The PBN1 Gene of Saccharomyces cerevisiae: An Essential Gene That Is Required for the Post-translational Processing of the Protease B Precursor

Corresponding author: Rajesh R. Naik, Department of Biological Sciences, Carnegie Mellon University, 4400 Fifth Avenue, Pittsburgh, PA 15213, rnaik{at}cmu.edu (E-mail).

Communicating editor: A. P. MITCHELL

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The vacuolar hydrolase protease B in Saccharomyces cerevisiae is synthesized as an inactive precursor (Prb1p). The precursor undergoes post-translational modifications while transiting the secretory pathway. In addition to N- and O-linked glycosylations, four proteolytic cleavages occur during the maturation of Prb1p. Removal of the signal peptide by signal peptidase and the autocatalytic cleavage of the large amino-terminal propeptide occur in the endoplasmic reticulum (ER). Two carboxy-terminal cleavages of the post regions occur in the vacuole: the first cleavage is catalyzed by protease A and the second results from autocatalysis. We have isolated a mutant, pbn1-1, that exhibits a defect in the ER processing of Prb1p. The autocatalytic cleavage of the propeptide from Prb1p does not occur and Prb1p is rapidly degraded in the cytosol. PBN1 was cloned and is identical to YCL052c on chromosome III. PBN1 is an essential gene that encodes a novel protein. Pbn1p is predicted to contain a sub-C-terminal transmembrane domain but no signal sequence. A functional HA epitope-tagged Pbn1p fusion localizes to the ER. Pbn1p is N-glycosylated in its amino-terminal domain, indicating a lumenal orientation despite the lack of a signal sequence. Based on these results, we propose that one of the functions of Pbn1p is to aid in the autocatalytic processing of Prb1p.

THE yeast vacuole contains an ensemble of hydrolases, including aminopeptidase I (ApI), proteinase A (PrA), proteinase B (PrB) and carboxypeptidase Y (CpY). These proteases participate in protein degradation and turnover. Vacuolar hydrolases are synthesized as inactive precursors and they reach the vacuole via the secretory pathway or the cytoplasm-to-vacuole pathway (JONES and MURDOCK 1994 ; JONES et al. 1997 ).

The vacuolar serine endoproteinase B (PrB) of Saccharomyces cerevisiae is a member of the subtilisin family of proteases (MOEHLE et al. 1987 ). PrB plays a major role in the degradation of proteins and peptides during starvation for nitrogen or carbon sources, autophagy, meiosis and sporulation, and in the constitutive and/or environmentally triggered degradation of endocytosed plasma membrane proteins (ZUBENKO and JONES 1981 ; TEICHERT et al. 1989 ; TAKESHIGE et al. 1992 ; JONES and MURDOCK 1994 ). Processing of precursors to several vacuolar hydrolases involves specific cleavages catalyzed by PrB. These include cleavages that generate the mature N termini of protease A (PrA), carboxypeptidase Y (CpY), carboxypeptidase S (CpS), aminopeptidase I (ApI), aminopeptidase Y (ApY), and PrB, as well as the mature C termini of yeast vacuolar alkaline phosphatase (ALP) and PrB itself (JONES and MURDOCK 1994 ; VAN DEN HAZEL et al. 1996 ; JONES et al. 1997 ).

PrB is encoded by the PRB1 structural gene in yeast and is initially synthesized as a large 76-kD inactive precursor (Prb1p). Prb1p undergoes extensive post-translational processing and modification as it traverses the secretory pathway (Figure 3A). The 76-kD precursor enters the endoplasmic reticulum (ER) via the post-translational translocation pathway. The signal peptide is cleaved by the signal peptidase as evidenced by its failure to occur in a sec11 mutant (V. L. NEBES and E. W. JONES, unpublished results). In addition, a single N-linked carbohydrate chain as well as O-linked carbohydrate groups are added. Next, the large, 260-amino-acid propeptide is cleaved in the ER by an intramolecular autocatalytic process. The autocatalytic cleavage does not occur if the active site residue Asp⁴⁵ or Ser²³⁹ is mutated, indicating that a functional PrB catalytic site is required for the ER autocatalytic processing step. The autocatalytic cleavage is intramolecular and cannot be catalyzed in trans (NEBES and JONES 1991 ). The 39-kD precursor formed in the ER is further modified in the Golgi, and the molecular mass increases to 40-kD, due to the elaboration of the Asn-linked side chain. Either in the prevacuolar compartment or, more likely, in the vacuole, PrA catalyzes the cleavage of the first C-terminal post region, P1, removing about 30 C-terminal amino acids to form a 37-kD precursor. The 40-kD form accumulates in PrA-deficient strains. The final proteolytic cleavage that removes the second post region, P2, containing the N-linked carbohydrate chain, is autocatalytic or can be cleaved by PrB itself in trans. The cleaved propeptide is targeted to the vacuole for degradation subsequent to the autocatalytic cleavage (MECHLER et al. 1982 ; MECHLER et al. 1988 ; MOEHLE et al. 1989 ; NEBES and JONES 1991 ; HIRSCH et al. 1992 ).

View larger version (11K): In this window In a new window Download PPT slide   Figure 1. Genetic scheme for isolating regulatory mutants. Cells (BJ6252) bearing the UASPRB1-URA3 plasmid, pBJ7934, were mutagenized with 3% EMS. Survivors were then screened for 5-FOA and Prb phenotypes. By the HPA overlay plate assay, Prb- colonies are unable to solubilize the HPA particles, hence remain covered (). Prb+ colonies are uncovered due to solubilization of the particles and are surrounded by a clear halo ().

View larger version (27K): In this window In a new window Download PPT slide   Figure 2. pbn1-1 mutation affects PrB levels. (A) Prb phenotypes of PBN1 (BJ8594), pbn1-1 (BJ8596), pbn1-1/pPBN1 (BJ8596 transformed with pBJ8701), pbn1-1/pPRB1 (BJ8596 transformed with pBJ2465), prb1 (BJ5991 transformed with vector) and prb1/PRB1 (BJ5991 transformed with pBJ2465), as assessed by HPA overlay plate assay. (B) Steady-state levels of lumenal hydrolases. Whole cell extracts were prepared from PBN1 (BJ8594), pbn1-1 (BJ8596), pbn1-1/pPBN1 (BJ8596 transformed with pBJ8701), and prb1 (BJ5991), and subjected to SDS-PAGE (extract protein and acrylamide concentrations for each hydrolase were: CpY, 5 µg, 10% gel; PrA, 15 µg, 10% gel; PrB, 10 µg, 12% gel) followed by immunoblot analysis. (C) Whole cell extracts from PBN1 (BJ8594), pbn1-1 (BJ8596), pbn1-1/pPBN1 (BJ8596 transformed pBJ8701), and pbn1-1/pPRB1 (BJ8596 transformed with pBJ2465), were assayed for PrB activity using Azocoll as a substrate.

View larger version (24K): In this window In a new window Download PPT slide   Figure 3. The post-translational processing of Prb1p. (A) A schematic representation of the post-translational modifications and processing of Prb1p in the yeast secretory pathway. Solid circle and open circle represent O- and N-linked glycosylations, respectively. The molecular mass of each intermediate is shown. (B) Kinetic maturation of Prb1p in a PBN1 (BJ8594) and pbn1-1 mutant (BJ8596). Autoradiograph of proteins labeled with Trans-35S in vivo, immunoprecipitated with anti-Prb1p and separated by SDS-PAGE. Spheroplasts were labeled for 2 min, excess cold methionine and cysteine were added and spheroplasts were harvested at the times shown after the start of the chase. The numbers on the left indicate the apparent molecular mass of each intermediate. Lane 1 was loaded with labeled molecular weight (MW) markers (91, 67, 43, and 30 kD).

The amino-terminal propeptide of PrB is 260 amino acids long and represents about 40% of Prb1p. Inspection of the amino acid sequence of the propeptide shows a relatively high frequency of charged amino acids and the potential for a coiled-coil domain (amino acids 193 to 209 counting from the N terminus) in the propeptide. This N-terminal propeptide is an essential part of the structure; in its absence further processing of the precursor does not occur unless the missing peptide is supplied in trans (NEBES and JONES 1991 ).

