Ubiquitin Is Required for Sorting to the Vacuole of the Yeast General Amino Acid Permease, Gap1*

In yeast, ubiquitin plays a central role in proteolysis of a multitude of proteins and serves also as a signal for endocytosis of many plasma membrane proteins. We showed previously that ubiquitination of the general amino acid permease (Gap1) is essential to its endocytosis followed by vacuolar degradation. These processes occur when NH 4 (cid:1) , a preferential source of nitrogen, is added to cells growing on proline or urea, i.e. less fa-vored nitrogen sources. In this study, we show that Gap1 is ubiquitinated on two lysine residues in the cytosolic N terminus (positions 9 and 16). A mutant Gap1 in which both lysines are mutated (Gap1 K9K16 ) remains fully stable at the plasma membrane after NH 4 (cid:1) addition. Furthermore, each of the two lysines harbors a poly-ubiquitin chain in which ubiquitin is linked to the lysine 63 of the preceding ubiquitin. The Gap1 K9 and Gap1 K16 mutants, in which a single lysine is mutated, are down-regulated in response to NH 4 (cid:1) although more slowly. In proline-grown cells lacking Npr1, a protein kinase involved in the control of Gap1 trafficking, newly synthesized Gap1 is sorted from the Golgi to the vacuole without passing through the plasma membrane

Ubiquitin is a 76-amino acid protein, which, in all eukaryotes, undergoes conjugation to a multitude of proteins. Although ubiquitination generally serves as a recognition signal for degradation by the proteasome (1,2), studies in yeast have shown that ubiquitination of plasma membrane proteins results in their endocytosis followed by vacuolar degradation (3). Proteins subject to this mechanism include the G-proteincoupled mating pheromone receptors Ste2 (4) and Ste3 (5) and several transporters: the ABC proteins Ste6 (6) and Pdr5 (7), the uracil permease Fur4 (8), the amino acid permease Gap1 (9,10), the tryptophan permease Tat2 (11), the galactose permease Gal2 (12), and the zinc transporter Zrt1 (13). Ubiquitination of most of these proteins has been shown to involve the ubiquitin-conjugating enzymes (E2) encoded by the UBC1-4 genes and an HECT-type ubiquitin ligase (E3) encoded by the essential NPI1/RSP5 gene (14). Ubiquitin has been shown to contain an endocytosis signal in the form of two surface patches surrounding two critical residues (Phe 4 and Ile 44 ) (15). However, the protein components of the endocytosis machinery involved in ubiquitin recognition remain unknown. It also remains undetermined as to whether ubiquitin also plays a role in the late steps of endocytosis and whether plasma membrane proteins undergo successive cycles of ubiquitination-de-ubiquitination during transit to the vacuole.
Here we have investigated the role of ubiquitin in the internal trafficking of the general amino acid permease (Gap1), 1 which is tightly regulated by nitrogen. On proline or urea medium, i.e. conditions of poor nitrogen supply, the GAP1 gene is transcribed to high levels (16), and the synthesized Gap1 permease accumulates at the plasma membrane in an active and stable form (17,18). Upon the addition of NH 4 ϩ (a preferential source of nitrogen), Gap1 is internalized by endocytosis and targeted to the vacuole for degradation. Ubiquitination of Gap1 is essential to this NH 4 ϩ -induced down-regulation (9,10). In the npi1 mutant, which displays an abnormally low level of the HECT-type ubiquitin ligase Npi1/Rsp5, or the npi2 mutant lacking the Npi2/Doa4 de-ubiquitinating enzyme, Gap1 is not ubiquitinated and stays at the plasma membrane after NH 4 ϩ addition (10,19). Furthermore, as shown for the uracil permease Fur4 (8), Gap1 is poly-ubiquitinated, the ubiquitin moieties being attached to the lysine 63 of the preceding ubiquitin (19) (henceforth called the lysine 63-linked poly-ubiquitin chain). Gap1 poly-ubiquitination is required for down-regulation of the permease at a maximal rate (19). The fate of newly synthesized Gap1 in the late secretory pathway is also under nitrogen control. On proline or urea medium, neosynthesized Gap1 is delivered to the plasma membrane, but in a medium containing glutamate (20) or NH 4 ϩ (18) as the sole nitrogen source, Gap1 is directly sorted from the Golgi to the vacuole without passing via the cell surface. A similar situation has been observed on proline medium with cells lacking Npr1, a protein kinase controlling both cell-surface and internal Gap1 (18) and apparently inactivated by phosphorylation when good nitrogen sources are available (21).
