Vac14 Controls Ptdins(3,5)P2 Synthesis and Fab1-Dependent Protein Trafficking to the Multivesicular Body

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

Vac14 Controls PtdIns(3,5)P2 Synthesis and Fab1-Dependent Protein Trafficking to the Multivesicular Body

Stephen K. Dove,1,4 Robert K. McEwen,1 least abundant of the inositol lipids [1, 2]. During hyper- Andrew Mayes,2,5 David C. Hughes,3 osmotic stress, stimulated synthesis rapidly elevates 2 1 Jean D. Beggs, and Robert H. Michell the yeast PtdIns(3,5)P2 concentration to a level similar to 1 School of Biosciences that of PtdIns(4,5)P2 [1, 3], the other ubiquitous PtdInsP2 University of Birmingham isomer that has a striking variety of functions in signaling Birmingham B15 2TT and other aspects of cell regulation [4–12]. Although the 2 Wellcome Trust Centre for Cell Biology Hog1p protein kinase pathway is responsible for many University of Edinburgh yeast responses to hyperosmotic stress, activation of

Edinburgh EH9 3JR PtdIns(3,5)P2 synthesis is transmitted through some differ- 3 Biology and Genetics Group ent and still unidentified osmotic response pathway [1]. Assisted Conception Unit Fab1 [13, 14] is the PtdIns3P 5-kinase that makes

Birmingham Women’s Hospital PtdIns(3,5)P2 [3, 15–20]. S. cerevisiae carrying mutations Birmingham B15 2TG in the FAB1 gene have very enlarged vacuoles, because

United Kingdom cells that cannot make PtdIns(3,5)P2 fail to recycle vacuole membrane back to late endosomes [18, 21]. Similar problems beset mammalian cells in which the homolo- Summary gous PtdIns3P 5-kinase is inhibited [22]. Fab1 and

PtdIns(3,5)P2 are also needed for the sorting of a subset Background: The PtdIns3P 5-kinase Fab1 makes of proteins into the internal vesicles of the multivesicular PtdIns(3,5)P , a phosphoinositide essential for retro- 2 body (MVB) [17, 18, 23]. Inability to make PtdIns(3,5)P2 grade trafficking between the vacuole/lysosome and the causes sterility in S. pombe, at least partly as a result late endosome and also for trafficking of some proteins of impaired mating factor synthesis [20]. into the vacuole via multivesicular bodies (MVB). No Little is known of how the PtdIns3P 5-kinase activity regulators of Fab1 were identified until recently. of Fab1 is regulated, either during normal membrane Results: Visual screening of the Eurofan II panel of S. trafficking or when it is activated hyperosmotically [1, cerevisiae deletion mutants identified YLR386w as a 3]. The Vac7 protein has been proposed as a regulator novel regulator of vacuolar function. Others recently of Fab1 [16, 24], but no VAC7 homolog has been identi- identified this ORF as encoding the vacuolar inheritance fied in higher organisms. Herein we identify a yeast pro- gene VAC14. Like fab1 mutants, yeast lacking Vac14 tein that is necessary both to maintain the normal have enlarged vacuoles that do not acidify correctly. “basal” PtdIns3P 5-kinase activity of Fab1 and for acti- FAB1 overexpression corrects these defects. vac14⌬ vation of its PtdIns3P 5-kinase activity in response to cells make very little PtdIns(3,5)P , and hyperosmotic 2 hyperosmotic stress. While this paper was being pre- shock does not stimulate PtdIns(3,5)P synthesis in the 2 pared, others showed that the vacuolar inheritance gene normal manner, implicating Vac14 in Fab1 regulation. VAC14 encodes this protein [25]. We also show that, like fab1⌬ mutants, vac14⌬ cells fail to sort GFP-Phm5 to the MVB and thence to the vacuole: irreversible ubiquitination of GFP-Phm5 overcomes this Results defect. In the BY4742 genetic background, loss of Vac14 causes much more penetrant effects on phosphoinosi- Identification of Swollen Vacuole tide metabolism and vacuolar trafficking than does loss Phenotype Genes of Vac7, another regulator of Fab1. Vac14 contains mo- The most penetrant phenotype of cells carrying defects tifs suggestive of a role in protein trafficking and inter- in novel components of the Fab1/PtdIns(3,5)P pathway acts with several proteins involved in clathrin-mediated 2 is likely to be a very enlarged fab1⌬-like vacuole that membrane sorting and phosphoinositide metabolism. has a relatively smooth outline. To search for previously Conclusions: Vac14 and Vac7 are both upstream activa- unrecognized proteins that might be involved in this tors of Fab1-catalysed PtdIns(3,5)P synthesis, with .EUROFAN II [26] S 4900ف pathway, we screened the 2 Vac14 the dominant contributor to the hierarchy of control. Vac14 is essential for the regulated synthesis of cerevisiae deletion mutants of nonessential genes, all PtdIns(3,5)P , for control of trafficking of some proteins of which are in the BY4742 genetic background, for this 2 ⌬ ⌬ to the vacuole lumen via the MVB, and for maintenance appearance. Over 50 mutants, including fab1 , vps34 ⌬ of vacuole size and acidity. [27], and pep12 [28], had this visual phenotype. The known functions of several other genes identified by Introduction this screen made them unlikely candidates for involvement in the Fab1/PtdIns(3,5)P2 pathway. The screen did not pick up vac7⌬ cells, which looked normal in this Phosphatidylinositol 3,5-bisphosphate [PtdIns(3,5)P2] was discovered in yeast and fibroblasts, where it is the genetic background (see below). We focused our attention on three genes that were 4Correspondence: [email protected] novel and of unassigned function, which we provision- 5Present address: Unilever Research, Colworth, Colworth House, ally termed SVP1, SVP2, and SVP3 (for Swollen Vacuole Sharnbrook, Bedford MK44 1LQ, United Kingdom. Phenotype). Putative orthologs of SVP1 and SVP2 are Current Biology 886

Figure 1. The Swollen Vacuole and Vacuole Acidification Phenotypes of fab1⌬ and vac14⌬ Cells, Compared with Normal Cells (A) Differential interference contrast (DIC) images are shown of wild-type, fab1⌬, and vac14⌬ cells, together with images of the same cells stained with FM4-64. (B) Quinacrine fluorescence images are superimposed on low-light DIC images: quinacrine only accumulates in acidified vacuoles. Wild-type cells stain brightly with this dye, indicating vacuole acidification. Quinacrine did not accumulate in the vacuoles of vac14⌬ cells, but GFP- Fab1 or GFP-Vac7 overexpression restored vacuole acidification. ubiquitous in eukaryotic genomes. SVP1, which en- CEN-based plasmid rescued all of these phenotypes. codes the prototypic member of a novel family of The very recent independent cloning of the protein en-

