Yol082p, a Novel CVT Protein Involved in the Selective Targeting of Aminopeptidase I to the Yeast Vacuole*

The yeast vacuolar enzyme aminopeptidase I (API) is synthesized in the cytoplasm as a precursor (pAPI). Upon its assembly into dodecamers, pAPI is wrapped by double-membrane saccular structures for its further transport within vesicles that fuse with the vacuolar membrane and release their content in the vacuolar lumen. Targeting of API to the vacuole occurs by two alternative transport routes, the cvt and the autophagy pathways, which although mechanistically similar spe-cifically operate under vegetative growth or nitrogen starvation conditions, respectively. We have studied the role of Yol082p, a protein identified by its ability to interact with API, in the transport of its precursor to the vacuole. We show that Yol082p interacts with mature API, an interaction that is strengthened by the amino extension of the API protein. Yol082p is required for targeting of pAPI to the vacuole, both under growing and short term nitrogen starvation conditions. Absence of Yol082p does not impede the assembly of pAPI into dodecamers, but precludes the enclosure of pAPI within transport vesicles. Microscopy studies show that during pYOL082 D C32 was derived from pYOL082 D C32-GFP by excising a Sal I- Sal I fragment containing the GFP coding region. The resulting plasmids, pGAL1-YOL082, pGFP-YOL082, pYOL082 D C32-GFP, and pYOL082 D C32, were used to transform the BY4741, D yol082 , and D apg12 strains using the lithium acetate procedure (14). The bait plasmid encoding precursor API fused in frame to the Gal4p DNA-binding domain was constructed as follows: the API open reading frame was PCR-amplified from pEMBLyex4-API (5) with a forward primer, 5 9 -GGTACCCCGGGAACCTCTAGA-3 9 , which incorporated an Sma I site, and a reverse primer, 5 9 -CGAATTCATCGTCGACT-GATCTCC-3 9 , which incorporated a Sal I site. The Sma I- Sal I fragment was then cloned into pGBT9 (CLONTECH) to create pGBT9-pAPI (ami-no acids 1–507). The API region encoding the prepro-amino extension (amino acids 1–44) was cloned using the same forward primer and a reverse primer, 5 9 -GTGCTCGAGGATCCACCACG-3 9 , creating a Bam HI site. The corresponding Sma I- Bam HI fragment was inserted into pGBT9 to generate pGBT9-preproAPI. The mature API coding region (amino acids 54–514) was amplified with a forward primer, 5 9 -TATGAGGATAGTCGACAGGAATTC-3 9 , generating a S al I site and a reverse universal primer; the resulting PCR product was cloned into pUC18, and a S al I- Pst I fragment was cloned into pGBT9 to give pGBT9-mAPI. To construct pACT2-YOL082 expressing the complete Yol082p fused in frame to the Gal4p activation domain, the YOL082 coding region was amplified from TOPO-YOL082 interactions

In the yeast Saccharomyces cerevisiae the vacuolar hydrolase leucine aminopeptidase I (API) 1 is synthesized in the cytoplasm as a precursor (pAPI) (1) and delivered to the vacuole by one of two alternative routes that operate under distinct physiological conditions: the cytoplasm to vacuole targeting (Cvt), in nutrient-rich conditions, and the autophagy (Apg) pathway, under starvation conditions (2). The Cvt pathway is constitutive and biosynthetic, while autophagy is nonselective and degradative and is induced to survive periods of nutrient limitation (3). However, the two pathways share many molecular components and both involve sequestration by double-membrane saccular structures of unknown origin that capture the load, close into vesicles, and then fuse with the vacuole (4). A major difference between these pathways appears to be the size and content of the transport vesicles. The Cvt vesicles exclude cytoplasm and are smaller than autophagosomes that engulf bulk cytoplasm and even organelles (2). Strikingly, despite all these differences, targeting of API to the vacuole is specific and saturable, both in vegetative growth conditions and under nitrogen deprivation (3), although the molecular details of its selective recognition and capture remain essentially unknown. Previous studies have shown that pAPI recognition by the transport machinery involves its prepro-amino extension (5,6) and cytoplasmic chaperones of the Ssa family (7,8). Furthermore, the amino extension is necessary and sufficient to target the reporter protein GFP to the vacuole (9). In this study we report that Yol082p, a protein shown to interact physically with pAPI in a two-hybrid screening performed with the whole yeast genome (10), mediates API loading into transport vesicles and targeting to the vacuole. We also show that Yol082p interacts with API by a process that does not only involve the prepro-amino extension but also the mature part of the API protein. Yol082p is distributed between the cytoplasm and distinct round mobile structures.