In this paper we show that Gap1 is ubiquitinated on two lysine residues in its extreme N terminus, at positions 9 and 16. Using the Gap1 K9K16 variant in which both lysine residues are mutated, we show that ubiquitination of Gap1 is required not only for down-regulation of the protein pre-accumulated at the cell surface but also for direct sorting of the protein from the late secretory pathway to the vacuole, as occurs in an npr1⌬ mutant. We further show that ubiquitination and degradation of both cell-surface and internal Gap1 requires Bul1 and Bul2, two proteins interacting with the Npi1/Rsp5 ubiquitin ligase. As this paper was being reviewed, it was reported by others (22) that sorting of Gap1 to the vacuole requires its polyubiquitination and that the specific role of Bul1 and Bul2 is to specify this modification (see "Discussion").

EXPERIMENTAL PROCEDURES
Strains, Growth Conditions, and Plasmids-All Saccharomyces cerevisiae strains used in this study (see Table I) are isogenic with ⌺1278b (23). Cells were grown in minimal buffered medium (24) with 3% glucose as the carbon source except when mentioned otherwise. In steady-state experiments, proline (10 mM) was the sole nitrogen source. In experiments of Gap1 neosynthesis, cells were grown exponentially on glutamine (5 mM) or NH 4 ϩ (100 mM) and transferred to proline medium to relieve GAP1 repression. In ubiquitin overexpression experiments, cells were grown on glutamine medium and transferred to preheated YNB (yeast nitrogen base without NH 4 ϩ or amino acids; Difco) medium containing 10 mM proline, 3% glucose, and 0.1 mM CuSO 4 to induce expression of ubiquitin. The 2 -based multi-copy plasmid YEp96 contains a synthetic yeast ubiquitine gene under the control of the copperinducible CUP1 promoter (25). The plasmids encoding the ubiquitin in which lysine residue 63 (UbK63R), or lysine residues 29, 48, and 63 (UbRRR) are replaced by arginine are derived from YEp96 (26). Centromeric plasmid YCpGAP1 (27) is based on the YCpFL38 plasmid (28). Plasmid YCpFL39 was used to complement trp1 auxotrophy (28).
Mutagenesis-Site-directed mutagenesis of GAP1 was performed using the Quick Change Site-directed Mutagenesis Kit (Stratagene) on plasmid YCpGAP1 as recommended by the supplier. The primers used for each construct are described in Table II. Each construct was checked entirely by sequencing.
Permease Assays-Gap1 activity was determined by measuring incorporation of 20 M 14 C-labeled citrulline as described by Grenson (29). To avoid competitive inhibition of citrulline transport by glutamine, cells grown on glutamine medium were filtered, washed, and transferred to preheated proline medium just before the transport assay. The permease was inactivated by adding preheated (NH 4 ) 2 SO 4 to the culture (final concentration, 10 mM).
Yeast Cell Extracts and Immunoblotting-Crude cell extracts (9) and membrane-enriched preparations were prepared as previously described (10). In Western blot experiments, protein concentrations were estimated by densitometry of the Pma1 signal (ImageMaster1D, Amersham Pharmacia Biotech). Equal quantities of protein were loaded on an 8% SDS-polyacrylamide gel in a Tricine system (30). After transfer to a nitrocellulose membrane (Schleicher and Schü ll), the proteins were probed with polyclonal antibodies raised against Gap1 (1: 10 000) or Pma1 (1: 1 000). Primary antibodies were detected with horseradish peroxidase-conjugated anti-rabbit IgG secondary antibody (Amersham Pharmacia Biotech) followed by enhanced chemiluminescence (Roche Molecular Biochemicals).

