Interactions of the chaperone Hsp104 with yeast Sup35 and mammalian PrP

[PSI+] is a genetic element in yeast for which a heritable change in phenotype appears to be caused by a heritable change in the conformational state of the Sup35 protein. The inheritance of [PSI+] and the physical state of Sup35 in vivo depend on the protein chaperone Hsp104 (heat shock protein 104). Although these observations provide a strong genetic argument in support of the "protein-only" or "prion" hypothesis for [PSI+], there is, as yet, no direct evidence of an interaction between the two proteins. We report that when purified Sup35 and Hsp104 are mixed, the circular dichroism (CD) spectrum differs from that predicted by the addition of the proteins' individual spectra, and the ATPase activity of Hsp104 is inhibited. Similar results are obtained with two other amyloidogenic substrates, mammalian PrP and beta-amyloid 1-42 peptide, but not with several control proteins. With a group of peptides that span the PrP protein sequence, those that produced the largest changes in CD spectra also caused the strongest inhibition of ATPase activity in Hsp104. Our observations suggest that (i) previously described genetic interactions between Hsp104 and [PSI+] are caused by direct interaction between Hsp104 and Sup35; (ii) Sup35 and PrP, the determinants of the yeast and mammalian prions, respectively, share structural features that lead to a specific interaction with Hsp104; and (iii) these interactions couple a change in structure to the ATPase activity of Hsp104.     

Antagonistic interactions between yeast chaperones Hsp104 and Hsp70 in prion curing

The maintenance of [PSI], a prion-like form of the yeast release factor Sup35, requires a specific concentration of the chaperone protein Hsp104: either deletion or overexpression of Hsp104 will cure cells of [PSI]. A major puzzle of these studies was that overexpression of Hsp104 alone, from a heterologous promoter, cures cells of [PSI] very efficiently, yet the natural induction of Hsp104 with heat shock, stationary-phase growth, or sporulation does not. These observations pointed to a mechanism for protecting the genetic information carried by the [PSI] element from vicissitudes of the environment. Here, we show that simultaneous overexpression of Ssa1, a protein of the Hsp70 family, protects [PSI] from curing by overexpression of Hsp104. Ssa1 protein belongs to the Ssa subfamily, members of which are normally induced with Hsp104 during heat shock, stationary-phase growth, and sporulation. At the molecular level, excess Ssa1 prevents a shift of Sup35 protein from the insoluble (prion) to the soluble (cellular) state in the presence of excess Hsp104. Overexpression of Ssa1 also increases nonsense suppression by [PSI] when Hsp104 is expressed at its normal level. In contrast, hsp104 deletion strains lose [PSI] even in the presence of overproduced Ssa1. Overproduction of the unrelated chaperone protein Hsp82 (Hsp90) neither cured [PSI] nor antagonized the [PSI]-curing effect of overproduced Hsp104. Our results suggest it is the interplay between Hsp104 and Hsp70 that allows the maintenance of [PSI] under natural growth conditions.     

Genesis and variability of [PSI] prion factors in Saccharomyces cerevisiae

We have previously shown that multicopy plasmids containing the complete SUP35 gene are able to induce the appearance of the non-Mendelian factor [PSI]. This result was later interpreted by others as a crucial piece of evidence for a model postulating that [PSI] is a self-modified, prion-like conformational derivative of the Sup35 protein. Here we support this interpretation by proving that it is the overproduction of Sup35 protein, and not the excess of SUP35 DNA or mRNA that causes the appearance of [PSI]. We also show that the "prion-inducing domain" of Sup35p is in the N-terminal region, which, like the "prion-inducing domain" of another yeast prion, Ure2p, was previously shown to be distinct from the functional domain of the protein. This suggests that such a chimeric organization may be a common pattern of some prion elements. Finally, we find that [PSI] factors of different efficiencies and different mitotic stabilities are induced in the same yeast strain by overproduction of the identical Sup35 protein. We suggest that the different [PSI]-containing derivatives are analogous to the mysterious mammalian prion strains and result from different conformational variants of Sup35p.     

