Mutations That Affect Vacuole Biogenesis Inhibit Proliferation of the Endoplasmic Reticulum in Saccharomyces cerevisiae

Corresponding author: Robin L. Wright, University of Washington, Box 351800, Seattle, WA 98195-1800., wrightr{at}u.washington.edu (E-mail)

Communicating editor: E. W. JONES

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

In yeast, increased levels of the sterol biosynthetic enzyme, 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase isozyme, Hmg1p, induce assembly of nuclear-associated ER membranes called karmellae. To identify additional genes involved in karmellae assembly, we screened temperature-sensitive mutants for karmellae assembly defects. Two independently isolated, temperature-sensitive strains that were also defective for karmellae biogenesis carried mutations in VPS16, a gene involved in vacuolar protein sorting. Karmellae biogenesis was defective in all 13 other vacuole biogenesis mutants tested, although the severity of the karmellae assembly defect varied depending on the particular mutation. The hypersensitivity of 14 vacuole biogenesis mutants to tunicamycin was well correlated with pronounced defects in karmellae assembly, suggesting that the karmellae assembly defect reflected alteration of ER structure or function. Consistent with this hypothesis, seven of eight mutations causing defects in secretion also affected karmellae assembly. However, the vacuole biogenesis mutants were able to proliferate their ER in response to Hmg2p, indicating that the mutants did not have a global defect in the process of ER biogenesis.

CELLS modulate many of their structural features in response to changes in physiology or environment. Control of membrane structure and function appears to be particularly responsive to changing metabolic conditions. For example, the area of endoplasmic reticulum (ER) increases threefold as naïve B cells differentiate into immunoglobulin secreting cells (WEIST et al. 1990 ). Likewise, the smooth ER in liver cells expands in response to drugs such as phenobarbital (JONES and FAWCETT 1966 ; BOLENDER 1974 ). Steroid hormone-producing cells such as those of the adrenal cortex also proliferate large amounts of smooth ER during their differentiation (BLACK 1972 ). Two proteins that play pivotal roles in drug detoxification and sterol biogenesis, the ER resident proteins cytochrome P450 and 3-hydroxy-3-methylglutaryl Coenzyme A (HMG-CoA) reductase, have themselves been shown to induce the proliferation of ER membranes when expressed at increased levels (CHIN et al. 1982 ; ANDERSON et al. 1983 ; WRIGHT et al. 1988 ; SCHUNCK et al. 1991 ; VOGEL et al. 1992 ; OHKUMA et al. 1995 ). Similar responses are elicited by cytochrome b5, Sec12p, the ribosome receptor, peroxisomal membrane proteins, and Pbn1p (VERGERES et al. 1993 ; NISHIKAWA et al. 1994 ; WANKER et al. 1995 ; ELGERSMA et al. 1997 ; NAIK and JONES 1998 ). However, despite these numerous examples of naturally occurring and experimentally induced ER proliferation, the cellular mechanisms that control regulated ER biogenesis are as yet unclear.

In response to Hmg1p, one of the two yeast HMG-CoA reductase isozymes, yeast assemble a nuclear-associated array of stacked ER membranes called karmellae (WRIGHT et al. 1988 ). The process of karmellae assembly provides a useful opportunity to explore how yeast cells alter ER membrane structure in response to specific physiological demands. However, the likelihood that studies of karmellae assembly may have general relevance is suggested by the conservation of cellular responses to HMG-CoA reductase. For example, expression of mammalian HMG-CoA reductase isozymes in yeast leads to karmellae production (WRIGHT et al. 1990 ). Proliferation and reorganization of the ER to produce "crystalloid ER" is induced in vertebrate tissue culture cells by experimental elevation of HMG-CoA reductase levels (ANDERSON et al. 1983 ). Treatment with HMG-CoA reductase inhibitors such as lovastatin leads to karmellae-like ER proliferation in the liver due to drug-induced increases in HMG-CoA reductase levels (SINGER et al. 1988 ). In addition, vertebrate cells that are expected to contain high levels of HMG-CoA reductase, such as steroid hormone-secreting cells in the adrenal gland, testes, or ovary, possess cell-type-specific proliferations of ER that are similar to karmellae (BLACK 1972 ; FAWCETT 1981 ). Thus, the ability of a cell to respond to HMG-CoA reductase levels by altering ER structure in specific ways appears to be universal.

To investigate the molecular events required for HMG-CoA reductase-induced membrane biogenesis, we carried out a genetic screen to identify trans-acting factors that affect karmellae biogenesis. We hypothesized that at least some of the genes involved in karmellae biogenesis would also play essential roles in regulating membrane homeostasis. Consequently, we screened several thousand temperature-sensitive mutants for their ability to assemble karmellae at permissive and nonpermissive temperatures. Mutants that were defective for karmellae biogenesis, yet still expressed high levels of Hmg1p, were retained for further analysis. Two independently isolated strains carried mutations in the same complementation group, which we discovered to be VPS16, a gene initially identified because of its role in vacuole biogenesis (DULIC and RIEZMAN 1990 ). Subsequent surveys of karmellae assembly in other vacuole biogenesis and secretory pathway mutants were consistent with the hypothesis that efficient flux of proteins and/or lipids from the ER is important for karmellae biogenesis.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Yeast strains and media:
The strains used in this study are shown in Table 1 and the plasmids are shown in Table 2. Most yeast strains are derived from the S288C strain, JRY527. However, we were unable to delete PEP5 or VPS29 in this strain and obtained pep5 and vps29 deletion strains from the Saccharomyces Genome Deletion Project (via Research Genetics, Huntsville, AL). These deletion strains are also in an S288C background, either BY4739 or BY4741 (BRACHMANN et al. 1998 ).

Yeast were grown either in minimal medium with appropriate amino acid and nucleotide base supplements (YM; SHERMAN et al. 1986 ) or in complete synthetic medium (CSM; Bio101, La Jolla, CA) lacking either uracil or histidine to select for retention of plasmids. The carbon source was 2% glucose or 2% galactose + 3% raffinose. Strains with plasmids containing the URA3-selectable marker were grown on rich minimal medium (YM medium plus 2% casamino acids with nucleotide base and amino acid supplements as needed). For induction of Hmg1p from the CUP1 promoter, overnight cultures growing on rich minimal medium were diluted to 1 OD₆₀₀/ml. The cultures were allowed to grow for 2 hr before addition of 0.25 mM CuSO₄. The cultures were incubated for 5 hr following copper addition and their ER structure was examined by 3,3'-dihexyloxacarbocyanine iodide (DiOC_6; Kodak, Rochester, NY) staining and fluorescence microscopy as described previously (KONING et al. 1996 ). Although many vacuole biogenesis mutants are sensitive to copper and other ions, these conditions produced no toxicity at permissive temperature. Standard methods were used for crossing strains and tetrad analysis (SHERMAN et al. 1986 ).