RESULTS
Direct Sorting of Neosynthesized Gap1 to the Vacuole Requires the Npi1/Rsp5 Ubiquitin-Protein Ligase-In wild-type cells grown on a medium containing urea or proline as sole nitrogen source, newly synthesized Gap1 is sorted from the Golgi to the plasma membrane where it accumulates in an active and stable form. In contrast, in mutant cells lacking the Npr1 kinase, neosynthesized Gap1 is sorted from the Golgi to the vacuole without passing via the plasma membrane (18). This direct sorting of Gap1 to the vacuole also occurs when wild-type cells are grown on a medium containing either glutamate (20) or NH 4 ϩ (18) as sole nitrogen source, this being consistent with Npr1 being inactive under these favorable nitrogen supply conditions (35).
It was previously reported that loss of Gap1 activity in npr1 mutants is suppressed by the npi1 mutation (17,18). This mutation results in a markedly reduced level of the HECT-type ubiquitin ligase Npi1/Rsp5, leading to non-ubiquitination of Gap1 (9,10,36). These observations raise the interesting possibility that direct sorting of Gap1 from the late secretory pathway to the vacuole involves ubiquitination of the permease. To investigate this possibility, we monitored the fate of newly synthesized Gap1 in wild-type, npr1⌬, npi1, and npr1⌬ npi1 cells (Fig. 1A). The cells were first grown on glutamine MATa KanMX2-GAL-GAP1 ura3 MATa npi2 trp1 ura3 medium to repress transcription of the GAP1 gene, then transferred to proline medium to relieve repression. In the wild type, this resulted in the appearance of a high intensity Gap1 signal on immunoblots and of high Gap1 activity in citrulline uptake assays (Fig. 1A). The npi1 strain displayed a similar phenotype. In the npr1⌬ mutant, in keeping with the observation that Gap1 is directly sorted from the secretory pathway to the vacuole (18), no Gap1 activity was measured and only a low quantity of Gap1 was detectable after the cells were shifted to proline medium. The npr1⌬ npi1 strain displayed a phenotype similar to that observed with the wild-type and npi1 strains, indicating that in the double mutant, neosynthesized Gap1 is  A, wild-type (W-T) (23344c), npr1⌬ (30788a), npi1 (27038a), and npr1⌬ npi1 (30788d) strains were grown on glutamine medium at 29°C. The cells were then transferred to proline medium. Upper panel, Gap1 activity (nmol.min Ϫ1 .mg prot Ϫ1 ) was assayed in cells growing on glutamine (Gln) and 3 h after transfer to proline medium (Pro) by measuring the incorporation of [ 14 C]citrulline (0.02 mM). The maximal Gap1 activity in the wild type (100%) corresponded to 20 nmol.min Ϫ1 .mg Ϫ1 protein.
Lower panel, immunoblot of Gap1 present in crude extracts prepared from cell samples taken from the same cultures before (Gln) and 3 h after transfer to proline medium (Pro). To make sure that equal amounts of protein were loaded, Pma1 was also immunodetected. B, 33201b), and GAL1-GAP1 npr1⌬ npi1 (ƒ, 33191a) were grown on raffinose-proline medium. Galactose (200 mM final concentration) was added to induce GAP1 expression. Gap1 activity (nmol.min Ϫ1 .mg prot Ϫ1 ) was assayed before and at various times after the addition of galactose by measuring the uptake of [ 14 C]citrulline (0.02 mM).