Plasmid Constructions-The plasmid pGAL1-YOL082 expressing the YOL082w gene under the control of the inducible GAL1 promoter in the pYES2.0 vector (Invitrogen) was constructed as follows: DNA of the YOL082 coding region containing universal termini with EcoRI and PvuII sites at the 5Ј end and an SmaI site at the 3Ј end was amplified by PCR using Universal Yeast open reading frame primers (Research Genetics, Inc.). The polymerase chain reaction product was ligated into TOPO-TA Cloning ® (Invitrogen) vector to give TOPO-YOL082. The EcoRI-SmaI fragment containing YOL082 was then cloned into the pYES2.0 vector. For the NH 2 -terminal fusion pGFP-YOL082, the YOL082 coding region was PCR-amplified from TOPO-YOL082 using the forward Universal Yeast open reading frame primer and a reverse primer introducing a XhoI site 3Ј to the stop codon. PvuII-XhoI-digested PCR product was then cloned into SmaI-XhoI-digested pGFP-N-FUS (13) generating an in frame fusion. The plasmid pYOL082⌬C32-GFP expressing a truncated version of Yol082p lacking 32 residues in its carboxyl terminus was created by cloning an EcoRI-ClaI fragment of TOPO-YOL082 (residues 1-383 of Yol082p) into pGFP-C-FUS (13) creating an in frame COOH-terminal fusion. Plasmid pYOL082⌬C32 was derived from pYOL082⌬C32-GFP by excising a SalI-SalI fragment containing the GFP coding region. The resulting plasmids, pGAL1-YOL082, pGFP-YOL082, pYOL082⌬C32-GFP, and pYOL082⌬C32, were used to transform the BY4741, ⌬yol082, and ⌬apg12 strains using the lithium acetate procedure (14).
The bait plasmid encoding precursor API fused in frame to the Gal4p DNA-binding domain was constructed as follows: the API open reading frame was PCR-amplified from pEMBLyex4-API (5) with a forward primer, 5Ј-GGTACCCCGGGAACCTCTAGA-3Ј, which incorporated an SmaI site, and a reverse primer, 5Ј-CGAATTCATCGTCGACT-GATCTCC-3Ј, which incorporated a SalI site. The SmaI-SalI fragment was then cloned into pGBT9 (CLONTECH) to create pGBT9-pAPI (amino acids 1-507). The API region encoding the prepro-amino extension (amino acids 1-44) was cloned using the same forward primer and a reverse primer, 5Ј-GTGCTCGAGGATCCACCACG-3Ј, creating a BamHI site. The corresponding SmaI-BamHI fragment was inserted into pGBT9 to generate pGBT9-preproAPI. The mature API coding region (amino acids 54 -514) was amplified with a forward primer, 5Ј-TATGAGGATAGTCGACAGGAATTC-3Ј, generating a SalI site and a reverse universal primer; the resulting PCR product was cloned into pUC18, and a SalI-PstI fragment was cloned into pGBT9 to give pGBT9-mAPI. To construct pACT2-YOL082 expressing the complete Yol082p fused in frame to the Gal4p activation domain, the YOL082 coding region was amplified from TOPO-YOL082 using a forward primer, 5Ј-CTGACCACCATGGACAACTCAAAG-3Ј, which added an NcoI site and a universal reverse primer. An NcoI-SmaI fragment of the PCR product was then cloned into NcoI-SmaI-digested pACT2.
Studies on the Processing of the pAPI into mAPI-To determine pAPI and mAPI protein levels, BY4741 and ⌬yol082 cells were grown overnight in SD medium to 1 OD 660 . Strains tested under nitrogen starvation conditions were then washed twice with distilled water and incubated in SD(ϪN) for different time periods. Cells were harvested and crude cell extracts prepared with glass beads in cold 50 mM Tris-HCl, pH 8.0, 5 mM EDTA and protease inhibitors (1 mM phenylmethylsulfonyl fluoride and pepstatin, aprotinin, and leupeptin at 1 g/ml each). The extracts were centrifuged at 4°C for 5 min at 500 ϫ g to eliminate cell debris and 10 g of protein studied by Western analysis with a rabbit polyclonal anti-API antibody (5) or a rabbit polyclonal antibody against the amino extension of pAPI (9), using the ECL detection system (Amersham Pharmacia Biotech). Cells transformed with pGAL1-YOL082 were grown in SRaf medium and induced for different time periods in SGal, prior to harvesting and preparation of the crude cell extracts. Cells transformed with pGFP-YOL082, pYOL082⌬C32-GFP, and pYOL082⌬C32 were grown in low methionine-SD (10 g/ml) to control the expression of the fusion protein from the MET25 promoter.