Deletion of vacuole biogenesis genes in JRY527 was accomplished using appropriate PCR products that carried portions of the 5' and 3' regions of the gene to be deleted and the geneticin-resistance marker, KanMX6 (WACH et al. 1994 , WACH et al. 1997 ; BAGANZ et al. 1997 ; BAHLER et al. 1998 ; BRACHMANN et al. 1998 ; PEARSON et al. 1998 ). Sequences of the primers used for producing the PCR product are available upon request. Note that the vps38 mutation simultaneously removed a portion of the 3' end of open reading frame (ORF) YLR361C. In addition, the pep7 also removes the promoter of the TIM1 gene, probably placing it under control of the PEP7 promoter. Disruption was confirmed via PCR, using primers that hybridized to sequences flanking the deleted gene and within the KanMX6 sequence.

Plasmids:
The galactose-inducible HMG1 plasmid, pAK266, has been previously described (KONING et al. 1996 ). The copper-inducible HMG1 plasmid, pAK220, was constructed by inserting a 3.75-kb SalI fragment containing the protein-coding and 3' regions of HMG1 into the XhoI site of pCGY1444 (a 2µ TRP1 plasmid containing the CUP1 promoter; ETCHEVERRY 1990 ) to create pEF164. A 5-kb SalI to NheI fragment from pEF164 containing the CUP1 promoter and HMG1 was inserted into the SalI and NheI sites of YEp24 (2µ URA3 plasmid) to create pAK220. Another 2µ plasmid containing a copper-inducible HMG1 gene and the HIS3 selectable marker, pAK444, was constructed by inserting the 5-kb SalI to NheI fragment from pAK220 into pRS423 (CHRISTIANSON 1992 ) cut with SalI and SpeI.

To clone the wild-type gene, vps16-101 mutant cells containing pAK444 were transformed with a genomic library (Sau3A partial digest) in pRS316 (SIKORSKI and HIETER 1989 ) provided by Bryan Jensen and Breck Byers (Department of Genetics, University of Washington). Transformants were grown at room temperature overnight and then shifted to 37° to select for rescue of the temperature-sensitive phenotype. Plasmids from two independent transformants were isolated after passage through Escherichia coli strain MC1066 (CASADABAN et al. 1983 ) and selection for uracil prototrophy. pAK472 contained a 13-kb insert, which was digested with ClaI and religated to produce pAK474 or digested with SpeI and religated to produce pAK473. pAK473 was digested with XbaI and religated to produce pAK494 and pAK495. Alternatively, pAK473 was digested with SacI and SpeI and ligated to pRS316 cut with the same enzymes to produce pAK496. pAK516 was constructed by ligating the 1.5-kb SpeI to ClaI fragment of pAK472 into pAK496 digested with the same enzymes. The 4.1-kb SacI to ClaI insert was ligated to pRS313 (SIKORSKI and HIETER 1989 ) to yield pAK522 in a HIS3 vector.

Mutagenesis and isolation of mutants:
Independent cultures of RWY410 (JRY527 containing pAK266, a galactose-inducible HMG1 gene on a centromere-containing plasmid) were mutagenized with UV light to 50% survival and grown in the dark for 3–4 days at room temperature (22°–24°). Colonies were replica plated to YM glucose medium and grown at 37° to identify temperature-sensitive colonies. Temperature-sensitive strains were recovered from the original plate, streaked onto solid medium to produce single colonies, and then patched onto YM glucose medium. The patches were replica plated to YM galactose + raffinose medium to induce karmellae assembly and incubated overnight at 37° or for 2 days at room temperature. For analysis of karmellae assembly, cells were picked from these plates into multiwell slides containing 2.5 µl of DiOC₆ solution (1 ml of yeast culture medium + 3 µl of a 10-mg/ml stock solution of DiOC₆ in ethanol) and screened for karmellae membranes using fluorescence microscopy. Prospective mutants were streaked on 5-fluoroorotic acid plates (BOEKE et al. 1984 ) to select for loss of the original plasmid (pAK266), retransformed with both pAK266 and a plasmid containing a copper-inducible HMG1 gene (pAK220), and retested for their ability to assemble karmellae.

Isolation of genomic DNA and PCR:
Genomic DNA was isolated from overnight cultures (5–10 ml) using glass bead lysis and phenol extraction as described in HOFFMAN 1996 . PCR was performed on 500 ng of genomic DNA or 20 ng of plasmid DNA in 10 mM Tris-HCl, pH 8.8, 1.5 mM MgCl₂, 50 mM KCl, 0.2 mM of each dNTP, and 0.5 units Taq polymerase (BAHLER et al. 1998 ). Alternatively, a small amount of cells grown overnight on a plate were smeared inside a PCR tube using a sterile 200-µl pipette tip and microwaved for 1 min at high power. Sterile water (23 µl) was added, followed by 1 µl of each oligonucleotide primer (from a 20-mM stock) and 1 PCR bead (Pharmacia Biotech, Piscataway, NJ). PCR products were amplified for 30 cycles at 94° for 30 sec, 55° for 30 sec, and 72° for 1 min followed by a 7-min extension step at 72°.

Immunoblot analysis:
Eight OD₆₀₀ units of cells were centrifuged at 800 x g for 3 min and the cell pellet was frozen at -80°. The frozen cell pellet was disrupted by glass bead lysis as previously described (KONING et al. 1996 ). A crude membrane fraction was obtained by differential centrifugation and the membrane pellet was resuspended in 20 mM HEPES, pH 7.4, 50 mM KOAc, 250 mM sorbitol, and 10% glycerol. To equalize total protein loaded, an aliquot of each sample was denatured by addition of SDS to 1% and the absorbance at 280 nm was measured. A total of 100 A₂₈₀ units was loaded into lanes on duplicate gels after heating the sample at 60° in 1x sample buffer (0.03 M Tris-HCl, pH 6.8, 2% SDS, 10% glycerol, 5% ß-mercaptoethanol, and 0.005% Bromphenol blue). One gel was blotted to nitrocellulose and probed with anti-Hmg1p or anti-Kar2p antibodies as described (LUM and WRIGHT 1995 ); the duplicate gel was stained with Coomassie blue. Jeff Brodsky (University of Pittsburgh, Pittsburgh, PA) provided the anti-Kar2p serum. Secondary antiserum was alkaline phosphatase-conjugated goat-anti-rabbit IgG from Promega (Madison, WI). Visualization of the secondary antiserum bound to the blot used the chromogenic substrates nitroblue tetrazolium and bromochloroindolylphosphate, using conditions within the linear range of detection (HARLOW and LANE 1988 ). A duplicate, Coomassie-stained gel and the immunoblot were digitized and analyzed using NIH Image 1.60 analysis software as described, using the stained gel to normalize for protein loading (LUM and WRIGHT 1995 ).