targeted to the cell surface rather than to the vacuole. To confirm this result, we placed the GAP1 gene under the control of the galactose-inducible GAL1 promoter and monitored the neosynthesis of Gap1 by adding galactose to cells grown on a raffinose-proline medium (Fig. 1B). As expected, this resulted in the progressive increase of Gap1 activity in the wild type, indicating that Gap1 was delivered to the plasma membrane, whereas the permease remained inactive in the npr1⌬ mutant. In the npr1⌬ pep12⌬ double mutant lacking the t-SNARE (target-soluble N-ethylmaleimide attachment protein receptor) protein Pep12 (required for transport of proteins from the Golgi to the late endosome/pre-vacuolar compartment) (37), Gap1 activity gradually appeared after galactose addition in a manner similar to that in the pep12⌬ mutant. This confirms that the pep12⌬ mutation largely suppresses the effect of the npr1⌬ mutation, even though the activity in the pep12⌬ strain is lower than in the wild type (18). The same phenotype was observed in the npr1⌬ npi1 strain (Fig. 1B). These results confirm those of Fig. 1A and show that sorting of Gap1 from the late secretory pathway to the vacuole requires the ubiquitin ligase Npi1/Rsp5. They also show that when Npi1/Rsp5 is lacking, at least part of the neosynthesized Gap1 is rerouted to the plasma membrane.
Direct Sorting of Neosynthesized Gap1 to the Vacuole Requires a Normal Pool of Ubiquitin-To further assess the role of ubiquitin in the direct sorting of Gap1 from the Golgi to the vacuole, we monitored the fate of neosynthesized Gap1 in npr1⌬ cells also lacking the Npi2/Doa4 ubiquitin hydrolase. This enzyme facilitates ubiquitin recycling from proteasometargeted substrates (38). In mutants affected in the Npi2/Doa4 ubiquitin hydrolase, the internal pool of ubiquitin is reduced severalfold (8,19,39); this impairs ubiquitination and the subsequent down-regulation of Gap1, which normally occur when NH 4 ϩ is added to proline-grown cells (19). Wild-type, npi2, npr1⌬, and npr1⌬ npi2 strains were grown on glutamine medium and then shifted to proline medium (Fig. 2). As expected, Gap1 remained inactive in the npr1⌬ strain, and the quantity of Gap1 detected was much lower than in the wild type. In the npr1⌬ npi2 strain, Gap1 was as active as in the wild type, and an even higher amount of Gap1 was detected after the cells were shifted to proline medium, indicating that the npi2 mutation results in rerouting of neosynthesized Gap1 to the plasma membrane. In the npr1⌬ npi2 strain overexpressing ubiquitin, a phenotype similar to that of the npr1⌬ strain was observed, confirming that the effect of the npi2 mutation can be overcome by increasing the internal ubiquitin pool. Hence, direct sorting to the vacuole of neosynthesized Gap1 in npr1⌬ cells requires a normal pool of ubiquitin, and if this pool is too limiting, Gap1 is rerouted to the plasma membrane. Lysine Residues at Positions 9 and 16 Are Essential to Downregulation of Gap1-The data presented above show that direct sorting to the vacuole of newly synthesized Gap1, as occurs in the npr1⌬ mutant, requires normal levels of both Npi1/Rsp5 ubiquitin ligase and monomeric ubiquitin. These results suggest that ubiquitination of Gap1 could be required for its sorting to the vacuole. To test this hypothesis, experiments were conducted to isolate a mutant form of Gap1 resistant to ubiquitination. Previous work has identified the lysine residues of several permeases to which ubiquitin is attached (11,13,14). In the case of the uracil permease, for instance, ubiquitin is covalently linked to two lysine residues in a PEST region at the extreme N terminus (positions 38 and 41) (40). When these residues are mutated, ubiquitination and endocytosis of the permease are impaired. Each residue, furthermore, is subject to poly-ubiquitination, the ubiquitin moieties