Trs85 is Required for Macroautophagy, Pexophagy and Cytoplasm to Vacuole Targeting in Yarrowia lipolytica and Saccharomyces cerevisiae

Abstract

Yarrowia lipolytica was recently introduced as a new model organism to study peroxisome degradation in yeasts. Transfer of Y. lipolytica cells from oleate/ethylamine to glucose/ammonium chloride medium leads to selective macroautophagy of peroxisomes. To decipher the molecular mechanisms of macropexophagy we isolated mutants of Y. lipolytica defective in the inactivation of peroxisomal enzymes under pexophagy conditions. Through this analysis we identified the gene YlTRS85, the ortholog of Saccharomyces cerevisiae TRS85 that encodes the 85 kDa subunit of transport protein particle (TRAPP). A parallel genetic screen in S. cerevisiae also identified the trs85 mutant. Here, we report that Trs85 is required for nonspecific autophagy, pexophagy and the cytoplasm to vacuole targeting pathway in both yeasts.

INTRODUCTION

There are three autophagy-related pathways that deliver cargo proteins and/or organelles in cytosolic double-membrane vesicles to the vacuole, the membrane compartment responsible for degradation, recycling and storage of cellular constituents in yeasts. These are macroautophagy of bulk cytosol also referred to here as nonspecific autophagy, selective macroautophagy of peroxisomes that is termed macropexophagy, and selective macroau-tophagy of precursor aminopeptidase I (prApe1) by the cytoplasm to vacuole targeting (Cvt) pathway.¹ All three of these pathways utilize topologically the same basic mechanism to enclose cargo material into autophagosomes, pexophagosomes and Cvt vesicles, respectively. A large number of molecular components that mediate different steps of autophagy-related pathways were identified in the last decade. Not surprisingly, all three pathways appear to share most of them.² This is also consistent with the common localization of most of the autophagy-related (Atg) proteins at the preautophagosomal structure (PAS).^3-5 Assembly of both Cvt vesicles and autophagosomes appears to be dependent on a functional endoplasmic reticulum (ER) and Golgi complex.^6-8 Therefore, the early secretory pathway was proposed to be the source of lipids for autophagosome and Cvt vesicle formation.^7,8 The question of whether the early secretory pathway contributes to macropexophagy, however, had not been addressed.

Although pexophagy appears to share a number of genes with macroautophagy, it occurs at a faster rate and to a much greater extent than nonspecific autophagy of bulk cytosol.⁹ The molecular background of such exclusive efficacy is not clear, because only one strictly pexophagy-specific gene was reported until now in yeasts, ATG26/UGT51. ATG26 encodes sterol glucosyltransferase, essential for pexophagy in Pichia pastoris, but dispensable for macroautophagy and both macroautophagy and the Cvt pathway in P. pastoris and S. cerevisiae, respectively;^10,11 however, its function in pexophagy may be restricted to P. pastoris or methylotrophic yeasts only, since ATG26 was not required for pexophagy in Yarrowia lipolytica.¹¹ This last study, in addition to other work that has revealed differences in the localization and possibly function of other Atg proteins among different yeasts¹² indicates the importance of analyzing autophagy-related pathways in alternative model systems.

Transfer of Y. lipolytica from acetate/oleate/ethylamine to glucose/ammonium sulfate medium induces selective autophagy of peroxisomes.¹³ Macropexophagy was proposed to be the major route of Y. lipolytica peroxisome degradation as determined by both electron¹³ and fluorescent¹⁴ microscopy. Y. lipolytica can use ethylamine as a sole nitrogen source due to the induction of peroxisomal amine oxidase (AMO). AMO was used previously as a marker enzyme in the isolation of pexophagy mutants,¹³ but none of the mutant strains has been further characterized. We isolated Y. lipolytica mutants affected in the degradation of peroxisomes (ref. ¹⁵, this study) using insertional mutagenesis¹⁶ and a modified AMO plate assay screening procedure.¹⁷ Here, we present the identification of the gene disrupted in two such mutants, both of which appeared to be affected in the YlTRS85 gene. A separate screen for mutants defective in the Cvt pathway in S. cerevisiae also identified ScTRS85. A deletion of this gene resulted in a block in nonspecific autophagy, pexophagy and import of prApe1. These results suggest a function for Trs85 as an Atg protein.