Cell Viability Studies-To examine the survival of the different yeast strains under nitrogen starvation conditions, cells were grown to 1 OD 660 in SD medium, washed once with distilled water, and resuspended in SD(ϪN). At different times aliquots were removed, and the appropriate dilutions were plated in triplicate onto YPD (1% yeast extract, 2% peptone, 2% glucose) plates.
Analysis of API Dodecamers by Sedimentation Velocity Centrifugation-Wild-type and ⌬yol082 cells grown in SD 1 medium overnight to 1 OD 660 were harvested and crude cell extracts prepared with glass beads as described above. The extracts were fractionated by rate-velocity centrifugation on a glycerol gradient (20 -50%) (17) and the gradient fractions scrutinized for API by Western blot using the anti-API antibody and the ECL technique.
Protease Protection Assays-Wild-type and ⌬yol082 cells grown to 1 OD 660 in SD medium were harvested and spheroplasted by treatment with Zymolyase 20-T (United States Biological). Briefly, cells were incubated in 0.1 M Tris/SO 4 buffer, pH 9.4, containing 20 mM dithiothreitol for 20 min at 30°C, collected, and treated with 0.2 mg/ml zymolyase in 1.2 M sorbitol, 20 mM KH 2 PO 4 , pH 7.4, for 30 min at 30°C. Spheroplasts were adjusted to 15 OD 660 /ml with SL medium (1 M sorbitol, 1% glucose, 1% proline, 0.17% YNB without amino acids and ammonium sulfate, with the appropriate auxotrophic requirements) (18) and 1 ml preincubated for 5 min at 30°C and metabolically labeled for 10 min using 100 Ci of [ 35 S]methionine/cysteine (Amersham Pharmacia Biotech) per OD 660 . Labeled spheroplasts were then diluted 10-fold in SL medium containing 8 mM methionine, 4 mM cysteine, and 0.2% yeast extract and chased for 2 min or 2 h. Chases were stopped by incubation for 2 min with 10 mM NaN 3 and the spheroplasts harvested and washed with SL medium. Lysates from spheroplasts were prepared by osmotic lysis in 1 ml of 200 mM sorbitol, 5 mM MgCl 2 , 20 mM Pipes, pH 6.8 (lysis buffer) for 20 min at 4°C. Unlysed spheroplasts were removed by centrifugation for 2 min at 500 ϫ g. Lysates were treated at 4°C for 30 min with 50 g/ml proteinase K (Life Technologies, Inc.) in the absence and presence of 0.2% Triton X-100. The proteolytic digestion was stopped with a mixture of protease inhibitors (5 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, and pepstatin, aprotinin, and leupeptin at 1 g/ml each). Protein was then precipitated with cold 10% trichloroacetic acid, using 100 g/ml bovine serum albumin as carrier, washed with acetone, and the API in the resuspended pellets immunoprecipitated as described previously (7), resolved by 10% SDS-PAGE, and studied by autoradiography.
Subcellular Fractionation and Membrane Flotation Analysis-To biochemically determine the subcellular distribution of the GFP-Yol082p, wild-type and ⌬yol082 cells transformed with the pGFP-YOL082 plasmid were grown in low methionine-SD medium, and 15 OD 660 cell equivalents were harvested, spheroplasted, and lysed as described above. The lysates were precleared by centrifugation at 500 ϫ g for 5 min to remove unlysed spheroplasts and the supernatants subjected to centrifugation at 17,000 ϫ g for 10 min at 4°C to give pellet and supernatant fractions (P17 and S17). The S17 fraction was subjected to further centrifugation at 100,000 ϫ g for 30 min at 4°C resulting in pellet and soluble fractions (P100 and S100). The P17 and P100 fractions were washed and resuspended in lysis buffer. Protein from all fractions was precipitated with trichloroacetic acid, solubilized with Laemmli buffer, resolved by 8% SDS-PAGE, and analyzed by Western blot with anti-GFP antibody (CLONTECH).