of the poly-ubiquitin chains being linked via the Lys 63 residue of ubiquitin (41). In the case of the tryptophan permease Tat2, a protein homologous in sequence to Gap1, mutation of the five lysine residues present in the 31 N-terminal amino acids preceding the first transmembrane domain were needed to protect the permease against down-regulation induced by rapamycin treatment (11). These observations prompted us to mutagenize lysine residues present in the cytosolic N terminus of Gap1. Two Gap1 mutants were thus constructed in which lysine residues were replaced with arginine, respectively, at positions 9 and 16 (Gap1 K9K16 ) and positions 51, 56, 60, and 63 (Gap1 K51-K63 ) (Fig. 3A). The gap1⌬ strain was transformed with centromere-based plasmids bearing the GAP1, GAP1 K9K16 , or GAP1 K51-K63 gene expressed under the natural GAP1 promoter. Citrulline uptake assays performed on transformed cells grown on proline medium revealed that both the GAP1 K9K16 and the GAP1 K51-K63 gene encode fully active Gap1 permeases. The addition of NH 4 ϩ to GAP1 cells resulted in the endocytosis of Gap1, as shown by progressive loss of Gap1 activity (Fig. 3B), followed by degradation of the permease, as shown by progressive disappearance of the Gap1 signal detected on immunoblots (Fig. 3C) (10). A similar situation was observed with cells expressing the GAP1 K51-K63 gene, indicating that lysine residues at positions 51, 56, 60, and 63 are not important for NH 4 ϩ -induced downregulation. In contrast, the permease encoded by the GAP1 K9K16 allele remained active and stable after NH 4 ϩ addition (Fig. 3, B and C), demonstrating that the lysine residue(s) at position(s) 9 and/or 16 is/are crucial to down-regulation of Gap1. To determine the contribution of each lysine residue, we constructed alleles GAP1 K9 and GAP1 K16 . In the corresponding proteins, the lysine residue at position 9 or 16, respectively, is replaced with arginine. Experiments showed that both of these Gap1 variants are fully active on proline medium (Fig. 3B). After NH 4 ϩ addition, both permeases were down-regulated, indicating that the lysine residues at positions 9 and 16 are each To confirm that equal amounts of proteins were loaded, Pma1 was also immunodetected. sufficient to promote NH 4 ϩ -induced down-regulation of Gap1. However, the Gap1 K9 and Gap1 K16 variants were down-regulated significantly more slowly than the native Gap1 protein, indicating that both lysine residues are required to promote maximal-rate down-regulation.
Lysine Residues 9 and 16 of Gap1 Are Acceptor Sites for Lys 63linked Poly-ubiquitin Chains-We next examined the ubiquitination state of Gap1, Gap1 K9K16 , Gap1 K9 , and Gap1 K16 . To facilitate detection of ubiquitin-permease conjugates, we performed immunoblotting with membrane-enriched fractions (10). We used cells overexpressing ubiquitin: gap1⌬ cells expressing the plasmid-borne GAP1, GAP1 K9 , GAP1 K16 , or GAP1 K9K16 allele were additionally transformed with a high-copy vector bearing the ubiquitin gene under the control of the inducible CUP1 promoter (26). The signal immunodetected in proline-grown cells expressing native Gap1 consisted of a major band at 60 kDa plus minor upper bands at higher molecular masses, corresponding to ubiquitin-conjugated forms of the permease (Fig. 4, upper panel) (10,19). In keeping with previous observations (10,19), the amount of ubiquitin-conjugated forms of Gap1 increased in the early minutes after NH 4 ϩ addition. Upper bands were also observed in cells expressing the GAP1 K9 or GAP1 K16 allele, showing that these Gap1 mutant forms are still ubiquitinated. No upper bands were detected in cells expressing the GAP1 K9K16 allele (Fig. 4, upper panel) even if the immunoblots were overexposed (not shown). These data show that Gap1 is ubiquitinated specifically on the lysine residues at positions 9 and 16.
