was used for amplification of all plasmids, except where cleavage with the restriction enzyme BclI was required. In this case, plasmids were transformed into E. coli GM2163, a dam-strain.
PCR and Synthesis of Oligonucleotide Probes-For both VACl deletion constructs, the vector pLW101, comprised of the yeast shuttle vector pRS316 (Sikorski and Hieter, 1989) and 8 kb' of VACl the URA3 DNA fragment and the TRPl fragment by means of DNA, was used. Restriction enzyme sites were placed at the ends of polymerase chain reaction (Higuchi, 1989). Amplification with 1.5 units of Taq polymerase (Perkin-Elmer Cetus) was performed at 53 "C in 80 p1 in 300 mM KCl, 100 mM Tris-C1, pH 8.4, 100 mM MgC12, 1 mg/ml bovine serum albumin with 5 X pmol of vector containing URA3 or TRPl, 2 mM dATP, 2 mM dCTP, 2 mM dTTP, 2 mM dGTP, and 20 pmol of each primer. For the URA3 construct, the oligonucleotides 5'-CCCGAATTCTCATGTTTGACAGCTTAT-CATCG-3' and 5'-CCCGGTCACCGCATTAAAGCTTTTTCTT-TCC-3' were synthesized. The 1.2-kb fragment obtained from the polymerase chain reaction synthesis was digested with BstEII and The abbreviations used are: kb, kilobasepair(s); FITC, fluorescein isothiocyanate.
EcoRI, purified on a 0.7% agarose gel, and directly ligated into the 11-kb fragment from the pLWlOl vector which had been digested to completion with BstEII and partially digested with EcoRI. For the TATTGGCACGCC-3' and 5'-CCCGGTGACCGCTTTTCAAAAG-TRPl construct, the oligonucleotides 5'-CCCTGAGCGCGTGAAAC-GCCTGCAG-3' were used as primers. In this case, the 0.9-kb fragment was first subcloned into pCRlOOO (Invitrogen), the resulting plasmid was amplified in E. coli GM2163, and the TRPl fragment released by digestion with BclI and BstEII. After purification of the 0.9-kb fragment on a 0.7% agarose gel, it was ligated into the 10.5-kb fragment of pLWlOl which had been digested to completion with BstEII and partially digested with BclI.
Sequencing Strategy-Sequencing and complementation analysis were performed on two sets of nested deletions. The 5-kb SacI-Sac11 and the 3-kb SacII-SpeI fragments were subcloned into the polylinker of pRS316 (Sikorski and Hieter, 1989), a yeast shuttle vector which was generously provided by Dr. Philip Hieter (Johns Hopkins University). The 5-kb insert was used for the creation of nested deletions starting at the Sac11 site, whereas the deletions of the 3-kb insert originated from the SpeI site. The respective clones were cut in the polylinker with KpnI and HindIII, and limited exonuclease digestion was performed utilizing the Erase-a-base kit (Promega). Aliquots were removed every 20 s. Samples of interest were transformed into DH5a. Clones with DNA of the appropriate size were identified by their migration on 0.7% agarose gels. Sequencing templates were prepared from 3 ml of culture using alkaline lysis (Sambrook et al., 1989) followed by adsorption and elution from an anion exchange column (Quiagen). Automated sequencing (Smith et al., 1986) was performed based on dideoxynucleoside triphosphate chain termination (Sanger et al., 1977) using fluorescently labeled M13 reverse primers (ABI) and an AB1 model 370A interfaced with a Hewlett-Packard Vectra computer. Manual dideoxy chain termination sequencing was also performed. The 1.5-kb PstI fragment was subcloned into pUC19. Template was prepared by modification of the alkaline lysis procedure (Kraft et al., 1988). Four primers were utilized for sequencing in this region, 5'-CCATTTGTCTAAGGAAGTTTG-3', GGTCGG-3', and 5'-GATCTGTTGATAGCAGTGCGC-3'. Doublestranded sequencing was performed using the Taq Trak Sequencing kit (Promega). Sequencing reactions were run on 6% polyacrylamide gels containing 7 M urea (Sambrook et al., 1989).