Presence of sauvagine-like epitopes in the interrenal gland of the bullfrog Rana catesbeiana

The objective of this study was to determine if a variety of hepatotoxicants could induce the level of heat shock protein 70I, and whether or not elevated levels of heat shock proteins (hsp's) could provide cytoprotection from those hepatotoxicants. Exposure of HepG2 cells to cytotoxic concentrations of bromobenzene, cadmium, cyclophosphamide, or diethylnitrosamine increased the level of hsp 70I protein and mRNA, while carbon tetrachloride and cocaine had no effect on hsp 70I or mRNA levels. To determine if induction of hsp 70I might afford protection against cytotoxicity, HepG2 cells were given a prior sublethal heat shock (sub-LHS) (43 degrees C for 1 hr) to induce hsp's and then challenged 24 hr later with the hepatotoxicants. Sub-LHS pretreatment diminished toxicity from bromobenzene, cadmium, cyclophosphamide, or diethyl-nitrosamine, but not carbon tetrachloride or cocaine. In cells treated with [14C]carbon tetrachloride or [3H]cocaine, no detectable covalent binding to proteins was observed; whereas, [14C]-bromobenzene treatment resulted in substantial covalent binding to cellular protein. The apparent absence of formation of reactive metabolite adducted proteins from cocaine and carbon tetrachloride may explain why no hsp 70I induction was observed with these agents. The correlation between hepatotoxicant induction of hsp 70I and cytoprotection afforded by sub-LHS pretreatment suggests that hsp 70I induction may represent an important cellular defense mechanism in the liver.     

Regulation of tubulin polypeptides and microtubule function: Luv1p [correction of Rki1p] interacts with the beta-tubulin binding protein Rbl2p

Genetic analysis of microtubule functions in the yeast Saccharomyces cerevisiae suggests that cells manage the levels and activities of the tubulin polypeptides. These reactions may be involved in protein folding, formation of the heterodimer, and maintenance of the appropriate balance between alpha- and beta-tubulin. One protein involved in these functions is Rbl2p, which forms a complex with beta-tubulin. Here we describe the identification of a novel yeast gene, LUV1 [corrected], that interacts genetically with RBL2. Deletion of rki1 causes conditional defects in microtubule assembly and cell growth. Luv1p [corrected] can be isolated in a complex containing Rbl2p. The results support the existence of cellular mechanisms for regulating microtubule function through the tubulin polypeptides.     

A critical role for amino-terminal glutamine/asparagine repeats in the formation and propagation of a yeast prion

The yeast [PSI+] factor propagates by a prion-like mechanism involving self-replicating Sup35p amyloids. We identified multiple Sup35p mutants that either are poorly recruited into, or cause curing of, wildtype amyloids in vivo. In vitro, these mutants showed markedly decreased rates of amyloid formation, strongly supporting the protein-only prion hypothesis. Kinetic analysis suggests that the prion state replicates by accelerating slow conformational changes rather than by providing stable nuclei. Strikingly, our mutations map exclusively within a short glutamine/asparagine-rich region of Sup35p, and all but one occur at polar residues. Even after replacement of this region with polyglutamine, Sup35p retains its ability to form amyloids. These and other considerations suggest similarities between the prion-like propagation of [PSI+] and polyglutamine-mediated pathogenesis of several neurodegenerative diseases.     