That Gap1 is ubiquitinated on only two acceptor sites was previously deduced from experiments aimed at determining the type of ubiquitin chain linked to Gap1 (19). Upon overexpression in an npi2 mutant of a ubiquitin mutant unable to form any kind of poly-ubiquitin chain, Gap1 forms attached to zero, one, or two ubiquitin moieties were detected (19). A similar pattern was specifically observed with cells overexpressing a ubiquitin variant (Ub K63R ) defective in formation of Lys 63linked chains, indicating that poly-ubiquitin chain formation on Gap1 involves linkage through the lysine 63 residue of ubiquitin (19). The data presented in Fig. 4 (lower panel) show that upon overexpression of Ub K63R instead of normal ubiquitin in cells expressing GAP1 K9 or GAP1 K16 genes, a single upper band accumulates above the main Gap1 signal. In contrast, two bands accumulated above the main Gap1 signal in cells ex-pressing wild-type GAP1. These results show that both lysine residues (positions 9 and 16) are ubiquitinated with Lys 63linked chains.
Ubiquitination of Gap1 Is Required for Direct Sorting from the Late Secretory Pathway to the Vacuole-The data presented in Fig. 3 show that non-ubiquitination of Gap1 at positions 9 and 16 renders the permease pre-accumulated at the plasma membrane resistant to NH 4 ϩ -induced down-regulation. To determine whether these lysine residues of Gap1 are also required for direct sorting of neosynthesized permease from the late secretory pathway to the vacuole, we transformed a gap1⌬ npr1⌬ strain with the plasmid-borne GAP1, GAP1 K9 , GAP1 K16 , or GAP1 K9K16 gene. Cells were grown on a medium containing NH 4 ϩ at high concentration (100 mM) so as to repress GAP1 gene expression and were then shifted to proline medium to relieve repression (Fig. 5). As expected, no Gap1 activity was measured in the gap1⌬ npr1⌬ strain expressing the wild-type GAP1 allele. The results were the same when the GAP1 K9 or GAP1 K16 allele was expressed in the same strain. High Gap1 activity was measured, however, in cells producing the Gap1 K9K16 variant. Hence the Gap1 K9K16 variant, defective in ubiquitination, is rerouted to the plasma membrane upon neosynthesis in an npr1⌬ mutant. This result confirms those presented above (Figs. 1 and 2) and shows that ubiquitination of Gap1 is essential to its direct sorting from the late secretory pathway to the vacuole. Ubiquitination of a single lysine residue (position 9 or 16) is apparently sufficient for the sorting of Gap1 to the vacuole, because the Gap1 K9 and Gap1 K16 mutants both behave like wild-type Gap1. We next assessed whether poly-ubiquitin chain formation is required for the sorting of Gap1 to the vacuole. For this, we monitored the activity and immunodetected levels of Gap1 neosynthesized in an npr1⌬ npi2 strain overexpressing either normal ubiquitin or a ubiquitin mutant (UbRRR) in which lysines at positions 29, 48, and 63 are replaced by arginine (Fig.  6). This ubiquitin variant is unable to form poly-ubiquitin chains in vivo (26). As illustrated above, overexpression of normal ubiquitin largely suppressed the effect of the npi2 mutation in the npr1⌬ strain, i.e. Gap1 remained poorly active and did not accumulate to high levels after the shift of the cells from glutamine to proline. Although a slightly higher activity and level of Gap1 were reproducibly observed when UbRRR instead of normal ubiquitin was overexpressed in the npr1⌬ npi2 strain, the effect of the npi2 mutation was also largely overcome by UbRRR expression. This indicates that poly-ubiquitin chain formation is not essential to the direct sorting of Gap1 to the vacuole.