PtdIns(3,5)P2 effector proteins, will be described else- coded by the vacuolar inheritance gene VAC14 shows where. its identity with SVP2 [25] (see below): Vac14 therefore The deleted ORF in svp2⌬ cells was YLR386w. svp2⌬ takes precedence as the name for Svp2. cells were enlarged and had a strikingly large vacuole that was not correctly acidified. They looked very similar to fab1⌬ cells (Figure 1A), suggesting that they might GFP-Vac7 and GFP-Fab1 Localize Correctly suffer from a defect in membrane recycling from the in vac14⌬ Cells vacuole to late endosomes similar to that seen in pre- GFP-Vac14 constructs have not yet been obtained that viously defined vacuolar function genes [14, 21]. Some rescue the vac14⌬ vacuole phenotype, so we could not svp2⌬ cells displayed an “open-figure eight” vacuole experimentally localize Vac14. We therefore assessed profile like that seen in vac7⌬ and fab1⌬ cells (data not the localization of GFP constructs of Fab1 and Vac7, shown), which points to a possible vacuole inheritance both of which are needed for protein traffic through defect [14]. svp2⌬ cells also displayed other fab1⌬-like PtdIns(3,5)P2-dependent pathways [14, 16, 25, 29–31]. phenotypes. Their growth was abnormally sensitive to In wild-type and in vac14⌬ cells, GFP-Fab1 and GFP- elevated temperature in some strain backgrounds (data Vac7 localized to FM4-64-positive vacuolar structures not shown), and they failed to acidify the vacuole cor- (Figure 2). This indicates that lack of Vac14 probably rectly (Figure 1B). causes vacuolar defects directly, rather than causing Expression of one copy of the SVP2 gene from a mislocalization of other essential components of the Vac14 Is a Fab1 Activator 887

Figure 2. Localization of GFP-Fab1 and GFP-Vac7 in Wild-Type and vac14⌬ Cells Panel (A) compares the localization of GFP-Fab1 in wild-type and vac14⌬ cells (middle images) with DIC images (top) and FM4-64 images (bottom). Panel (B) presents a similar comparison for cells expressing GFP-Vac7. Both GFP-tagged proteins accumulate in FM4-64-positive structures in both wild-type and vac14⌬ cells. The vac14⌬ cells shown expressing GFP-Vac7 are of a small percentage (10%–20%) of cells in which overexpression largely rescued the vac14⌬ vacuole phenotype, as these gave the strongest GFP-Vac7 signal.

Fab1/PtdIns(3,5)P2/MVB pathway and thus some sec- ment in membrane trafficking (Figure 3). Its N-terminal ondary vacuolar dysfunction. domain includes at least one HEAT-like repeat (residues In these experiments, the overexpression of GFP- 80–117) and six “dileucine pairs” (actually LeuLeu or Fab1 that was needed to assess its localization cor- LeuVal) between residues 149 and 363. HEAT-like re- rected the retrograde membrane trafficking defect in peats are found in many membrane-trafficking proteins, 60%–80% of vac14⌬ cells, as indicated by normalization including huntingtin and Vps15 [33, 34]: Vps15 uses of vacuole morphology (Figure 2). The vacuole acidifica- three HEAT-like repeats to associate with Vps34. Vac14 tion defect was corrected in all cells. Thus, the vac14⌬ and its putative orthologs include up to three HEAT-like acidification defect is more readily corrected than is the repeats in the well-conserved N terminus, although it is retrograde traffic defect: there is normal acidification only in the human ortholog that these are all sufficiently even of those enlarged vacuoles that resist FAB1 over- conserved to be immediately recognized by search pro- expression. In wild-type cells, GFP-Fab1 overexpres- grams. A section near the N terminus of the human sion caused the vacuole to become more condensed protein is notable because this stretch of HsVac14 was than normal (data not shown). GFP-Vac7 overexpression isolated in a screen for binding partners of the HTLV-1 had less effect on the phenotypes of vac14⌬ cells: it encoded protein Tax1 [35]. completely reduced vacuolar enlargement in up to Dileucine pairs are involved in adaptin binding during of cells and also corrected the acidification clathrin-mediated vesicle formation [36, 37], and acidic %20–%10ف defect in all vac14⌬ cells (see Figure 2). dileucine clusters also bind GGA family proteins essential for vacuolar protein trafficking [38]. The potentially Vac14 Is the S. cerevisiae Representative important N-terminal 350 residues of Vac14, which show of a Widespread Protein Family modest homology with the N-terminal clathrin binding Vac14 is an 880 residue protein encoded by ORF domains of adaptins, are quite well conserved in all of YLR386w. Genes that encode putative Vac14 orthologs the Vac14-related proteins. Residues 434–610 of Vac14 are widely distributed (for example, in plants, humans, show some similarity to VHS domains that serve to bind and other mammals, a fly, and a nematode) [25]. acidic-cluster dileucine motifs. Vac14 and the C. elegans ortholog both include a predicted transmembrane domain (centered on residue Vac14 May Associate with Clathrin via AP Proteins 434 in Vac14 and close to the same position in CeVac14), Two-hybrid analysis using a LexA-Vac14 construct as the prediction of which persists even after the applica- bait underlined the likely involvement of Vac14 in mem- tion of a rigorous algorithm designed to minimize the brane trafficking, maybe through interactions with adap- detection of “false-positive” putative membrane-span- tins and so indirectly with clathrin. The strongest interac- ning sequences [32]. However, the proteins from other tions detected were with Vac14 itself, suggesting that it species do not include similarly placed putative trans- might dimerize. Interactions were also detected with membrane sequences, and biochemical evidence indi- several proteins involved in membrane trafficking and/ cates that Vac14 is not membrane inserted [25]. or phosphoinositide metabolism. These included Nyv1, Vac14 contains sequence motifs that suggest involve- a vacuolar SNARE; Fig4, a phosphoinositide phospha- Current Biology 888

Figure 3. The Domain Organization of Vac14 and Its Homologs A schematic diagram of the S. cerevisiae VAC14 sequence, showing the locations of identified sequence motifs. Alignments of the full sequences of VAC14 and homologs in other eukaryotes are shown in the Supple- mentary Material.