MATERIALS AND METHODS

Media and growth conditions

S. cerevisiae cultures were grown at 30°C in YPD (1% yeast extract, 2% peptone, 2% glucose) or SMD (0.67% yeast nitrogen base, 2% glucose, amino acids and vitamins) media. For nitrogen starvation, SD-N (0.17% yeast nitrogen base without amino acids and ammonium sulfate, 2% glucose) medium was used. Solid media were identical with the addition of agar to 2% final concentration. Y. lipolytica strains were cultured at 28°C in YPD (1% yeast extract, 1% peptone, 1% glucose) medium, YND (1.7 g/l YNB without amino acids and ammonium sulfate supplemented with 1% glucose and 0.5% ammonium sulfate) and YND(-N) (identical to YND except for omission of the nitrogen source), YOE (1.7 g/l YNB without amino acids and ammonium sulfate, 50 mM phosphate buffer, pH 6.8 (PiB), 0.05% yeast extract, 1% oleate, 25 mM ethylamine) and GA (1.7 g/l YNB without amino acids and ammonium sulfate, PiB, 1% glucose, 25 mM ammonium chloride) media. We prepared 1% oleic acid in 0.025% Tween-80 as a 20-fold sonicated stock emulsion. For growth on solid media, strains were cultured on 2% agar plates using YND and YEE (1.7 g/l YNB without amino acids and ammonium sulfate supplemented with 0.05% yeast extract, 0.5% ethanol, 0.2% ethylamine-HCl).

Strains and plasmids

To delete the chromosomal TRS85 locus in S. cerevisiae strain YTS159 the entire coding region was replaced by transformation with the URA3 gene, which was amplified from the template plasmid pUG72¹⁸ by PCR using primers containing 42 nucleotides identical to the regions flanking the TRS85 open reading frame (ORF); primer sequences will be made available upon request. The pCuTRS85(416) plasmid was constructed by cloning the entire coding region of the TRS85 gene plus 155 nucleotides of the 3′ untranslated region into the pCu(416) vector¹⁹ between BamHI and HindIII restriction sites after the CUP1 promoter. The plasmid encoding GFP-ATG8 under the control of the CUP1 promoter was pCuGFPAUT7(416).²⁰

To generate Y. lipolytica mutants defective in amine oxidase (AMO) inactivation, the zeta-URA3 mutagenesis cassette (MTC) was amplified with the primers MTC1 and MTC2,¹⁷ specific for the right and left borders of the MTC, respectively, using the plasmid JMP5¹⁶ as template. Y. lipolytica was transformed with MTC by the LiAc/LiCl method.²¹ For each transformation assay we used 0.5 μg of MTC generating up to 5,000 Ura⁺ transformants per μg of DNA that were selected on YND plates. Mutants deficient in AMO inactivation (Ain) were isolated by a plate assay screening procedure:¹⁷ Ura⁺ transformants from YND plates were replica-plated onto YEE plates and incubated for 18 h. The plates were then carefully overlaid with 7-8 ml of AMO inactivation mixture containing 0.3% agar, 3% glucose and 1% ammonium sulfate, and incubated for 10 h at 28°C. Then the plates were overlaid with 7-8 ml of AMO assay mixture containing 0.3% agar in 100 mM phosphate buffer, pH 7.0, 0.05% o-dianisidine as a chromogen, 0.5% cetyltrimethylammonium bromide as a permeabilizing agent, 2.3 U/ml peroxidase and 4 mM ethylamine as the substrate for AMO. Colonies with high residual activity of AMO were stained red and identified after 14 h of incubation at 28°C. Genomic DNA preparation from Y. lipolytica cells²¹ and Southern blot analysis¹⁶ were done as described previously.

For complementation studies in Y. lipolytica we constructed the plasmid JMP61(hph) by replacement of the Y. lipolytica URA3 gene with the Klebsiella pneumoniae hygromycin B phosphotransferase (hph) gene on the plasmid JMP61.²² The hph gene under the control of the synthetic promoter hp4d was liberated from the plasmid JMP115²³ as a BamHI (blunted)-XhoI fragment and ligated to the larger ClaI (blunted)-SalI fragment of JMP61. Then the plasmid JMP61(hph)BamHI with the zeta-hph empty cassette was constructed by removal of the POX2 promoter and LIP2 prepro sequence as a BamHI fragment from the plasmid JMP61(hph). The JMP61(hph)-YlTRS85 plasmid with the zeta-hph-YlTRS85 cassette was constructed by replacement of the POX2 promoter and LIP2 prepro sequence on the plasmid JMP61(hph) with the BamHI-KpnI fragment containing the YlTRS85 ORF with 492 nucleotides of the promoter and 150 nucleotides of the terminator regions, which was amplified from the H222 wild type genomic DNA by PCR; primer sequences will be made available upon request. Ain16 and Ain19 mutants were transformed with 0.3 μg of NotI-digested JMP61-(hph)BamHI and JMP61(hph)YlTRS85 plasmids by the LiAc/LiCl method.²¹ The transformants with zeta-hph and zeta-hph-YlTRS85 cassettes were selected on YPD+hygromycin plates with 50 μg/ml of hygromycin B.

Amplification and sequencing of MTC borders

To identify the genes disrupted in Ain16 and Ain19 mutants, MTC borders were amplified by convergent PCR. Genomic DNA of the mutants was triple digested with restriction enzymes producing blunt ends, DraI, MscI and FspI, and ligated with specific adapter oligonucleotides.²⁴ Left and right borders were amplified and PCR fragments were gel purified and sequenced. Primer sequences will be made available upon request.