Analogous to Prb1p, the yeast prohormone processing protease kexin precursor (Kex2p) contains a 77-residue propeptide that is cleaved in the ER by an intramolecular autocatalytic mechanism and requires a functional Kex2p active site for autocatalysis (GLUSCHANKOF and FULLER 1994 ). The propeptide is essential for the folding and maturation of Kex2p. Unlike Prb1p and Kex2p, the propeptides of PrA, ApI, and CpY are cleaved in the vacuole (JONES et al. 1997 ). This striking difference in the processing of Prb1p amd Kex2p suggests that the ER provides a unique environment that supports the autoprocessing event.

In this paper we report the identification of a novel essential gene PBN1 that is required for the post-translational processing of Prb1p. In a pbn1-1 mutant, Prb1p enters the ER lumen, but fails to undergo the autoprocessing. Furthermore, the unprocessed Prb1p is degraded in the cytosol. Pbn1p is a type I integral membrane glycoprotein of the ER that is lumenally oriented despite the lack of a signal sequence.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Materials:
Restriction enzymes and all other DNA-modifying enzymes were purchased from Boehringer Mannheim Biochemicals (Indianapolis), Promega (Madison, WI), or New England Biolabs (Beverly, MA). Taq DNA Polymerase was purchased from Fisher Scientific (Pittsburgh, PA). Lyticase L-8012, ß-glucuronidase G-7770, 3-aminotriazole, and Ponceau S were obtained from Sigma Chemical Company (St. Louis, MO). Protein A Sepharose CL4B was purchased from Pharmacia (Piscataway, NJ). 5-fluro-orotic acid (5-FOA) was obtained from Toronto Research Chemicals, Inc. (Toronto, Canada). Trans³⁵S was purchased from ICN Biochemicals (Costa Mesa, CA). Goat anti-rabbit IgG-HRPO conjugate was purchased from Bio-Rad (Hercules, CA). Nitrocellulose membranes "Optitran" type HA85 were obtained from Schleicher & Schuell (Keene, NH). Anti-HA 12CA5 monoclonal antibodies were purchased from Boehringer Mannheim Biochemicals. Antibodies to Sec61p and Kar2p were kind gifts from J. BRODSKY and M. ROSE, respectively. Fluorescein-conjugated goat anti-mouse IgG antibody and rhodamine-conjugated goat anti-rabbit secondary antibody were purchased from Molecular Probes (Eugene, OR). Tunicamycin, Azocoll, and Hide Powder Azure (HPA) were purchased from Calbiochem (La Jolla, CA). Oligonucleotide primers were obtained from Ransom Hill Bioscience (Ramona, CA).

Media and strains:
The strains and plasmids used in this work are listed in Table 1. Growth media used were YEPD, synthetic complete, synthetic medium containing galactose or raffinose as the carbon source, or omission media as previously described (JONES et al. 1982 ). Escherichia coli strains were grown in Luria Broth medium; for plasmid selection, ampicillin (100 µg/ml) was added (SAMBROOK et al. 1989 ). All plasmids were propagated in the strain LM1035 (CEPKO et al. 1984 ). All yeast strains were derived in our laboratory from strain X2180-1B (MAT gal2 SUC2) or from crosses between strains in our isogenic series and strains congenic to strain X2180-1B unless otherwise stated.

DNA manipulations:
Standard procedures for plasmid DNA isolations, restriction enzyme digestion, ligations, gel purification and transformation into bacteria were performed as described (SAMBROOK et al. 1989 ). Yeast transformation was performed using the lithium acetate/dimethyl sulfoxide procedure (HILL et al. 1991 ). Yeast genomic DNA was prepared by the method of HOFFMAN and WINSTON 1987 as was plasmid DNA isolated from yeast to be shuttled into E. coli. Polymerase chain reaction (PCR) was performed using standard methods in a Perkin-Elmer Cetus Thermal Cycler. DNA sequence was determined using the Sequenase Kit (U.S. Biochemicals, Cleveland, OH). DNA used for sequencing was prepared using Wizard Plus Minipreps DNA purification system (Promega, Madison, WI).

Genetic analysis:
Genetic crosses, sporulation, tetrad analysis and complementation analysis were performed using standard procedures as previously described (HAWTHRONE and MORTIMER 1960 ).

Mutagenesis:
Cultures of BJ6252 bearing pBJ7934 were grown to stationary phase in SC-Leu to a density of 8 x 10⁷ cells/ml. Cells were harvested and washed twice in 50 mM potassium phosphate buffer pH 7.0 and resuspended in 5 ml of the same medium. Cells were mutagenized using 3% (w/v) ethyl methanesulfonate (EMS) as described (PRESTON et al. 1992 ). The mutagenized cells were plated directly onto SC-leucine plates. About 18% of the cells survived this treatment. After three days incubation at 30°, the plates were replica plated to one plate of SC-leucine containing 5-FOA and a second containing YEP-Glycerol (YEPG). The colonies on YEPG were tested for PrB activity by means of the HPA overlay plate assay after two days of growth at 30° (see below). Colonies that were 5-FOA^R and Prb^- were chosen for further characterization.

HPA overlay plate assay for Protease B activity:
HPA overlay plate assay was performed as described by JONES 1991 .

Biochemical assays:
ß-galactosidase activity was assayed as described by KIPPERT 1995 . Activities are represented in Miller units (MILLER 1972 ). PrB activity in stationary phase cells was determined with Azocoll (Calbiochem, San Diego, CA) as a substrate as described by JONES 1991 ; specific activity is defined as (1.91)(A_520nm)/(mg protein/min) as previously described (JONES 1991 ).

Preparation of yeast whole cell extracts and immunoblotting:
Cell-free protein extracts were prepared and immunoblot analysis was carried out as described previously (NAIK et al. 1997 ). Affinity-purified rabbit anti-PrB open reading frame (ORF), anti-CpY and anti-PrA antibodies were used as the primary antibodies. Immune complexes were visualized according to the manufacturer's instructions by use of secondary goat anti-rabbit horseradish peroxidase conjugate activity (Bio-Rad, Hercules, CA).

Spheroplast labeling and immunoprecipitation:
Radiolabeling of spheroplasts and immunoprecipitations was carried out as described in WEBB et al. (1997). Labeling of temperature-sensitive strains required the following changes: (1) all steps of spheroplast formation were done at room temperature, (2) spheroplasts were preincubated at the restrictive temperature for 20–30 min prior to labeling, and (3) labeling was carried out at the restrictive temperature unless otherwise stated. If tunicamycin was used, it was added from a 1 mg/ml dimethyl sulfoxide stock to a final concentration of 20 µg/ml 15 min prior to adding the label. The immunoprecipitates were separated by SDS-PAGE and subjected to autoradiography.

Subcellular fractionation:
Subcellular fractionation by differential centrifugation was carried out as previously described (BECHERER et al. 1996 ). To determine the nature and association of HA::Pbn1p with membranes, cell extracts were prepared and processed as described by FELDHEIM et al. 1993 .

Cloning by complementation:
The wild-type PBN1 gene was cloned by complementation of the Prb^- phenotype. A CEN LEU2 based yeast genomic library (obtained from P. HIETER) was transformed into strain BJ8591 (pbn1-1 leu21 ura3-52 lys2-801) and transformants were selected on SC-leu plates. The plates were incubated at 30° for 3 days. Colonies were then replica-plated onto YEPG and incubated for 2 days at 30°; colonies were tested for PrB activity by means of the HPA overlay plate assay. Plasmids from colonies that gave a Prb⁺ phenotype were isolated. Plasmids were then shuttled into E. coli and back into yeast and retested for complementing activity. The plasmids were sequenced to define the ends of the insert DNA using primers MHCP506 (5'-CAGCAACCGCACCTGTG-3') and MHCP507 (5'-GACTACGCGATCATG-3') that flank the BamHI site of YCp50. The sequence obtained was compared to the yeast genome using the BLAST program. The full DNA sequence was obtained from the Saccharomyces Genome Database (SGD) (http://genome-www.stanford.edu/Saccharomyces/) and a restriction map of the region was generated using the DNA Strider program.

