The Rsp5 Ubiquitin Ligase Binds to and Ubiquitinates Members of the Yeast CIN85-Endophilin Complex, Sla1-Rvs167*

Sla1 and Rvs167 are yeast proteins required for receptor internalization and organization of the actin cytoskeleton. Here we provide evidence that Sla1 and Rvs167 are orthologues of the mammalian CIN85 and endophilin proteins, respectively, which are required for ligand-stimulated growth factor receptor internalization. Sla1 is similar in domain structure to CIN85 and binds directly to the endophilin-like Rvs167. Akin to CIN85, Sla1 interacts with synaptojanins and a ubiquitin ligase that regulates endocytosis. This ubiquitin ligase, Rsp5, binds directly to both Sla1 and Rvs167. The interaction between Rsp5 and Rvs167 is mediated through Rsp5 WW domains and P X Y motifs in the central Gly-Pro-Ala-rich domain of Rvs167. Rvs167 P X Y motifs are required for Rsp5-dependent monoubiquitination of Rvs167 on Lys 481 in the Src homology 3 (SH3) domain. Mutation of Lys 481 3 Arg causes cells to grow slowly on medium containing 1 M NaCl, although this phenotype is not due to the defect in ubiquitination caused by the K481R mutation. We propose that Rsp5 interaction with Sla1-Rvs167 promotes Rvs167 ubiquitination and regulates activity of this protein complex. Rvs167 ubiquitination

Sla1 and Rvs167 are yeast proteins required for receptor internalization and organization of the actin cytoskeleton. Here we provide evidence that Sla1 and Rvs167 are orthologues of the mammalian CIN85 and endophilin proteins, respectively, which are required for ligand-stimulated growth factor receptor internalization. Sla1 is similar in domain structure to CIN85 and binds directly to the endophilin-like Rvs167. Akin to CIN85, Sla1 interacts with synaptojanins and a ubiquitin ligase that regulates endocytosis. This ubiquitin ligase, Rsp5, binds directly to both Sla1 and Rvs167. The interaction between Rsp5 and Rvs167 is mediated through Rsp5 WW domains and PXY motifs in the central Gly-Pro-Ala-rich domain of Rvs167. Rvs167 PXY motifs are required for Rsp5-dependent monoubiquitination of Rvs167 on Lys 481 in the Src homology 3 (SH3) domain. Mutation of Lys 481 3 Arg causes cells to grow slowly on medium containing 1 M NaCl, although this phenotype is not due to the defect in ubiquitination caused by the K481R mutation. We propose that Rsp5 interaction with Sla1-Rvs167 promotes Rvs167 ubiquitination and regulates activity of this protein complex. Rvs167 ubiquitination is not required for general function of Rvs167, but may control specific Rvs167 SH3 domain-protein interactions or negatively regulate SH3 domain activity.
A diverse protein network is required to sort and internalize cell surface cargo into primary endocytic vesicles that bud from the plasma membrane. Many protein-protein interactions between members of this network have been mapped, suggesting a dynamic process that must be exquisitely temporally and spatially regulated. Phosphorylation is an important regulator of interaction between endocytic proteins (1,2), particularly as a modification that can disassemble endocytic complexes to prepare for another cycle of cargo sorting and vesicle formation. Recently, interaction with ubiquitin ligases and post-translational modification by ubiquitin have also emerged as important regulators of endocytic proteins (3)(4)(5)(6).
The specificity of ubiquitination reactions is regulated by ubiquitin ligases (E3s). Several ubiquitin ligases play key roles in regulating transport through the endocytic pathway, in particular ubiquitin ligases of the Nedd4 and Cbl families. These proteins ubiquitinate multiple components of the endocytic machinery (e.g. see Refs. 4,7,8,and 9) and modify cell surface proteins with a ubiquitin sorting signal that targets cargo for internalization and degradation in the lysosome (10,11). Ubiquitin-dependent endocytosis controls the cell surface activity of a large variety of proteins (reviewed in Ref. 12). Ubiquitin is the principal signal for the entry of plasma membrane cargo into primary endocytic vesicles in yeast and can also function as an internalization signal in mammalian cells (13,14). Ubiquitin-dependent internalization relies on proteins that orchestrate the selection of endocytic cargo, rearrangements of the underlying actin cytoskeleton at the plasma membrane, and deformation and scission of the plasma membrane. Of these proteins, epsin, Eps15, ␤-arrestin, and CIN85 are known to be ubiquitinated. Specifically, they are modified with monoubiquitin (4,7,(15)(16)(17)(18)(19).