tase that has very recently been implicated in regulation in Ͼ90% of vac14⌬ cells, showing that fab1⌬ and of the Fab1 pathway (see the Discussion); the ␥-adaptin vac14⌬ cells suffer from functionally similar MVB sorting of the AP protein Apl4; Yap180, another clathrin adaptin defects. By contrast, GFP-Phm5p sorts correctly to the that is involved in endocytosis; and a protein kinase vacuolar interior in all vac7⌬ cells (Figure 4C: see the related to mammalian protein kinase B (see Table 1 for Discussion for further consideration of Vac7). references). Some other binding partners were identified In fab1⌬ cells, the MVB sorting defect can be cor- but are less likely to be related to vacuolar function (data rected by irreversible ubiquitination of the GFP-Phm5 not shown). cargo protein [23]—ubiquinated GFP-Phm5 arrives safely in the vacuole lumen (Figure 4E). The same is true Vac14 Regulates Protein Sorting of vac14⌬ cells (Figure 4F). into and via the MVB A characteristic defect of fab1⌬ cells is failure to correctly sort certain proteins from the trans-Golgi network (TGN) to the vacuole via the MVB: these include carboxypeptidase S [17, 39, 40] and the vacuolar polyphosphate phosphatase Phm5 [23, 41, 42]. The initial targetting of proteins into this trafficking pathway involves their ubiquitination in the Golgi by the Ubc4/Tul1 ubiquitin ligase complex [23]. We monitored possible involvement of Vac14 in this pathway by following GFP-Phm5 sorting. Normally, Phm5 is first incorporated into the membranes of the vesicles that bud into the MVB lumen. The mature MVB then fuses with the vacuole, these vesicles are released into the vacuole, and their digestion liberates Phm5 into the vacuole lumen. This is illustrated by the intravacuolar localization of GFP-Phm5 in wild-type cells (Figure 4; see [23]). In fab1⌬ cells, GFP-Phm5 stalls at the outer membrane of the MVB and is therefore found on the vacuole limiting membrane after the MVB has fused with the vacuole [23]. The same GFP-Phm5 distribution is seen

Table 1. Two-Hybrid Interactions of Vac14 Number of Clones Gene Protein SGD-ORF

2 unnamed Protein kinase B homolog YDR466w 1 YAP180 Adaptin [53, 54] YGR241c 5 NYV1 Vacuolar SNARE [55–57] YLR093c 10 VAC14 Vac14 YLR386w Figure 4. Defective Trafficking of GFP-Phm5 in fab1⌬ and vac14⌬

1 FIG4 PtdIns(3,5)P2 phosphatase [24, 58] YNL325w Cells 3 APL4 ␥-adaptin [59] YPR029c (A) GFP-Phm5 is in wild-type cells sorted to the internal vesicles of MVB and thence to the vacuolar interior. LexA-Vac14 constructs that include the entire Vac14 sequence were (B) In fab1⌬ cells, GFP-Phm5 trafficking stalls at the exterior of the used as bait in two-hybrid studies (for methods, see [51]). Those vacuole (as does carboxypeptidase S [23, 40]). recovered proteins that may be implicated in protein sorting and/ (C) In vac7⌬ cells, Phm5 reached the vacuole interior normally. or membrane trafficking events, which comprised about half of the (D) In vac14⌬ cells, GFP-Phm5 trafficking stalls in the same manner 13 proteins identified by the Svp2/Vac14 screen, are shown. Column as for fab1⌬ cells. 1 indicates how many times a particular protein was retrieved. Refer- (E) When irreversibly ubiquitinated GFP-Phm5 is used, Ubi-GFP- ences for the functions of each of these proteins can be found in Phm5 trafficking to the vacuole lumen was restored in fab1⌬ cells. [53] and [54] for YAP180, [55–57] for NYV1, [24] and [58] for FIG4, (F) vac14⌬ cells also trafficked the Ubi-GFP-Phm5 construct nor- and [59] for APL4. mally. Vac14 Is a Fab1 Activator 889

Figure 5. Defective Control of PtdIns(3,5)P2 Synthesis in vac14⌬ Cells Panels (A) (wild-type cells) and (B) (vac14⌬ cells) show HPLC traces of the deacylated phosphoinositides extracted from [3 H]inositol-labeled cells stressed for 10 min with 0.9 M NaCl. Panel (C) summarizes both basal and hyperosmotically stimulated phosphoinositide labeling in these cells (expressed as dpm ϫ 10Ϫ3, normalized to labeling of PtdIns of 107 dpm). This experiment is representative of three that gave results within 5% of the above values.

Failure of the Control of PtdIns(3,5)P2 Synthesis normal vacuole morphology (see above), were near nor- in vac14⌬ Cells mal (data not shown). As also reported by others [24],

Isolated Fab1 phosphorylates PtdIns3P to PtdIns(3,5)P2, we found that hyperosmotic activation of PtdIns(3,5)P2 and cells lacking Fab1 make no PtdIns(3,5)P2: Fab1 and synthesis fails in vac7⌬ cells (data not shown). its ortholog Ste12 are the sole PtdIns3P 5-kinases in S. cerevisiae and S. pombe, respectively [3, 15–17, 20]. Discussion Two potential causes of fab1⌬-like phenotypes are either a cellular PtdIns(3,5)P2 deficit caused by low or The Fab1/PtdIns(3,5)P2 Pathway absent Fab1 PtdIns3P 5-kinase activity or the loss of PtdIns(3,5)P2, found in all eukaryotes so far studied [1], some PtdIns(3,5)P2-responsive effector or other down- has several proposed functions. In S. cerevisiae, stream component. To determine whether the vac14⌬ PtdIns(3,5)P2 has been implicated in retrograde mem- defect lies upstream or downstream of Fab1-catalysed brane trafficking from the vacuole to the late endosome

PtdIns(3,5)P3 synthesis, we examined PtdIns(3,5)P2 me- [3, 15, 16, 19], in the sorting of some proteins into the tabolism in vac14⌬ cells. Cells were labeled with MVB [17, 18], and in regulation of vacuolar inheritance 3 [ H]myo-inositol and the extracted glycerophospho- and pH [14]. Cells that lack PtdIns(3,5)P2 grow poorly at inositides were deacylated and analyzed by anion- elevated temperatures [14, 16, 29, 31], probably because exchange HPLC [3, 43]. of problems both with vacuole membrane remodeling

The PtdIns(3,5)P2 content of vac14⌬ cells was very and with control of cell integrity (S.K.D., unpublished low and variable—only a small fraction of that of wild- data). PtdIns(3,5)P2 is needed for vesicular transport in type cells (Figure 5)—and their PtdIns3P complement mammalian cells [22] and is essential for mating in S. was modestly elevated (Figure 5). By contrast, vac14⌬ pombe, probably because of a role in intracellular traf- cells contained normal quantities of PtdIns4P and ficking of the mating factor and its receptor [20].