Membrane flotation experiments were performed by the method described by Noda et al. (19), modified as follows: precleared spheroplasts lysates were centrifuged for 5 min at 5,000 ϫ g and the pellet, resuspended in 100 l of 15% Ficoll prepared in lysis buffer, overlaid with 1.2 ml of 13% Ficoll and then with 100 l of 2% Ficoll, and centrifuged for 10 min at 4°C at 17,000 ϫ g. Fractions of 230 l were collected, subjected to trichloroacetic acid precipitation, resolved by 8% SDS-PAGE, and blots probed with the anti-API antibody. Stripped blots were tested for CPY with specific antibodies.
Microscopy Studies-Cells expressing GFP-Yol082p or Yol082⌬C32p-GFP were grown to 1.2 OD in low methionine-SD medium and when required incubated for 2 h in SD(ϪN) medium. Upon concentration by low speed centrifugation the cells were spread thin on glass slides, air-dried, and studied by fluorescence microscopy using a Radiance 2000 confocal microscope (Bio-Rad).

Yol082p
Interacts with API in a Two-hybrid Assay-Yol082p was identified as a protein able to interact with API in a comprehensive analysis of protein-protein interactions of the yeast proteome using the two-hybrid system (10). The Yol082 protein has a predicted molecular mass of 47.5 kDa and shows significant homology, 33% identity, to another S. cerevisiae putative protein of unknown function, referred to as Yol083p. No other homologues were found in the data bases. Analysis of the predicted sequence of Yol082p suggests that it is a hydrophilic protein with no predictable transmembrane domains. The null mutant of YOL082w, which as described was viable (20), showed wild-type growth rate in SD medium with a doubling time of 150 min for both wild-type and ⌬yol082 strains, indicating that Yol082p function is not required for vegetative growth. To further characterize its interaction with API, Yol082p was fused in frame with the Gal4p activation domain carried in the pACT2 vector and its interaction with the preproamino extension of API, pAPI and mAPI, separately cloned into the bait plasmid pGBT9, studied using the two-hybrid system. Yol082p did not interact with the prepro domain, but interacted with mAPI, as shown by the ability of the co-transformed cells to grow on selective plates (Fig. 1). The interaction with mAPI was enhanced by the presence of the amino extension as evidenced by its stronger interaction with pAPI (Fig. 1). The specificity of the above interactions was confirmed by the observation that none of the constructions co-transformed with the empty vectors allowed growth in the restrictive medium.
Yol082p Is Involved in Vacuolar API Import/Processing-The interaction of API with Yol082p led us to study whether it was involved in the transport of API to the vacuole. To test this, we first compared the vacuolar processing of API in a yeast strain carrying a deletion of the YOL082w gene with that in its isogenic wild-type strain BY4741. The results showed that in ⌬yol082 cells the majority of API is in its precursor form pAPI ( Fig. 2A), both in cells vegetatively growing or starved of nitrogen for 2 h. In contrast, in the BY4741 strain, the entire API detected was in the processed form, mAPI. Furthermore, CPY, a vacuolar protease that is transported to the vacuole through the secretory pathway, was correctly processed in wild-type and mutant cells, indicating normal functioning of the secretory pathway in these strains. These results indicated that Yol082p was required for transport of pAPI to the vacuole and that this requirement was not bypassed by the 2-h period of nitrogen starvation. The involvement of Yol082p in the transport of pAPI to the vacuole was then directly tested by studying the conversion of pAPI into mAPI in ⌬yol082 cells transformed with a multicopy plasmid carrying the YOL082w gene under the control of the GAL1 promoter (Fig. 2B). It was observed that these cells gradually converted pAPI into mAPI as they were incubated in galactose from 0 to 6 h. The overexpression of Yol082p in the wild-type background did not interfere with the processing of pAPI. This was interesting because overexpression of Cvt9p, a protein that appears to function primarily under vegetative growth conditions, has been shown to produce a dominant negative phenotype for API import when overexpressed (21). Control mutant cells transformed with the empty vector were, on the other hand, unable to process the precursor. Altogether these results clearly showed that the conversion of pAPI into mAPI, an event associated with the transport of pAPI from the cytoplasm to the vacuole, is facilitated by Yol082p.