CIN85 (Cbl-interacting protein of 85 kDa) and its close homologue, CMS (Cas ligand with multiple SH3 domains), are part of the endocytic network and bind to many proteins involved in endocytosis and signaling (20). A number of observations suggest that CIN85/CMS regulates the actin cytoskeleton and may act as one link between actin polymerization and receptor internalization (e.g. see Refs. 21 and 22). CIN85 has three N-terminal SH3 1 domains followed by a proline-rich region and a coiled-coil region that mediates dimerization. The Cbl ubiquitin ligase was recently found to recruit a complex of CIN85 with another endocytic protein, endophilin, to facilitate the internalization of activated, ubiquitinated growth factor receptors (5,23). Furthermore, CIN85 is monoubiquitinated at an unknown site(s) by both Nedd4 and Cbl (7,18).
Endophilin and amphiphysin are proteins that share a number of properties and are required for the internalization step of endocytosis. Both proteins have N-terminal BAR (Bin, amphiphysin, Rvs) domains that mediate lipid interactions in vitro and a C-terminal SH3 domain that binds to dynamin and to the polyphosphoinositide phosphatases, synaptojanins (24,25). Each protein can independently promote tubulation or vesiculation of membranes (26,27), and both proteins are predicted to coordinate membrane deformation with actin cytoskeleton dynamics (28,29). Endophilin and amphiphysin are localized to the highly curved necks that link a deeply invaginated vesicle to the donor membrane (30,31), and they are thought to function at a late step in vesicle budding together with dynamin to promote vesicle scission. Like CIN85, endophilin is implicated in regulation of the actin cytoskeleton. Endophilin binds to and regulates N-WASP, an activator of actin polymerization (29). In addition, after treatment of cells with growth factor, the CIN85-endophilin complex recruits cortactin, another regulator of actin polymerization (21,29). The yeast endophilin/amphiphysin homologue, Rvs167 (reduced viability upon starvation 167), shares structural features of both mammalian proteins and functions in both endocytosis and actin regulation (32,33).
Here we identify endocytic proteins that bind to Rsp5, the sole yeast homologue of the Nedd4 family of ubiquitin ligases. Rsp5, like most Nedd4 family members, contains a N-terminal C2 domain, WW protein-protein interaction domains, and a C-terminal catalytic HECT (homology to E6-AP C terminus) domain. We find that Rsp5 co-precipitates with the interacting proteins Sla1 and Rvs167 required for receptor internalization and organization of the actin cytoskeleton. We further investigate the similarity of yeast Sla1-Rvs167 to the mammalian CIN85-endophilin complex and the consequences of Rsp5 interaction with Sla1 and Rvs167.

EXPERIMENTAL PROCEDURES
Strains, Media, and Reagents-Yeast strains used in this study are listed in Table I. Genetic manipulations were performed by standard techniques, and cell transformations were performed by the lithium acetate method (34). The rvs167⌬::TRP1 disruption has been described (33). Cells were propagated in synthetic minimal medium (34), YPUAD medium (35), casamino acids medium (0.67% yeast nitrogen base, 0.5% vitamin assay casamino acids, 2% glucose supplemented with 50 mg/liter adenine, histidine, and tryptophan), or YNB selective medium (0.67% yeast nitrogen base, 2% glucose supplemented with commercially purchased selective amino acid mixes) as indicated in the figure legends. Sodium chloride (1 M) was added to 2% agar-containing medium prior to autoclaving. The growth medium of rsp5⌬ (LHY2492) and congenic RSP5 (LHY2491) cells was supplemented with 0.2% Nonidet P-40 and 2 mM oleic acid (free acid; Sigma) to rescue the lethality of the rsp5⌬ mutation.
Purification of polyclonal Ste2 antibodies was performed as previously described (36). His 6 antibodies were obtained from Amersham Biosciences, and GST antibodies were from Molecular Probes, Inc. (Eugene, OR). Rvs167, Sla1, and hexokinase antisera were gifts from Howard Riezman (University of Geneva, Switzerland), Gregory Payne (UCLA School of Medicine, Los Angeles, CA), and Gottfried Schatz (Biozentrum, University of Basel, Switzerland), respectively, and monoclonal HA (12CA5) antibodies were kindly provided by Robert Lamb (Northwestern University, Evanston, IL). Rsp5 for antibody production was purified from Escherichia coli expressing GST-Rsp5 from the pGEX-6P-1 vector (Ben Sehgal and Hilary Godwin, Northwestern University). The fusion protein was isolated on glutathione-Sepharose beads, and Rsp5 was cleaved with Prescission protease (Amersham Biosciences). Purified Rsp5 was used for injection of two rabbits (Covance Research Products Inc., Denver, PA).