Biochemical studies of autophagy and pexophagy

Nonspecific autophagy was monitored in S. cerevisiae by measuring uptake of Pho8Δ60 through an alkaline phosphatase assay.²⁵ A second method monitored processing of GFP-Atg8. Wild type (BY4742) and trs85Δ cells were transformed with pCuGFPAUT7(416) and grown in SMD lacking uracil to OD₆₀₀ = 1.0. Cells (5 ml) were harvested, washed once in SD-N medium and resuspended in 5 ml SD-N for 1, 2 and 4 h. At each time point, 1 ml of sample was removed and subjected to TCA precipitation, then analyzed by Western blotting. GFP-Atg8 fusion protein and free GFP were detected with a monoclonal antibody against GFP (Covance Research Products, Berkeley, CA). Pexophagy was analyzed by following the degradation of Fox3 as described elsewhere.⁹

The survival of Y. lipolytica cells under nitrogen starvation conditions in liquid cultures was examined as described previously.²⁶ For biochemical studies, Y. lipolytica cells from YPD cultures in the mid-log growth phase were washed twice with PiB and inoculated into YOE medium at OD₆₀₀ = 0.3 for 16 h (until the early-log growth phase). Then cells were washed twice with PiB and transferred to fresh GA medium at OD₆₀₀ = 1-2. Growth rates, enzymatic activities and protein levels were followed in culture samples taken after 0, 3, 6, 9 and 12 h of glucose adaptation. For enzymatic assays and immunoblotting, cells were washed twice with PiB at 4°C and resuspended in 0.2 ml of ice-cold PiB with 1 mM PMSF, 0.3 g of glass beads, and frozen at -20°C. Cell-free extracts were prepared by vortexing for 15 min at 4°C using a Fisher Vortex Genie 2 at speed 7.5. Protein concentrations were determined by the Lowry method,²⁷ using bovine serum albumin as the standard. AMO activity was measured in 50 mM phosphate buffer, pH 7.0, using 10 mM ethylamine as the substrate. This reaction led to the production of hydrogen peroxide, which was metabolized by horseradish peroxidase (2.5 U/ml) resulting in the oxidation of 1.1 mM 2,2′-azino-bis-(3-ethylbenz-thiazoline-6-sulphonic acid) that was followed by spectrometry at 420 nm.²⁸ The SDS-PAGE²⁹ and immunoblotting³⁰ were performed as previously described. Rabbit anti-thiolase (anti-THI) antibodies were kindly provided by Dr. Richard A. Rachubinski (Department of Cell Biology, University of Alberta, Edmonton, Canada). Rabbit anti-GroEL antibodies with cross-reactivity against the mitochondrial Hsp60 homologue of Y. lipolytica were kindly provided by Dr. Marten Veenhuis (Groningen Biomolecular Sciences and Biotechnology Institute, University of Groningen, Haren, The Netherlands). Antigen-antibody complexes were detected by enhanced chemiluminescence.

Studies of pexophagy by fluorescence microscopy

Y. lipolytica cells were transformed with 0.3 μg of NotI-digested pYEG1hyg-POX3-EYFP plasmid¹⁴ by the LiAc/LiCl method.²¹ The transformants with integrated peroxisomal fluorescent protein-encoding cassette 2 (PFC2; 14) were selected on YPD+hygromycin with 200 μg/ml of hygromycin B for H222 transformants and 50 μg/ml of hygromycin B for transformants of Ain16 and Ain19 strains. The induction of pexophagy for fluorescence microscopy, double-fluorescence labeling of peroxisomes and vacuoles, observation of peroxisome-vacuole dynamics with a single filter set and acquisition of photomicrographs were done as described previously.¹⁴

Cell labeling and immunoprecipitation

S. cerevisiae cells were grown in SMD medium to OD₆₀₀ = 1.0. Cells (10 ml) were collected and resuspended in 150 μl SMD and labeled with 100 μCi of Tran-³⁵S-label (ICN, Costa Mesa, CA) for 5 min and then subjected to a nonradioactive chase by adding 6.5 ml medium (SMD supplemented with 0.2% yeast extract, 2 mM methionine, and 1 mM cysteine) and incubating at 30°C. Samples were taken at the indicated time points, precipitated with 10% trichloroacetic acid and washed with acetone. Protein extracts were generated by glass bead lysis and subjected to immunoprecipitation as described previously.³¹ Rabbit antisera against Prc1, Pep4, Ape1 and Fox3 have been previously described.^9,31,32

Protease-sensitivity analysis

The protease-sensitivity experiment was carried out essentially as described previously.³³

RESULTS

Peroxisomes are not degraded in the Y. lipolytica Ain16 and Ain19 mutants

Over twenty proteins that are specific to autophagy have been identified in the past seven years³⁴ and these likely represent the majority of the components involved in autophagy, pexophagy and the cytoplasm to vacuole targeting pathway. For two reasons, however, we decided to pursue an additional screen for autophagy mutants. First, biases inherent in any one screen may result in certain mutants not being identified. Second, most of the mutants were identified in screens that were carried out to identify mutants in nonspecific autophagy and the Cvt pathway; however, one gene, ATG25, has already been identified from analyses of pexophagy in Hansenula polymorpha that lacks a homolog in S. cerevisiae, and another, ATG26, from Pichia pastoris that appears to be specific to pexophagy. Accordingly, we undertook a screen for pexophagy-defective mutants in the yeast Yarrowia lipolytica.