Southern Analysis-For the demonstration that URAB was integrated near the VACl locus, genomic DNA was isolated (Rose et al., 1990) from URAB diploids, digested to completion with NruI, and separated on a 0.7% agarose gel in Tris-EDTA-acetate buffer (Sambrook et al., 1989). The fragments were transferred to Nytran (Schleicher and Schuell) following the manufacturer's instructions. Prehybridization and hybridization were carried out as described previously (Sambrook et al., 1989). The probe was the yeast integrating vector, YIp5, which was labeled with [(u-~'P]~CTP (50 pCi, 3000 Ci/mmol) with a random priming kit (Boehringer Mannheim). For the demonstration that VACl was replaced with TRP1, the same procedures were followed except that genomic DNA was digested with ClaI and a 1.5-kb PstI fragment, containing one-third of the VACl open reading frame plus flanking sequences, served as the probe.
Microscopy-Conditions for light microscopy, fluorescence microscopy, photography, and labeling of yeast with dichlorocarboxyfluorecein diacetate (Pringle et al., 1989;Weisman et al., 1990) and the endogenous d e 2 fluorophore (Weisman et al., 1987) have been described. For labeling yeast vacuoles with fluorescein isothiocyanate (FITC) (Preston et al., 1987;de Mesquita et al., 1991), 1 ml of yeast growing in YEPD to an ODm of 0.1-0.5 were collected by centrifugation (10,000 X g, 15 s) and resuspended in 1 ml of YEPD (50 mM citrate, phosphate, pH 4.5) to which 1 pl of 4 pg/ml FITC in dimethyl sulfoxide was added. Cells were incubated for 10 min at 37 "C, centrifuged (10,000 X g, 15 s), and resuspended in 100 pl of synthetic dextrose minimal medium complete (Rose et al., 1990).
Immunofluorescence-Antiserum which recognizes the acidic domain of Sec7p (Franzusoff et al., 1991) was the generous gift of Drs. Alex Franzusoff (University of Colorado Health Sciences Center, Denver) and Randy Schekman (University of California, Berkeley). The serum was preadsorbed with an equal volume of suspended wildtype yeast which had been killed by incubation at 70 "C for 1 h. Cells and serum were incubated at 23 'C for 1 h. The conditions for immunofluorescence were as described previously (Franzusoff et al., 1991) except that a 1:200 dilution of the primary antiserum was utilized.

RESULTS
Cloning the VACl Gene-The uacl-1 mutant, strain LWY148, exhibits little growth on YEPD media at 37 "C and no growth on YEP glycerol at 37 "C (Weisman et aL, 1990). We took advantage of this tight conditional phenotype to clone the VACl gene by complementation. LWY148 was transformed with a yeast genomic library (Carlson and Botstein, 1982) carried on the multicopy plasmid YEp24. Of 60,000 URA transformants, 81 isolates were obtained in which growth on glycerol at 37 "C was tightly linked to presence of the plasmid. Eighteen of these clones were mapped with several restriction enzymes and all clones were found to overlap in a region of 7 kb (Fig. lA). We found that this 7-kb region provided full complementation of both the growth defect and the vacuole segregation defect when introduced into uacl-1 on a single-copy plasmid. To demonstrate that this complementing fragment originated from the VACl locus and did not encode another yeast gene which could suppress the uacl-1 phenotype, the 2-kb BstEII fragment from within the 7-kb complementing region was subcloned into YIp5, digested with KpnI (leaving approximately 1 kb on either side) and integrated into LWY147, a strain which is wild-type for VACl and contains the uracil auxotrophy ura3-52. Introduction of URA into the chromosome at the VACl locus was demonstrated by Southern analysis (data not shown). When the diploid formed from this modified strain and LWY148 was sporulated and tetrads dissected, it was found that the URA marker segregated away from the uacl-1 vacuole segregation defect in each of the 26 tetrads analyzed. This demonstrates that the original chromosomal location of the isolated VACl clone is tightly linked with the uacl-1 gene.