Conversion of pAPI into mAPI Becomes Yol082p-independent upon Prolonged Periods of Nitrogen Starvation-In yeast, nitrogen deprivation induces a degradative process known as autophagocytosis. During autophagocytosis, bulk cytoplasm and organelles are packaged into double-membrane vesicles, termed autophagosomes, for further delivery to the vacuole. Under these conditions the majority of pAPI is targeted to the vacuole by autophagosomes (Apg pathway), but the maturation kinetics of pAPI is indistinguishable from that recorded under growing conditions (3). This is in contrast to the slow degradation of cytoplasmic markers (3), thus suggesting API transport by the Apg pathway remains selective. Because the Cvt and the Apg pathways share many of their molecular components (3,22), we, therefore, investigated the role of Yol082p in the transport of pAPI to the vacuole by the Apg pathway.
One of the distinctive phenotypes of the apg mutants is their reduced viability during starvation. We, therefore, compared the viability of wild-type and ⌬yol082 cells in SD(ϪN) medium with that of a ⌬apg12 mutant strain, a mutant defective in autophagocytosis and sensitive to nitrogen starvation (23). We observed no loss of viability in the wild-type and ⌬yol082 cells during 10 days of nitrogen starvation, while ⌬apg12 cells progressively lost viability during the incubation (Fig. 3). We interpreted this difference as an indication that Yol082p was not essential for autophagy. This result was somewhat surprising, since, as described above (see Fig. 2A), we found that API was not processed when the mutant cells were starved of nitrogen for 2 h. Therefore, we studied the effect of extending the nitrogen starvation period on the processing of pAPI. For this purpose, ⌬yol082 cells were starved of nitrogen for 12, 14, 16, 18, and 24 h and the conversion of pAPI into mAPI analyzed and compared with that in wild-type and ⌬apg12 cells. It was observed that the extension of the starvation caused the slow conversion of pAPI into mAPI, with as much as 15% of the precursor being processed in the first 12 h and 35% in 24 h. This observation was in contrast with the inability of the ⌬apg12 cells to process pAPI under the same conditions (Fig.  4). We interpreted these results as evidence that under prolonged nitrogen starvation conditions pAPI was passively transported to the vacuole through the autophagy pathway.
pAPI Remains in the Cytoplasm of ⌬yol082 Cells-To characterize the transport step blocked in ⌬yol082 cells, we first examined the assembly of pAPI into dodecamers, an event that precedes its wrapping by the saccular structures and its tar-geting to the vacuole (17). For this purpose, the assembly state of pAPI in ⌬yol082 cells was studied by examining its mobility in glycerol gradients (Fig. 5). We observed that pAPI migrated as the pAPI dodecamers extracted from wild-type cells. It was, therefore, concluded that Yol082p was not involved in API oligomerization.
To study whether pAPI was captured by cvt vesicles in ⌬yol082 cells, we compared its recovery with membrane fractions prepared by differential centrifugation of wild-type and ⌬yol082 cells on Ficoll gradients (Fig. 6). For this purpose, osmotic lysates obtained from spheroplasted wild-type and mutant cells were centrifuged at 5,000 ϫ g and the resulting pellets subjected to flotation in Ficoll as described under "Experimental Procedures." Upon flotation on Ficoll, membraneassociated proteins are recovered with the light-density fractions, soluble proteins with the denser ones, and large protein complexes with the pellet (18). As expected, most of the API extracted from the wild-type cells floated with the membranes (F1), whereas the majority of the pAPI extracted from ⌬yol082 cells, which remained as pAPI, was recovered with the pellet (F7) and a small percentage (10%) with the membranes. On the other hand, under these same conditions the CPY extracted from ⌬yol082 cells was recovered with the membranes, a result consistent with its expected ability to reach the vacuole.
The floating of a small part of pAPI with the membranes from ⌬yol082 cells led us to study whether it was sequestered within vesicles. For this purpose, spheroplasts prepared from wild-type and ⌬yol082 cells were pulse-labeled for 10 min with [ 35 S]methionine/cysteine and chased for 2 min or 2 h in medium without radioactivity. Then, the lysates from spheroplasts were incubated at 4°C for 30 min without or with 50 g/ml proteinase K in the absence or presence of Triton X-100, prior to immunoprecipitation with the anti-API antibody. It was observed that pAPI remained fully accessible to the protease in the absence of detergent, during the 2-h chase period ( Fig. 7). This result indicated that pAPI is not sequestered within vesicles and suggests that the population of pAPI that floats with membranes is probably associated with open membrane structures or, less likely, bound to the surface of vesicles.