Plasmid pGEX-6P-2 (Amersham Biosciences) was used for production of GST-Sla1. Plasmids encoding GST fused to the N terminus of Sla1 (codons 1-420; LHP2098) or to a central fragment (codons 420 -720; LHP2099) were generated by PCR amplification of the appropriate DNA using BamHI site-containing oligonucleotides. The PCR products were digested and ligated into BamHI-digested pGEX-6P-2. The in-frame fusions were verified by automated sequencing.
Plasmids used for production of GST and GST-Rsp5 were pGEX-4T (Amersham Biosciences) and pGST-RSP5 (a gift from Jon Huibregtse, University of Texas, Austin, TX). A plasmid encoding GST fused to the three WW domains of RSP5 (pGST-3ϫWW, LHP703) was generated by PCR amplification of the WW domains (codons 228 -430) using a 5Ј EcoRI site-containing oligonucleotide and a 3Ј XhoI site-containing oligonucleotide. The PCR product was digested and ligated into EcoRI-and XhoI-digested pGEX-PKT (37).
INP52 was amplified from yeast genomic DNA using 5Ј PstI and 3Ј BamHI site-containing oligonucleotides, digested and ligated into PstI-and BamHI-digested YCplac22 to yield plasmid LHP1459. The NotI site was inserted at the 5Ј-end of INP52 in LHP1459 via QuikChange TM mutagenesis. A triple HA epitope was ligated into the NotI site to generate pHA-INP52 (LHP1463). A PstI-SacI HA-INP52 fragment was subcloned into a LEU2-marked multicopy plasmid (YEplac181) digested with PstI and SacI to generate LHP2077.
Preparation of Cell Lysates for Immunoblotting-For detection of Ste2, cell extracts were prepared by glass bead lysis and SDS/urea extraction as previously described (35,36). For detection of ubiquitinated Rvs167, cell lysates were prepared by alkaline lysis and trichloroacetic acid precipitation as described (38).
Native Co-immunoprecipitations-Cells were lysed in native immunoprecipitation buffer (0.2 M sorbitol, 50 mM potassium acetate, 25 mM potassium chloride, 10 mM HEPES, 1 mM EDTA, pH 7.0) containing yeast protease inhibitors by vortexing the cell suspension with acid-washed glass beads 4 -6 times with intervening rests on ice. Whole cell lysates were extracted with 2 mg/ml ␤-D-maltoside for 30 -60 min on ice and clarified by centrifugation at 20,000 ϫ g for 10 min at 4°C. A fraction of each cleared lysate was reserved for analysis, and the remainder was incubated with Rsp5 antiserum followed by incubation with protein A-Sepharose beads (Amersham Biosciences). Beads were collected at 100 ϫ g and washed four times in native immunoprecipitation buffer containing yeast protease inhibitors. Bound proteins were eluted by boiling in SDS sample buffer (39), resolved by SDS-PAGE, and analyzed by immunoblotting.
GST Fusion Protein Precipitation Experiments-GST pulldown experiments with yeast lysates were performed as previously described with minor modifications (35). Lysates were prepared in MES lysis buffer with the addition of 1% Triton X-100. The lysates were extracted with 2 mg/ml ␤-D-maltoside and incubated with glutathione-Sepharose beads containing equivalent amounts (ϳ20 g each) of GST, GST-3ϫWW, or GST-Rsp5 for 2 h at room temperature or overnight at 4°C. After the binding incubation, beads were washed four times in MES lysis buffer, and bound proteins were eluted by boiling in SDS sample buffer. Eluted proteins were analyzed by immunoblotting.
E. coli cells were induced to express His 6 -tagged Rvs167 proteins with 0.1 mM isopropyl-␤-D-thiogalactopyranoside for 1-3 h. Cells expressing full-length His 6 -Rvs167 were lysed by pulsed sonication in phosphate-buffered saline containing protease inhibitors. The lysate was extracted with 1% Triton X-100 and clarified by centrifugation at 4°C. A fraction of the cleared lysate was reserved for analysis, and the remainder was split into equal aliquots and incubated with GST, GST-3ϫWW, or GST-Rsp5 glutathione-bound Sepharose beads for 2 h at 4°C. After several washes in phosphate-buffered saline containing protease inhibitors and 0.2% Triton X-100, bound proteins were eluted by boiling in SDS sample buffer. Cells expressing His 6 -tagged GPA and SH3 fragments were lysed by the same procedure in 50 mM sodium phosphate buffer, 300 mM sodium chloride with 1 mM phenylmethylsulfonyl fluoride and 1 g/ml pepstatin. The tagged proteins were purified on TALON metal affinity resin according to the manufacturer's recommendations for native purification (Clontech) and eluted with 200 mM imidazole in the same buffer. Purified proteins were diluted into an E. coli cell lysate (prepared as described above) to equivalent concentrations. Binding to GST and GST-3ϫWW was performed as described for full-length His 6 -Rvs167. Samples were resolved by SDS-PAGE and stained with Coomassie Brilliant Blue G-250 (Bio-Rad) or analyzed by immunoblotting.