The AMO plate assay screening procedure¹⁷ was used as a first step in the selection of mutants deficient in peroxisome degradation. Tagged mutants were generated by insertional mutagenesis¹⁶ by inserting the PCR amplified zeta-URA3 mutagenesis cassette (MTC) into the genome of the Y. lipolytica H222-S4 strain. We tested more than 30,000 Ura⁺ transformants and identified 31 strains with reproducible defects in AMO inactivation on plates.¹⁵ We found that YOE (oleic acid, ethylamine) medium induced the highest levels of peroxisomal enzymes in Y. lipolytica liquid cultures (data not shown). Subsequent transfer of cells into GA (glucose, ammonium chloride) medium led to efficient induction of pexophagy that could be followed by the rates of inactivation of AMO (Fig. 1A and B) and isocitrate lyase (ICL; data not shown) and degradation of thiolase (THI; Fig. 1C). The stable level of the mitochondrial Hsp60 protein showed the specificity of pexophagy after the transfer of cells from YOE to GA (Fig. 1C). Two mutants, Ain16 and Ain19, exhibited delayed inactivation of both AMO (Fig. 1A and B), and ICL (data not shown), and degradation of THI (see Fig. 1C, for Ain19) after the transfer of YOE-induced cells into GA medium. In addition, both strains displayed a decreased viability under conditions of nitrogen starvation relative to the wild type strain, indicating a defect in nonspecific autophagy (Fig. 1D).

Figure 1.

The Ain16 and Ain19 mutants are defective in pexophagy and nonspecific autophagy. (A-C) Transfer of Y. lipolytica H222 wild type cells from YOE (oleic acid, ethylamine) to GA (glucose, ammonium chloride) medium for 12 h led to specific peroxisome degradation as detected by inactivation of peroxisomal AMO, degradation of peroxisomal THI and the stable level of mitochondrial Hsp60. (A and B) Inactivation of AMO was affected in the Y. lipolytica pexophagy mutants Ain16, Ain19 and their random integrative transformants (tr.1 and tr.2) with zeta-hph empty cassette. Transformation with the zeta-hph-YlTRS85 cassette restored the inactivation of AMO in both mutant strains. The activities were corrected for growth of the cells in GA medium. (C) Degradation of peroxisomal THI in the Ain19 mutant. (D) Ain16 and Ain19 mutants are starvation-sensitive. Cells were grown in YND medium, shifted to YND(-N) and incubated for the indicated number of days at 28°C, then spread on YPD plates. The number of viable colonies was determined as a percentage of those present at day zero essentially as described previously.²⁶

To study the stage of peroxisome degradation affected in Ain16 and Ain19 strains, we observed their peroxisome-vacuole dynamics by fluorescence microscopy under pexophagy conditions. For this purpose the wild type strain H222 and Y. lipolytica Ain16 and Ain19 mutants were transformed with a peroxisomal fluorescent protein-encoding cassette 2 (PFC2) and several random transformants of each strain were selected on YPD+hygromycin plates. Peroxisomes were labeled with a PFC2-encoded fusion of acyl-CoA oxidase 3 and enhanced yellow fluorescent protein.¹⁴ All transformants exhibited normal growth and peroxisomal fluorescence patterns in YOE medium (data not shown). After the transfer of cells from YOE to GA medium the vacuolar membranes were stained red with the FM 4-64 dye. Peroxisomes of the Y. lipolytica wild type strain were delivered to the vacuoles as seen by a decrease in punctate staining in the cytosol and a transient increase in fluorescence within the vacuole lumen (Fig. 2). This decrease in cytosolic punctate staining presumably reflected a macroautophagic mechanism, because the sites of close proximity of peroxisomes and vacuoles did not coincide with the sites of vacuolar membrane invagination.¹⁴ In most cases vacuoles surrounded the lipid bodies but not peroxisomes. Considering that microautophagy of peroxisomes is an easily recognized series of events in double labeling experiments,³⁵ we could exclude this mode of pexophagy in our studies. In contrast to the H222 wild type strain, almost all of the peroxisomes remained outside the vacuoles in Ain16 (data not shown) and Ain19 (Fig. 2) cells even after 9 h of adaptation to glucose, and the vacuole lumens remained dark (nonfluorescent). This result indicated that macropexophagy was blocked at the stage of pexophagosome formation or fusion of pexophagosomes with the vacuolar membrane. We noted, however, that the vacuoles appeared to be fragmented into smaller vesicles in both Y. lipolytica Ain16 and Ain19 mutants (see Fig. 2 for Ain19) indicating a possible defect in vacuole biogenesis. Thus, we could not rule out a nonspecific fusion defect that resulted from the aberrant vacuole morphology, which may have reflected a mutation in proteins known to be involved in homotypic vacuole fusion.³⁶

Figure 2.