Microscopy:
For electron microscopy, cells were grown as described above, fixed with 2% glutaraldehyde followed by 2% potassium permanganate, and embedded in Spurr's resin as previously described (PROFANT et al. 1999 ). For immunofluorescence, cells were grown as described above, fixed in 3.7% formaldehyde in medium for 1 hr, and prepared for immunofluorescence using antibodies against Hmg1p and Kar2p as described (KONING et al. 1996 ). Examination of Hmg1:green fluorescent protein (GFP) localization in living cells was carried out in cells that had been cultured on galactose-containing medium for 15 hr at 30°. Growth at the lower temperature was necessary, since wild-type GFP fluoresces poorly at higher temperatures. At 30°, the vsp16 mutants grew very slowly, completing approximately one doubling. To visualize vacuole dynamics, cells were pregrown as described above, Hmg1p expression was induced with copper sulfate, and cells were grown for 3.5 hr at either 26° or 37° before incubation with FM4-64 for 1.5 hr. FM4-64 staining was performed as described in VIDA and EMR 1995 .

Allele rescue and sequencing:
The vps16-101 and vps16-102 alleles were rescued from the yeast genome by plasmid gap repair (ROTHSTEIN 1991 ). Specifically, pAK522 was digested with StuI and BstEII to eliminate the VPS16 coding region. The digested DNA was gel purified twice and used to transform wild-type, vps16-101, and vps16-102 strains. His⁺ transformants were screened for temperature sensitivity and other vps16-associated phenotypes. The plasmid was recovered from the yeast cells, amplified in E. coli, and digested to determine if the gapped plasmid had been repaired as expected. Plasmid DNA was purified for sequencing using the QIAGEN (Chatsworth, CA) MidiPrep. Sequencing was done with the ABI PRISM dye terminator cycle sequencing ready reaction kit (Perkin-Elmer, Norwalk, CT) using a series of primers that covered the VPS16 coding region in 300-bp intervals. Sequencing reactions were analyzed on an ABI377 automated DNA sequencer (Applied Biosystems, Foster City, CA). The 5' and 3' flanking ends of pAK473 and pAK477 (a plasmid from an independent transformant that also complemented vps16-101) were sequenced using the Taq dye primer cycle sequencing kit (Applied Biosystems).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

A fluorescence-microscopy screen of temperature-sensitive mutants identified karmellae biogenesis mutants:
Yeast cells with increased levels of Hmg1p assemble karmellae membranes, yet have no other readily discernible phenotype. In addition, karmellae are asymmetrically segregated at mitosis, such that the mother cell retains karmellae membranes and the daughter cell inherits an unelaborated ER structure (WRIGHT et al. 1988 ). Thus, within a logarithmically growing population, only 30–60% of the cells contain karmellae (KONING et al. 1996 ). These features dictate that identification and analysis of mutations that affect karmellae biogenesis must include direct observation of a population of cells. To isolate mutants defective in karmellae biogenesis, cells containing a galactose-inducible HMG1 gene on a plasmid (RWY410) were mutagenized and temperature-sensitive colonies were isolated on noninducing (glucose-containing) medium. The ability of these temperature-sensitive strains to assemble karmellae was then visually screened at permissive and nonpermissive temperature by fluorescence microscopy using the lipophilic, vital dye DiOC₆ (KONING et al. 1993 ).

Of 2500 temperature-sensitive strains screened, 215 initially showed some alteration in karmellae biogenesis. When cured of the plasmid and then retransformed with the original plasmid containing the galactose-inducible HMG1, 20% of the mutants no longer displayed an alteration in karmellae biogenesis, indicating that the karmellae phenotype resulted from mutations in the original plasmid. These strains were not investigated further. Since the generation of karmellae membranes requires increased levels of Hmg1p, any chromosomal mutation that affected expression from the GAL1 promoter would be scored as defective in karmellae biogenesis. Almost one-half (45%) of the originally identified mutations were presumed to be gal-, since they were able to generate karmellae membranes when HMG1 was expressed from the CUP1 promoter. Of the 22 potential karmellae assembly mutants, 11 assembled less than one-half of expected karmellae levels when HMG1 was expressed from either the GAL1 or the CUP1 promoter and were retained as karmellae-defective mutants. All 11 mutations were karmellae defective at both permissive and nonpermissive temperatures, were recessive, and, except for D1Z and D20i, were in different complementation groups.

Because they were in the same complementation group, we initially focused on mutants D1Z and D20i. As in the other mutants examined, the temperature sensitivity and karmellae biogenesis defects in DiZ and D20i mutants cosegregated (data not shown). In wild-type yeast cells containing the copper-inducible HMG1 gene (pAK220), 35% of the cells contained karmellae membranes after 5 hr of copper induction at permissive temperature (24°). At 37°, 62% of wild-type cells contained karmellae membranes. In contrast, karmellae were present in 7% of D1Z cells at permissive temperature and in 2% of the population at restrictive temperature. D20i cells displayed similar defects in karmellae assembly. Mutants D1Z and D20i also had densely stained cytoplasm at both permissive and restrictive temperature and lacked any obvious vacuole (Fig 1B, Fig C, Fig E, and Fig F).

View larger version (134K): In this window In a new window Download PPT slide   Figure 1. Karmellae assembly was abnormal in D1Z mutant cells. Expression of Hmg1p was induced in wild-type cells and D1Z cells containing a multicopy pCUP1::HMG1 plasmid (pAK220) by growth for 5 hr in 0.5 mM CuSO4 at room temperature (RT) or 37°. Cells were then stained with the lipophilic dye DiOC6 and examined by confocal microscopy. Wild-type (WT) cells cultured at either 26° (A) or 37° (D) contain prominent karmellae membranes associated with the nucleus (arrows). Karmellae (arrow) are rare in D1Z cells grown at 26° (B) and essentially absent in D1Z cells grown at 37° (E). Likewise, karmellae are present in few D20i cultures, either at 26° (C) or at 37° (F). Both mutant strains at both temperatures have a densely stained cytoplasm due to lack of a large vacuolar structure.

Hmg1p levels remained elevated in D1Z and D20i mutants:
The level of Hmg1p in D1Z and D20i strains was compared to that of wild-type cells (Fig 2). Both D1Z and D20i mutants had elevated levels of Hmg1p, yet the percentage of cells with karmellae membranes in the mutant populations was much lower than expected given the level of Hmg1p in the cell. This defect in karmellae assembly was more pronounced in cultures grown at 37°, the nonpermissive temperature for the mutant strains.

View larger version (39K): In this window In a new window Download PPT slide   Figure 2. D1Z and D20i expressed similar levels of Hmg1p as wild type. Expression of Hmg1p was induced in wild-type (squares) and in mutant D1Z (circles) and D20i (triangles) strains containing pAK220 (pCUP1::HMG1) by addition of 0.25 mM CuSO4 and growth for 5 hr at 26° (open) and 37° (shaded). Wild-type cells were also grown without addition of CuSO4 to provide additional information concerning the relationship between the amount of Hmg1p and the extent of karmellae assembly. (A) Crude cellular membranes were then immunoblotted using antiserum against the Hmg1p catalytic domain. (B) Comparison of the relative amounts of Hmg1p and the presence of karmellae in wild-type (squares) and in mutant D1Z (circles) and D20i (triangles) strains.