Immobilization of and Binding to Bacterially Expressed His 6 -tagged Proteins-His 6 -tagged Sla1 polypeptides were immobilized on TALON metal affinity resin by preparing bacterial lysates as described above and binding to the beads as previously described (40). Pull-down assays with the immobilized polypeptides were performed as previously described with several modifications (35). Bound proteins (10 g) were incubated with 0.5 mg of yeast lysate proteins containing either Rvs167-HA or HA-Vps9 (9) prepared as described above. After the incubation, the beads were washed two times in MES lysis buffer plus 10 mM imidazole and two times in MES buffer plus 20 mM imidazole. Bound proteins were eluted by boiling in SDS sample buffer.
A yeast lysate containing HA-Inp52 was prepared as previously described (41), except that the lysis buffer contained 3 mg/ml bovine serum albumin and 1% Nonidet P-40. In addition, beads with immobilized His 6 -tagged polypeptides were preincubated in buffer containing 1 mg/ml bovine serum albumin in Tris-buffered saline, pH 7.5, for 1 h at 4°C. After the binding incubation, beads were washed two times in lysis buffer, 10 mM imidazole, and 1% Nonidet P-40 and two times with lysis buffer, 20 mM imidazole, and 1% Nonidet P-40.

RESULTS
The Rsp5 Ubiquitin Ligase Binds Directly to Rvs167 and Sla1-Genetic studies suggested that Rsp5 regulates components of the receptor internalization machinery (3). To identify proteins involved in endocytosis that bind to Rsp5, we generated polyclonal rabbit antiserum against full-length native Rsp5. On immunoblots, Rsp5 antiserum recognized a single protein of the expected size in wild type cells but not in rsp5⌬ cells (Fig. 1A). We then performed native immunoprecipitations with anti-Rsp5 on lysates prepared from wild type yeast strains, some of which expressed HA-tagged versions of different endocytic proteins. Although interactions with a number of endocytic proteins were observed inconsistently, Rvs167-HA and Sla1 reproducibly co-precipitated with Rsp5 (Fig. 1B). The Rsp5-Rvs167 interaction was further confirmed by yeast twohybrid experiments (Table II). During the course of this study, a systematic analysis of interactions in the yeast proteome identified a complex containing both Rsp5 and Rvs167, providing a third independent confirmation of this protein-protein interaction (42).
Rvs167 and Sla1 are proteins required for receptor internalization and regulation of the actin cytoskeleton organization (32,33,43,44), and both bind to many other proteins involved in actin polymerization and endocytosis. Sla1 carries three N-terminal SH3 domains, central SHD1 (Sla1-homology domain 1) and coiled-coil domains, and a C-terminal repeat domain (Fig. 1C). Sla1 associates with other endocytic proteins at the plasma membrane early in the formation of an endocytic vesicle (45), but the molecular mechanisms by which Sla1 acts are unknown. Rvs167 is an endophilin/amphiphysin-like protein that is one of two yeast proteins containing a BAR domain. The Rvs167 BAR domain is at the N terminus, followed by a Gly-Pro-Ala-rich (GPA) region of 176 amino acids and a Cterminal SH3 domain (Fig. 1C). Rvs167 forms a BAR domainmediated heterodimer with the other yeast BAR protein, Rvs161 (46,47).
To test whether the observed physical interaction between Rsp5 and Rvs167 was direct, we expressed recombinant His 6tagged Rvs167 in bacteria and tested for interaction with recombinant GST-Rsp5 fusion proteins. Rvs167 bound specifically to immobilized GST fusion proteins containing full-length Rsp5 or a fragment containing the WW domains alone ( Fig.  2A). To localize the Rsp5 interaction domain in Rvs167, we expressed His 6 -tagged Rvs167 fragments comprising each of the defined domains (BAR, GPA, and SH3; see Fig. 1C). Whereas the SH3 domain showed no interaction with either GST or the WW domains, the GPA domain bound specifically to immobilized WW domains (Fig. 2B). It was not possible to definitively test interaction of the BAR domain by this assay, because the recombinant BAR domain bound to GST alone. However, binding of BAR to GST-3ϫWW domains was similar to its interaction with GST, suggesting that the BAR domain does not contribute significantly to the interaction of Rvs167 with WW domains. These data demonstrate that the GPA domain of Rvs167 interacts directly with Rsp5 WW domains.