Bul1 and Bul2, Two Additional Factors Required for Ubiquitination and Down-regulation of Cell-surface Gap1, Are Also Required for Direct Sorting of the Permease to the Vacuole-Bul1 and Bul2 are highly similar proteins (51% identity) shown in two-hybrid system, cosedimentation, and immunoprecipitation experiments to interact with the Npi1/Rsp5 ubiquitin ligase (42,43). There is evidence that this interaction involves a PY-motif in the Bul proteins (XPPXY) and some or all of the three repeats of the WW(P) domain in Npi1/Rsp5. As Bul1 is not a substrate of Npi1/Rsp5, it has been proposed that the Bul proteins function together with Npi1/Rsp5 in protein ubiquitination (42,43). To assess the role of Bul1 and Bul2 in controlling Gap1 trafficking, bul1⌬ and bul2⌬ single mutants and a bul1⌬ bul2⌬ double mutant were isolated. We first analyzed the influence of the bul1⌬ and bul2⌬ mutations on Gap1 activity and stability after the addition of NH 4 ϩ to cells grown on proline medium. In the bul1⌬ bul2⌬ mutant, Gap1 was totally protected against the NH 4 ϩ -induced inactivation and degradation observed in the wild-type strain (Fig. 7, A and B). We next assessed the level of Gap1 ubiquitination in this mutant. In keeping with previous observations (10,19), immunoblots of membrane-enriched extracts of wild-type cells harvested after NH 4 ϩ addition showed an increased intensity of the upper bands corresponding to ubiquitin-conjugated forms of Gap1 (Fig. 7C). These upper bands were barely detectable in the bul1⌬ bul2⌬ double mutant, indicating that ubiquitination of Gap1 is largely defective in this mutant (Fig. 7C). Similar experiments with the bul1⌬ and bul2⌬ single mutants revealed that Gap1 is still ubiquitinated and largely sensitive to NH 4 ϩinduced down-regulation in these strains (not shown). Hence, the Bul1 and Bul2 proteins appear to share a redundant function essential to Gap1 ubiquitination and subsequent downregulation of cell-surface Gap1. We finally tested whether the direct sorting of neosynthesized Gap1 to the vacuole occurring in an npr1⌬ strain also involves the Bul proteins. The npr1⌬ strain and an npr1⌬ bul1⌬ bul2⌬ triple mutant were grown on high NH 4 ϩ medium and were then shifted to proline medium (Fig. 7D). As expected, no Gap1 activity was detected in the npr1⌬ mutant. In contrast, the Gap1 activity of the npr1⌬ bul1⌬ bul2⌬ strain was as high as that of the wild type, showing that neosynthesized Gap1 is rerouted to the plasma membrane in this mutant. This result confirms that a deficiency in Gap1 ubiquitination not only protects Gap1 present at the plasma membrane against NH 4 ϩ -triggered down-regulation but also prevents Gap1 neosynthesized in an npr1⌬ strain from being sorted to the vacuole. DISCUSSION In this report we show that ubiquitination of the yeast Gap1 permease takes place on two lysine residues in the extreme N terminus of the protein (positions 9 and 16), a region recently shown to be cytosolic (44). Furthermore, each lysine residue carries a chain consisting of at least two ubiquitin moieties linked to each other via the lysine 63 residue of ubiquitin. A similar situation has been described for the uracil permease (Fur4), which is poly-ubiquitinated by Lys 63 -linked chains on residues 38 and 41 (40). The Gap1 K9K16 variant, in which both lysine residues are replaced with arginine, is fully protected against endocytosis and subsequent degradation, which normally occur when NH 4 ϩ is added to cells growing on proline or urea as the sole nitrogen source. The Gap1 K9 and Gap1 K16 variants, in which a single lysine residue is changed, are downregulated after NH 4 ϩ addition. This indicates that ubiquitination of Gap1 on a single lysine residue (at position 9 or 16) is sufficient for effective NH 4 ϩ -triggered endocytosis and degradation. However, down-regulation of both Gap1 K9 and Gap1 K16 is slightly slower than wild-type Gap1, indicating that ubiquitination of Gap1 on both lysines is required for maximal rate down-regulation. We have also shown that, in addition to the Npi1/Rsp5 ubiquitin ligase, at least one of the two highly similar proteins found to interact with Npi1/Rsp5, i.e. Bul1 or Bul2 (42,43), is essential to NH 4 ϩ -induced ubiquitination and downregulation of Gap1. Taken together, these results suggest that a complex containing Npi1/Rsp5 and at least one of the Bul proteins promotes binding of ubiquitin to lysine residues 9 and 16 of Gap1. Further experiments will be needed to determine whether subsequent poly-ubiquitination of Gap1 via the Lys 63 residue of ubiquitin (19) is also mediated by the Npi1/Rsp5-Bul system or whether another ubiquitin ligase enzyme is involved.