Peroxisomes are not delivered to the vacuoles in the Y. lipolytica Ain19 mutant strain. Peroxisome-vacuole dynamics were followed by fluorescence microscopy after the transfer of cells from YOE (oleic acid, ethylamine) to GA (glucose, ammonium chloride) medium. Peroxisomes were labelled yellow-green with the fusion of acyl-CoA oxidase 3 and enhanced yellow fluorescent protein and vacuolar membranes were stained red with FM 4-64 as described in Materials and Methods.

The Y. lipolytica Ain16 and Ain19 mutants are both disrupted in the TRS85 gene

To identify the mutated gene that caused the pexophagy defect in the Ain16 and Ain19 mutants, we proceeded with an analysis of the MTC integration events. The genomic DNA of Ain16 and Ain19 strains was analyzed as described in Materials and Methods. We amplified and sequenced 370 nucleotides and 570 nucleotides of the left borders from Ain16 and Ain19, respectively. Both DNA fragments appeared to be parts of the same open reading frame that was obtained during the sequencing project of the Y. lipolytica genome.³⁷ BLAST analysis of the predicted protein product revealed that the MTC disrupted the homolog of S. cerevisiae TRS85, the 85 kDa subunit of transport protein particle (TRAPP).³⁸ The ScTRS85 gene was originally identified as GSG1 due to an indirect effect on sporulation.^39,40 Recently, ScTrs85 was shown to be the component of two forms of TRAPP, TRAPP I and TRAPP II, which mediate ER-to-Golgi and Golgi transport events, respectively.⁴¹ The complete Y. lipolytica TRS85 gene encodes a predicted protein of 573 amino acids displaying 22% identity and 41% similarity to Trs85 of S. cerevisiae. Although the score of homology is not very high, there are no other homologues of ScTRS85 in the Y. lipolytica genome. Insertion of the MTC in YlTRS85 occurred 171 nucleotides and 1106 nucleotides downstream of the predicted translational start codon in Ain16 and Ain19, respectively. At the amino acid level, the MTC disrupted YlTrs85 downstream of A57 in Ain16 and downstream of R368 in Ain19, producing a truncated protein in each case. The same overall size of EcoRV MTC borders in Ain16 and Ain19 obtained by Southern blotting (data not shown) excluded the possibility of genomic DNA deletion during MTC insertion and suggested that the pexophagy phenotype was related to disruption of YlTRS85 in both strains. The latter conclusion was also confirmed by complementation studies that showed the complete restoration of AMO inactivation in both Y. lipolytica trs85 mutants by transformation with a cassette containing the YlTRS85 gene (Fig. 1A and B).

The S. cerevisiae trs85Δ mutant is defective in selective autophagy

The identification of YlTRS85 as the mutated gene in the Ain16 and Ain19 mutants suggested that a defect in vacuole biogenesis was not the cause of the observed pexophagy defect. Nonetheless, the fragmented vacuole phenotype made it problematic to determine whether the pexophagy defect was a direct result of the mutation in YlTRS85. Independent of the screen described above, we had screened the systematic S. cerevisiae gene deletion library for strains that were defective in the transport of precursor aminopeptidase I (prApe1; 42). One of the mutants that displayed a strong block in processing of the precursor protein was deleted for the ScTRS85 gene. In light of the above results, we examined this phenotype more carefully. Protein extracts were prepared from isogenic wild type and mutant cells grown under vegetative conditions and examined by Western blot. Wild type cells displayed predominantly the mature form of Ape1, indicating efficient delivery to the vacuole and removal of the propeptide (Fig. 3A). In contrast, the Sctrs85Δ cells accumulated only the precursor form of the protein indicating a complete block in vacuolar delivery or proteolytic processing (Fig. 3A). The Ape1 phenotype of the Sctrs85Δ mutant was restored to that similar to the wild type strain by transformation with a plasmid encoding ScTRS85, indicating that the initial defect was due to the deletion of ScTRS85 (Fig. 3A).

Figure 3.

(A) The Cvt pathway is blocked in the Sctrs85Δ strain. The wild type (BY4742), Sctrs85Δ strain and the Sctrs85Δ strain harboring the pRS416 empty vector or pCuTRS85(416) plasmid were cultured in SMD medium to mid-log phase. Samples were collected and the proteins were precipitated with TCA and subjected to SDS-PAGE followed by Western blotting with antiserum to Ape1 as described in Materials and Methods. The positions of precursor and mature Ape1 are indicated. (B) Vacuoles in the Sctrs85Δ strain have a morphology similar to wild type. Cells were grown in YPD and vacuoles stained with FM 4-64 as described in Materials and Methods. DIC, differential interference contrast. (C) Pep4 (PrA) and Prc1 (CpY) are processed normally in the Sctrs85Δ strain. Wild type and Sctrs85Δ strains were grown to mid-log phase and then subjected to a pulse-chase analysis at the indicated time points to monitor the kinetics of Pep4 and Prc1 processing as described in Materials and Methods. The positions of precursor and mature forms of both proteins are indicated.