PtdIns(4,5)P2. In wild-type cells, hyperosmotic stress stimulates PtdIns(3,5)P2 synthesis 10- to 30-fold [1, 3], Vac7 and the Control of PtdIns(3,5)P2 Synthesis but the deficient PtdIns(3,5)P2 complement of vac14⌬ Despite the striking cellular effects of PtdIns(3,5)P2 defi- cells only doubled when the cells were hyperosmotically ciency, Fab1 was until recently the only identified protein challenged (Figure 5). of the PtdIns(3,5)P2 regulatory pathway. While preparing So vac14⌬ cells make PtdIns3P at a normal rate, but this paper, we learned that Vac7, a vacuolar transmem- this is only converted to PtdIns(3,5)P2 very slowly, and brane protein, regulates the PtdIns3P 5-kinase activity the cells fail correctly to sort PtdIns(3,5)P2 synthesis of Fab1 [24]. VAC7 deletion causes a fab1⌬-like pheno- when stressed. It seems that cells need Vac14 both to type in some strain backgrounds, so it had earlier been maintain a normal basal rate of PtdIns(3,5)P2 synthesis proposed as an activator of Fab1 [14, 16, 24, 29–31]. and to hyperosmotically activate the PtdIns3P 5-kinase Reported vac7⌬ phenotypes include defects in both activity of Fab1. This depressed PtdIns(3,5)P2 synthesis basal and stimulated PtdIns(3,5)P2 synthesis [16, 24]. is presumably catalysed by fully functional Fab1 that However, BY4742 yeast, the strain in which the Euro- lacks its normal stimulatory input, which also accounts fan II deletion mutants were constructed, maintain al- for accumulation of supranormal concentrations of the most normal basal PtdIns(3,5)P2 levels when they are precursor PtdIns3P. made vac7⌬ (S.K.D., unpublished data). Few of these In contrast to fab1⌬ and vac14⌬ cells, basal levels of vac7⌬ cells have obviously swollen vacuoles, and they

PtdIns(3,5)P2 in BY4742 vac7⌬ cells, most of which retain traffic GFP-Phm5p to the vacuole normally (Figure 5C). Current Biology 890

Despite these relatively normal properties when un- stressed, hyperosmotic stimulation of Fab1-catalysed

PtdIns(3,5)P2 synthesis by BY4742 cells still needs Vac7 (S.K.D., unpublished data). Vac7 function must therefore be partly dispensable for those functions of Fab1 and

PtdIns(3,5)P2 that do not need the stress-provoked en- hancement of PtdIns(3,5)P2 production. We could find no homolog of Vac7 in any database—even in S. pombe, which shows stress-activated PtdIns(3,5)P2 production similar to that in S. cerevisiae [1]. Basal synthesis of

PtdIns(3,5)P2 must therefore be maintained indepen- dently of Vac7-like proteins in most organisms.

A Simple Visual Genetic Screen that Identifies

Participants in the Fab1/PtdIns(3,5)P2 Pathway Availability from the Eurofan II program (http://www. uni-frankfurt.de/fb15/mikro/euroscarf/index.html) [26] of a comprehensive panel of deletion mutants of nonessential yeast genes in one genetic background offers Figure 6. Model of the Multiple Roles in Yeast Biology of PtdIns(3,5)P2, Formed by Fab1 under the Control of Vac14 and, to a simple route to direct identification of genes whose a Lesser Extent, Vac7 absence causes readily recognizable phenotypes— The existing evidence points to diversity in the effector mechanisms including swollen vacuoles akin to those of fab1⌬ cells. leading from PtdIns(3,5)P2 to the various indicated biological end- The results reported here confirm that at least one of points. these genes encodes a protein that controls the PtdIns3P 5-kinase activity of Fab1. We shall report else- where on genes that encode PtdIns(3,5)P2 effectors that Figure 6 schematically summarizes the fact that execute downstream functions. PtdIns(3,5)P2, with its synthesis primarily controlled by Vac14, is essential for at least four aspects of yeast Svp2/Vac14 Regulates the PtdIns3P 5-Kinase biology. Available evidence indicates that these pheno- Activity of Fab1 types are at least to some degree distinct and separable Fab1 and its orthologs, the “type III PtdInsP kinases” in their PtdIns(3,5)P2 sensitivities, possibly being exe- cuted by more than one effector protein. that make PtdIns(3,5)P2, are found in all eukaryotes [3], so we reasoned that there must be activator(s) of Fab1 Type III PtdInsP kinases related to Fab1 and well- other than Vac7—maybe even in S. cerevisiae. When conserved Vac14 orthologs are similarly widespread in we identified the Fab1-regulatory protein that we initially eukaryotes, so Vac14 homologs are probably ubiquitous termed Svp2, it was immediately apparent that its or- Fab1 regulators. The two-hybrid screen identified no thologs might be ubiquitous regulators of Fab1 or- direct Vac14-Fab1 interaction, even though it detected thologs (see the Supplementary Material available with other Fab1 interactions (R. Van-Nues and J.D.B., unpub- this article online). We then learned that others had iden- lished data): whether Vac14 directly activates Fab1 di- tified the same protein as Vac14 by a classical genetic rectly remains a problem for the future. approach [25]. The fab1⌬ and vac14⌬ phenotypes are very similar— Models of Vac14 Action including swollen vacuoles, vacuole acidification de- Two observations point to involvement of Vac14 in some fects, and a failure correctly to sort certain proteins to type of clathrin-based vesicle trafficking system: the the vacuole lumen via the MVB. However, vac14⌬ cells Vac14 sequence includes motifs that point to such a are less heat sensitive than fab1⌬ cells, at least in the role, and the two-hybrid studies detected an interaction BY4742 background. This is presumably because of the with the ␥-adaptin of the AP-1 complex. Moreover, retro- difference between fab1⌬ cells that lack PtdIns(3,5)P2 grade trafficking of the hybrid RS-ALP protein from the and vac14⌬ cells that retain a small residual vacuole to late endosomes, a process that requires