The D1Z and D20i mutations were complemented by plasmids containing VPS16:
To clone the wild-type copy of the gene defective in D1Z and D20i, strain D1Z containing a copper-inducible HMG1 gene in a 2µ HIS3 vector (pAK444) was transformed with a centromeric URA3 genomic library and grown at restrictive temperature. Plasmids that complemented both the temperature-sensitive growth and karmellae biogenesis defects were isolated from two independent transformants. The sequence at the ends of the insert in these plasmids revealed that they both contained DNA from the same region of chromosome XVI. Fragments of the 13-kb insert in one of the complementing plasmids, pAK472, were subcloned to identify the smallest complementing region. One subclone containing only the VPS16 gene complemented both the temperature-sensitive growth and karmellae biogenesis defects of both D1Z and D20i. Consequently, it appeared that D1Z and D20i had lesions in the VPS16 gene and we have named these new alleles of VPS16 as follows: D1Z, vps16-101; and D20i, vps16-102.

The vps16-101 and vps16-102 mutations encoded truncated versions of Vps16p:
The karmellae biogenesis-defective alleles vps16-101 and vps16-102 were recovered from the genome by plasmid gap repair and sequenced. In vps16-101, a transversion at nucleotide 1778 from a T to A produced a premature stop codon, resulting in truncation of the protein at amino acid 592 compared to the full-length protein of 798 amino acids. In vps16-102, a deletion of nucleotides 1715 and 1716 produced a frameshift, resulting in translation of 11 altered amino acids before reaching a premature stop codon that truncated the protein at amino acid 582. Therefore, both alleles of vps16 with defects in karmellae biogenesis encoded similarly truncated proteins lacking the last one-quarter of Vps16p.

The vps16-101 and vps16-102 mutants have abnormal vacuole and karmellae ultrastructure:
VPS16 had been previously identified because of its role in the assembly and function of the vacuole (DULIC and RIEZMAN 1990 ). To compare the phenotype of vps16-101 and vps16-102 mutants to previous reports of vps16 mutants, we examined the ultrastructure of our mutants. Neither vps16-101 nor vps16-102 had normal vacuole structures at 26° or at 37° (Fig 3). Instead, these cells contained multiple electron-translucent and electron-dense membrane-bound compartments as well as multilamellar membrane-enclosed structures similar to those seen previously in class C mutants (BANTA et al. 1990 ; RIEDER and EMR 1997 ). At permissive temperature, the karmellae membranes produced in these mutants, although much less abundant, were similar in morphology to karmellae produced in wild-type cells. However, at 37°, many of the vps16-101 and vps16-102 mutant cells contained disorganized membranes that were associated with the nucleus or present in the cytoplasm (Fig 3D).

View larger version (143K): In this window In a new window Download PPT slide   Figure 3. vps16-101 and vps16-102 have abnormal vacuole and karmellae ultrastructure at nonpermissive temperature. The ultrastructure of karmellae and the vacuole in WT (RWY353), vps16-101, and vps16-102 cells was compared. (A) Wild-type cells grown at 26°. (B) Wild-type cells grown at 37°. (C) vps16-101 cells grown at 26°. (D) vps16-101 cells grown at 37°. (E) vps16-102 cells grown at 26°. (F) vps16-102 cells grown at 37°. (Top row) WT cells contain a prominent vacuole (V) and karmellae membranes (arrowheads) in an ordered array around the nucleus (N). (Middle row) vps16-101 cells have no large vacuole and instead contain multiple small membrane-bound compartments. Karmellae membranes are rare and often disorganized, especially at 37°. (Bottom row) vps16-102 cells have a phenotype similar to that of vps16-101 cells.

Vacuole dynamics were abnormal in vps16-101 and vps16-102:
Dynamics of endocytic delivery of membrane to the vacuole can be characterized by staining with the styryl dye, N-(3-triethylammoniumpropyl)-4-(r-diethylaminophenylhexatrienyl) pyridinium dibromide (FM4-64; VIDA and EMR 1995 ). This dye initially inserts into the plasma membrane and is then endocytosed, yielding a punctate pattern that disappears coincident with fusion of the vesicles with the vacuolar membrane. FM4-64 staining allows visualization of vacuolar morphology dynamics including the formation of vacuole segregation structures and vacuole fission and fusion events in living cells (VIDA and EMR 1995 ; ZHENG et al. 1998 ). Hmg1p expression was induced by growth in rich minimal medium supplemented with 0.25 mM copper sulfate for 3.5 hr at 26° and 37°, and then the cells were stained with FM4-64 for 1.5 hr to allow endocytosis of the dye. This timing resulted in a total of 5 hr exposure to the respective temperatures following copper addition, consistent with previous experiments. The vacuole in wild-type cells (RWY353) consisted of one or more relatively large structures (Fig 4A and Fig D). Wild-type cells had significantly larger vacuoles at higher temperatures than at lower temperatures, as has been previously observed (MEADEN et al. 1999 ). In contrast, no normal vacuolar structure was visible in vps16-102 cells treated similarly (Fig 4C and Fig F). Instead, these cells resembled the FM4-64 staining pattern of vps16 cells (Fig 5B and Fig E; see also VIDA and EMR 1995 ; ZHENG et al. 1998 for pep3 cells). Interestingly, labeling of the nuclear envelope was observed in the mutants, suggesting that the endocytic pathway may be altered in vps16 mutants.

View larger version (148K): In this window In a new window Download PPT slide   Figure 4. The vacuole of vps16-101 and vps16-102 alleles was severely disrupted. Cells were incubated with the styryl dye FM4-64 for 1.5 hr following preincubation and copper induction at 26° or 37° for 3.5 hr. (A and D) The vacuolar membranes in WT cells (RWY353) incubated at 26° and 37° stain prominently with FM4-64. Karmellae membranes in WT cells do not stain with FM4-64. (B and E) vps16-101 cells have a dispersed cytoplasmic FM4-64 stain. Karmellae membranes in these cells are also stained. (C and F) vps16-102 cells have a staining pattern similar to vps16-1. These cells also have additional punctate areas of stain.

View larger version (106K): In this window In a new window Download PPT slide   Figure 5. The localization of Hmg1p was not altered in vps16 mutants. (A.1–D.1) Immunofluorescent localization of Hmg1p at 26° and at 37°. (A.2–D.2) The corresponding 4',6-diamidino-2-phenylindole nuclear stain is shown directly below. At the lower temperature, the protein is concentrated in the nuclear envelope, with some staining in the peripheral ER elements near the plasma membrane. At the higher temperature, an increase in peripheral ER staining is present. The localization patterns of wild type and mutant are indistinguishable. (E and F) The localization of an Hmg1:GFP fusion protein expressed from the GAL1 promoter in living cells at 30°. At this temperature, vps16 strains grow very slowly and do not assemble high levels of karmellae. Again, the localization of Hmg1p is similar in wild-type and vps16 mutant cells: The protein is concentrated in the nuclear envelope, frequently in an asymmetric pattern. Considerably less staining of the peripheral ER is present in both strains expressing Hmg1:GFP than in the immunofluorescent images.