Because the Y. lipolytica mutant had a fragmented vacuole phenotype, we analyzed the morphology of the vacuole in the Sctrs85Δ strain. Although a higher percentage of the Sctrs85Δ cells had a fragmented vacuole phenotype, the total number of cells with normal vacuoles was not substantially different from the wild type strain (Fig. 3B); the wild type and Sctrs85Δ strains had 28% and 13% of the cells displaying one vacuole, 49% and 48% displaying 2-4 vacuoles and 23% and 39% displaying more than four vacuoles, respectively. As a separate means of assessing the competency of the vacuole, we decided to examine the transport and processing of two resident vacuolar hydrolases, Pep4 and Prc1, which are delivered to the vacuole via the CPY pathway through the use of transient transport vesicles.⁴³ S. cerevisiae wild type and trs85Δ cells were analyzed by pulse/chase labelling and immunoprecipitation as described in Materials and Methods. The Sctrs85Δ cells showed essentially identical kinetics for the maturation of both hydrolases (Fig. 3C) suggesting that the vacuole in the Sctrs85Δ mutant cells was competent for fusion with transport vesicles. Furthermore, the normal processing of both precursor proteins indicated that the Sctrs85Δ vacuoles retained adequate proteolytic activity to remove the prApe1 propeptide.

The above results suggested that prApe1 was not delivered to the vacuole in the Sctrs85Δ mutant strain. An alternative possibility, however, was that the Sctrs85Δ mutant was defective in the breakdown of the subvacuolar vesicle, the Cvt body, that results from the fusion of the double-membrane Cvt vesicle with the vacuole.⁴⁴ To differentiate between these possibilities, we carried out a protease-sensitivity analysis. Control Scpep4Δ and Scatg1Δ cells and Sctrs85Δ mutant cells were converted to spheroplasts and lysed under conditions that retain the integrity of subcellular compartments. Exogenous protease was added and the state of Ape1 was determined by Western blot. If prApe1 were present within a completed cytosolic vesicle or a subvacuolar vesicle it would be protected from exogenous protease, whereas prApe1 present in a partial vesicle in the cytosol would be sensitive to removal of the propeptide.³³ The Scpep4Δ mutant is defective in subvacuolar vesicle breakdown; in this strain prApe1 was protected from exogenous protease in the absence of detergent (Fig. 4), reflecting its localization within Cvt bodies present in the vacuole lumen. In contrast, prApe1 was protease-sensitive in the absence of detergent in the Scatg1Δ strain, which is defective in the formation of Cvt vesicles. We found that prApe1 was also protease-sensitive in the Sctrs85Δ strain (Fig. 4) indicating that there was a defect in formation of the Cvt vesicle rather than in breakdown of the Cvt body. Because the Sctrs85Δ strain did not have a severely fragmented vacuole we decided to pursue our analysis of the role of Trs85 in S.cerevisiae.

Figure 4.

The Sctrs85Δ mutant accumulates prApe1 in a protease-sensitive form. Spheroplasts isolated from Scpep4Δ, Sctrs85Δ and Scatg1Δ cells were lysed in osmotic lysis buffer. An aliquot was removed for a total lysate control (T). Supernatant (S) and pellet (P) fractions after a 5,000xg centrifugation were collected, and the pellet fractions were subjected to protease treatment in the absence or presence of 0.2% Triton X-100 as described in Materials and Methods. The resulting samples were subjected to immunoblot analysis with antiserum against Ape1.

Trs85 is required for pexophagy in S. cerevisiae

The import of prApe1 is one type of specific autophagy. The degradation of peroxisomes also occurs through a specific mechanism. We used the peroxisomal matrix protein Fox3 as a marker for peroxisomes to determine whether pexophagy in S. cerevisiae was also dependent on Trs85. Wild type, Scpep4Δ and Sctrs85Δ cells were grown in medium containing oleic acid as the sole carbon source to induce peroxisome proliferation. Cells were then shifted to SD-N medium to initiate pexophagy. Protein extracts were prepared at different times and analyzed by Western blot. Fox3 was almost completely degraded in wild type cells after 6 h (Fig. 5). The Scpep4Δ strain is defective in vacuolar hydrolase activity and unable to degrade peroxisomes, thus serving as a negative control; the Fox3 protein level remained unchanged throughout the 24 h time course (Fig. 5). Similar to the results with the Scpep4Δ strain, the Sctrs85Δ mutant displayed essentially no change in the level of Fox3 (Fig. 5) indicating a defect in pexophagy.

Figure 5.

The Sctrs85Δ strain is defective in pexophagy. Wild type, Scpep4Δ and Sctrs85Δ strains were grown in YTO medium to induce peroxisomes and then were shifted to SD-N to induce pexophagy as described in Materials and Methods. Identical volume samples from each indicated time point were collected and subjected to TCA precipitation. Protein samples were processed for Western blotting using polyclonal antiserum against Fox3. Fox3 was degraded in the wild type strain but remained stable in the Sctrs85Δ strain and a vacuolar protease-deficient strain Scpep4Δ.