PtdIns(3,5)P2 complement. GFP-Fab1 and GFP-Vac7 are Vac14, is defective in yeast expressing some mutant correctly localized at the vacuole in vac14⌬ cells, so alleles of clathrin (R.C. Piper, 2000, American Society the vac14⌬ phenotype is not caused indirectly by the of Cell Biology Meeting Book, abstract). AP-1-mediated

PtdIns(3,5)P2-synthesizing apparatus failing to localize effects of Vac14 cannot form the only link between correctly. Despite having an intact FAB1 gene, vac14⌬ clathrin, the Vac14/Fab1/PtdIns(3,5)P2 pathway, and ret- cells display defective control of PtdIns(3,5)P2 synthesis rograde membrane trafficking from the vacuole, since (Figure 6), so Vac14 is likely to be an upstream regulator apl1-4 mutants have no phenotype, and Fab1 overex- of Fab1: Fab1 makes little PtdIns(3,5)P2 in its absence. pression suppresses the swollen vacuole defect of FAB1 overexpression and to a lesser degree VAC7 over- vac14⌬ cells. Given the redundancy that characterizes expression rescues many vac14⌬ phenotypes (Figure clathrin biology, with AP-1 and Gga1/Gga2 proteins both 2), presumably by tending to normalize the basal rate contributing to clathrin-based Golgi-to-late-endosome of PtdIns(3,5)P2 synthesis despite the absence of activa- trafficking [38, 44, 45], it is possible that the vacuole- tion by Vac14. to-late-endosome pathway shows similar redundancy. Vac14 Is a Fab1 Activator 891

Moreover, Vac14 might interact directly with clathrin: its from infected mammalian cells: viral proteins somehow sequence contains several dileucine pairs, two of which mediate mistargeting of the MVB machinery to the cell are in acidic patches but do not conform to the VHS surface [48–51]. Inhibition of ubiquitination prevents the binding consensus. sorting of retroviral gag proteins into the forming virions, and irreversible ubiquitination of these proteins overcomes this inhibition. It seems improbable that a virus A Mechanism for Vac14-Mediated Control would coincidentally obtain its envelope by subverting of PtdIns(3,5)P Synthesis? 2 ubiquitin-dependent MVB protein sorting and that one Fig4 was recently implicated as a negative regulator of its key proteins, Tax1, should interact with a protein of PtdIns(3,5)P -mediated events in cells [24], so the 2 implicated in this same process without there being detected interaction between Vac14 and Fig4, which is some mechanistic link between these two phenomena. at least partly a peripheral vacuolar protein, may be of The alternative notion, that HsVac14 may recruit HsFab1 some significance. FIG4, first identified as a gene and PtdIns(3,5)P as essential players in viral packaging needed for mating in S. cerevisiae, encodes a member 2 or budding, looks worthy of examination. of the Sac1 family of phosphoinositide phosphatases. It might serve as a PtdIns(3,5)P phosphatase [24], 2 Experimental Procedures though no phosphoinositide phosphatase activity in vitro has yet been detected (F. Cooke, W. Hughes, and Quinacrine, G418, and TBAHS were from Sigma. Methionine was P. Parker, personal communication). “Unactivated” Vac14 from Amersham. FM4-64 was from Molecular Probes. The Eurofan might normally bind Fig4 and recruit it to a position on II yeast strain collection was from EUROSCARF (Institute for Micro- biology, Johann Wolfgang Goethe-University Frankfurt, Marie- the vacuole where it can hydrolyse vacuolar PtdIns(3,5)P2, and Vac14 that is ‘activated’ by hyperosmotic stress Curie-Strasse 9, D-60439 Frankfurt, Germany). Yeast strains, in the BY4742 genetic background, were as indicated below. Most meth- might then release Fig4 to the cytosol, so enhancing ods for the growth of yeast and for assay of vacuolar staining were and sustaining a stress-stimulated elevation of cellular as described previously [1, 3, 43]. The two-hybrid analyses were PtdIns(3,5)P2 levels. Alternatively, Vac14 may directly part of the program described in [52]. activate Fab1. In this model, Fig4 would be recruited to Vac14 to terminate the PtdIns(3,5)P2 signal after the Vacuole Morphology Screen effector function(s) of this lipid have been fulfilled. The Eurofan II S. cerevisiae haploid deletion mutants were grown to exponential phase in 96-well microtiter plates at 30ЊC for 5–16 hr with shaking in 200 ␮l of YEPD medium containing 300 ␮g/ml G418 Vac14 and MVB Protein Sorting and 8% glucose (to suppress glucose starvation-induced vacuole (and Viral Budding?) enlargement). Cells were recovered and examined by phase contrast ف Vac14 is needed for the sorting of a particular subset microscopy. The screen, which took 7 days, was conducted with the experimenter blind to which mutant each well contained. of proteins, including Phm5 and carboxypeptidase S, to the vacuole via the MVB. Having sorted correctly as far Growth of Yeast and Phosphoinositide Analysis as the MVB lumen, the Vac14-dependent proteins need Radiolabeling and extraction of inositol lipids were as previously a functional Vac14/Fab1/PtdIns(3,5)P2 pathway for in- described [1, 3, 43] except that yeast were stimulated in their growth corporation into the MVB internal vesicles. Irreversible medium at physiological cell densities (1–3 ϫ 106 cell/ml): cells can ubiquitination overcomes this requirement [23]. Since become glucose stressed and unresponsive to hyperosmotic stress both Cps1 and Phm5 are ubiquitinated normally in a at higher densities. They were killed by addition of two volumes of a 100:1 (v/v) mixture of methanol and 11.5 M HCl and centrifuged: fab1⌬ strain [23], the simplest interpretation of these centrifugation before killing can stimulate PtdIns(3,5)P2 synthesis (our data is that PtdIns(3,5)P2, either directly or through an as unpublished data). To improve the recovery and stability during deacyl- yet uncharacterised effector protein, somehow controls ation of PtdInsPs, the subsequent one-phase CHCl3:CH3OH:H2O mix- the timing of protein deubiquitination by the Doa4 (or ture was split using 1 M HCl containing 5 mM tetrabutylammonium other) ubiquitin isopeptidase [46], maybe by influencing hydrogen sulfate (TBAHS), 20 mM Na2HPO4, 1 mM EDTA, and 1 mM its localization or activity. In this model, Cps1 and Phm5 EGTA (rather than 0.6 M HCl, as described previously). are prematurely deubiquitinated in fab1⌬ cells before MVB Trafficking Assay they are sorted into the forming MVB vesicles. Why ⌬ Yeast strains were transformed with plasmids encoding GFP-Phm5 or proteins like Ste2 are not similarly affected in fab1 cells an irreversibly ubiquitinated GFP-Phm5 (Ubi-GFP-Phm5; from Prof H is unknown but may reflect differences in the ubiquitin- Pelham, LMB, Cambridge) [23]. They were cultured to 1 ϫ 107 cells/ protein linkage of cell surface proteins versus those traf- ml in SC-Ura with 3% glucose and all amino acids at 30–50 mg/l ficking in from the Golgi: Ste2 is ubiquitinated by the Rps5 and examined on a Nikon Eclipse E600 microscope with an XF100-3 ubiquitin ligase, whereas Phm5 is ubiquitinated by the filter cube (Omega Optical). Images were acquired with an ORCA digital camera (Hamamatsu, Japan) and processed in Adobe Pho- Tul1 protein. Evidence exists that proteins ubiquitinated toshop. by Rps5 may carry a Lys63-linked ubiquitin chain, in contrast to those ubiquitinated by other ligases [47]. Quinacrine Staining Putative Vac14 orthologs include a human protein that Yeast cells were grown to 1–3 ϫ 107 cells/ml, centrifuged (10,000 ϫ aligns well with Vac14 (see the Supplementary Material). g ϫ 30 s), and incubated for 5 min at room temperature in 100 mM Intriguingly, a partial sequence of this protein was earlier HEPES/KOH (pH 7.5), 3% glucose, 200 ␮M quinacrine. They were cloned from T lymphocytes as TRX, a ubiquitously ex- sedimented, washed in HEPES-Glucose buffer without quinacrine, pressed binding partner for the multifunctional Tax1 pro- and viewed as above except that images were captured that com- bined low-light DIC and quinacrine fluorescence. Quinacrine fluores- tein of the lymphotrophic retrovirus HTLV-1 [35]. Recent cence overwhelmed the 10-fold lower intensity GFP signal in the work has revealed that a subverted ubiquitin-dependent experiments in which vacuole acidity was determined in vac14⌬ MVB sorting machinery underlies retroviral budding cells overexpressing GFP-Fab1 or Vac7. Current Biology 892