Complementation Analysis-To identify the location of the VACl gene within the original 7-kb region, subclones were tested for complementing activity. Two overlapping frag-

CAAAAGTTGAAAAGAAGCCATTGGGAlUVVLTTCAAAAAAGGGAAGAGCTGCTGTCACACATGTGGAAGGACTTTGAATAACAATATTGGC Q K L K R S H W E K F K K G K S C C H T C G R T L N N N I G GCCATTAATTGTAGGAAATGTGGTAAACTGTATTGCAGAAGGCATCTTCCTAATATGATTAAACTTAATCTTTCCGCACAGTATGACCCC
A

GACCATGGGTTTAAT~GACATATCTTTATTATTCGATCATTCGCGG~TCTCCACGCTAATTGGTCCGC D H G F N
ments of interest were identified, Sad-Sac11 and SucII-SpeI (Fig. 1). These were subcloned in opposite orientations into a CEN plasmid, pRS316 (Sikorski and Hieter, 1989), and digested with EroIII from the Sac11 and SpeI sites, respectively. A set of nested deletions from each direction was obtained, and sequencing of the relevant clones was performed. Sequence of the VACl Gene-The sequence of the VACl gene and the derived protein sequence is presented in Fig. 2. The 1545 bases encode a protein of 515 amino acids. This extended open reading frame is the only one among the six possible reading frames. Two possible TATA boxes (Oliver and Warmington, 1989), TATAAAT and TATATAT, are located 205 and 256 base pairs upstream of the first ATG in the open reading frame (Fig. 2). We found that the first 108 amino acids are not required for complementation of the uucl-1 defect in vacuole segregation (Fig. 3, clone A ) . In two examples where the second methionine is missing (Fig. 3,   clones B and D), no complementation occurs. In one special case (Fig. 3, clone C), when this second methionine plus an additional two amino acids are missing, full complementation still occurs. However, it seems likely that this complementation results from the formation of a fusion protein with E. coli @-galactosidase. The observation that the second methionine is sufficient for full complementation suggests that the amino-terminal third of the protein is not required for VACl protein function. Alternatively, the second methionine may be the initiating methionine. However, there is no TATA box located between the first and second ATG. It is also possible that two proteins are produced from a single transcript. Note that a construct with only five bases upstream of the AUG of the second methionine yielded full complementation of the uacl-1 vacuole segregation defect. This suggests that sufficient transcription is occurring from the neighboring E. coli sequences (perhaps the T3 promoter or the p-galactosidase promoter). These findings suggest that a very low level of transcription is needed for full complementation of uacl-1, i.e. that the VACl protein is only required in low copy number.
The FASTA and TFASTA programs of Pearson and Lipman (1988) were used to compare the derived VACl amino acid sequence of 515 residues with sequences in GenBank, Swiss EMBL, and the NBRF protein libraries. In addition, the DNA sequence was compared with sequences in GenBank and NBRF DNA libraries. No significant homologies with any known sequences were observed. The protein has no long stretches of hydrophobic amino acids, suggesting that it is not an integral membrane protein. In addition, there is no NHzterminal leader sequence. Vaclp has three zinc fingers. The first, starting at amino acid eight, has the pattern Cys-XZ-Cys-X3-Phe-XS-Leu-Xz-His-X4-His and is the type first identified in TFIIIA (Miller et al., 1985). The second two, starting at Cys-78 and Cys-221, have the pattern Cys-Xz-Cys-Xlz-Cys-X2-Cys-X4-Cys, and are similar to the sequence first identified in adenovirus ElA, which is Cys-XZ-Cys-X~~-Cys-X~-cys.