Rsp5 contains type I WW domains, defined by affinity for PPXY motif-containing ligands (48). The Rvs167 GPA domain carries the sequence P 398 P 399 AY and two imperfect PPXY motifs, P 334 SY and P 372 QY (Fig. 3A). To analyze the role of these motifs in Rsp5 interaction, we generated a series of mutations in the Rvs167 PPXY and PXY sequences. All of the mutant proteins were expressed at normal levels in cells carrying a chromosomal disruption of RVS167 (Fig. 3A). In addition, the mutant alleles restored growth of rvs167⌬ cells on high salt medium at all temperatures and rescued temperature-sensitive growth of rvs167⌬ cells on medium containing caffeine, 2 suggesting that the mutant proteins fold normally and complement at least some aspects of Rvs167 function. The P398A,P399A mutation significantly diminished but did not abolish interaction with Rsp5 in vitro (Fig. 3B) and inhibited the ability of Rvs167 to co-precipitate with Rsp5 in yeast lysates (Fig. 3C). Complete deletion of this motif (⌬PPPAY) also did not abolish Rsp5 binding, suggesting that the two imperfect PXY motifs upstream in the GPA domain contribute to the 2 R. Dunn and L. Hicke, unpublished results. interaction. Mutations of one (3P3 A) or two (4P3 A) of the upstream PXY motifs in combination with the P398A,P399A mutation further reduced binding, indicating that the GPA domain contains redundant proline-based interaction motifs for Rsp5.
Rvs167 and Sla1 have been reported to interact in the yeast two-hybrid system (49); thus, it is possible that our observed co-precipitation of Sla1 with Rsp5 (Fig. 1B) was due to an indirect interaction via Rvs167. To test this possibility, we performed native Rsp5 immunoprecipitations in wild type and rvs167⌬ cells. Sla1 precipitates with Rsp5 regardless of the presence of Rvs167 (Fig. 4A). To determine whether the Sla1-Rsp5 interaction is direct, we expressed three His 6 -tagged fragments of Sla1 in E. coli: Sla1 1-420, which includes the three SH3 domains, Sla1 420 -720, which contains the SHD1 and coiled-coil domains, and Sla1 720 -1240, which harbors 26 repeats of the approximate consensus sequence TGGX 2-6 PQ and a Gln-rich region. Sla1 420 -720 bound specifically to recombinant GST-Rsp5 purified from bacterial lysates (Fig. 4B), indi-  6 -tagged Rvs167 was incubated with GST, GST-3ϫWW, or GST-Rsp5 immobilized on beads for 2 h at 4°C. Bound proteins and a fraction of the input lysate were analyzed by SDS-PAGE and Coomassie Blue staining. The identity of the Rvs167 species was confirmed by its reactivity with both anti-His 6 and anti-Rvs167 antibodies. 3 B, interaction of recombinant fragments of Rvs167 with Rsp5 WW domains. Bacterially expressed His 6 -tagged GPA and SH3 domains were purified on and eluted from a metal affinity resin. To prevent nonspecific binding, the purified domains were diluted into E. coli lysates at equivalent concentrations and incubated with immobilized GST and GST-3ϫWW as in A. Bound proteins and a fraction of the input lysate were resolved on a 16.5% Tris-Tricine-SDS gel and analyzed by Coomassie Blue staining (top panel) and immunoblotting with Rvs167 antiserum (bottom panel). cating that the Sla1-Rsp5 interaction is direct and is mediated by the central region of Sla1.