Plasmid Construction 10. Godi, A., Pertile, P., Meyers, R., Marra, P., Di Tullio, G., Iurisci, C., pGFP-Fab1 and pGFP-Vac7 were constructed by PCR amplification Luini, A., Corda, D., and De Matteis, M.A. (1999). ARF mediates of the respective ORFs from yeast genomic DNA using the Expand recruitment of PtdIns-4-OH kinase-beta and stimulates synthe- proofreading DNA polymerase (Roche). pGFP-Fab1 was created by sis of PtdIns(4,5)P2 on the Golgi complex. Nat. Cell Biol. 1, ligating a 6.8 kb fragment encoding the entire FAB1 ORF from plas- 280–287. mid pCRblunt FAB1 (a gift of Dr. F. Cooke, UCL), cut with BglII 11. Insall, R.H., and Weiner, O.D. (2001). PIP3, PIP2, and cell move- and XhoI, into the BamHI and SalI sites of pUG36 (NCBI accession ment—similar messages, different meanings? Dev. Cell 1, AF298791, a gift of Dr. J.H. Hegemann, Dusseldorf). The sequences 743–747. across the junctions of pGFP-Fab1 were (FAB1 sequences in bold 12. Simonsen, A., Wurmser, A.E., Emr, S.D., and Stenmark, H. and vector sequences italicised, with junctions underlined), at the (2001). The role of phosphoinositides in membrane transport. 5Ј end of FAB1, ACTAGTGGA TCTATGTCATCGGAA; at the 3Ј end Curr. Opin. Cell Biol. 13, 485–492. of FAB1, GAAGGAAATTAACTCGACCTCGAG. pGFP-Vac7 was cre- 13. Yamamoto, A., DeWald, D.B., Boronenkov, I.V., Anderson, R.A., ated by ligating a 3.5 kb PCR fragment encoding the VAC7 ORF Emr, S.D., and Koshland, D. (1995). Novel PI(4)P 5-kinase homo- with BamHI and Sal I sites immediately flanking the ORF sequences. logue, Fab1p, essential for normal vacuole function and mor- This was ligated into BamHI/SalI cut pUG36. The sequences at the phology in yeast. Mol. Biol. Cell 6, 525–539. junctions of this construct were, at the 5Ј end of VAC7, ACTAGTG 14. Bonangelino, C.J., Catlett, N.L., and Weisman, L.S. (1997). GATCCATGACAGAAGAA; at the 3Ј end of VAC7, GGTAAGAAGT Vac7p, a novel vacuolar protein, is required for normal vacuole GAGTCCACCTCGAG. The transformed yeast were grown to 1 ϫ inheritance and morphology. Mol. Cell. Biol. 17, 6847–6858. 107 cells/ml in SC-Ura-Met (the promoter in pUG36 is methionine 15. Cooke, F.T., Dove, S.K., McEwen, R.K., Painter, G., Holmes, regulated) and visualized as above. A.B., Hall, M.N., Michell, R.H., and Parker, P.J. (1998). The stress-activated phosphatidylinositol 3-phosphate 5-kinase Supplementary Material Fab1p is essential for vacuole function in S. cerevisiae. Curr. Supplementary Material describing the sequence comparison of Biol. 8, 1219–1222. Vac14 and its putative orthologs in diverse eukaryotes is available 16. Gary, J.D., Wurmser, A.E., Bonangelino, C.J., Weisman, L.S., at http://images.cellpress.com/supmat/supmatin.htm. and Emr, S.D. (1998). Fab1p is essential for PtdIns(3)P 5-kinase activity and the maintenance of vacuolar size and membrane Acknowledgments homeostasis. J. Cell Biol. 143, 65–79. 17. Odorizzi, G., Babst, M., and Emr, S.D. (1998). Fab1p PtdIns(3)P We thank Dr. Lois Weisman and Dr. Robert Piper (University of 5-kinase function essential for protein sorting in the multivesicu- Iowa), Professor Hugh Pelham (LMB, Cambridge, UK), Professor lar body. Cell 95, 847–858. Peter Parker and Dr. T. Jefferies (ICRF, London, UK), Dr. Frank 18. Odorizzi, G., Babst, M., and Emr, S.D. (2000). Phosphoinositide Cooke (University College London, UK), and Dr. Miklos Cserzo (Uni- signaling and the regulation of membrane trafficking in yeast. versity of Birmingham, UK) for helpful discussions and for sharing Trends Biochem. Sci. 25, 229–235. unpublished data. This work was supported by grants from the Royal 19. Burd, C.G., Babst, M., and Emr, S.D. (1998). Novel pathways, Society and Wellcome Trust. S.K.D. is a Royal Society University membrane coats and PI kinase regulation in yeast lysosomal trafficking. Semin. Cell Dev. Biol. 9, 527–533. Research Fellow. 