Plasmid constructions:
pBJ8701 was constructed as follows: Using the sequence obtained from the SGD, primers against regions approximately 513 nucleotides upstream (ORF52-2 5'-GGATTTGGATCCGATATTTGC-3') and 233 nucleotides downstream (ORF52-1 5'-GATGGTGTCTAGAAGAATG-3') of YCL052c were synthesized. PCR on genomic DNA yielded a 1.994-kb amplification product with restriction sites at each end introduced by the primers. The amplification product was digested with BamHI and XbaI, and the restriction fragment was gel purified and then cloned into the corresponding sites of pRS315. The BamHI-XbaI fragment from the resulting plasmid, pBJ8701, was cloned into other single- and high-copy plasmids of the pRS series (SIKORSKI and HIETER 1989 ).

The UAS_PRB1-URA3 reporter plasmid, pBJ7934, was constructed by cloning the 1520-bp upstream sequence of PRB1 into pBJ7115. Primers PRB1-UAS1 (5'-CCACGATCTAGACCCCGTTGTCC-3') and PRB1-UAS2 (5'-GTATTTTGGATCCTCATCTTTGC-3') were used to amplify the UAS fragment of PRB1. The PCR fragment was digested with BamHI and XbaI and cloned into the same sites of pBJ7115.

A disrupted allele of PBN1 (pBJ8828) was constructed by replacing the HindIII-NsiI fragment of PBN1 with the HIS3 gene. pBJ8705 was digested with HindIII-NsiI to drop out a 588-bp internal fragment of PBN1 and replaced with a HindIII-PstI HIS3 fragment from pBJ7143. To disrupt the PBN1 gene, the pbn1::HIS3 fragment was PCR amplified using primers ORF52-1 and ORF52-2 from pBJ8828, and was introduced into the chromosome of a diploid strain by homologous recombination. The transformants were plated onto SC-histidine. The integration events were confirmed by DNA blot analysis. Three different transformants were selected and sporulated.

To clone PBN1 into the dihybrid vectors (FIELDS and SONG 1989 ), an EcoRI-XhoI fragment from pBJ8706 was cloned into the EcoRI-XhoI and EcoRI-SalI sites of pACTII and pAS1-CYH2, respectively. The propeptide region of Prb1p was cloned into pAS1-CYH2 using a PCR fragment with NdeI and BglII restriction sites flanking the propeptide coding sequence; the PCR primers RN1-Pro (5'-AGCATCTCTGCTCATATGGTCATCCC-3') and VN7 (5'-CGAGACGCCTAAGGAA AG-3') were used to amplify the propeptide region from pBJ3637. The fragment obtained by digesting the amplified product with restriction enzymes NdeI and BglII was ligated into the NdeI and BamHI sites of pAS1-CYH2. Primers PRB1-RN2 (5'-GTCGAGGATCCAGAATTTGAC-3') and PRB1-RN3 (5'-CCGACTTGGATCCTCGAGAACG-3') were used to PCR amplify the catalytic domain from pBJ4680. The amplified product was digested with BamHI and XhoI and ligated into the same sites of pACTIINcoI. The full-length Prb1p lacking the first 13 amino acids was cloned into pACTII by ligating an AvrII-BamHI fragment from pBJ4780 into NcoI-BamHI sites of pACTII (pBJ9153). All fusion proteins could be detected in immunoblots.

Epitope tagging of PBN1:
Construction of an HA-tagged version of PBN1 was carried out as follows: an EcoRI site was introduced by PCR immediately before the first ATG codon of the PBN1 ORF. An XhoI site was introduced by PCR approximately 65 bp downstream of the stop codon. The PCR amplification product obtained by using the ORF52-HA1 (5'-GCTGAGAATTCGGTGACAAGA-3') and ORF52-HA2 (5'-GAAAACTCGAGTGCACAGTAG-3') primers was digested using EcoRI and XhoI, and the restriction fragment was ligated into the same enzyme sites in the HA-tagging vector pRD54 GAL1::HA in pRS316 (constructed by R. DESHAIES), yielding the plasmid pBJ8706 with a single copy of the HA epitope fused in frame at the 5' end of the PBN1 ORF; the ATG for the fusion is provided by the HA epitope. This HA-tagged version of PBN1 (pBJ8706) could be detected in immunoblots and also by immunoprecipitation from radiolabeled whole cell protein extracts; no corresponding protein band was observed in strains carrying the vector alone. The 670-bp GAL1 promoter in pBJ8706 was completely replaced with the native upstream region of PBN1 (without adding any extraneous sequence) as follows: the 540-bp region upstream of the PBN1 ORF was amplified from pBJ8701 using PCR primers ORF52HA-422 (5'-GCACATCTAGAGTATGCG-3') and ORF52HA+6 (5'-CACCATGGCTCAGCTAGAAACG-3') with restriction sites, XbaI and NcoI; the fragment was cloned into the corresponding sites of pBJ8706 resulting in the plasmid pBJ8707. For introduction of the triple HA tag, a 121-bp DNA fragment with flanking EcoRI sites encoding the triple HA epitope (provided by A. SRIVASTAVA) was digested with EcoRI and then ligated into the same site between the resident single HA tag and the first ATG of the PBN1 ORF in pBJ8707. Orientation of the triple tag was checked by PCR followed by testing the ability of the plasmid to complement pbn1-1. The plasmid pBJ9194 satisfied both criteria. However, the pBJ8707 and pBJ9194 versions of PBN1 could not be detected in immunoblots or by immunoprecipitation from radiolabeled whole cell protein extracts.

Immunofluorescence microscopy:
Immunofluorescence microscopy was essentially done as described by PRINGLE et al. 1991 . Cells were grown overnight in SC-uracil containing raffinose as the sole carbon source. The next day cells were inoculated into 25 ml of SC-uracil containing galactose as the sole carbon source and allowed to grow for 5–8 hr (OD_600nm = 0.5–1.0). In some cases, galactose induction was carried out for 16 hr, by growing the cells in galactose overnight and then transferring the cells into 25 ml of SC-uracil + galactose. Cells were collected and processed for immunofluorescence microscopy. The staining of HA-tagged Pbn1p was observed by using the 12CA5 monoclonal antibody (25 µg/ml) as the primary antibody. Decoration of 12CA5 was performed by the addition of 5 µg/ml fluorescein isothiocyanate-conjugated goat anti-mouse secondary antibody. Visualization of anti-Kar2p antibody was achieved by the addition of 10 µg/ml of rhodamine-conjugated goat anti-rabbit secondary antibody. For double-labeling experiments of Kar2p and HA::Pbn1p, permeabilized cells were incubated with both primary antibodies. After washing, the second incubation was done with 5 µg/ml fluorescein isothiocyanate-conjugated goat anti-mouse secondary and 10 µg/ml of rhodamine-conjugated goat anti-rabbit secondary antibody. As controls, reactions with only one of the primary antibodies were assayed to confirm that there was no leakage of emission between the two channels of observation. The cells were observed using a Nikon EFD-3 microscope.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Isolation of pbn1-1:
The primary objective of our genetic scheme was to identify genes that are involved in the transcriptional regulation of PRB1. Figure 1 shows the strategy that was used in selecting for mutants. A ura3-52 strain (BJ6252) was transformed with the plasmid, pBJ7934, bearing the URA3 gene fused to 1250 bp of PRB1 upstream sequence: UAS_PRB1-URA3 reporter plasmid. The Leu⁺ transformants bearing the URA3 promotor fusions are Ura⁺. This strain was mutagenesized using 3% EMS and plated on selective complete minus leucine (SC-Leu) plates. After three days, these plates were replica-plated to 5-FOA medium (SC-Leu + 5-FOA) to select for mutants that are unable to express the URA3 gene and to YEPG for the HPA overlay plate assay for PrB activity. Colonies that were Prb^- and FOA^R were most likely candidates for the mutations of interest in genes that are positive regulators of PRB1 expression. These mutants were chosen for further analysis. The next step was to exclude plasmid-borne mutations and mutations linked to the PRB1 locus. This was achieved by crossing the mutants to a prb1 ura3 strain (BJ5902). Diploids that were FOA^S and Prb⁺, indicating recessive mutation(s), were the desired mutants.