Long non-stop reading frames on the antisense strand of heat shock protein 70 genes and prion protein (PrP) genes are conserved between species

Several mammalian genes, including heat shock protein (Hsp70) and prion protein (PrP) genes, have been reported to have long open reading frames (ORFs) or non-stop reading frames (NRFs) in the antisense direction. A simple explanation would be that these long antisense reading frames, which are usually in the same triplet frame as the coding strand, are the fortuitous byproduct of a high overall [G+C] content with concomitant preference for G/C over A/T in the third codon position, a preference for RNY type codons (purine/any nucleotide/pyrimidine), and/or a bias against serine and leucine, the only amino acids with codons that can be read as stop codons in the antisense direction. The PrP genes and most heat shock genes with long antisense NRFs (aNRFs) are indeed relatively [G+C] rich but do not show a bias against serine and leucine. In several vertebrates investigated, at least one of the Hsp70 genes has a long antisense reading frame, and we found that some, though not all, putative stop codons in long Hsp70 antisense reading frames were due to sequencing errors. The PrP gene contains an extended antisense open reading frame in all 45 eutherian mammals tested, but not in a marsupial and in a bird. In the PrP gene, the long, protein-coding exon also harbors the antisense nonstop reading frame. In both Hsp70 and PrP genes, the putative antisense protein sequence is well conserved. Even though there is no clear evidence in Hsp70 or PrP genes for the existence of the respective antisense proteins, we speculate that such antisense proteins serve to regulate the genuine Hsp and PrP proteins under special circumstances. Alternatively, regulation might occur at the RNA level, and the antisense RNA would merely lack stop codons to prevent its rapid degradation by an mRNA quality control mechanism that is triggered by premature stop codons. We note that both Hsp and PrP are involved in physiological or pathological protein aggregation phenomena, that scrapie prions have been reported to modify the expression or localization of heat shock proteins, and that in yeast, propagation of a prion-like state (PSI+) depends on a heat shock (Hsp104) protein.     

Interaction between yeast Sup45p (eRF1) and Sup35p (eRF3) polypeptide chain release factors: implications for prion-dependent regulation

The SUP45 and SUP35 genes of Saccharomyces cerevisiae encode polypeptide chain release factors eRF1 and eRF3, respectively. It has been suggested that the Sup35 protein (Sup35p) is subject to a heritable conformational switch, similar to mammalian prions, thus giving rise to the non-Mendelian [PSI+] nonsense suppressor determinant. In a [PSI+] state, Sup35p forms high-molecular-weight aggregates which may inhibit Sup35p activity, leading to the [PSI+] phenotype. Sup35p is composed of the N-terminal domain (N) required for [PSI+] maintenance, the presumably nonfunctional middle region (M), and the C-terminal domain (C) essential for translation termination. In this study, we observed that the N domain, alone or as a part of larger fragments, can form aggregates in [PSI+] cells. Two sites for Sup45p binding were found within Sup35p: one is formed by the N and M domains, and the other is located within the C domain. Similarly to Sup35p, in [PSI+] cells Sup45p was found in aggregates. The aggregation of Sup45p is caused by its binding to Sup35p and was not observed when the aggregated Sup35p fragments did not contain sites for Sup45p binding. The incorporation of Sup45p into the aggregates should inhibit its activity. The N domain of Sup35p, responsible for its aggregation in [PSI+] cells, may thus act as a repressor of another polypeptide chain release factor, Sup45p. This phenomenon represents a novel mechanism of regulation of gene expression at the posttranslational level.     

Virulence factors of Cryptococcus neoformans

Cryptococcus neoformans is an encapsulated yeast which causes cryptococcosis, a disease typified by an initial pulmonary infection which can disseminate to cause a life threatening meningoencephalitis. Although the disease may occur in individuals who show no evidence of immunosuppression it has had it most significant impact as an infection in patients with AIDS. Research into the potential virulence factors of this yeast has recently attracted particular attention. Capsule synthesis has been the focus of most interest and it is now established as a major virulence determinant. The mechanisms by which the capsule and capsular material effect the immune response have now largely been elucidated, and the genes underlying capsular synthesis are now under investigation. The isolation of mutants incapable of melanogenesis have implicated this process in the pathogenesis of C. neoformans infections, and evidence suggests that the production of melanin protects the yeast against oxidant induced damage. There is also some genetic evidence for the potential involvement of temperature tolerance and mating types in the virulence of this encapsulated yeast. The roles of other potential C. neoformans virulence determinants are more speculative; these include proteinase production, release of polyol metabolites, interaction with hormones, adherence and production of mannoproteins. The involvement of housekeeping enzyme systems in the maintenance of infection by C. neoformans is now also under active investigation.     