Trs85 is required for nonspecific autophagy in S. cerevisiae

Pexophagy and the Cvt pathway share most of the same molecular machinery as nonspecific autophagy; however, some of the components are specific to each pathway.¹ Accordingly, we used three different assays to test whether ScTrs85 was required for nonspecific autophagy. Atg8 is required for Cvt vesicle formation and autophagosome expansion.⁴⁵ The Atg8 protein lines both sides of the forming autophagosome and a portion of the protein remains associated with the completed autophagosome. This population of Atg8 is degraded in the vacuole lumen following fusion with the vacuole. GFP-Atg8 can functionally replace the endogenous Atg8 protein. When GFP-Atg8 is delivered to the vacuole via autophagy the Atg8 portion of the protein is degraded, releasing free GFP that is more resistant to proteolysis.⁴⁶ Cleavage of GFP-Atg8 provides a sensitive measure for autophagosome fusion with the vacuole. We monitored the release of GFP from GFP-Atg8 in wild type and Sctrs85Δ cells. In the wild type strain free GFP appeared by the one h time point and increased over time, with a concomitant decrease in the level of the full-length fusion protein (Fig. 6A). In contrast, free GFP was not produced in the Sctrs85Δ mutant indicating a block in autophagy. We also monitored delivery of the GFP-Atg8 protein through fluorescent microscopy. In wild type cells GFP-Atg8 could be seen at the pre-autophagosomal structure, the site of autophagosome formation, and within the vacuole lumen after four h in SD-N medium (Fig. 6B). GFP-Atg8 could be seen at the PAS but did not appear in the vacuole lumen in the Sctrs85Δ strain (Fig. 6B) indicating a block in autophagic transport in agreement with the analysis of GFP-Atg8 processing.

Figure 6.

GFP-Atg8 is not efficiently transported into the vacuole in the Sctrs85Δ strain. (A) GFP-Atg8 is not degraded in the Sctrs85Δ strain. Wild type and Sctrs85Δ cells transformed with the pCuGFPAUT7(416) plasmid were grown in selective minimal medium. Cells at mid-log phase were shifted to SD-N for the indicated times. Equal volume samples from each time point were TCA precipitated and processed for Western blotting. The full-length GFP-Atg8 fusion protein and free GFP moiety were detected with antibody against GFP. (B) The Sctrs85Δ strain does not accumulate vacuolar GFP-Atg8. The same two strains used in (A) were grown as described above, then shifted to SD-N for 4 h. Cells were collected and monitored by fluorescence microscopy as described in Materials and Methods. Arrows mark some of the sites corresponding to the PAS. DIC, differential interference contrast.

Finally, we examined the uptake of a bulk cytosolic marker, Pho8Δ60. Pho8Δ60 lacks the N-terminal transmembrane domain that allows entry of wild type Pho8 into the secretory pathway.⁴⁷ As a result, Pho8Δ60 can only be delivered to the vacuole via autophagy. Delivery results in removal of the C-terminal propeptide that can be monitored by following an increase in enzymatic activity.²⁵ We measured Pho8Δ60-dependent alkaline phosphatase activity in wild type, Scatg1Δ and Sctrs85Δ cells following a shift to SD-N to induce autophagy. All three strains showed a similar basal level of activity when grown in rich medium. After four h in starvation conditions, there was a large increase in activity in the wild type strain (Fig. 7). The Scatg1Δ mutant is completely defective in autophagy and maintained a similar level of Pho8Δ60 activity before and after shifting to SD-N medium (Fig. 7). Similarly, the Sctrs85Δ mutant showed essentially no increase in activity under autophagy-inducing conditions (Fig. 7). Together, these results indicate that ScTrs85 is defective in nonspecific autophagy in agreement with the decreased viability seen in the Yltrs85 strains (Fig. 1D).

Figure 7.

Nonspecific autophagy is blocked in the Sctrs85Δ strain. Wild type (YTS159), Scatg1Δ (YTS161) and Sctrs85Δ (YJH3) strains were grown in SMD and shifted to SD-N for 4 h. Autophagy was monitored by measuring the Pho8Δ60-dependent alkaline phosphatase (AlP) activity as described in Materials and Methods. The activity of the wild type strain in SD-N was set as 100% activity and the other strains were normalized relative to this level. Error bars represent the standard deviation from three separate experiments.

DISCUSSION

In this study, we describe the identification of the TRS85 gene that was isolated in independent screens in Y. lipolytica and S. cerevisiae. In Y. lipolytica, trs85 mutants Ain16 and Ain19 were isolated using a plate screen for strains with delayed AMO inactivation under pexophagy conditions. For both mutants the pexophagy phenotype was confirmed in liquid cultures by following the inactivation of peroxisomal AMO and degradation of THI in cell-free extracts (Fig. 1A-C) and by following peroxisome-vacuole dynamics by fluorescence microscopy (Fig. 2) after the transfer of cells from YOE (oleic acid, ethylamine) to GA (glucose, ammonium chloride) medium that leads to efficient induction of pexophagy in the Y. lipolytica wild type strain. In S. cerevisiae, the mutant was isolated in a screen for Cvt pathway-defective strains that accumulated the precursor form of the vacuolar hydrolase Ape1.