We identified one mutant colony that was FOA^R and Prb^- but the diploid obtained by mating to a prb1 ura3 strain gave a Prb⁺ 5-FOA^R phenotype. This suggested that this mutant harbored two mutations, one of which was genomic, the other plasmid-borne. This mutant was designated pbn1-1 (Protease B Negative) for its Prb^- phenotype as assessed by the HPA overlay assay (Figure 2A). The pbn1-1 mutant was cured and retransformed with fresh plasmid pBJ7934. This newly transformed pbn1-1 mutant was unable to grow on SC-Leu + 5-FOA, thus indicating that the 5-FOA^R phenotype of the original mutant had been due to a plasmid-borne mutation. The PRB1 gene, even on a 2µ plasmid, failed to complement the Prb^- phenotype of the pbn1-1 mutant; thus, increasing the copy number of the PRB1 gene fails to overcome the Prb^- phenotype of the pbn1-1 mutant (Figure 2A).

The pbn1-1 mutant was crossed to a wild-type strain of opposite mating type (BJ6253); the heterozygous diploid obtained was then sporulated. All 33 tetrads segregated 2 Prb⁺:2 Prb^-, indicating that pbn1-1 segregates as a single mutation.

pbn1-1 affects only PrB levels:
To determine the effects of the pbn1-1 mutation on the levels of PrB under steady-state conditions, we performed immunoblot analysis on whole cell extracts from pbn1-1. Under steady-state conditions, no mature PrB or processing intermediates were seen in the pbn1-1 mutant extract (Figure 2B). However, the levels of lumenal hydrolases protease A (PrA) and carboxypeptidase Y (CpY) were unaffected and the precursors were processed to the forms that accumulate in PrB-deficient cells. Furthermore little or no PrB activity was detected by the Azocoll assay in the mutant (Figure 2C). PrB activity in the pbn1-1 mutant was <10% that of its wild-type parent. Increasing PRB1 copy number in a pbn1-1 mutant did not lead to increased PrB activity. Thus the pbn1-1 mutation affects only PrB, but not PrA and CpY levels.

Post-translational processing defect in the pbn1-1 mutant:
Since the pbn1-1 mutant showed no processing intermediates or mature PrB accumulating under steady-state conditions, we performed a kinetic experiment to follow the maturation of PrB in the pbn1-1 mutant. We had earlier confirmed by Northern blot analysis that the pbn1-1 mutation does not affect the transcriptional levels of PRB1 (data not shown). As shown in Figure 3A, the PrB precursor undergoes extensive post-translational processing and modification as it traverses the secretory pathway.

In wild-type cells, Prb1p is completely processed through a series of intermediates to the mature 31-kD species within the 30-min chase period (Figure 3B). Under the same conditions, in the pbn1-1 mutant the 76-kD precursor was detected, but no other processed forms of Prb1p were observed. The 76-kD precursor was eventually degraded within the 30-min chase period. No evidence for autocatalysis to give the 39-kD and 40-kD species is evident; however, overexposure of the autoradiographs revealed a faint 40-kD band, but no other smaller intermediates were observed. This result indicates a defect in the intramolecular autoprocessing step of Prb1p in the pbn1-1 mutant. In contrast, the kinetic maturation of the PrA and CpY precursors in the pbn1-1 mutant was indistinguishable from that of wild-type strain (data not shown). Thus the Prb^- phenotype in the pbn1-1 mutant is due to a defect in the post-translational processing of Prb1p.

Translocation and N-glycosylation of Prb1p in the pbn1-1 mutant:
Although we were able to detect the 76-kD precursor in the pbn1-1 mutant, this observation does not differentiate between the cytosolic 76-kD precursor and the ER lumenal 76-kD precursor (see Figure 3A). Because the estimated molecular mass of signal sequence removed from Prb1p is similar to that of a single N-glycosyl side chain (about 2 kD), the cytosolic and the glycosylated ER-lumenal species of Prb1p are hard to distinguish by SDS-PAGE.

To determine whether Prb1p was cytosolic or lumenal in the pbn1-1 mutant, we treated spheroplasts with tunicamycin prior to metabolic labeling. Because tunicamycin inhibits synthesis of the glycosyl donor (KUO and LAMPEN 1974 ; LEHLE and TANNER 1976 ), proteins fail to be N-linked glycosylated in its presence. When the pbn1-1 mutant was treated with tunicamycin prior to radiolabeling and immunoprecipitation with anti-Prb1p, we observed an increase in the mobility of the Prb1p due to a decrease in molecular mass by about 2 kD (Figure 4). Prb1p in the untreated mutant migrated more slowly. Thus the Prb1p species seen in the pbn1-1 mutant is N-glycosylated.

View larger version (27K): In this window In a new window Download PPT slide   Figure 4. Prb1p is translocated into the ER and is N-glycosylated. Spheroplasts of PBN1 (BJ8594), pbn1-1 (BJ8596) sec62-1 (BJ9195), and sec62-1 pbn1-1 (BJ9196) were preincubated at 37° for 20 min and for an additional 10 min in the presence of tunicamycin (20 µg/ml) prior to metabolic labeling with Trans-35S. Spheroplasts were labeled for 2 min and chased with an excess of cold methionine and cysteine for an additional 5 min. The spheroplasts were lysed and the labeled proteins were immunoprecipitated with anti-Prb1p. As a control, wild-type cells were subjected to the same regimen.

Translocation of Prb1p into the ER lumen occurs post-translationally and requires Sec61p and Sec62p. The translocation of Prb1p is blocked in either a sec61 or sec62 mutant at the nonpermissive temperature, leading to an accumulation of the cytosolic, unglycosylated 76-kD precursor (MOEHLE et al. 1989 ; R. R. NAIK and E. W. JONES, unpublished results). The molecular mass of the untranslocated Prb1p that accumulates in a sec62-1 mutant is unaffected by treatment with tunicamycin. In a pbn1-1 sec62-1 double mutant the molecular mass of Prb1p was identical to that seen in a sec62-1 mutant, indicating that sec62 is epistatic to pbn1 (Figure 4). Furthermore, in the presence of tunicamycin, the unglycosylated Prb1p species in a pbn1-1 mutant is ~2 kD smaller than the untranslocated species in a sec62-1 mutant. This size difference is presumably due to the processing of Prb1p signal sequence, indicating that Prb1p was translocated in the pbn1-1 mutant. Thus translocation into the ER and N-glycosylation of Prb1p are unaffected in the pbn1-1 mutant.

Degradation of Prb1p in pbn1-1 is cytosolic:
In the pbn1-1 mutant, the 76-kD precursor is degraded during the 30-min chase. In principle the degradation of Prb1p in the pbn1-1 mutant could occur in any of three ways: (1) by secretion and degradation of Prb1p in the extracellular medium, (2) by transport to and degradation of Prb1p in the vacuole, or (3) by retro-translocation and degradation of Prb1p in the cytosol. We tested each of these three possibilities.

As shown in Figure 5A, we were unable to detect the 76-kD Prb1p in the extracellular fraction of the pbn1-1 mutant during the 30-min chase. Prb1p was retained in the intracellular fraction and degraded during the chase period. This indicates that the disappearance of 76-kD Prb1p in the kinetic experiments was not due to its secretion to the extracellular medium in the pbn1-1 mutant. In order to differentiate between the two pathways for intracellular degradation of Prb1p, vacuolar versus cytosolic, we constructed pep4::HIS3 pbn1-1 and sen3-1 pbn1-1 double mutants. Degradation of many short-lived proteins can be attributed to vacuolar proteolysis and is dependent on PrA, encoded by the PEP4 gene (JONES et al. 1997 ). Sen3p encodes a regulatory subunit of the 26S proteasome and is required for cytosolic ubiquitin-dependent protein degradation. The sen3-1 mutant shows severe defects in the degradation of ubiquitinated proteins (DEMARINI et al. 1995 ). If degradation is vacuolar, we expect the pep4::HIS3 pbn1-1 strain to show decreased degradation; if degradation is cytosolic via the proteasome, we expect the sen3-1 pbn1-1 strain to show decreased degradation.