Interestingly, another gene involved in normal vacuole biogenesis, VPS18, has recently been identified which contains a single zinc finger Cy~-X~-Cys-X,~-Cys-X~-Cys-X4-Cys VPS18 shows 70% identity in this region (Robinson et al., 1991) with yet a third gene involved in vacuole biogenesis, ENDlIPEP5 (Dulic and Riezman, 1989;Woolford et al., 1990). Deletion of the Entire VACl Coding Region-To construct a yeast strain in which the entire VACl coding region has been deleted, 1617 base pairs between the BstEII site and the closer BclI site were deleted and replaced with an 866-base pair fragment containing the TRPl marker (Fig. 1B). The construction of this strain, termed uacl-A2, was confirmed by Southern analysis. Genomic DNA from both the uacl-A2/ VACl heterozygous diploid and a VACl/VACl control were digested with excess ClaI and electrophoresed through a 0.7% agarose gel. The fragments were transferred to a membrane which was hybridized with the 1.5-kb PstI fragment containing one-third of the VACl open reading frame plus flanking sequence (Fig. lA). As expected, the heterozygous diploid showed two bands (Fig. 4, lane I), the wild-type fragment a t 2.1 kb (Fig. lA) and an additional band greater than 4 kb (Fig. 1B). In the wild-type control, only the 2.1-kb fragment is present (Fig. 4, lane 2). In Fig. 4, lane 1, the 2.1-kb fragment is much darker than the 4.7-kb fragment, because the latter has significantly less homology with the 1.5-kb PstI fragment used as the hybridization probe.
Deleting the entire VACl gene results in a viable yeast strain which is defective in vacuole segregation (Fig. 5). The strain is viable on YEPD at both 23 and 37 "C, although the doubling time in liquid culture at 23 "C is one-half that observed for wild-type cells and growth is very poor a t 37 "C. Only microcolonies can form when this strain is grown on YEP glycerol plates a t 37 "C.
A second strain, uucl-Al, in which two-thirds of the coding region was removed, leaving only amino acids 1-152, was also constructed (Fig. IC). The vacuole segregation defect in each of these strains were identical to each other and similar to, FIG. 5. Comparison of vacuole segregation in a VACl a n d a uacl-A2 strain. Yeast were grown overnight at 23 "C in YEPD to an OD, of 0.1-0.5. Cells were collected from 1 ml of each culture in an Eppendorf Microfuge (15,000 X g, 15 s), and the pellets were resuspended in 1 ml of YEPD containing 3.9 pg of FITC. Cells were incubated for 10 min, harvested as above, and resuspended in 100 pl of synthetic dextrose minimal medium complete (Rose et al., 1990).
Cell suspensions (1.5 PI) were applied to a glass slide pretreated with 1 mg/ml concanavalin A and photographed. A, wild-type cells labeled with FITC. B, uacl-A2 cells labeled with FITC. although more extreme than, that reported previously for uacl-1 (Fig. 4 of Weisman et al., 1990). Whereas the mother cells contain vacuoles that are approximately the same size as wild-type vacuoles, the bud vacuoles are either much smaller than normal or not detectable with either FITC or dichlorocarboxyfluorescein diacetate. Another phenotype which distinguishes the wild-type cells from the uacl-A2 mutant is that the vacuoles of wild-type mother cells are often irregularly shaped; two examples of wild-type cells with segregation structures can be seen (Fig. 5A). In the mutant, the vacuole in the mother cell is rounded, and no segregation structures can be seen (Fig. 5B). This difference in vacuole morphology and number of segregation structures between wild-type and mutant cells has also been reported for ups3 .
Disruption of VPS3 has been reported to produce a yeast strain with the same phenotype as udcl-1 . We find that the size and distribution of vacuoles in the bud was identical in ups3-A1 and uacl-A2 and that introduction of VACl on a multicopy plasmid does not suppress the ups3-A1 vacuole segregation defect (data not shown). Alleles of ups34, which have also been reported to cause a defect in vacuole segregation (Herman and Emr, 19901, are also not complemented by overexpression of VAC1. Vacuole inheritance in uacl-Al, as measured with the ade2 fluorophore in vegetatively growing cells and transfer of the ade2 fluorophore in yeast zygotes, was nearly identical to that reported for the original uacl-1 mutant strain (Weisman et al., 1990). Thus, although deleting the VACl gene has a profound affect Characterization of VACl which tethers the vacuole membrane to microtubules, either directly or via kinesin or dyenin-like motor proteins. This postulate may be tested both by immunofluorescence, using the VACl sequence as an aid to the generation of antibody, and by fractionation and reconstitution of the in vitro reaction of formation of tubulovesicular structures.