Sla1 and Rvs167 Bind Directly to Each Other and Are Similar to the Mammalian CIN85-Endophilin Complex-Above, we described physical interactions between the Rsp5 ubiquitin ligase and two yeast proteins required for receptor internalization. In mammalian cells, the Cbl ubiquitin ligase binds to the CIN85-endophilin complex, a component of the endocytic machinery important for growth factor receptor internalization. Sla1 is the yeast protein most like CIN85 in domain structure (Fig. 1C), sharing 34% similarity over its first 600 amino acids. Like CIN85, Sla1 is a multivalent adaptor that binds a large number of other proteins involved in endocytosis and actin organization, including Pan1, End3, Las17, and Sla2 (50 -52). Rvs167 is similar in structural organization to mammalian endophilins, although Rvs167 has a central GPA domain not found in endophilins (Fig. 1C). In addition, Rvs167 and Sla1 have been observed to interact in a high throughput yeast two-hybrid analysis (49). These observations suggest that Sla1 and Rvs167 may be functionally similar to CIN85 and endophilin, respectively. To test this idea, we first determined whether Rvs167 and Sla1 bind directly to each other. The three His 6tagged fragments of Sla1 previously used to test Rsp5 interaction were expressed in E. coli, immobilized on metal affinity beads, and incubated with yeast lysates prepared from cells expressing either Rvs167-HA or, as a control, another HAtagged protein required later in the endocytic pathway (HA-Vps9). Rvs167 bound to Sla1 1-420 and weakly to Sla1 420 -720 and Sla1 720 -1240 (Fig. 5A). HA-Vps9 did not bind to any of the Sla1 fragments (Fig. 5A). 3 We were unable to test direct binding of Rvs167 to Sla1 720 -1240 because GST-Sla1 720 -1240 was not stably expressed in E. coli; however, recombinant His 6 -Rvs167 bound to GST-Sla1 1-420 and weakly to GST-Sla1 420 -720 (Fig. 5B), indicating that Rvs167 binds to Sla1 directly, perhaps via multiple sites. The observation that rvs167⌬ and sla1⌬ mutations are synthetically lethal provides evidence for a functional relationship between Rvs167 and Sla1 (53). Thus, Sla1, like CIN85, interacts physically and functionally with an endophilin-like protein.
Rvs167 Is Monoubiquitinated by Rsp5-Rsp5 is required to modify plasma membrane proteins with a ubiquitin internalization signal (reviewed in Ref. 10), although it is not known how Rsp5 recognizes its membrane substrates. We considered the possibility that Rvs167 or Sla1 may link Rsp5 to plasma membrane endocytic cargo to facilitate cargo ubiquitination. To test this idea, we analyzed ligand-stimulated ubiquitination of 3 S. D. Stamenova and L. Hicke, unpublished data.

FIG. 4. Rsp5 binds directly to the central region of Sla1.
A, Sla1 co-precipitates with Rsp5 from yeast lysates in the absence of Rvs167. Native precipitations (IP) with Rsp5 antiserum were performed from RVS167 (LHY2363) and rvs167⌬ (LHY2086) cell lysates and analyzed as described for Fig. 1. The anti-Rsp5 immunoblot (IB) was included to demonstrate that equivalent amounts of Rsp5 were present in each precipitate. B, binding of recombinant Sla1 fragments to immobilized GST-Rsp5. E. coli lysates containing recombinant His 6 -tagged Sla1 fragments were incubated with GST or GST-Rsp5 beads for 1 h at 4°C. Bound proteins and a fraction of the input lysate were analyzed by SDS-PAGE followed by immunoblotting with anti-His 6 antibodies.

FIG. 5. Sla1 binds to Rvs167 and synaptojanin.
A, Rvs167-HA in yeast lysates binds to fragments of Sla1. Lysates prepared from cells expressing Rvs167-HA (LHY2363) or HA-Vps9 (negative control, LHY2726) were incubated with bacterially expressed Sla1 fragments immobilized on metal affinity beads. Eluted proteins were detected by SDS-PAGE followed by Coomassie Blue staining to visualize equivalent loading of the Sla1 fragments or by immunoblotting with HA antibodies to detect bound Rvs167 or Vps9. B, the N-terminal and central domains of Sla1 bind directly to Rvs167. Immobilized His 6 -Rvs167 domains or beads alone were incubated with lysates from E. coli expressing the indicated GST-Sla1 fragments. Eluted proteins were analyzed by SDS-PAGE and immunoblotting with GST antibodies. C, a yeast synaptojanin binds to the N terminus of Sla1. A lysate prepared from cells expressing HA-Inp52 (LHY4851) was incubated with bacterially expressed Sla1 1-420 immobilized on metal affinity beads. A fraction of the lysate and eluted proteins were analyzed by SDS-PAGE followed by immunoblotting with anti-HA antibodies to detect Inp52. the ␣-factor receptor (Ste2) in wild type cells, rvs167⌬ cells, and sla1⌬ cells (Fig. 6, A and B). In wild type cells, ␣-factor stimulated receptor ubiquitination. Ubiquitinated forms of the receptor accumulated in the end4-1 endocytosis mutant, whereas ubiquitinated receptors were undetectable in rsp5-1 cells, consistent with previous observations (35,36). Ste2 was ubiquitinated normally in both rvs167⌬ and sla1⌬ cells. Therefore, Rvs167 and Sla1 are not required for Rsp5-dependent ubiquitination of endocytic cargo.