Hmg1p displayed a normal ER localization pattern in vps16 mutants:
To determine whether vps16-101 and vps16-102 mutants might be unable to assemble karmellae because they did not properly target Hmg1p to the nuclear envelope/ER, the localization of Hmg1p was examined by indirect immunofluorescence. The Hmg1p in both mutants was present in the nuclear envelope and peripheral ER in a similar pattern as in wild-type cells (Fig 5). One noticeable difference was the more diffuse staining pattern of Hmg1p at 37° in the vps16 strain. This difference may reflect the altered morphology of proliferated ER membranes in some cells as seen by electron microscopy (Fig 3).

We also examined the localization of an Hmg1:GFP fusion protein in living wild-type and vps16 cells. This analysis required that the cells be cultured at 30° rather than at 37°, since the wild-type GFP protein used in this analysis fluoresces poorly at higher temperatures (LIM et al. 1995 ). In addition, expression of Hmg1:GFP was under control of the GAL1 promoter rather than the CUP1 promoter as in the immunofluorescence analysis. Cells were shifted to galactose-containing medium and cultured at 30° for 15 hr, similar to the conditions used to isolate the original vsp16-101 and vps16-102 mutants. At this temperature, vps16 mutants completed approximately one doubling while wild-type doubled approximately four times. In both the mutant and wild type, Hmg1:GFP was localized largely in the nuclear envelope, frequently in an asymmetric pattern with increased staining on one side (Fig 5E and Fig F). Unlike the immunofluorescence patterns, staining of peripheral ER elements was not frequently observed (an example is shown in Fig 5E). The different staining pattern probably reflects the lower expression produced from the GAL1 promoter vs. the CUP1 promoter. Similar to the immunofluorescence analysis, the localization of Hmg1:GFP in vps16 was similar to that of wild-type cells. Based on these observations, it appeared unlikely that the inability of vps16 mutants to assemble karmellae was due to abnormalities in Hmg1p localization. In addition, these results demonstrate that Hmg1p can become asymmetrically localized in the nuclear envelope without triggering karmellae assembly.

Overexpression of truncated Vps16-102p did not interfere with karmellae assembly in wild-type cells:
Vps16p is a rare protein that is present in a high-molecular-weight complex (HORAZDOVSKY and EMR 1993 ; SATO et al. 2000 ; WURMSER et al. 2000 ; PETERSON and EMR 2001 ). To test whether overproduction of the truncated mutant proteins might have a dominant negative effect on karmellae biogenesis, a multicopy plasmid containing vps16-102 (pAK537) was transformed into wild-type cells. Wild-type cells expressing Vps16-102p remained capable of generating karmellae membranes, indicating that the vps16-102 mutation was recessive even when overexpressed (Table 3). The effect of excess Vps16-102p on karmellae and vacuolar biogenesis in the vps16-102 mutant was also tested. Interestingly, the excess Vps16-102p partially complemented both the karmellae biogenesis defect and the vacuolar morphology defect (Table 3). However, complementation of both phenotypes did not necessarily occur in the same cells. Cells were observed that lacked a visible vacuole but contained karmellae membranes; conversely, other cells contained a visible vacuole but lacked karmellae membranes.

The observation that both vps16-101 and vps16-102 encoded similarly truncated proteins suggested that these particular alleles might have novel effects. Consequently, we compared the phenotype of vps16-101 and vps16-102 point mutants to vps16 null mutants in which the entire coding sequence was replaced with HIS3 or with KanMX6. The vps16 strains had similar phenotypes as the point mutants, except that karmellae assembly defects appeared temperature sensitive in the deletion mutant rather than constitutive as in strains with the point mutations. Following induction of Hmg1p expression at permissive temperature, a vps16 population assembled karmellae in 25% of the cells whereas a vps16-101 or vps16-102 population assembled karmellae in 5% of the cells. Our original tetrad analysis had suggested that genes capable of modifying karmellae assembly at permissive temperature might be segregating in our crosses of the vps16 point mutants to wild type. Consequently, for analysis of karmellae assembly in other vacuole biogenesis mutants (see below), vps16 strains were used as negative controls.

Karmellae assembly was defective in other vacuole biogenesis mutants:
VPS16 is a member of the class C vacuole biogenesis genes; mutations in these genes result in severely fragmented vacuoles that are not acidified (BANTA et al. 1988 ; RAYMOND et al. 1992 ). To determine whether or not other class C vacuole biogenesis genes were required for karmellae biogenesis, we examined the ability of pep3, pep5, and vps33 mutants to properly assemble karmellae membranes in response to increased levels of Hmg1p. The pep3, pep5, and vps33 strains had phenotypes similar to the vps16 strain and were defective for karmellae biogenesis at 37° (Table 4). However, the karmellae assembly defect was less pronounced in the pep5 strain, which was able to assemble karmellae at 20% of expected levels at nonpermissive temperature (in contrast to the other class C mutants that assembled only 4–7% of expected levels). It is possible that this result reflected differences in strain background, since we were unable to generate pep5 or vps29 deletions in our strain background and obtained these strains from the Saccharomyces Genome Deletion Project (via Research Genetics).

The inability of all class C deletion mutants to assemble karmellae efficiently at nonpermissive temperature suggested that normal vacuole structure might be important for karmellae assembly at high temperatures. To test this hypothesis, deletion mutants in other vacuole biogenesis genes were evaluated for their ability to assemble karmellae at permissive and nonpermissive temperatures. Class A mutants have vacuoles that are morphologically similar to those of wild-type cells and are acidified (BANTA et al. 1988 ; RAYMOND et al. 1992 ). The three class A deletion mutants that were tested (vps8, vps29, and vps38) assembled lower levels of karmellae than wild type, but significantly higher levels than the class C mutants (Table 4). These results supported the hypothesis that abnormal vacuole structure might be an important factor in the severe karmellae assembly defects of class C mutants.

To test this hypothesis further, we took advantage of the intermediate vacuolar phenotype of class B mutants, which have highly fragmented, but acidified vacuoles (BANTA et al. 1988 ; RAYMOND et al. 1992 ). The prediction is that, if vacuole structure, per se, is important for karmellae biogenesis, class B mutants should have karmellae assembly defects that are intermediate between class A mutants, which have relatively normal vacuole morphology, and class C mutants, which have severe vacuole morphology defects. One class B mutant, vps5, was as karmellae defective as class C mutants, assembling only 13% of expected levels of karmellae at 37°. In contrast, another class B mutant, vps41, was as karmellae proficient as class A mutants. Although the VAM3 gene has not been placed into the existing classification of vacuole biogenesis mutants, the defects of vam3 are consistent with it being another class B gene, since vam3 mutants have fragmented, acidified vacuoles (WADA et al. 1992 , WADA et al. 1997 ; SRIVASTAVA and JONES 1998 ). Karmellae assembly in vam3 mutants was temperature sensitive to a similar level as the class C mutant, pep5. Thus, the karmellae-assembly defects in class C mutants were not due to their profound abnormalities in vacuole structure, since intermediate alterations of vacuole structure did not uniformly display intermediate karmellae assembly defects.