The Y. lipolytica Ain16 and Ain19 mutants appeared to be disrupted at two different locations of the same gene that corresponded to the Y. lipolytica homologue of S. cerevisiae TRS85. ScTRS85 encodes the 85 kDa subunit of TRAPP I and TRAPP II, protein complexes that are required for ER-to-Golgi and Golgi transport, respectively.^38,41 The Yltrs85 mutants had a fragmented vacuole that made it problematic to demonstrate that the pexophagy defect was a direct result of the absence of YlTrs85. In contrast, the Sctrs85Δ mutant displayed a relatively minor defect in vacuole morphology (Fig. 3B). In addition, this strain showed normal processing of two vacuolar hydrolases that transit through the CPY pathway (Fig. 3C), a vacuolar targeting route that is mostly independent from the autophagic pathway. Thus, the vacuoles of the Sctrs85Δ mutant retain the ability to fuse with donor compartments such as the autophagosome, and also possess the proteolytic activity necessary to process newly delivered zymogens such as prApe1. The Sctrs85Δ mutant, however, accumulated only the precursor form of Ape1 (Fig. 3A) in a proteaseaccessible state (Fig. 4), indicating a block in the Cvt pathway. Similarly, this mutant was defective in pexophagy (Fig. 5) and nonspecific autophagy (Figs. 6 and 7) similar to the Y. lipolytica mutant strains (Figs. 1 and 2).

S. cerevisiae TRAPP I consists of 7 different subunits (Bet5, Trs20, Bet3, Trs23, Trs31, Trs33 and Trs85) and TRAPP II contains an additional 3 subunits (Trs65, Trs120 and Trs130).⁴¹ We can speculate why only trs85 mutants were isolated in our screens. In S. cerevisiae, only 3 of the 10 TRAPP subunits are dispensable for growth: Trs33, Trs65 and Trs85.³⁸ However, the loss of Trs33 and Trs65 did not alter the assembly of the complex and membrane traffic.⁴¹ Consequently, only trs85 unconditional mutants could be directly selected for a pexophagy or prApe1accumulation phenotype. Nevertheless, to our knowledge this is the first report that implicates the TRS85 gene in autophagy-related pathways. It has been suggested that the early secretory pathway is needed for nonspecific autophagy but not for the Cvt pathway in yeast;^6,7 however, a recent report indicates that the early secretory pathway is also required for the Cvt pathway, suggesting a role in supplying the membrane during specific types of autophagy.⁸ We suspect that the early secretory pathway is also involved in pexophagy; however, we cannot conclude that result from the present data. The finding that the Sctrs85Δ strain showed no defect in the kinetics of Pep4 and Prc1 processing (Fig. 3C) suggests normal delivery through the early secretory pathway.

The function of Trs85 in autophagy has not been identified yet, but from the evidence that the PAS is normally formed but Atg8 is not transported into the vacuole in the Sctrs85Δ mutant (Fig. 6) and peroxisomes are not delivered to the vacuole in the Yltrs85 mutants (Fig. 2), we can speculate that it functions before the vesicle docking and fusion stage, and that it is probably involved in vesicle formation/completion, similar to most of the other Atg proteins. The finding that the Sctrs85Δ strain accumulates protease-sensitive prApe1 supports this model (Fig. 4). Finally, we note that it was reported that in a TRS85 deletion strain, ER-Golgi transport of Prc1 is blocked at 37°C (Sacher et al, 2001), but this phenotype was not seen in our analysis; the Sctrs85Δ strain used in this study displayed normal processing of Prc1 and Pep4 at both 30 (Fig. 3C) and 37 (data not shown) degrees. This discrepancy may reflect differences due to the strain background. Further study is required to address the function of Trs85 and to examine whether other TRAPP subunits are also involved in the Cvt, autophagy and pexophagy pathways.

Table 1.

Yeast strains used in this study

Strain	Genotype	Source
S. cerevisiae
BY4742	MATa his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0	ResGen™
atg1Δ	BY4742 atg1Δ::KAN	ResGen™
pep4Δ	BY4742 pep4Δ::KAN	ResGen™
trs85D	BY4742 trs85Δ::KAN	ResGen™
YJH3	YTS159 trs85Δ::URA3	This study
YTS159	BY4742 pho8Δ60 pho13Δ::KAN	This study
YTS161	BY4742 pho8Δ60 pho13Δ::KAN atg1Δ::URA3	This study
Y. lipolytica
H222	MATA ylT1-free wild type	(21)
H222-S4	H222 ura3-302::SUC2	(16)
Ain16	H222-S4 trs85-1::zeta-URA3	(15; this study)
Ain19	H222-S4 trs85-2::zeta-URA3	(15; this study)