Our data further show that ubiquitin plays an essential role in another pathway of Gap1 trafficking, direct sorting of neosynthesized Gap1 from the late secretory pathway to the vacuole. This direct sorting to the vacuole occurs in cells in which Npr1 is inactive, i.e. in the npr1⌬ mutant grown on proline medium and in wild-type cells growing under good nitrogen supply conditions (18,20). We have shown here that if, under these conditions, ubiquitination of Gap1 is defective (as a result of an npi1, npi2, bul1⌬ bul2⌬, or Gap1 K9K16 mutation), the permease is rerouted to the plasma membrane. We conclude that in the wild-type strain grown on proline medium, Gap1 is targeted to the plasma membrane, but that loss of the Npr1 function results in sorting of Gap1 to the vacuole (18). This sorting requires ubiquitination of the permease on at least one of the two lysine residues 9 and 16. In the absence of Gap1 ubiquitination in an npr1⌬ mutant, Gap1 is retargeted to the plasma membrane. The mechanisms by which the Npr1 kinase positively regulates the sorting of internal Gap1 to the cell surface remain undetermined. Phosphorylation of Gap1 is not strictly dependent on Npr1, suggesting that the effect of Npr1 is indirect (18). The precise role of ubiquitin in the targeting of Gap1 to the vacuole also remains undetermined. Ubiquitin might serve as a signal for packaging Gap1 into vesicles bound for the vacuole, for instance vesicles budding from the late Golgi and/or ones formed by invagination of the prevacuole/late endosome membrane to form a multivesicular body (45). When this ubiquitination is defective, Gap1 could be recycled from this compartment to the plasma membrane.
Our data thus contribute to the growing body of evidence that ubiquitin plays an important role in the sorting of membrane proteins in the endosomal and/or late Golgi system(s) (46). In the same line, an unexpected link between ubiquitin and the endosomal system was recently evidenced by the partial association of the Npi2/Doa4 ubiquitin hydrolase enzyme with the late endosome (47). Furthermore, suppressor mutations bypassing the requirement for the Npi2/Doa4-de-ubiquitinating enzyme appear to affect genes coding for components of the vacuolar protein-sorting (Vps) pathway (47). In another study, a mutant form of the tryptophan permease (Tat2) incapable of binding ubiquitin appeared to be stabilized under FIG. 7. Bul1 and Bul2 are involved in the down-regulation and direct sorting of Gap1 to the vacuole. A, the wild-type (f, 23344c) and the OS27-1 (OE, bul1⌬ bul2⌬) strain were grown on proline medium. At time 0, NH 4 ϩ (20 mM) was added. Gap1 activity (nmol.min Ϫ1 .mg prot Ϫ1 ) was assayed before and at various times after NH 4 ϩ addition by measuring the uptake of [ 14 C]citrulline (0.02 mM). B, immunoblot of Gap1 present in crude extracts from cell samples collected from the same cultures before and various times after NH 4 ϩ addition. C, the wild-type (W-T) (23344c) and the OS27-1 (bul1⌬ bul2⌬) strains were grown on proline medium and transferred to proline supplemented with CuSO 4 (0.1 mM) 2 h before the addition of NH 4 ϩ . Cells were collected before and 5 min after NH 4 ϩ addition. Membrane-enriched fractions were prepared, and Gap1 was detected by Western immunoblotting. D, the wild type (f, 23344c) and strains OS27-1 (, bul1⌬ bul2⌬), 30788a (OE, npr1⌬), and JA410 (ࡗ, bul1⌬ bul2⌬ npr1⌬) were grown on NH 4 ϩ (100 mM) medium, and cells were transferred to proline. Gap1 activity (nmol.min Ϫ1 .prot Ϫ1 ) was assayed before and at various times after transfer by measuring the uptake of [ 14 C]citrulline (0.02 mM).