We also examined the effects of class D and E mutations on karmellae assembly (Table 4). Class D mutants have a single large vacuole and are defective for vacuolar inheritance and acidification (RAYMOND et al. 1992 ). In addition, certain class D alleles display genetic interactions with or encode proteins that physically interact with class C genes/gene products (SRIVASTAVA et al. 2000 ; PETERSON and EMR 2001 ). Class E mutants have an enlarged, acidified prevacuolar/endosome compartment (RAYMOND et al. 1992 ). All three class D deletion mutants tested, pep7, vps34, and vps45, as well as the single class E mutant tested, vps24, were defective in karmellae assembly at nonpermissive temperature. Consequently, the severe karmellae assembly defects in a subset of vacuole biogenesis mutants were not caused by abnormal vacuole structure or abnormal acidification.

Expression from the CUP1 promoter occurs even in the absence of additional copper. Consequently, in our wild-type strain, sufficient Hmg1p is produced in the absence of copper to induce karmellae in 18% of the population grown at 26°. Thus, we can also examine whether vacuole biogenesis mutants were unable to assemble new karmellae or also had defects in retention of preexisting karmellae. Strains that were apparently unable to maintain the preexisting karmellae (i.e., displayed karmellae in significantly <18% of the population at nonpermissive temperature) include vps5, pep3, vps16, vps33, vps34, and vps45. In contrast, pep5, pep7, vps24, and vam3 mutants were able to retain the karmellae that had been assembled at the time of copper addition and shift to nonpermissive temperature. Only vps8, vps29, vps38, and vps41 mutants were able to assemble additional karmellae following the shift to high temperature.

The inability to assemble karmellae was not an immediate consequence of defects in ER-to-vacuole traffic, but displayed phenotypic lag:
To test whether the karmellae assembly defects were direct consequences of a block in delivery of proteins to the vacuole, we examined mutants containing an allele of PEP3 that is temperature sensitive for function (RIEDER and EMR 1997 ). Strains with the pep3TSF allele properly target vacuolar proteins and have normal vacuoles at permissive temperature, but undergo a rapid block in vacuolar protein targeting upon a shift to 37°. When karmellae were induced simultaneously with the shift to nonpermissive temperature, 38% of pep3TSF cells were able to assemble karmellae, compared to 62% of wild-type cells. In contrast, only 3% of pep3 cells assembled karmellae after a shift to nonpermissive temperature. Thus, the inability to assemble karmellae was not an immediate result of blocking protein delivery to the vacuole. This result was not surprising, since the effects of pep3TSF on vacuole morphology and function also display phenotypic lag, such that normal morphology is observed for at least 5 hr following shift to nonpermissive temperature (ROBINSON et al. 1988 ).

Vacuole biogenesis mutations may affect ER function:
The inability of vacuole mutants to assemble karmellae might simply reflect morbidity of these strains at elevated temperatures. Consequently, it was important to determine whether temperature-sensitive growth was tightly correlated with defective karmellae assembly. Serial dilutions of deletion mutants, without plasmids, were plated onto YPD medium and growth was examined after incubation for 3 days at 26° or at 37° (Fig 6, top row).

View larger version (136K): In this window In a new window Download PPT slide   Figure 6. The patterns of sensitivity to drugs suggest that karmellae-defective mutants may be hypersensitive to tunicamycin. Fivefold serial dilutions of deletion mutants, beginning at a culture concentration of 1 OD600/ml, were spotted onto YPD media containing inhibitors of different cellular functions. Drugs or inhibitors were added from concentrated stocks to cooled YPD medium immediately before pouring. Except as noted, all plates were incubated at 26°. The photographs shown were taken following growth for 3 days. To the right are listed the percentages of cells with karmellae in a population grown for 5 hr at 37°. The uppercase letters to the left denote the classes of the vacuole biogenesis mutations, on the basis of classification by BANTA et al. 1988 and RAYMOND et al. (1988). Strains used are wild type (JRY527), vps8 (RWY1385), vps38 (RWY1522), vps5 (RWY1525), vps41 (RWY1263), vps16 (RWY1485), pep3 (RWY1123), vps33 (RWY1119), pep7 (RWY1484), vps34 (RWY1127), vps45 (RWY1257), vps24 (RWY1462), vam3 (RWY1535), vps29 (deletion 975), and pep5 (deletion 817). All strains are isogenic to JRY527, except vps29 and pep5, which are both in the BY4741 strain background. Of the drugs tested, tunicamycin and DTT hypersensitivity displayed the best correlation with the inability of a strain to efficiently assemble karmellae. Note that pep7 and vps34 strains grew more poorly than other strains even under permissive conditions and were hypersensitive to most drugs. Interestingly, many vacuole biogenesis mutants were more resistant than wild type to miconazole, a drug that inhibits ergosterol synthesis. The known or suspected targets of the drugs are as follows: A23187, calcium ionophore (IIDA et al. 1990 ); benomyl, inhibits microtubule function (THOMAS et al. 1985 ); ß-mercaptoethanol, inhibits protein disulfide bond formation; caffeine, may affect phosphodiesterase and/or calcium levels; calcofluor, inhibits chitin synthase (THOMAS et al. 1985 ; VALDIVIESO et al. 1991 ); cerulenin, inhibits fatty acid synthesis (INOKOSHI et al. 1994 ); clotrimazole, inhibits ergosterol synthesis (SUD and FEINGOLD 1981 ); cycloheximide, inhibits translation; cyclosporine, binds to proline isomerase and inhibits calcineurin (FOOR et al. 1992 ); 2-deoxyglucose, affects glucose metabolism (HERVE et al. 1992 ); DTT, inhibits protein disulfide bond formation (JAMSA et al. 1994 ); hydrogen peroxide, induces oxidative stress; hydroxyurea, inhibits DNA synthesis (RITTBERG and WRIGHT 1989 ); hygromycin B, inhibits protein synthesis (DEAN 1995 ); lovastatin, inhibits ergosterol synthesis (LORENZ and PARKS 1990 ); miconazole, inhibits cytochrome P450 and ergosterol synthesis (VANDEN BOSSCHE et al. 1989 ); nocodazole, inhibits microtubule polymerization (PILLUS and SOLOMON 1986 ); nystatin, inhibits ergosterol synthesis (MCCAMMON et al. 1984 ; SKAGGS et al. 1996 ) and affects vacuole morphology (BHUIYAN et al. 1999 ); oligomycin, inhibits mitochondrial ATPase (UH et al. 1990 ); tunicamycin, inhibits protein glycosylation (KUO and LAMPEN 1974 ; ZOU et al. 1995 ).