To investigate the possibility that Sla1 and Rvs167 are substrates of Rsp5, we performed a variety of experiments to detect ubiquitinated forms of both proteins. We first examined these proteins in concentrated lysates prepared from wild type cells and cells deficient in Rsp5 or the related ubiquitin ligases Hul4 and Hul5. Immunoblotting with anti-Rvs167 antibodies revealed a higher molecular mass, 66-kDa species of Rvs167 that was absent in rsp5⌬ cells but was unaffected by deletions of HUL4 or HUL5 (Fig. 7A). The 66-kDa form migrated ϳ8.5 kDa above the primary Rvs167 species, consistent with the molecular weight of a single ubiquitin moiety. To confirm that the Rsp5-dependent modification of Rvs167 was a ubiquitin conjugate, Rvs167 species were analyzed in cells expressing plasmidborne ubiquitin, Myc epitope-tagged ubiquitin, or a vector control. Ubiquitin overexpression induced the higher molecular mass form of Rvs167, and the mobility of this form shifted when ubiquitin carried a Myc epitope (Fig. 7B). Induced ubiquitination was highly specific, because modification of HAtagged Arc15, another PPXY motif-containing protein required for actin regulation and receptor internalization, was not detected in the same lysates. 2 These data indicate that Rvs167 is modified by monoubiquitin in vivo in an Rsp5-dependent reaction. We have not been able to detect ubiquitination of Sla1 using similar assays.
Because the PXY and PPXY sequences in the Rvs167 GPA domain are required for Rsp5 interaction, we tested whether point mutations in these sequences reduce its ubiquitination. Disruption of the PPXY motif (P398A,P399A) significantly re-duced the ubiquitinated Rvs167 species, and additional loss of upstream PXY motifs (3P3 A and 4P3 A) further diminished the modification (Fig. 7C). Furthermore, mutations in the Rsp5 WW domains, which bind to Rvs167, also reduced Rvs167 ubiquitination. Deletion of the entire C2 domain or point mutation of each WW domain individually had little effect on Rvs167 ubiquitination; however, mutation of all three WW domains simultaneously severely diminished ubiquitination (Fig. 7D). 2 These observations together indicate that there is strong correlation between the requirements for Rvs167-Rsp5 binding and Rvs167 monoubiquitination.
Site and Consequences of Rvs167 Monoubiquitination-Rsp5 binds to Rvs167 through PXY motifs in the GPA domain. However, because the GPA domain contains no lysines, it cannot be the site of ubiquitination. To determine the site of Rvs167 monoubiquitination, we mutated the single lysine in the Cterminal SH3 domain (Lys 481 ) or the 32 lysine residues at the N terminus individually or in pairs to arginine. Mutation of lysines in the N terminus (such as K290R immediately upstream of the GPA domain) did not affect the level of the ubiquitinated Rvs167 species, but monoubiquitinated Rvs167 was reduced by the K481R mutation and was not detectable upon deletion of the entire SH3 domain (Fig. 8A and unpublished data). These results are consistent with recently published mass spectrometry data (54,55), indicating that the site of Rvs167 ubiquitination is Lys 481 , the penultimate amino acid in the protein.
The SH3 domain of Rvs167 binds to proteins important for endocytosis and/or regulation of the actin cytoskeleton, includ- ing Las17 and Abp1. Deletion of the Rvs167 SH3 domain does not affect localization of a fluid phase endocytic marker (47) or the kinetics of ␣-factor internalization. 3 Furthermore, rvs167 K481R , rvs167 ⌬SH3 and rvs167 4P3A strains internalized receptors bearing a ubiquitin or linear peptide internalization signal as well as wild type cells. 3 These observations suggest that the Rvs167 SH3 domain and its modification by ubiquitin are dispensable for endocytosis.
We next tested whether Lys 481 and its ubiquitination are important for the ability of Rvs167 to support the growth of yeast cells on medium containing a high concentration of salt. rvs167 K481R and rvs167 ⌬SH3 cells grew poorly on this medium, whereas rvs167 P398A,P399A cells grew as well as wild type cells (Fig. 8B). These results indicate that Lys 481 is important for a subset of Rvs167 functions. However, because mutations in the PPXY motif that also impair ubiquitination at Lys 481 (see Fig.  8A) do not cause salt sensitivity, defective monoubiquitination at this site probably does not account for the salt-sensitive growth defect in the K481R mutant.

DISCUSSION
Rvs167 and Sla1 are two yeast proteins that act as part of the endocytic machinery and regulate assembly of the actin cytoskeleton. Each protein is important for fluid phase endocytosis and ␣-factor receptor internalization (33,44). Although it was previously suggested that deletion of SLA1 has little effect on internalization of ␣-factor by wild type Ste2 (56), we find that ␣-factor internalization by wild type Ste2 is reduced 6-fold in sla1⌬ cells, 3 similar to another published report (44).