View larger version (51K): In this window In a new window Download PPT slide   Figure 5. Degradation of Prb1p in the pbn1-1 mutant is cytosolic. (A) Spheroplasts of pbn1-1 (BJ8596) were labeled with Trans-35S for 2 min followed by addition of excess unlabeled methionine and cysteine, and samples were taken at the indicated times. Spheroplasts were lysed and labeled proteins were immunoprecipitated with anti-Prb1p from cell associated (I) and medium (E) fractions. The immunoprecipitated samples were subjected to SDS-PAGE and autoradiography. (B) Spheroplast of pbn1-1 (BJ8596), pbn1-1 pep4::HIS3 (BJ9048), and pbn1-1 sen3-1 (BJ9146) were labeled with Trans-35S for 2 min and subsequently chased for the indicated times. Spheroplasts were lysed and labeled proteins were immunoprecipitated with anti-Prb1p. The immunoprecipitated samples were subjected to SDS-PAGE and autoradiography.

In the pep4::HIS3 pbn1-1 double mutant, Prb1p was degraded at a rate comparable to that in the pbn1-1 mutant, suggesting that degradation of Prb1p in a pbn1-1 mutant is not dependent on vacuolar proteases (Figure 5B). In contrast, the rate of degradation of Prb1p was reduced in the sen3-1 pbn1-1 double mutant (note especially the 12- and 45-min points), and Prb1p could be immunoprecipitated from the extracts of the pbn1-1 sen3-1 mutant that had been chased for 45 min. The reduction of Prb1p degradation in the sen3-1 pbn1-1 double mutant suggests that the 26S proteasome is responsible for degradation of Prb1p in the pbn1-1 mutant. Therefore, these results suggest that Prb1p is exported back into the cytosol for degradation by the 26S proteasome in the pbn1-1 mutant.

Cloning of the PBN1 gene:
The wild-type PBN1 gene was cloned by complementation of the Prb^- phenotype of pbn1-1. The mutant strain BJ8591 was transformed with a CEN LEU2 based yeast genomic library. Transformants were selected on SC-Leu plates. The Leu⁺ transformants were replica plated to YEPG for the HPA overlay plate assay for complementation of the Prb^- phenotype. Of 13,000 colonies screened by the HPA assay, two colonies gave a Prb⁺ phenotype. The library plasmids recovered from these transformants complemented the pbn1-1 mutant upon shuttling and proved to be identical. DNA sequencing of the ends of the inserted yeast sequences in clone pC1c revealed an 11-kb fragment from the left arm of chromosome III, containing six complete ORFs (Figure 6). Subcloning of this region indicated that the plasmids containing YCL052c were able to complement the Prb^- phenotype of pbn1-1. Using the sequence obtained from the SGD, the PBN1 gene was cloned by genomic PCR. Primers were designed such that the amplified insert contained sufficient upstream (625-bp) and downstream (233-bp) regions to ensure proper expression of the gene. Restriction sites were introduced into the primers to facilitate cloning of the amplification product into a yeast plasmid vector pRS315. The resulting plasmid pBJ8701 was able to complement the Prb^- phenotype of the pbn1-1 mutant as examined by the HPA plate assay (Figure 2A). In addition, restoration of PrB activty in a pbn1-1 mutant transformed with pBJ8701 was confirmed biochemically (Figure 2C).

View larger version (14K): In this window In a new window Download PPT slide   Figure 6. Cloning of PBN1. Physical maps of the constructed plasmids and their ability to complement the Prb- phenotype of a pbn1-1 mutant (BJ8586). The original complementing clone, pC1c showing the location of YCL052c on chromosome III. pBJ8701 carries the smallest fragment that complements the pbn1-1 mutation. The region from nucleotide + 144 (ATG is the reference point) to + 732 of the ORF was deleted and replaced with the HIS3 gene. The abbreviations are as follows: B, BamHI; C, ClaI; H, HindIII; N, NcoI; S, NsiI; X, XbaI. Parentheses indicate restriction enzyme sites engineered by PCR.

To determine whether the PBN1 gene corresponds to YCL052c, YCL052c was cloned into an integrating vector, pRS305, carrying the LEU2 marker. The resulting plasmid, pBJ8702, was linearized with NsiI to direct integration to the YCL052c chromosomal locus of a wild-type strain (BJ6253). This integrant was then crossed to the pbn1-1 mutant. All 20 tetrads obtained from sporulating the heterozygous diploid segregated 2 Leu⁺ Prb⁺: 2 Leu^- Prb^-, indicating tight linkage of LEU2 and the wild-type allele for pbn1-1, namely PBN1. Thus, YCL052c is the bona fide PBN1 gene.

The pbn1-1 allele was recovered from BJ8591 by gap repair. The gap repaired plasmid was sequenced and compared to the wild-type allele. The pbn1-1 allele contains a single amino acid substitution, a glycine (GGA) to arginine (AGA) change at position 182 of the Pbn1p amino acid sequence.

Features of Pbn1p:
The PBN1 (YCL052c) ORF encodes an uncharacterized protein of 416 amino acids with a predicted M_r of 48 kD (Yeast Proteome Database). Hydropathy analysis by the method of KYTE and DOOLITTLE 1982 revealed a 19-amino-acid (residues 386–404), sub-C-terminal hydrophobic region predicted to be a transmembrane domain (Figure 7A). The region between amino acids 62 and 78 is predicted to adopt an -helical coiled-coil conformation (Figure 7B). The probability that this region of Pbn1p would form coiled-coil was calculated using the COILS 2.1 program (MCLACHLAN and STEWART 1975 ; MCLACHLAN and KARN 1982 ; COHEN and PARRY 1986 ; LUPAS et al. 1991 ). Pbn1p also contains six potential N-glycosylation sites (Asn24, Asn85, Asn120, Asn212, Asn365 and Asn386) that lie within the portion of the polypeptide N-terminal to the predicted transmembrane domain (Figure 7C). Pbn1p does not contain a clearly defined signal sequence. The C-terminal tail of Pbn1p is highly charged and contains several lysine residues. No other significant similarity to any other proteins or defined functional motifs is evident in the Pbn1p sequence.

View larger version (26K): In this window In a new window Download PPT slide   Figure 7. Features of Pbn1p. (A) Hydropathy plot of the predicted Pbn1p polypeptide. (B) Coiled-coil formation probability generated using COILS 2.1 program (MTIK matrix, a and d positions weighted, 14-residue window). (C) Cartoon of Pbn1p, a 416-amino-acid polypeptide with an N-terminal coiled-coil domain (), a sub-C-terminal transmembrane domain (), and the six predicted N-glycosylation sites (*).

PBN1 is an essential gene:
We constructed a pbn1::HIS3 allele in which amino acids 40–243 of the PBN1 coding region were replaced with the HIS3 marker (Figure 6). This allele was introduced into the BJ6252 x BJ5414 diploid strain, and a heterozygous diploid for this allele was constructed by the one-step gene disruption method. Sporulation of the heterozygous diploid gave tetrads (n = 24) with only two viable spore clones/tetrad. All surviving spore clones were His^- Prb⁺. The sporulation of the heterozygous diploid bearing PBN1 on a plasmid carrying URA3 as the selectable marker (pBJ8827) yielded viable spore clones that were His⁺ Prb⁺ Ura⁺; however, these spore clones were unable to grow on selective medium containing 5-FOA. Therefore the PBN1 gene is essential for vegetative growth.