delta-N-methylarginine is a novel posttranslational modification of arginine residues in yeast proteins

We have found a novel modification of protein arginine residues in the yeast Saccharomyces cerevisiae. Intact yeast cells lacking RMT1, the gene encoding the protein omega-NG-arginine methyltransferase, were labeled with the methyl donor S-adenosyl-L-[methyl-3H]methionine. The protein fraction was acid-hydrolyzed to free amino acids, which were then fractionated on a high resolution sulfonated polystyrene cation exchange column at pH 5.27 and 55 degreesC. In the absence of the omega-NG, NG-[3H]dimethylarginine product of the RMT1 methyltransferase, we were able to detect a previously obscured 3H-methylated species that migrated in the region of methylated arginine derivatives. The [3H]methyl group(s) of this unknown species were not volatilized by treatment with 2 M NaOH at 55 degreesC for up to 48 h, suggesting that they were not modifications of the terminal omega-guanidino nitrogen atoms. However, this base treatment did result in the formation of a new 3H-methylated derivative that co-chromatographed with delta-N-methylornithine on high resolution cation exchange chromatography, on reverse phase high pressure liquid chromatography, and on thin layer chromatography. From these data, we suggest that the identity of the original unknown methylated residue is delta-N-monomethylarginine. The presence of this methylated residue in yeast cells defines a novel type of protein modification reaction in eukaryotes.     

Glucose repression/derepression in budding yeast: SNF1 protein kinase is activated by phosphorylation under derepressing conditions, and this correlates with a high AMP:ATP ratio

BACKGROUND: Genetic studies of Saccharomyces cerevisiae have shown that Snf1p and Snf4p, which together form the SNF1 complex, are essential for gene derepression on removal of glucose from the medium. However the metabolic signal(s) involved, and the exact role of SNF1, have remained enigmatic. Recently, the AMP-activated protein kinase (AMPK) was shown to be the mammalian homologue of SNF1. AMPK is activated by the elevation of the cellular AMP:ATP ratio, which occurs during cellular stress in mammalian cells. The mechanism of activation involves phosphorylation of AMPK by an upstream protein kinase (AMPKK). We have investigated whether a similar mechanism might explain the role of SNF1 in yeast in the response to the stress of glucose starvation. RESULTS: The protein kinase activity of SNF1 was dramatically and rapidly activated by phosphorylation on removal of glucose from the medium. SNF1 was not activated directly by AMP, but could be inactivated by protein phosphatases and reactivated by mammalian AMPKK. We also demonstrated that an endogenous SNF1-reactivating factor, most likely an upstream protein kinase, is present in yeast extracts. Under a variety of different growth conditions, there was a correlation between cellular adenine nucleotide levels and the activation state of SNF1. CONCLUSIONS: Apart from the lack of direct allosteric activation of SNF1 by AMP, the regulation of the mammalian AMPK and yeast SNF1 protein kinase cascades is highly conserved. Adenine nucleotides are now good candidates for metabolic signals which indicate the lack of glucose in the medium, triggering activation of SNF1 and derepression of glucose-repressed genes.     

AMPA receptors and bacterial periplasmic amino acid-binding proteins share the ionic mechanism of ligand recognition

In order to identify key structural determinants for ligand recognition, we subjected the ligand-binding domain of the alpha-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid (AMPA)-selective glutamate receptor GluR-D subunit to site-directed mutagenesis. Based on the analysis of the [3H]AMPA-binding properties of the mutated binding sites, we constructed a revised three-dimensional model of the ligand-binding site, different in many respects from previously published models. In particular, our results indicate that the residues Arg507 and Glu727 represent the structural and functional correlates of Arg77 and Asp161 in the homologous bacterial lysine/ornithine/arginine-binding protein and histidine-binding protein, and directly interact with the alpha-carboxyl and alpha-amino group of the bound ligand, respectively. In contrast, Glu424, implicated previously in ionic interactions with the alpha-amino group of the agonist, is unlikely to have such a role in ligand binding. Our results indicate that glutamate receptors share with the bacterial polar amino acid-binding proteins the fundamental mechanism of amino acid recognition.     