GFP-Yol082p Rescues the Conversion of pAPI into mAPI in ⌬yol082 Cells-To further learn about the role of Yol082p in API transport and its cellular distribution, we constructed a fusion protein in which the GFP was fused in frame to the amino end of Yol082p. Next, GFP-Yol082p was expressed in the ⌬yol082 mutant cells and its ability to convert pAPI into mAPI studied. The production of mAPI species in these cells indicated that GFP-Yol082p was functional in vivo (Fig. 8). The complementation of the defect in pAPI processing was, however, not observed when GFP was fused to the carboxyl end of Yol082⌬C32p, a mutant protein developed by deleting the last 32 residues of Yol082p. Moreover, when expressed in wild-type cells Yol082⌬C32p-GFP inhibited the processing of pAPI. These results and the inability of Yol082⌬C32p to complement the defective processing of pAPI in ⌬yol082cells supported the idea that Yol082⌬C32p and Yol082⌬C32p-GFP were inactive, but correctly localized.
GFP-Yol082p Rescues the Association of API with Mem-branes-To further examine whether Yol082p was required for the association of API with membranes, we studied the flotation of the API contained in extracts from ⌬yol082 cells transformed with pGFP-YOL082 on Ficoll gradients (Fig. 6). The recovery of API as mAPI with the membrane-enriched fractions (F1) indicated that GFP-Yol082p was able to rescue the association of API with membranes as well as its processing in the vacuole. This result, together with the recovery of the API extracted from untransformed ⌬yol082 cells with the pellet fraction (F7), as pAPI, strongly suggested that Yol082p plays a critical role in the wrapping of pAPI by membranes. GFP-Yol082p Accumulates in Round, Mobile Cytoplasmic Structures-The subcellular distribution of GFP-Yol082p was investigated by studying its partition between fractions from spheroplasted ⌬yol082 cells obtained by differential centrifugation (data not shown). It was observed that ϳ15% of the protein sedimented at 17,000 ϫ g, with the rest remaining soluble even after centrifugation for 1 h at 100,000 ϫ g.
To gain further insight into the localization of Yol082p, we studied, by confocal microscopy, the distribution of GFP-Yol082p in smears from wild-type and ⌬yol082 cells (Fig. 9). The study revealed that GFP-Yol082p was distributed between the cytoplasm and a variable number of round structures, hereon referred to as YR structures (Yol082p Round struc-FIG. 5. Yol082p is not involved in the assembly of pAPI into dodecamers. Extracts from wild-type and ⌬yol082 cells were fractionated by sedimentation velocity centrifugation using a 20 -50% glycerol gradient and the distribution of the pAPI species studied by Western blot on the gradient fractions using the anti-API antibody. Molecular mass standards were ovalbumin (Ovo, 45 kDa), catalase (Cat, 240 kDa), apoferritin (Apo, 450 kDa), and thyroglobulin (Thy, 669 kDa).
FIG. 6. Loss of API from the vacuole-rich fraction obtained from ⌬yol082Cells is restored by GFP-Yol082p. Spheroplasts from wild-type cells, and from ⌬yol082 cells transformed or not with pGFP-YOL082, were lysed osmotically, precleared, centrifuged for 10 min at 4°C at 5,000 ϫ g, and the pellets (P5) further fractionated by differential centrifugation on a Ficoll step gradient as described under "Experimental Procedures." An aliquot from P5, the gradient fractions (F1-F6) and the gradient pellet (F7) were resolved by SDS-PAGE and analyzed by Western blot, first with the anti-API antibody and then, after stripping the blots, with the anti-CPY antibodies.

FIG. 7. pAPI remains protease-accessible in ⌬yol082 extracts.