Here we report that Rvs167 and Sla1 interact directly with each other and with the Rsp5 ubiquitin ligase. Our observations strongly suggest that Sla1 is a functional homologue of mammalian CIN85 and the closely related protein, CMS. Although CIN85 is smaller than Sla1, the proteins are 34% similar over the entire length of CIN85, and each protein carries three N-terminal SH3 domains with similar spacing between the domains (see Fig. 1C). Sla1 is more similar in sequence, domain structure, and function to CIN85/CMS than to intersectin, previously suggested as a Sla1 functional homologue (56,57). Both CIN85/CMS and Sla1 appear to be scaffolding proteins that bind to a large variety of proteins involved in endocytosis, actin regulation, and, in the case of CIN85, signaling. Like CIN85/CMS, Sla1 is required for the internalization of signaling receptors, and we show that Sla1 binds to an endophilin-like protein and synaptojanins.
Rvs167 is similar to the mammalian amphiphysins and endophilins, and we propose that Rvs167 is an orthologue of both proteins. Rvs167, amphiphysins, and endophilins all regulate endocytosis and the actin cytoskeleton. Rvs167 binds to the other yeast BAR domain protein, Rvs161, analogous to amphiphysin isoforms that dimerize through their BAR domains (1,58,59). Like endophilin, Rvs167 binds to the actin assembly regulator N-WASP (Las17) through its C-terminal SH3 domain (29,47,59,60). Unlike the amphiphysins and endophilins, Rvs167 does not carry clathrin and clathrin adaptor (AP)binding motifs and does not appear to bind either of these proteins. This difference may exist because clathrin adaptors are not required for receptor internalization in yeast (61,62), and clathrin itself is not strictly required (63).
The interaction of Rsp5 with Rvs167 and Sla1 is not required for cargo ubiquitination, but an Rsp5-Rvs167 interaction is important for monoubiquitination of Rvs167. PXY and PPXY motifs in Rvs167 are necessary to bind to Rsp5 WW domains in vitro and for Rvs167 monoubiquitination in vivo, indicating that the Rvs167-Rsp5 interaction is physiologically relevant. Rsp5 binds to the central region of Sla1 (amino acids 420 -720). This domain does not have PXY motifs but does carry an SHD1 domain that interacts with NPF motifs (56). Rsp5 carries an NPF sequence in the HECT domain (amino acids 513-515); however, mutation of this sequence to APA does not inhibit Sla1 interaction or receptor internalization, and the Sla1 SHD1 domain is not required for co-precipitation of Sla1 with Rsp5. 3 Therefore, it is unlikely that the Sla1 SHD1 and Rsp5 NPF sequences mediate interaction between the proteins, and at this time it is not known how Sla1-Rsp5 binding occurs.
CIN85 appears to be constitutively monoubiquitinated by Nedd4 and undergoes ligand-induced ubiquitination by Cbl (7). Ubiquitination of endophilin has also been reported (64). We have observed Rvs167 monoubiquitination that is dependent on the Nedd4 homologue Rsp5, and this event is not stimulated by the addition of ␣-factor ligand. 3 Although we have not detected Sla1 ubiquitination in the absence or presence of ␣-factor, this event may occur at low levels or only in response to specific stimuli. Mutations in Rvs167 PXY motifs (Pro 3 Ala mutations) severely inhibit binding to Rsp5 in vitro and Rvs167 ubiquitination in vivo; however, we have not observed internalization, growth, or actin cytoskeleton phenotypes associated with these mutations.
Ubiquitination occurs on the penultimate amino acid of Rvs167, Lys 481 , in the SH3 domain. Lys 481 is not required for normal ␣-factor internalization kinetics but is important for growth on high salt medium. Because these defects are not observed with rvs167 P3A mutations that inhibit Lys 481 ubiquitination, this residue is likely to be important for protein-protein interactions in addition to being a ubiquitination site. It is unlikely that Lys 481 is important structurally, because it is in a location that is surface-exposed on known SH3 structures. Ubiquitination at Lys 481 may negatively regulate Rvs167 functions or its ability to interact with an SH3-binding partner. Alternatively, Lys 481 may be a site for other Lys-dependent modifications, such as acetylation, sumoylation, or methylation.
In sum, we have identified a system of protein interactions between the Rsp5 ubiquitin ligase and the yeast CIN85-endophilin homologues, Sla1-Rvs167. Rsp5 is required to monou- biquitinate Rvs167. Ubiquitination occurs on the SH3 domain of Rvs167 and may regulate SH3 binding to proteins that control actin polymerization.