The transfer of retinol from serum retinol-binding protein to cellular retinol-binding protein is mediated by a membrane receptor

The hypothesis that the cellular uptake of retinol involves the specific interaction of a plasma membrane receptor with serum retinol-binding protein (RBP) at the extracellular surface followed by ligand transfer to cytoplasmic cellular retinol-binding protein (CRBP) has been investigated. The experimental system consisted of the [3H]retinol-RBP complex, Escherichia coli-expressed recombinant apo-CRBP containing the 10 amino acid long streptavidin-binding peptide sequence at its C terminus (designated as CRBP-Strep) and permeabilized human placental membranes. [3H]Retinol transfer from RBP to CRBP-Strep was monitored by measuring the radioactivity associated with CRBP-Strep retained by an immobilized streptavidin resin. Using this assay system, we have demonstrated that optimal retinol uptake is achieved with holo-RBP, the membrane receptor and apo-CRBP. The effects are specific: other binding proteins, including beta-lactoglobulin and serum albumin, despite their ability to bind retinol, failed to substitute for either RBP or apo-CRBP. The process is facilitated by membranes containing the native receptor suggesting that this protein is an important component in the transfer mechanism. Taken together, the data suggest that the RBP receptor, through specific interactions with the binding proteins, participates (either directly or via associated proteins) in the mechanism which mediates the transfer of retinol from extracellular RBP to intracellular CRBP.     

The underlying pathway structure of biochemical reaction networks

Bioinformatics is yielding extensive, and in some cases complete, genetic and biochemical information about individual cell types and cellular processes, providing the composition of living cells and the molecular structure of its components. These components together perform integrated cellular functions that now need to be analyzed. In particular, the functional definition of biochemical pathways and their role in the context of the whole cell is lacking. In this study, we show how the mass balance constraints that govern the function of biochemical reaction networks lead to the translation of this problem into the realm of linear algebra. The functional capabilities of biochemical reaction networks, and thus the choices that cells can make, are reflected in the null space of their stoichiometric matrix. The null space is spanned by a finite number of basis vectors. We present an algorithm for the synthesis of a set of basis vectors for spanning the null space of the stoichiometric matrix, in which these basis vectors represent the underlying biochemical pathways that are fundamental to the corresponding biochemical reaction network. In other words, all possible flux distributions achievable by a defined set of biochemical reactions are represented by a linear combination of these basis pathways. These basis pathways thus represent the underlying pathway structure of the defined biochemical reaction network. This development is significant from a fundamental and conceptual standpoint because it yields a holistic definition of biochemical pathways in contrast to definitions that have arisen from the historical development of our knowledge about biochemical processes. Additionally, this new conceptual framework will be important in defining, characterizing, and studying biochemical pathways from the rapidly growing information on cellular function.     

Insulin signaling in the yeast Saccharomyces cerevisiae. 2. Interaction of human insulin with a putative binding protein

A putative insulin-binding protein (Kd = 0.5 +/- 0.2 microM for human insulin) was partially purified from solubilized plasma membranes of Saccharomyces cerevisiaeby wheat germ agglutinin and insulin affinity chromatographies. The binding affinities of various mutant insulin analogues correlated well with their capacities to activate glycogen synthase and SNF1 kinase in glucose-induced yeast spheroplasts, the ranking of their relative efficacies in yeast and in isolated rat adipocytes being similar. Using a bifunctional cross-linker and two different experimental protocols, a 53-kDa polypeptide contained in the insulin-binding protein preparation was specifically affinity cross-linked to [125I]monoiodo[B26]insulin. The relative rankings of the insulin analogues with respect to inhibition of cross-linking and binding to the partially purified insulin-binding protein were identical. Incubation of intact yeast spheroplasts with [125I]monoiodo[AI4]insulin led to specific and time-dependent association of the radiolabeled insulin with the cell surface followed by its internalization and degradation. These processes were considerably delayed by low temperature and energy depletion of the spheroplasts, suggesting involvement of the ATP-dependent endosomal apparatus. These data provide evidence for the existence of a low-affinity insulin-binding protein in the plasma membrane of Saccharomyces cerevisiae.     