ER-localization of epitope-tagged Pbn1p:
To determine the intracellular localization of Pbn1p, an epitope-tagged version of Pbn1p was constructed by introducing the HA epitope at the N terminus of the protein coding sequence (pBJ8706). This modified HA::Pbn1p was functional, as it was able to complement the Prb^- phenotype of pbn1-1. When the functional HA::Pbn1p was overproduced via the GAL1 promoter and localized by immunofluorescence using a monoclonal antibody (12CA5) to the HA epitope, a distinctive ER staining pattern was seen, typified by the perinuclear staining and staining around the plasma membrane (Figure 8C). This staining pattern resembles that found for Kar2p, an ER lumenal chaperone. Double-labeling experiments with Kar2p, an ER protein, and HA::Pbn1p indicated that the two proteins colocalize. The induction of pBJ8706 for greater than 16 hr in the presence of galactose led to extensive proliferation of the ER membranes (Figure 8D), a feature observed for some other ER proteins, such as HMG-CoA reductase and cytochromes when overexpressed (WRIGHT et al. 1988 ; VERGERES et al. 1993 ).

View larger version (118K): In this window In a new window Download PPT slide   Figure 8. Immunofluorescence localization of overexpressed HA::Pbn1p. pbn1-1 (BJ8596) cells carrying pBJ8706 were grown in selective medium containing galactose rather than glucose as carbon source. Cells were fixed and processed for immunofluorescence using antibodies against the HA epitope and Kar2p, an ER marker protein. (A) DAPI staining to localize the nucleus; (B) immunofluorescence due to Kar2p; (C) immunofluorescence of HA::Pbn1p induced for 6 hr; (D) HA::Pbn1p induced for 16 hr. No immunofluorescence was observed in pbn1-1 cells carrying the vector alone.

Attempts to visualize the HA::Pbn1p fusion when expressed at normal levels (on a CEN plasmid under the control of its native promoter) or even after the insertion of three additional epitopes proved unsuccessful. In addition antibodies raised against Pbn1p detect the Pbn1p antigen in immunoblots only when the protein is overexpressed. One possible explanation for this observation is the low level of expression of PBN1; ß-galactosidase activity levels expressed from a UAS_PBN1-lacZ reporter fusion are very low (data not shown).

Pbn1p is an integral membrane protein:
In conjunction with the results obtained by indirect immunofluorescence, we further confirmed the ER localization of HA::Pbn1p by subcellular fractionation. Cell-free extracts were subjected to differential centrifugation to separate the various organelles. Lysates were fractionated into a 13,000 x g pellet (P13), a 100,000 x g pellet (P100) and a 100,000 x g supernatant (S100). The majority of the 60-kD HA::Pbn1p was found in the P13 membrane fraction (Figure 9A). As a control, nearly all of the ER integral membrane protein Sec61p was also found in the P13 fraction, which is consistent with its being an ER-membrane-associated protein (STIRLING et al. 1992 ).

View larger version (40K): In this window In a new window Download PPT slide   Figure 9. Subcellular fractionation of Pbn1p. (A) pbn1-1 cells (BJ8596) carrying pBJ8706 or vector alone were grown in the presence of galactose to OD600nm between 0.5 and 0.8 and converted to spheroplasts. The spheroplasts were lysed and fractionated by differential centrifugation. The lysate was centrifuged at 13,000 x g, for 15 min yielding a pellet fraction (P13) and supernatant fraction (S13). The S13 fraction was subjected to a second centrifugation at 100,000 x g for 30 min resulting in a pellet fraction (P100) and a supernatant fraction (S100). Fractions were separated by SDS-PAGE followed by immunoblot analysis using 12CA5 against HA::Pbn1p and affinity-purified anti-Sec61p. Empty vector served as negative control. (B) Membrane fractions were prepared and treated with either buffer, 0.1 M Na2CO3 (pH 11), 0.5 M NaCl, 0.8 M urea, or 1% Triton X-100. After incubation on ice for 20 min, all samples were separated into supernatant (S) or pellet (P) fractions by centrifugation at 100,000 x g for 30 min, subjected to SDS-PAGE, and immunoblotted with anti-HA (12CA5).

The hydropathy analysis of Pbn1p shown in Figure 7 revealed a single 19-amino-acid sub-C-terminal hydrophobic region that would be of sufficient length to span a lipid bilayer. In addition, PredictProtein program (www.emblheidelberg.de/predictprotein/predictprotein.html) predicted with high probability that amino acids 386–404 will form a transmembrane domain. To determine whether Pbn1p is an integral membrane protein, the P13 pellet was extracted with 0.1 M Na₂CO₃ pH 11, 0.5 M NaCl, 0.8 M urea, or 1% Triton X-100. As only the 1% Trition X-100 treatment extracted Pbn1p from the pellet, we infer that Pbn1p is an integral membrane protein (Figure 9B).

Pbn1p topology based on glycosylation:
The predicted molecular mass of Pbn1p is 48 kD, whereas the protein migrates as a 60-kD species. As one possible source of the increased mass might be glycosylation, we examined expression of HA::Pbn1p in the presence or absence of tunicamycin. The cell extracts were immunoprecipitated with anti-HA (12CA5), resolved by SDS-PAGE and visualized by autoradiography. Immunoprecipitates from cells expressing HA::Pbn1p in the absence of tunicamycin contained the 60-kD species (Figure 10). Treatment of cells with tunicamycin resulted in the appearance of a faster migrating band with a molecular mass of 50 kD, quite close to the calculated 48 kD. Thus Pbn1p appears to carry at least four glycosyl side chains. All the six potential N-glycosylation sites are within amino acids 64–386, N-terminal to the transmembrane domain. Therefore, taken together these data indicate that the amino terminus lies within the ER lumen despite the lack of a signal sequence.

View larger version (24K): In this window In a new window Download PPT slide   Figure 10. Pbn1p is N-glycosylated. pbn1-1 cells (BJ8569) carrying pBJ8706 were grown in Wickeram's medium containing galactose as the carbon source for 6 hr (OD600 = 0.6). Spheroplasts were preincubated in the presence of tunicamycin (20 µg/ml) for 10 min and then labeled with Trans-35S for 10 min, followed by a 10 min chase with an excess of unlabeled methionine and cysteine. Spheroplasts were lysed and labeled proteins were immunoprecipitated with anti-HA. The immunoprecipitated samples were subjected to SDS-PAGE and autoradiography. pbn1-1 carrying empty vector was used as the negative control.

A sec61 mutant accumulates unglycosylated Pbn1p:
The experiments on the characterization of Pbn1p suggested that it is an ER integral membrane protein. Moreover, the amino terminus is lumenally oriented, despite lacking a signal sequence. The unusual topology of Pbn1p beckoned us to determine whether its insertion into the ER required known components of the translocation machinery. We determined whether the incorporation of Pbn1p into the ER membrane was blocked in sec mutants by making use of temperature-sensitive alleles of SEC61, SEC62, and SEC65 to block translocation at the restrictive temperature. Sec61p is a key component of the translocation apparatus, and mutations in the SEC61 gene affect the translocation of all proteins tested (STIRLING et al. 1992 ). Sec62p is required for post-translation translocation, while Sec65p is required for cotranslational translocation of proteins (DESHAIES and SCHEKMAN 1987 ; STIRLING and HEWITT 1992 ). pBJ8706 was transformed into the sec mutants and the effect of these mutations on the translocation of Pbn1p into the ER was determined (Figure 11). A block in the translocation of Pbn1p should result in the accumulation of the 50-kD unglycosylated form. Mutations in SEC62 and SEC65 had no effect on the translocation of Pbn1p; because it migrated as a 60-kD species, it had received its four N-glycosyl side chains. In contrast, an unglycosylated, presumably cytosolic, 50-kD species of Pbn1p can be seen in the sec61 mutant at the non-permissive temperature. Thus, introduction of a translocation block by sec61 prevents Pbn1p from receiving its N-glycosyl side chains. This result indicates that Sec61p plays a role in the insertion of Pbn1p into the ER membrane; Sec62p and Sec65p are most likely not involved.