Most karmellae assembly-defective strains (vps16, pep3, vps33, pep7, vps34, and vps45) were temperature sensitive for growth. However, other karmellae assembly-defective strains grew well at 37° (vps5) or displayed limited growth (vps24, vam3, and pep5). Consequently, temperature sensitivity was not a prerequisite condition for the presence of severe karmellae assembly defects at elevated temperature. This result makes it unlikely that the karmellae assembly defects were due simply to cell morbidity at elevated temperature.

An alternative hypothesis is that vacuole biogenesis mutations with severe karmellae assembly defects might have alterations in ER structure or function that affect karmellae assembly at elevated temperatures. If so, these mutants might be hypersensitive to drugs that target ER functions. To test this hypothesis, we grew the vacuole biogenesis mutants on YPD medium in the presence of the protein glycosylation inhibitor, tunicamycin (Fig 6, bottom right). Tunicamycin inhibits the activity of the ER protein, Alg7p, which catalyzes the first step in formation of dolichol-linked oligosaccharides needed for N-linked protein glycosylation (BARNES et al. 1984 ). This inhibition leads ultimately to increased synthesis of ER chaperones such as Kar2p, via the unfolded protein response (UPR; MORI et al. 1992 ; KOHNO et al. 1993 ; COX and WALTER 1996 ; MORI et al. 1996 ; SHAMU and WALTER 1996 ; WELIHINDA and KAUFMAN 1996 ). With the exception of vps5, the karmellae assembly-defective mutants were hypersensitive to tunicamycin (Fig 6 and Fig 7). Thus, tunicamycin hypersensitivity appeared to be a useful predictor of whether a vacuole biogenesis mutant was karmellae-assembly defective. This predictive value did not extend beyond vacuole biogenesis mutants, since both ire1 and hac1 mutants, which are tunicamycin sensitive and unable to activate the UPR, are able to assemble wild-type levels of karmellae (LARSON et al. 2002 ).

View larger version (49K): In this window In a new window Download PPT slide   Figure 7. Temperature-sensitive secretory mutants examined were generally unable to assemble karmellae at nonpermissive temperature. Karmellae assembly was analyzed in logarithmic cultures following growth for 2 hr at permissive (26°) or nonpermissive (37°) temperature. The data represent an average of two independent experiments in which at least 200 cells were scored. sec63 and sec12 mutants were unable to assemble karmellae at either permissive or nonpermissive temperature. In contrast, sec14, which encodes a phospholipid exchange factor involved in Golgi traffic, assembled karmellae efficiently at both permissive and nonpermissive temperatures.

Many mutations that affect vacuole form and function result in a variety of drug sensitivities. Thus, rather than reflecting altered ER function, the hypersensitivity to tunicamycin might simply indicate general sensitivity of those strains to drugs or other insults. To test this hypothesis, we examined the growth of deletion mutants in the presence of a variety of drugs, both with ER and with other cellular targets (Fig 6). Clearly, the hypothesis that karmellae-defective mutants are uniformly sensitive to drugs is not supported. For example, vam3 and pep5 were actually more resistant to miconazole than the wild-type control were.

Assembly of ER in response to Hmg2p was largely unaffected by the vacuole biogenesis mutations:
Increased expression of Hmg2p induces proliferation of stacked ER membranes that, although similar to karmellae, have distinct morphology (KONING et al. 1996 ). To test whether vacuole biogenesis mutants were also defective for the assembly of Hmg2p-induced ER membranes, we transformed the deletion mutants with a plasmid that encodes a copper-inducible HMG2 gene (pCL437). When expression of Hmg2p was elevated by copper addition, all of the mutants assembled at least 53% of expected ER proliferations at nonpermissive temperature (Table 5). Thus, vacuole biogenesis mutations do not globally interfere with all forms of ER proliferation.

Assembly of karmellae was also affected by secretory mutations:
On the basis of the data thus far, we considered it likely that inefficient clearing of vacuole proteins from the ER in the vacuole biogenesis mutants might produce altered protein composition within the ER, rendering it incompatible for karmellae assembly. This hypothesis predicts that blocking the ER-to-plasma membrane branch of the secretory pathway might also produce changes in ER composition that inhibit karmellae assembly. To test this hypothesis, we examined karmellae assembly at permissive and nonpermissive temperatures in eight temperature-sensitive secretory mutants (CLEVES and BANKAITIS 1992 ; SCHEKMAN 1994 ) that affect different stages of secretion (Fig 7). For these studies, cells were examined by DiOC₆ staining after 2 hr of induction of HMG1 expression by the addition of copper, rather than 5 hr as in previous studies. Earlier examination was necessary because the organization of the ER in many secretory mutants, particularly sec1 and sec6, was obscured by intense cytoplasmic staining at later times following shift to nonpermissive temperature.

Seven of the eight secretory mutants we examined were severely defective in karmellae assembly at nonpermissive temperature, including sec63, sec12, sec17, sec18, sec7, sec1, and sec6 (Fig 7). At permissive temperature, all mutants except sec63 and sec12 were capable of karmellae assembly. Thus, the majority of secretory mutants displayed temperature-sensitive defects in karmellae assembly similar to those of vacuole biogenesis mutants. In contrast, sec63 and sec12 mutants were constitutively defective in karmellae assembly and sec14 mutants were able to assemble karmellae at both permissive and nonpermissive temperatures. On the basis of plating assays, the sec14 strain appears to be as temperature sensitive for growth as the other secretory mutants that we examined (our unpublished results). Consequently, the temperature-sensitive growth of the secretory mutants did not account for the karmellae assembly defects, as was observed for the vacuole biogenesis mutants.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

In yeast, proteins that leave the Golgi apparatus have two major fates: Vacuolar proteins are delivered to the vacuole and secreted or plasma membrane proteins are delivered to the plasma membrane. With few exceptions, mutations that affect either of these branches of the secretory pathway also compromise karmellae assembly at high temperature. Since many of these mutants are also temperature sensitive for growth, the inability to assemble karmellae may simply be another characteristic of an already "sick" cell. Several observations weaken this possibility. First, if the inability to assemble karmellae simply reflects cell morbidity, karmellae assembly mutants should have been commonly encountered in our screen of random temperature-sensitive strains. Instead, karmellae assembly mutants were quite rare, comprising <1% of the several thousand temperature-sensitive strains screened. In addition, the "sick cell" hypothesis predicts that cells unable to assemble karmellae might be sick for a variety of reasons. Instead, nearly all of the karmellae assembly mutants that we obtained secreted carboxypeptidase Y (our unpublished results), a hallmark of defects in vacuole assembly. Thus, our screen for temperature-sensitive (TS) mutants with karmellae assembly defects appeared to efficiently identify vacuole biogenesis mutants. (Although secretory mutants should have been present in our original screen, they were undoubtedly discarded because their ER structure is not observable by DiOC₆ staining after 12 hr of growth at nonpermissive temperature as used in our screen. Under these conditions, secretory mutants have a uniform, intensely bright staining pattern that is indistinguishable from that of cells that have simply died.)