13C- and 15N-labeled peptide substrates as mechanistic probes of oligosaccharyltransferase

The carboxamide moiety that links the carbohydrate and protein moieties in N-linked glycoproteins has been unambiguously determined to arise intact from asparagine by the use of chemically synthesized Bz-[4-13C, 15N]Asn-Leu-Thr-NH2 as an oligosaccharyltransferase (OST) substrate. Bz-[4-13C]Asn-Leu-Thr-NH2 was also synthesized and used to evaluate a proposed mechanism of OST catalysis similar to that of glutamine-dependent amidotransferases using 15NH4OAc as a potential external nucleophile. Analysis of NMR and MS spectra of the isotopically labeled peptides and the resulting biosynthesized glycopeptides indicates that free 15NH3 is not lost from the doubly labeled substrate during catalysis nor can exogenous 15NH3 intercept any of several postulated enzyme-bound species. These results indicate that OST-catalyzed glycosylation does not follow a mechanism involving the transient generation of exchangeable "NH3". Thus, in contrast to several glutamine-dependent amidotransferases, OST catalysis does not lead to transient scission of the asparagine beta-carboxamide C-N bond. Together with previously published results, these data argue against nucleophilic activation of the asparagine beta-carboxamide moiety being the underlying chemical mechanism for OST-catalyzed glycosylation of peptides.     

Study on the activation of phospholipases A2 by purinergic agonists in rat submandibular ductal cells

Extracellular ATP and benzoyl-ATP (Bz-ATP) increased the release of [3H]arachidonic acid ([3H]AA) from prelabeled rat submandibular gland (RSMG) ductal cells respectively two- and threefold. Both agonists also increased the release of [3H]AA from acini but at a lower level (+50% and +100% respectively). Carbachol had no significant effect on either cellular population. In ductal cells phorbol myristate acetate, an activator of protein kinase C, slightly increased the basal release of [3H]AA but did not affect the release of [3H]AA in response to ATP. Staurosporine, an inhibitor of protein kinases, inhibited the response to the purines. The removal of calcium from the extracellular medium decreased the response to ATP and Bz-ATP. Only barium could partly substitute for calcium to restore the purinergic response. Zinc inhibited the release of [3H]AA. Permeabilization of the cells with streptolysin O (SLO) activated the calcium-independent phospholipase A2 activity (iPLA2). The iPLA2, not the calcium-dependent PLA2 (cPLA2), released [3H]oleic acid ([3H]OA) from RSMG ductal cells. It is concluded that RSMG ducts have a higher PLA2 activity when compared to acini. This activity is accounted for by iPLA2 and cPLA2. Both enzymes are activated by P2X agonists by a staurosporine-sensitive mechanism. Cells permeabilized with SLO or membranes from Escherichia coli as a substrate are not good models to study the regulation of these enzymes. In intact RSMG ductal cells the two activities can be distinguished by rather specific inhibitors, by different ionic conditions and also by the fatty acid used to label the cells.     