The VACl gene encodes a protein of 515 amino acyl residues which is without strong homolog in the three major FIG. 6. Golgi segregation is normal in vacl-Al. Cultures of sequence data banks. The novelty of the vAcl protein seboth wild-type and u~c1-A.l yeast were grown on YEPD at 23 "C to an ODw of 1. After formaldehyde was added to 5%, the cultures were quence is not as to date few genes in shaken for 20 min and incubated at 23 "C for an additional 4 h. The organelle segregation have been isolated. The only clue to cells were prepared for immunofluorescence as described previously function from the sequence is the fact that Vaclp contains (Franzusoff et al., 1991), utilizing an antiserum which recognizes the three zinc fingers. The first zinc finger is almost identical acidic domain of Sec7p. Examples of small buds containing Golgi with the zinc finger motif first identified in xempus he& apparatus can be seen in both wild-type ( A ) and uacl-A1 ( B ) yeast.
TFIIIA. Not only does this VACl zinc finger have the conserved cysteines and histidines with the correct spacing, but normal size sometime after cytokinesis and before bud emer-have been DroDosed to be the hallmarks of true transcription on WWXole segregation, the bud WXXdes eventually grow to a it also has the conserved phenylalanine and leucine which gence.
Segregation of the Golgi Apparatus Is Normal in Vucl-Al-Since most mutants which are unable to sort proteins from the Golgi to the vacuole are still normal for vacuole segregation into the bud, why is a small class of mutants defective in both processes? If the primary defect were in Golgi segregation and if the presence of a Golgi is required for each aspect of normal vacuole maintenance, then the conjunction of defects in vacuole segregation, vacuole protein sorting, and Golgi segregation should be observed. This is not the case for uacl, since Golgi segregation appears to be normal in uacl-A1 (Fig.  6) and is also normal in ups3 (data not shown).

DISCUSSION
The sequence of the VACl protein has neither the apolar domains characteristic of integral membrane proteins nor a NH2-terminal leader sequence, characteristic of proteins which enter the secretory pathway. However, two lines of evidence suggest that Vaclp is not freely diffusible in the cytosol. In zygotes, the transfer of d e 2 fluorophore from a parental vacuole into the bud, founding the new bud vacuole, is defective when both parents of the zygote are uacl-1 (Weisman and Wickner, 1988). This transfer normally occurs after bud emergence, well after nuclear fusion and cytoplasmic mixing have occurred. However, when one of the parents of the zygote is uucl-1 and the other is VAC1, the vacuole from the uacl-1 parent cannot participate in intervacuole tubulovesicular traffic, whereas the vacuole derived from the VACl parent shows no defect. This observation suggests that the Vaclp is bound to cellular structures such as membranes or cytoskeleton and not freely diffusible.
A second independent line of evidence comes from recent studies of an in vitro reaction* which reflects processes of vacuole inheritance. In this reaction, the vacuoles of semiintact cells produce striking extensions with the appearance of long tubules or strings of vesicles. This reaction requires the VACl and VAC2 proteins and incubation at 23-37 "C. When both the semi-intact cells and cytosol are prepared from uucl-1 or uacl-A2 strains, no reaction is seen at 37 "C. However, uacl-1 cytosol supports the formation of tubulovesicular structures by vacuoles from wild-type semi-intact cells, whereas wild-type cytosol complements uacl-1 semi-intact cells. These data support the inference from the in vivo studies, cited above, that the VACl protein is partially bound to cytoskeleton or membrane. We hypothesize, as a first postulate of Vaclp function, that it is part of the complex B. Conradt, J. Shaw, T. Vida, S. D. Emr, and W. Wickner, manuscript in preparation.