View larger version (53K): In this window In a new window Download PPT slide   Figure 11. Translocation of Pbn1p is blocked in the sec61-2 mutant. Wild-type (BJ8594) pbn1-1 (BJ8569), sec61-2 (BJ3947), sec62-1 (BJ6879), and sec65-1 (BJ6988) strains carrying pBJ8706 were grown at 23° in Wickeram's medium containing galactose as the carbon source for 6 hr (OD600 = 0.6). Spheroplasts were preincubated in the presence of tunicamycin (20 µg/ml) for 10 min at 37° followed by an additional 30 min incubation at 37°. Spheroplasts were labeled with Trans-35S for 10 min, followed by a 10 min chase with an excess of unlabeled methionine and cysteine. Spheroplasts were lysed and labeled proteins were immunoprecipitated with anti-HA. The immunoprecipitated samples were subjected to SDS-PAGE and autoradiography. The untranslocated Pbn1p species is indicated by an asterisk.

The pbn1-1 allele does not affect Kex2p processing:
Kex2p also undergoes an autoproteolytic cleavage of its propeptide in the ER, similar to Prb1p. We decided to test whether the pbn1-1 mutation affected the autoprocessing of Kex2p. In wild-type cells, the Kex2p precursor undergoes cotranslational signal sequence cleavage and addition of N-linked and O-linked modifications in the ER leading to the 129-kD I₁ form. The amino-terminal propeptide is cleaved, yielding in the ER, a 120-kD I₂ form. During the later stages of its lifetime, Kex2p undergoes elaborate modifications of its O-linked chains in the Golgi, causing a gradual increase in the molecular weight to an approximately 126-kD J form (WILCOX and FULLER 1991 ). In the wild type, processing of the 129-kD I₁ form to the 126-kD J form, was observed within the 30-min chase period (Figure 12). The pbn1-1 mutant showed similar kinetics for the maturation of Kex2p. These data suggest that the pbn1-1 allele does not affect the autocatalytic processing of Kex2p.

View larger version (33K): In this window In a new window Download PPT slide   Figure 12. Kinetic maturation of Kex2p in PBN1 (BJ8594) and pbn1-1 (BJ8596) strains. Spheroplasts were labeled for 2 min, excess cold methionine and cysteine were added and spheroplasts were harvested at the times shown after the start of the chase. The autoradiograph is of proteins labeled with Trans-35S in vivo, immunoprecipitated with anti-Kex2p, and separated by SDS-PAGE.

Pbn1p interacts with the propeptide of Prb1p:
The feature that distinguishes the processing of Prb1p from that of other vacuolar protease precursors is the autoproteolytic processing of its propeptide in the ER. The ER localization of Pbn1p and the effect of the pbn1-1 mutation on the autocatalytic processing of Prb1p in the ER led us to determine whether Pbn1p and Prb1p interact with each other. Attempts to coimmunoprecipitate HA::Pbn1p and Prb1p were unsuccessful. Based on the predicted amino acid sequences, Prb1p contains four cysteine residues, three in the catalytic domain and one in the propeptide, whereas Pbn1p contains five cysteine residues. Immunoprecipitations carried out in the absence of reducing agents did not reveal any high molecular mass species, indicating that the Pbn1p and Prb1p are not covalently linked by disulfide bridges (data not shown). However, a transient interaction between Pbn1p and Prb1p would most likely not be detected using the above methods. As seen in Figure 3B, the 76-kD precursor was completely processed to the 40-kD species within 2 min of chase, indicating that the autocatalytic event is rapid.

We used the two-hybrid method to test for interaction between the pro regions of Prb1p and Pbn1p (FIELDS and SONG 1989 ; DURFEE et al. 1993 ). We used the DNA-binding domain of Gal4p fused to the propeptide of Prb1p and the Gal4p activation domain fused to Pbn1p and tested for activation of two reporter genes fused to the GAL1_UAS:lacZ and HIS3. The results of these experiments are presented in Figure 13. The propeptide and catalytic domain of Prb1p were individually tested for interaction with Pbn1p (see Figure 13A). In addition, we tested for the interaction between the propeptide and catalytic domain of Prb1p as a control, since previous results indicated that the propeptide interacts with the catalytic domain in trans in vivo (NEBES and JONES 1991 ). As shown in Figure 13B, when both the propeptide and catalytic domain were present in the diploid, 82 units of ß-galactosidase activity were obtained (sector E). The positive control Snf1p and Snf4p gave a ß-galactosidase activity of 47 units (sector A). Both diploids were able to grow on synthetic medium containing 25 mM 3-aminotriazole (3-AT). Thus both reporter genes were activated. The propeptide failed to interact either with vector containing only GAL4 sequences (data not shown) or with the Snf1p control (sector B). These observations indicate that the propeptide and the catalytic domain strongly interact with each other. These observations support the earlier observations and, futhermore, confirm the specificity of the interaction.

View larger version (48K): In this window In a new window Download PPT slide   Figure 13. Two-hybrid interaction between the propeptide of Prb1p and Pbn1p. (A) Schematic representation of the fusion plasmids used in the two-hybrid experiments. (B) Plasmids present in the diploid (Y187 x Y190) are indicated for each sector (pACT, GAL4 activation domain; pAS1, GAL4 DNA = binding domain). (C) Growth of diploids bearing the different plasmid combinations on SC minus Trp, Leu media, and on SC minus Trp, Leu, His media containing 25 mM 3-AT. The levels of ß-galactosidase activity (presented in Miller units) produced by plasmid combinations are indicated around the outside of the circle.

Interestingly, the propeptide was also capable of interacting with Pbn1p. The interaction between Pbn1p and propeptide resulted in 36 units of ß-galactosidase activity and growth of the diploid bearing plasmids encoding the Pbn1p and propeptide fusions on synthetic medium containing 25 mM 3-AT (sector C). This interaction between the propeptide and Pbn1p was specific, because the negative control Snf4p (sector D) or vector alone did not interact with Pbn1p (data not shown). No interaction between the catalytic domain and Pbn1p was observed; <0.5 units of ß-galactosidase activity and no growth of a diploid bearing plasmids encoding the Pbn1p and catalytic domain fusions on synthetic medium containing 25 mM 3-AT was seen (Sector G). Thus the activation of lacZ and HIS3 genes was not due to fortuitous activation by either protein fusion alone or by nonspecific interactions. Thus our results from the two-hybrid experiments indicate that Pbn1p specifically interacts with the propeptide domain of Prb1p in vivo.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

In this study we report the identification and characterization of Pbn1p, an ER membrane protein that is involved in the post-translational processing of the vacuolar protease B precursor. Our results using the two-hybrid system demonstrate that Pbn1p interacts with the propeptide of Prb1p.

The propeptide acts as an intramolecular chaperone that aids in the folding and maturation of Prb1p. It prevents N-glycosylation of the catalytic domain and allows the precursor to be processed by PrA, presumably by catalyzing a conformational change (NEBES and JONES 1991 ). In the pbn1-1 mutant we observed little or no appearance of the 40-kD form of PrB, a product that results from the autocatalytic cleavage event. Thus Prb1p fails to undergo its autoprocessing in the ER. The unprocessed precursor polypeptide is exported to the cytosol, where it is eventually degraded by the 26S proteasome. Since recent evidence suggests that misfolded secretory proteins are exported from the ER to the cytosol for degradation (HILLER et al. 1996 ; WERNER et al. 1996 ; SOMMER and WOLF 1997 ), the export of Prb1p into the cytosol for degradation in the pbn1-1 mutant suggests that it is recognized as a misfolded protein by the ER quality control mechanism.

The failure of Prb1p to be autoproteolytically processed in the ER of the pbn1-1 mutant suggests that the precursor polypeptide has not assumed the correct conformation that favors autocatalysis. No autoproteolytic processing was observed when PRB1 was translated in vitro using cell-free extracts or in vivo when translocation into the ER was blocked in a sec61 mutant and when PRB1 is expressed in E. coli (MECHLER et al. 1988 ; MOEHLE et al. 1989 ). The ER might provide factor(s) that aid Prb1p in attaining the proper conformation for autoprocessing. The evidence presented here clearly suggests that in order for the autocatalytic cleavage to occur in the ER, the propeptide must interact with Pbn1p to achieve a conformation that allows autocatalysis and prevents export to the cytosol from the ER.