Spheroplasts from cells of the wild-type and the ⌬yol082 mutant strain were metabolically labeled for 10 min using [ 35 S]methionine/cysteine and then chased for either 2 min or 2 h. After their disruption by osmotic shock the resulting lysates were incubated at 4°C for 30 min without or with 50 g/ml proteinase K, in the absence or presence of 0.2% Triton X-100. Radiolabeled API protein was immunoprecipitated using the anti-API antibody conjugated to protein A-Sepharose and the precipitates resolved by SDS-PAGE and analyzed by autoradiography. tures), with an average size between 0.13 and 0.27 m. These structures were not detected when wild-type or mutant cells were transformed with the empty plasmid pGFP-N-FUS. Furthermore, incubation of the cells for 2 h in SD(ϪN) medium to activate the autophagy pathway resulted in loss of the YR structures from most of the cells. On the contrary, when the study was repeated in ⌬apg12 cells, with impeded autophagocytosis, a general decrease in cytoplasmic staining was observed, whereas the YR structures were clearly visible in all the cells. Interestingly, incubation of these cells for 2 h in SD(ϪN) medium did not result in the loss of the YR structures as observed before in wild-type and ⌬yol082 cells. The effect of the inactivation of the autophagy pathway on the distribution of GFP-Yol082p led us to study the distribution of the Yol082⌬C32p-GFP mutant protein that, as described above, was unable to complement the defect in the vacuolar targeting and processing of pAPI. The expression of Yol082⌬C32p-GFP in wild-type cells provoked a dramatic shift in the distribution of the protein, as shown both by the disappearance of the cytoplasmic staining and the increase in size of many YR structures. Again, as observed in ⌬apg12 cells the incubation of these cells in SD(ϪN) medium did not change the pattern of Yol082⌬C32p-GFP distribution.
Analysis of the localization of the YR structures at different time intervals by time-lapse confocal microscopy ( Fig. 10) showed that they were highly mobile, as shown both by their lateral displacement and their continuous popping on and off from the plane under focus. On average a 0.27-m YR structure was found to move at a rate of 1 nm/s. GFP-Yol082p, but not Yol082⌬C32p-GFP or Yol082⌬C32p, complements the defect in the vacuolar import and processing of pApI produced by disruption of YOL082. Cells of the wild-type and the mutant ⌬yol082 strains, transformed with either pGFP-YOL082, pYOL082⌬C32-GFP, or pYOL082⌬C32, were grown to logarithmic phase in SD medium and blots from their crude extracts probed with the anti-API antibody.
FIG. 9. Cellular distribution of Yol082p. Changes in distribution of Yol082p between the cytoplasmic pool and YR structures produced by growing conditions and the inhibition of the transport activity are shown. Wild-type, ⌬apg12, and ⌬yol082 strains transformed with pGFP-YOL082 or truncated pYOL082⌬C32-GFP were grown in SD medium to 1.2 OD 660 and aliquots further incubated for 2 h in SD(ϪN) medium, as indicated in the panels. The cellular distributions of both fusion proteins were studied by fluorescence microscopy on air-dried cell smears. Insets are 2.7-fold magnification of the areas next to the white arrows. Bars: 6.64 and 2.3 m.

DISCUSSION
In this study we have shown that Yol082p, identified in a whole-genome analysis of protein-protein interactions as an API-interacting protein (10), is required for vacuolar targeting and conversion of pAPI into mAPI, both in vegetative growth and under short term nitrogen starvation conditions.
The interaction of Yol082p with pAPI and mAPI, but not with the prepro-amino extension of the precursor, is particularly interesting given the role of the latter in the transport of pAPI to the vacuole and the observation that it is necessary and sufficient for the transport of the reporter protein GFP from the cytoplasm to the vacuole (9). This observation suggests that transport of API to the vacuole requires additional transport determinants localized in the mature part, outside its amino extension. In addition, the stronger two-hybrid interaction of Yol082p with pAPI, as compared with mAPI, suggests that either Yol082p interacts physically with the amino extension in the context of the native protein or, alternatively, that the extension is required for proper folding and exposure of the transport determinants contained in the mature part of API to Yol082p. Clearly, further research is required to determine whether the determinants involved in its interaction with Yol082p are specific of API.
Our studies on the processing of pAPI in wild-type and ⌬yol082 cells show that Yol082p is required for targeting and conversion of pAPI into mAPI in the vacuole, both under vegetative growth and short periods of nitrogen starvation. The rescue of the API processing defect in ⌬yol082 cells transformed with Yol082p expressed from an inducible promoter unambiguously shows the involvement of Yol082p in the vacuolar import of pAPI.