factors (Evans and Hollenberg, 1988). The only conserved amino acid of this zinc finger motif which is missing from VACl is a phenylalanine or tyrosine two amino acids before the first cysteine. In Vaclp this residue is a valine. The major difference between Vaclp and the known transcription factors is that this zinc finger occurs only once in VAC1, yet occurs two to nine times in the transcription factors (Evans and Hollenberg, 1988). The other two zinc fingers in Vaclp are very similar to the type first reported for adenovirus E1A. The major difference in this case is that E1A has 13 amino acids between the two sets of cysteines, whereas in VACl there are 12 amino acids. Although E1A does not appear to bind to a specific site on DNA (Berk, 1986), it is intimately involved in transcription and this cysteine-rich region is clearly important for its function (Lillie et al., 1987). Likewise, most other proteins in which this motif has been identified are DNA associated proteins (Frankel and Pabo, 1988).
These observations raise a second possibility of Vaclp function, that it is a nuclear transcription factcr. Our enthusiasm for this postulate is tempered by the fact that, although the sequence of at least nine genes involved in vacuole biogenesis is known Herman et al., 1991;Raymond et al., 1990;Dulic and Riezman, 1989;Woolford et al., 1990;Wada et dl., 1990;Banta et al., 1990;Robinson et al., 1991; this report), three of these genes have been found to contain zinc fingers. It seems unlikely that all are involved in vacuole biogenesis-related transcription. Rather, zinc may be involved in either the structure of Vaclp, its interactions with other proteins, or its catalytic function. As more types of proteins with this motif are discovered, the various roles of zinc fingers may emerge.
Aside from their effect on vacuole segregation, the uacl mutants secrete carboxypeptidase Y instead of localizing it to the vacuole, as initially required by the selection (Weisman et al., 1990). Recently, two other ups mutants with a uac phenotype have also been reported, ups3  and ups34 . What these three mutants have in common is a defect in vacuole segregation as well as localization of carboxypeptidase Y to the cell surface instead of the vacuole. Each mutant grows, albeit much more slowly, in the absence of their respective wild-type gene products. Interestingly, all three mutants have only a partial block in vacuole segregation. As was reported with uacl-1, these cells acquire a normal size vacuole before producing a bud despite the fact that they inherited little or no parental vacuolar material. Vps3p and Vps34p are partially associated with membranes or cytoskeleton, as judged by differential centrifugation of lysates (Raymond et dl., 1990;Herman and . .

Molecular Characterization
of VACl 623 . Although ups3 and ups34 mutants have a similar phenotype, no genetic interactions between these mutants have been observed. Our studies of the behavior of Vaclp in vitro also suggest that it is probably a peripheral membrane protein or associated with cytoskeleton.' The lack of an obvious membrane spanning domain or leader sequence in Vaclp is consistent with such a localization. In this paper we report that uacl-A2 is wild-type with regards to Golgi segregation. This observation extends our initial findings that both mitochondrial segregation and nuclear division are normal in uacl. There is reason to believe that mutants could be isolated that are defective in both Golgi and vacuole segregation, since vacuole segregation and Golgi segregation occur at a similar point in the cell cycle (Redding et al., 1991;Weisman et al., 1987). It is possible that Vaclp is responsible for sorting a protein directly involved in vacuole segregation to the cytoplasmic face of the vacuole membrane. Alternatively, Vaclp may be independently involved in Golgi to vacuole protein sorting and vacuole segregation. We favor this second hypothesis. First, of approximately 50 mutants defective in carboxypeptidase Y localization to the vacuole, only two have been reported to be defective in vacuole segregation (Raymond et al., 1990, Herman and. If mistargeting generally leads to a segregation defect, then this phenotype should have been more common. Furthermore, in uitro studies show that wild-type cytosol (but not uacl-1 cytosol) can complement the ability of uacl-1 vacuoles to form tubulovesicular structures.' These data are most simply explained by the Vaclp being a peripherally bound element of the cytoskeleton or vacuole. More direct studies, employing antibody and entailing purification of functional Vaclp, will be needed to test this postulate.