Another observation that is inconsistent with the sick cell hypothesis is that the inability to assemble karmellae was not invariably correlated with temperature-sensitive growth. For example, sec14 is temperature sensitive, yet assembles normal levels of karmellae at high temperatures. In addition, vps5 is not temperature sensitive, yet assembles low levels of karmellae at high temperatures. Finally, all of the vacuole biogenesis mutants examined were reasonably proficient in assembling ER proliferations in response to Hmg2p, a protein with considerable sequence homology and identical catalytic activity to Hmg1p (BASSON et al. 1986 ; BASSON 1988 ). Thus, vacuole biogenesis mutants that are incapable of assembling karmellae still retain the ability to remodel their ER in response to at least one other protein. Consequently, if these mutants cannot assemble karmellae because they are sick, they must be sick in a specific and interesting way.

Mutations with less effect on karmellae assembly (see Fig 8) were generally in genes with roles in retrograde traffic from the prevacuolar compartment to the Golgi or in anterograde traffic to the vacuole via the alternate pathway that bypasses the prevacuolar compartment (BRYANT et al. 1998 ). The exceptions are again VPS5 and SEC14. One possibility is that Vps5p, in addition to functioning in retrograde traffic from the endosome to the Golgi, might also function at other stages of the vacuole-targeting pathway. Sec14p, the major phosphatidylinositol/phosphatidylcholine transfer protein in yeast, appears to coordinate phospholipid metabolism with Golgi function (BANKAITIS et al. 1989 , BANKAITIS et al. 1990 ). sec14 mutations can be suppressed by changes in phospholipid synthesis or uptake (CLEVES et al. 1991 ; XIE et al. 2001 ). Perhaps the increased Hmg1p enzyme activity that accompanies karmellae assembly produces a compensatory alteration in membrane sterol or phospholipid content that partly suppresses the sec14 mutant phenotype, enabling karmellae assembly to proceed normally. If so, this compensation is insufficient to suppress the temperature-sensitive phenotype of this strain.

View larger version (66K): In this window In a new window Download PPT slide   Figure 8. Vacuole biogenesis and secretion mutants with severe karmellae assembly affect many different stages of the secretory pathway. Genes whose loss of function results in <25% of wild-type karmellae levels at high temperatures are indicated by red boxes; genes whose loss results in >40% of wild-type karmellae levels are indicated by green boxes. The step at which the gene product acts is indicated by the position of the box. Several proteins act at multiple steps of the secretory pathway. For example, Sec18p [yeast homolog of the n-ethylmaleimide-sensitive factor (NSF)] and Sec17p [yeast homolog of the soluble NSF attachment protein (SNAP)] function at essentially every membrane fusion event in cells, although they are indicated only once in the diagram. Sec14p and Sec7p are also required for movement of proteins through the Golgi apparatus itself. Vps8p, Pep3p, and Pep5p are involved in anterograde and retrograde traffic of proteins between the Golgi and prevacuolar compartment; Pep3p and Pep5p also function in fusion of vesicles with the vacuole. Pep3p is also involved in endocytosis.

Many possible explanations might account for the karmellae assembly defects in most secretory and vacuole biogenesis mutants. For example, vacuoles may perform an essential function at elevated temperature that is also needed for karmellae assembly. However, our working hypothesis proposes that inefficient flux through the secretory and vacuole biogenesis pathways leads to altered ER composition and/or function. At lower temperatures, rates of insertion and removal of proteins and/or lipids remain sufficiently balanced to enable ER function to continue. However, at elevated temperatures, the increased rate of protein translocation into an ER with a compromised efflux may produce a molecular context in which karmellae assembly cannot occur and, perhaps, cause lethality in certain of the vacuole biogenesis mutants. The sensitivity of many of the vacuole biogenesis mutants to ER stresses such as tunicamycin and dithiothreitol (DTT), which inhibit protein processing in the ER, and resistance to miconazole, which inhibits an ER-localized step in ergosterol synthesis, is consistent with this prediction. Although this drug-sensitivity profile suggests that these mutants may be unable to activate the unfolded protein response (see CHAPMAN et al. 1998 ), an abnormal UPR response cannot account for their karmellae assembly defect, since karmellae assembly is normal in ire1 and hac1 mutants that cannot activate the UPR, nor does karmellae assembly itself activate the UPR (LARSON et al. 2002 ). In addition, activation of the UPR does not interfere with karmellae assembly (R. L. WRIGHT, unpublished results).

Although we have not resolved the mechanisms by which altered vacuole biogenesis and secretory functions inhibit karmellae assembly, the fact that merely expressing high levels of Hmg1p is not sufficient to induce karmellae assembly provides an important insight into the process of ER remodeling and biogenesis. For example, these results make it unlikely that karmellae assembly results directly from interactions between Hmg1p proteins themselves, as the "zippered membrane" hypothesis suggests (GONG et al. 1996 ; SCHULKE et al. 1997 ; PROFANT et al. 1999 ). Instead, a proper environment of lipid and/or protein appears to be essential for karmellae assembly. Balanced flux through the two major branches of the yeast secretory pathway may be important for establishment or maintenance of this environment. Since mutants unable to efficiently assemble karmellae were able to respond properly to Hmg2p, our results also reinforce the hypothesis that distinct ER subdomains proliferate in response to Hmg1p vs. Hmg2p (KONING et al. 1996 ). Thus, at least two separable pathways for ER structural modification in yeast must exist.

	FOOTNOTES

¹ Present address: Tilligen, 1000 Seneca St., Suite 200, Seattle, WA 98101.
² Present address: Woodring College of Education, Western Washington University, Bellingham, WA 98225.
³ Present address: Rosetta Inpharmatics, 12040-115th Ave., NE, Kirkland, WA 98034.

	ACKNOWLEDGMENTS

The authors thank Elizabeth Jones and anonymous reviewers for helpful comments on the manuscript. R.W. dedicates this manuscript to the indomitable spirit of the diverse, dedicated, and caring people of the United States of America. This work was supported by National Institutes of Health grant GM45726 and National Science Foundation grant 78287.

Manuscript received August 14, 2001; Accepted for publication January 14, 2002.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ANDERSON, R. G., L. ORCI, M. S. BROWN, L. M. GARCIA-SEGURA, and J. L. GOLDSTEIN, 1983  Ultrastructural analysis of crystalloid endoplasmic reticulum in UT-1 cells and its disappearance in response to cholesterol. J. Cell Sci. 63:1-20[Abstract].

BAGANZ, F., A. HAYES, D. MARREN, D. C. GARDNER, and S. G. OLIVER, 1997  Suitability of replacement markers for functional analysis studies in Saccharomyces cerevisiae. Yeast 13:1563-1573[Medline].

BAHLER, J., J. Q. WU, M. S. LONGTINE, N. G. SHAH,