20. Morishita, M., Morimoto, F., Kitamura, K., Koga, T., Fukui, Y., Maekawa, H., Yamashita, I., and Shimoda, C. (2002). Phosphati- Received: April 3, 2002 dylinositol 3-phosphate 5-kinase is required for the cellular re- Revised: April 25, 2002 sponse to nutritional starvation and mating pheromone signals Accepted: April 25, 2002 in Schizosaccharomyces pombe. Genes Cells 7, 199–215. Published: June 4, 2002 21. Bryant, N.J., and Stevens, T.H. (1998). Vacuole biogenesis in Saccharomyces cerevisiae: protein transport pathways to the References yeast vacuole. Microbiol. Mol. Biol. Rev. 62, 230–247. 22. Ikonomov, O.C., Sbrissa, D., and Shisheva, A. (2001). Mamma- 1. Dove, S.K., Cooke, F.T., Douglas, M.R., Sayers, L.G., Parker, lian cell morphology and endocytic membrane homeostasis re- P.J., and Michell, R.H. (1997). Osmotic stress activates phos- quire enzymatically active phosphoinositide 5-kinase PIKfyve. phatidylinositol-3,5-bisphosphate synthesis. Nature 390, J. Biol. Chem. 276, 26141–26147. 187–192. 23. Reggiori, F., and Pelham, H.R. (2002). A transmembrane ubiqui- 2. Whiteford, C.C., Brearley, C.A., and Ulug, E.T. (1997). Phosphati- tin ligase required to sort membrane proteins into multivesicular dylinositol 3,5-bisphosphate defines a novel PI 3-kinase path- bodies. Nat. Cell Biol. 4, 117–123. way in resting mouse fibroblasts. Biochem. J. 323, 597–601. 24. Gary, J.D., Sato, T.K., Stefan, C.J., Bonangelino, C.J., Weisman, 3. McEwen, R.K., Dove, S.K., Cooke, F.T., Painter, G.F., Holmes, L.S., and Emr, S.D. (2002). Regulation of Fab1 PtdIns(3)P A.B., Shisheva, A., Ohya, Y., Parker, P.J., and Michell, R.H. 5-kinase pathway by the Vac7 protein and Fig4, a polyphos- (1999). Complementation analysis in PtdInsP kinase-deficient phoinositide phosphatase family member. Mol. Biol. Cell yeast mutants demonstrates that Schizosaccharomyces pombe 13, 1238–1251. and murine Fab1p homologues are phosphatidylinositol 25. Bonangelino, C.J., Nau, J.J., Duex, J.E., Brinkman, M., Wurmser, 3-phosphate 5-kinases. J. Biol. Chem. 274, 33905–33912. A.E., Gary, J.D., Emr, S.D., and Weisman, L.S. (2002). Osmotic 4. Berridge, M.J., Cobbold, P.H., and Cuthbertson, K.S. (1988). stress-induced increase of phosphatidylinositol 3,5-bisphos- Spatial and temporal aspects of cell signalling. Philos. Trans. phate requires Vac14p, an activator of the lipid kinase Fab1p. R. Soc. Lond. B Biol. Sci. 320, 325–343. J. Cell Biol. 156, 1015–1028. 5. Lemmon, M.A., and Ferguson, K.M. (2000). Signal-dependent 26. Dujon, B. (1998). European Functional Analysis Network (EURO- membrane targeting by pleckstrin homology (PH) domains. Bio- FAN) and the functional analysis of the Saccharomyces cerevis- chem. J. 350, 1–18. iae genome. Electrophoresis 19, 617–624. 6. Carpenter, C.L., and Cantley, L.C. (1996). Phosphoinositide ki- 27. Schu, P.V., Takegawa, K., Fry, M.J., Stack, J.H., Waterfield, nases. Curr. Opin. Cell Biol. 8, 153–158. M.D., and Emr, S.D. (1993). Phosphatidylinositol 3-kinase en- 7. Michell, R.H. (1997). The multiplying roles of inositol lipids and coded by yeast VPS34 gene essential for protein sorting. Sci- phosphates in cell control processes. Essays Biochem. 32, ence 260, 88–91. 31–47. 28. Becherer, K.A., Rieder, S.E., Emr, S.D., and Jones, E.W. (1996). 8. Cullen, P.J., Cozier, G.E., Banting, G., and Mellor, H. (2001). Novel syntaxin homologue, Pep12p, required for the sorting of Modular phosphoinositide-binding domains—their role in sig- lumenal hydrolases to the lysosome-like vacuole in yeast. Mol. nalling and membrane trafficking. Curr. Biol. 11, R882–R893. Biol. Cell 7, 579–594.