Localization and in vitro mutagenesis of the active site in the Saccharomyces cerevisiae mRNA capping enzyme

The yeast mRNA capping enzyme is composed of 52 (alpha) and 80 kDa (beta) polypeptides, which are responsible for its mRNA guanylyltransferase and RNA 5'-triphosphatase activities, respectively. We isolated the gene encoding the alpha subunit (CEG1) and showed that CEG1 is essential for yeast cell growth [Shibagaki et al., (1992) J. Biol. Chem. 267, 9521-9528]. In this study, CEG1 was expressed in Escherichia coli and the alpha subunit protein was purified to near homogeneity. A [32P]GMP-bound tryptic peptide derived from the recombinant enzyme-[32P]GMP covalent reaction intermediate was converted to a [32P]phosphoryl-peptide through periodate oxidation followed by beta-elimination. Hydrolysis of the [32P]phosphoryl-peptide with alkali resulted in [32P]N epsilon-phospholysine as the only phosphoamino acid, indicating that GMP in the enzyme-GMP complex is bound to a lysine residue via a phosphoamide linkage. Microsequencing of the [32P]GMP-peptide showed that the GMP binding site was located in the region between amino acids 60 and 75, which contained an internal trypsin-resistant lysine at position 70. CEG1 was subjected to site-directed mutagenesis and the mutant proteins were expressed in E. coli. Substitution of His or Ile for Lys70 entirely abolished the enzyme-GMP formation activity, and this mutation was lethal to yeast in vivo, supporting the notion that the active site in the alpha subunit is located at Lys70. Replacement of Lys70 with Arg reduced the ability to form the enzyme-GMP complex; however, yeast cells bearing this allele were not viable. A series of mutations, including 8 amino acid replacements and 3 insertions, near the active site (Lys70-Thr-Asp-Gly motif) were also introduced and the mutant polypeptides were examined for catalytic activity in vitro as well as yeast cell viability in vivo. There was a good correlation between the in vitro and in vivo functions of the mutant proteins, except when Asp72 was replaced with Glu, which allowed formation of the enzyme-GMP complex but failed to support cell growth. The results with Lys70 to Arg and Asp72 to Glu substitutions indicated that guanylyltransfer to RNA and/or additional roles besides cap formation per se are impaired in these mutant proteins.     

The vesicle transport protein Vps33p is an ATP-binding protein that localizes to the cytosol in an energy-dependent manner

Molecular mechanisms of vesicle transport between the prevacuolar compartment and the vacuole in yeast or the lysosome in mammalian cells are poorly understood. To learn more about the specificity of this intercompartmental step, we have examined the subcellular localization of a SEC1 homologue, Vps33p, a protein implicated to function in transport between the prevacuolar compartment and the vacuole. Following short pulses, 80-90% of newly synthesized Vps33p cofractionated with a cytosolic enzyme marker after making permeabilized yeast cells. However, during a chase, 20-40% of Vps33p fractionated with permeabilized cell membranes in a time-dependent fashion with a half-time of approximately 40 min. Depletion of cellular ATP increased the association rate to a half-time of approximately 4 min and caused 80-90% of newly synthesized Vps33p to be associated with permeabilized cell membranes. The association of Vps33p with permeabilized cell membranes was reversible after restoring cells with glucose before permeabilization. The N-ethylmaleimide-sensitive fusion protein homologue, Sec18p, a protein with known ATP binding and hydrolysis activity, displayed the same reversible energy-dependent sedimentation characteristics as Vps33p. We determined that the photosensitive analog, 8-azido-[alpha-32P]ATP, could bind directly to Vps33p with low affinity. Interestingly, excess unlabeled ATP could enhance photoaffinity labeling of 8-azido-[alpha-32P]ATP to Vps33p, suggesting cooperative binding, which was not observed with excess GTP. Importantly, we did not detect significant photolabeling after deleting amino acid regions in Vps33p that show similarity to ATP interaction motifs. We visualized these events in living yeast cells after fusing the jellyfish green fluorescent protein (GFP) to the C terminus of full-length Vps33p. In metabolically active cells, the fully functional Vps33p-GFP fusion protein appeared to stain throughout the cytoplasm with one or two very bright fluorescent spots near the vacuole. After depleting cellular ATP, Vps33p-GFP appeared to localize with a punctate morphology, which was also reversible upon restoring cells with glucose. Overall, these data support a model where Vps33p cycles between soluble and particulate forms in an ATP-dependent manner, which may facilitate the specificity of transport vesicle docking or targeting to the yeast lysosome/vacuole.