The block in API vacuolar processing shown by ⌬yol082 cells incubated for short periods in SD(ϪN) medium is in contrast to the ability of mutants with impeded API transport, such as apg13, vac8, cvt3, aut2, and aut7, to overcome the pAPI accumulation soon after their shift to medium without nitrogen (2,24,25). Furthermore, it is interesting that the inhibition of the pAPI processing in ⌬yol082 cells is partially reversed by the extension of the nitrogen starvation period. Under these conditions, mAPI is detected after 12 h in SD(ϪN) medium, and maturation proceeds slowly to reach a plateau at ϳ30% in 16 h. These observations strongly suggest that Yol082p is involved in the rapid and specific capture of pAPI by the autophagosomes developed after a short starvation period, but not in the slow and unspecific capture that occurs with the engulfment of large portions of the cytoplasm after prolonged starvation.
Apg mutants (26) often show a correlation between the effect of the mutation in autophagosome biogenesis and the loss in viability under nitrogen starvation, so that mutants with defective autophagosome nucleation die earlier upon nitrogen deprivation (25). In this context it is, therefore, interesting that although Yol082p appears to function in an early step in the pathway of API transport, the ⌬yol082 mutant is completely starvation-resistant. This resistance to starvation suggests again that Yol082p is not essential for autophagosome biogenesis.
Regarding the API transport step in which Yol082p is involved, we have shown that pAPI assembly into dodecamers takes place normally in the mutant cells. We also show that in the absence of Yol082p, the interaction of the oligomerized pAPI with the sequestering double-membrane sacs appears to be affected. The proteinase K protection assay performed with metabolically labeled protein reveals that, in the mutant strain, the newly synthesized pAPI remains unprotected in the cytoplasm after 2 h of its synthesis, which is in contrast to the wild-type. This difference suggests that Yol082p may work in the recognition of pAPI by the wrapping membranes or, alternatively, by closing these into vesicles. The recovery of the protease-sensitive pAPI extracted from apg5, apg7, apg9, aut7/apg8, and cvt3 cells defective in biogenesis of the API transport vesicles (19,(27)(28)(29) with the membranes that float on Ficoll, and the exclusion of the bulk of the pAPI extracted from ⌬yol082 cells from the membrane fraction, rather supports the first of the two above alternatives.
Confocal fluorescence microscopy studies, performed using the fluorescent protein GFP-Yol082p, show the existence of a pool of protein homogeneously distributed throughout the cytoplasm in equilibrium with a second pool organized into one or more round-shaped structures, which we have called YR. Fur- thermore, the equilibrium is dramatically shifted by changes in the cell growing conditions, the ability of the cells to use the autophagy pathway, and the functional activity of GFP-Yol082p. Although we lack direct evidence on the meaning of the above changes in distribution, some of these observations suggest that the pool organized into YR structures could be actively engaged in the transport of pAPI to the vacuole by the Cvt pathway. With regard to this, the disappearance of the YR structures from cells incubated for 2 h in SD(ϪN) medium strongly suggests that they are not needed under conditions in which the autophagy pathway is activated. The difference between the rapid processing of pAPI in mutants with a defective Cvt pathway and the inability of ⌬yol082 cells to process pAPI, when they were incubated in SD(ϪN) medium, strongly suggests that transport of pAPI by the autophagy pathway requires the disassembly of the YR structures. This may be related to the decrease in the cytoplasmic pool of GFP-Yol082p and the enhanced visibility of YR structures observed in ⌬apg12 cells incubated with SD or SD(ϪN) medium, changes that could reflect a blockage of the Cvt/Apg pathways after the transport step mediated by Yol082p. Also interesting is the disappearance of the cytoplasmic protein and the increase in size of YR structures observed in wild-type cells expressing the functionally inactive Yol082⌬C32p-GFP. This shift in equilibrium could again be consistent with our view that YR structures are involved in protein transport through the Cvt pathway, since the functionally inactive protein may cause the jam of the transport machinery and/or the accumulation of the transported material. A second alternative is the possibility that the deletion of the last 32 residues may shift the protein equilibrium toward the YR structures, which could also explain the insensitivity of these to the incubation of cells in SD(ϪN) medium. Obviously, these two possibilities are not mutually exclusive.
Because of its behavior as a Cvt protein we propose to rename Yol082p as Cvt19p.
Further research is required to characterize the transport step mediated by Yol082p, a step that appears to lay after the assembly of pAPI into dodecamers and before the wrapping of Cvt complexes by the saccular structures.