9. Martin, T.F. (2001). PI(4,5)P2 regulation of surface membrane 29. Gomes de Mesquita, D.S., van den Hazel, H.B., Bouwman, J., traffic. Curr. Opin. Cell Biol. 13, 493–499. and Woldringh, C.L. (1996). Characterization of new vacuolar Vac14 Is a Fab1 Activator 893

segregation mutants, isolated by screening for loss of protein- 50. Demirov, D.G., Ono, A., Orenstein, J.M., and Freed, E.O. (2002). ase B self-activation. Eur. J. Cell Biol. 71, 237–247. Overexpression of the N-terminal domain of TSG101 inhibits 30. Bryant, N.J., Piper, R.C., Weisman, L.S., and Stevens, T.H. HIV-1 budding by blocking late domain function. Proc. Natl. (1998). Retrograde traffic out of the yeast vacuole to the TGN Acad. Sci. USA 99, 955–960. occurs via the prevacuolar/endosomal compartment. J. Cell 51. Hicke, L. (2001). Protein regulation by monoubiquitin. Nat. Rev. Biol. 142, 651–663. Mol. Cell. Biol. 2, 195–201. 31. Muren, E., Oyen, M., Barmark, G., and Ronne, H. (2001). Identifi- 52. Fromont-Racine, M., Mayes, A.E., Brunet-Simon, A., Rain, J.C., cation of yeast deletion strains that are hypersensitive to brefel- Colley, A., Dix, I., Decourty, L., Joly, N., Ricard, F., Beggs, J.D., din A or monensin, two drugs that affect intracellular transport. et al. (2000). Genome-wide protein interaction screens reveal Yeast 18, 163–172. functional networks involving Sm-like proteins. Yeast 17, 32. Cserzo, M., Eisenhaber, F., Eisenhaber, B., and Simon, I. (2002). 95–110. On filtering false positive transmembrane protein predictions. 53. Wendland, B., and Emr, S.D. (1998). Pan1p, yeast eps15, func- Prot. Eng., in press. tions as a multivalent adaptor that coordinates protein-protein 33. Budovskaya, Y.V., Hama, H., DeWald, D.B., and Herman, P.K. interactions essential for endocytosis. J. Cell Biol. 141, 71–84. (2002). The C terminus of the Vps34p phosphoinositide 3-kinase 54. Huang, K.M., D’Hondt, K., Riezman, H., and Lemmon, S.K. is necessary and sufficient for the interaction with the Vps15p (1999). Clathrin functions in the absence of heterotetrameric protein kinase. J. Biol. Chem. 277, 287–294. adaptors and AP180-related proteins in yeast. EMBO J. 18, 34. Andrade, M.A., Petosa, C., O’Donoghue, S.I., Muller, C.W., and 3897–3908. Bork, P. (2001). Comparison of ARM and HEAT protein repeats. 55. Pelham, H.R. (1999). SNAREs and the secretory pathway-les- J. Mol. Biol. 309, 1–18. sons from yeast. Exp. Cell Res. 247, 1–8. 35. Mireskandari, A., Reid, R.L., Kashanchi, F., Dittmer, J., Li, W.B., 56. Ungermann, C., Nichols, B.J., Pelham, H.R., and Wickner, W. and Brady, J.N. (1996). Isolation of a cDNA clone, TRX encoding (1998). A vacuolar v-t-SNARE complex, the predominant form a human T-cell lymphotrophic virus type-I Tax1 binding protein. in vivo and on isolated vacuoles, is disassembled and activated Biochim. Biophys. Acta 1306, 9–13. for docking and fusion. J. Cell Biol. 140, 61–69. 36. Vowels, J.J., and Payne, G.S. (1998). A dileucine-like sorting 57. Fischer von Mollard, G., and Stevens, T.H. (1999). The Saccharo- signal directs transport into an AP-3-dependent, clathrin-inde- myces cerevisiae v-SNARE Vti1p is required for multiple mem- pendent pathway to the yeast vacuole. EMBO J. 17, 2482–2493. brane transport pathways to the vacuole. Mol. Biol. Cell 10, 37. Liu, S.H., Marks, M.S., and Brodsky, F.M. (1998). A dominant- 1719–1732. negative clathrin mutant differentially affects trafficking of mole- 58. Hughes, W.E., Cooke, F.T., and Parker, P.J. (2000). Sac phos- cules with distinct sorting motifs in the class II major histocom- phatase domain proteins. Biochem. J. 350, 337–352. patibility complex (MHC) pathway. J. Cell Biol. 140, 1023–1037. 59. Winzeler, E.A., Shoemaker, D.D., Astromoff, A., Liang, H., Ander- 38. Mullins, C., and Bonifacino, J.S. (2001). Structural requirements son, K., Andre, B., Bangham, R., Benito, R., Boeke, J.D., Bussey, for function of yeast GGAs in vacuolar protein sorting, alpha- H., et al. (1999). Functional characterization of the S. cerevisiae factor maturation, and interactions with clathrin. Mol. Cell. Biol. genome by gene deletion and parallel analysis. Science 285, 21, 7981–7994. 901–906. 39. Cowles, C.R., Odorizzi, G., Payne, G.S., and Emr, S.D. (1997). The AP-3 adaptor complex is essential for cargo-selective transport to the yeast vacuole. Cell 91, 109–118. 40. Odorizzi, G., Cowles, C.R., and Emr, S.D. (1998). The AP-3 complex: a coat of many colours. Trends Cell Biol. 8, 282–288. 41. Ogawa, N., DeRisi, J., and Brown, P.O. (2000). New components of a system for phosphate accumulation and polyphosphate metabolism in Saccharomyces cerevisiae revealed by genomic expression analysis. Mol. Biol. Cell 11, 4309–4321. 42. Reggiori, F., and Pelham, H.R. (2001). Sorting of proteins into multivesicular bodies: ubiquitin-dependent and -independent targeting. EMBO J. 20, 5176–5186. 43. Dove, S.K., and Michell, R.H. (1999). Analysis of the polyphos- phorylated inositol lipids of S. cerevisiae. In Signalling Research: A Practical Approach, G. Milligan, ed. (London: Oxford Univer- sity Press), pp. 255–261. 44. Hirst, J., Lui, W.W., Bright, N.A., Totty, N., Seaman, M.N., and Robinson, M.S. (2000). A family of proteins with gamma-adaptin and VHS domains that facilitate trafficking between the trans- Golgi network and the vacuole/lysosome. J. Cell Biol. 149, 67–80. 45. Costaguta, G., Stefan, C.J., Bensen, E.S., Emr, S.D., and Payne, G.S. (2001). Yeast Gga coat proteins function with clathrin in Golgi to endosome transport. Mol. Biol. Cell 12, 1885–1896. 46. Amerik, A.Y., Nowak, J., Swaminathan, S., and Hochstrasser, M. (2000). The Doa4 deubiquitinating enzyme is functionally linked to the vacuolar protein-sorting and endocytic pathways. Mol. Biol. Cell 11, 3365–3380. 47. Rotin, D., Staub, O., and Haguenauer-Tsapis, R. (2000). Ubiquiti- nation and endocytosis of plasma membrane proteins: role of Nedd4/Rsp5p family of ubiquitin-protein ligases. J. Membr. Biol. 176, 1–17. 48. Garrus, J.E., von Schwedler, U.K., Pornillos, O.W., Morham, S.G., Zavitz, K.H., Wang, H.E., Wettstein, D.A., Stray, K.M., Cote, M., Rich, R.L., et al. (2001). Tsg101 and the vacuolar protein sorting pathway are essential for HIV-1 budding. Cell 107, 55–65. 49. Patnaik, A., Chau, V., and Wills, J.W. (2000). Ubiquitin is part of the retrovirus budding machinery. Proc. Natl. Acad. Sci. USA 97, 13069–13074.