The mechanism for the autoprocessing of the bacterial subtilisin propeptide requires that the C terminus of the propeptide must be in close proximity to the active site triad in the catalytic domain. Upon attaining this conformation, the precursor polypeptide is able to undergo its autoproteolytic cleavage (SHINDE and INOUYE 1995 ). Therefore, because of the similarity of Prb1p to bacterial subtilisins and because autocatalysis is known to occur in the ER (MOEHLE et al. 1989 ), Prb1p presumably undergoes a similar conformation change in the ER. Once inside the ER, Prb1p is capable of attaining an autoproteolytically active conformation with the help of Pbn1p, and the precursor is transiently activated. The fact that the Prb1p propeptide interacts with Pbn1p by the two-hybrid assay raises the possibility that the autoproteolytically active conformation may be achieved through this interaction. The interaction between the propeptide and Pbn1p might prevent the escape of free propeptide from the ER after autocatalysis, thereby allowing the catalytic domain to interact with the propeptide in trans: possibly this interaction leads to silencing of the activated protease in the ER. It is known that the coexpression of the propeptide and the catalytic domain as separate molecules restores PrB activity in a prb1 mutant (NEBES and JONES 1991 ), but does not do so in the pbn1-1 mutant (data not shown). This observation suggests that Pbn1p may provide an additional function downstream of the autocatalytic event.

Our favored model to account for the observations presented in this paper is that Pbn1p contributes to the autoprocessing of Prb1p by interacting with the propeptide in the ER to drive Prb1p into an autoproteolytically competent state (Figure 14). We propose that the propeptide of Prb1p interacts with Pbn1p as it emerges on the lumenal side of ER. After the entirety of Prb1p has been completely translocated into the ER, the propeptide is autocatalytically cleaved from the precursor. Once autocatalysis has occurred, the free propeptide bound to Pbn1p is transferred to the catalytic domain. The two-hybrid results presented in this article suggest that the catalytic domain strongly interacts with the propeptide in trans. This interaction could lead to silencing of the protease and prevent premature activation during transit through the secretory pathway. Possibly it is this interaction that effects the conformational change that allows PrA to process the C-terminal end of the precursor in the vacuole. We know that the catalytic domain leaves the ER and undergoes further modifications and processing prior to its arrival at the vacuole (NEBES and JONES 1991 ; HIRSCH et al. 1992 ), and that the propeptide is subsequently degraded by proteases in the vacuole (HIRSCH et al. 1992 ). The proposed interaction between the propeptide and the 39-kD intermediate raises the possibility that targeting to the vacuole by one of the two interacting peptides may be by "piggybacking" one polypeptide on the other, that one vacuolar targeting signal suffices for both polypeptides.

View larger version (12K): In this window In a new window Download PPT slide   Figure 14. Proposed model for Pbn1p function. Prb1p enters the ER post-translationally (1), the propeptide interacts with Pbn1p as it enters the lumen (2). The autocatalytic cleavage of the propeptide occurs after Prb1p is completely translocated into the ER (3). After autocatalysis, the propeptide is transferred from Pbn1p to the catalytic domain (4). The propeptide-catalytic domain complex is transported out of the ER for further processing (5).

Although localization studies using overproduced proteins must be interpreted with caution, the facts that the pbn1-1 mutation affects the ER-dependent processing of Prb1p and that Prb1p is recognized as a misfolded protein resulting in export from the ER for degradation by the proteasome clearly support an ER localization of Pbn1p. In addition, the fact that PBN1 is an essential gene (PRB1 is not) clearly indicates that Pbn1p is not on the branch pathway between the Golgi and the vacuole (no genes for this branch are essential to date) and accords with a function for Pbn1p in the central secretory pathway.

The analysis of Pbn1p function presented here has used a single pbn1 mutation. The requirement of PBN1 for cell survival suggests that Pbn1p plays a more global role and that there are likely to be additional interacting partners. Attempts to isolate conditional alleles of PBN1 using conventional methods have so far been unsuccessful. Further analysis of Pbn1p and its interacting proteins and analysis of additional alleles of PBN1 should allow us to more definitively determine the role of Pbn1p in the ER.

Degradation in the cytosol is dependent on large proteolytic complexes, the 26S proteasomes, which are responsible for the majority of the protein turnover in the cytosol (DEMARINI et al. 1995 ; HILLER et al. 1996 ; PLEMPER et al. 1997 ; SOMMER and WOLF 1997 ). Our results clearly indicate that degradation of Prb1p in the pbn1-1 mutant involves the proteasome, since degradation is greatly reduced in the sen3-1 pbn1-1 mutant. A prerequisite for degradation of at least some aberrant proteins by the proteasome is that the exported proteins be ubiquitinated (HILLER et al. 1996 ). Two ER membrane-bound ubiquitin-conjugating enzymes, Ubc6p and Ubc7p, have been identified in yeast. The deletion of UBC6 and UBC7 leads to a defect in the degradation of some aberrant proteins as well as in the degradation of MAT2 and Hmg2p (reviewed in CHEN et al. 1993 ; HAMPTON and BHAKTA 1997 ; SOMMER and WOLF 1997 ). However, WERNER et al. 1996 , demonstrated that ER-associated degradation of a mutant form of human -1-proteinase inhibitor, A1PiZ, a known substrate of ER-associated degradation in yeast and mammalian cells, was independent of the Ubc6p and Ubc7p ubiquitin-conjugating enzymes. This observation suggests that either ubiquitination may occur through the activity of other ubiquitin-conjugating enzymes or grossly misfolded proteins might not require ubiquitination for degradation by the proteasome. We observed a moderate reduction in the rate of Prb1p degradation in a ubc6 ubc7 pbn1-1 triple mutant, but it was not as dramatic as that observed in a sen3-1 pbn1-1 mutant (data not shown).

Pbn1p is classified as a type I membrane protein with a large lumenal domain and a short cytoplasmic domain despite the absence of an N-terminal signal sequence. Ordinarily, type I integral membrane proteins of the ER, like Wbp1p and Ost1p, contain an N-terminal signal sequence (KARAOGLU et al. 1995 ). It is predicted that signal-anchor proteins assume a type I membrane (N_lumen-C_cyt) topology when the segment C-terminal to the signal-anchor is positively charged with respect to the N-terminal flanking region (HAEUPTLE et al. 1989 ; HARTMANN et al. 1989 ). This is the case for Pbn1p. We have demonstrated that Pbn1p is an integral membrane protein with a lumenal N-terminal domain; therefore, the sub-C-terminal hydrophobic domain of Pbn1p is presumed to serve as a membrane anchor and the positively charged segment C-terminal to the transmembrane domain to be a cytoplasmic tail. These results accord well with predictions.

The insertion of membrane proteins is believed to occur via the same machinery used for translocation of secretory proteins, but the mechanism is poorly understood. In vitro experiments using proteins with a single membrane anchor (signal-anchor proteins) have shown that the hydrophobic segment can be cross-linked to Sec61p, suggesting that insertion of proteins into the ER membrane is dependent on Sec61p (MARTOGLIO et al. 1995 ). Proteins are translocated into the ER through a protein channel (the translocon pore). Sec61p is the pore-forming subunit of the protein translocation channel in the ER membrane (HANEIN et al. 1996 ). The insertion of Pbn1p into the ER membrane is dependent on Sec61p but independent of Sec62p and Sec65p function. Since signal recognition particle (SRP) interacts with the signal sequence of a nascent polypeptide that is still bound to the ribosome (STIRLING and HEWITT 1992 ; RAPOPORT et al. 1996 ), the absence of an N-terminal signal sequence in Pbn1p might account for the lack of dependence on Sec65p, an SRP component. It is unlikely that Pbn1p is inserted cotranslationally into the ER membrane, because it does not contain a signal sequence and also because its membrane anchor is so close to the C terminus that it would still be buried in the ribosome when termination of translation occurred; therefore membrane insertion must be post-translational. The translocation of Pbn1p remained unaffected in a sec62 mutant at the nonpermissive temperature; Sec62p is involved in the post-translational translocation of proteins into the ER (DESHAIES and SHEKMAN 1987). The C-terminal tail-anchored membrane proteins that possess an N-terminal cytoplasmic domain and that do not contain hydrophobic signal se