QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

Vol. 18, Issue 5, 1803-1815, May 2007

This Article Abstract Full Text (PDF) All Versions of this Article: E06-11-1000v1 18/5/1803    most recent Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Copic, A. Articles by Schekman, R. Search for Related Content PubMed PubMed Citation Articles by Copic, A. Articles by Schekman, R.

Ent3p and Ent5p Exhibit Cargo-specific Functions in Trafficking Proteins between the Trans-Golgi Network and the Endosomes in Yeast

Alenka opi^*,, Trevor L. Starr^*,, and Randy Schekman^*

*Howard Hughes Medical Institute and Department of Molecular and Cell Biology, and Graduate Group in Microbiology, University of California at Berkeley, Berkeley, CA 94720

Submitted November 9, 2006; Revised January 26, 2007; Accepted February 21, 2007
Monitoring Editor: Sandra Lemmon

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
The phosphoinositide-binding proteins Ent3p and Ent5p are required for protein transport from the trans-Golgi network (TGN) to the vacuole in Saccharomyces cerevisiae. Both proteins interact with the monomeric clathrin adaptor Gga2p, but Ent5p also interacts with the clathrin adaptor protein 1 (AP-1) complex, which facilitates retention of proteins such as Chs3p at the TGN. When both ENT3 and ENT5 are mutated, Chs3p is diverted from an intracellular reservoir to the cell surface. However, Ent3p and Ent5p are not required for the function of AP-1, but rather they seem to act in parallel with AP-1 to retain proteins such as Chs3p at the TGN. They have all the properties of clathrin adaptors, because they can both bind to clathrin and to cargo proteins. Like AP-1, Ent5p binds to Chs3p, whereas Ent3p facilitates the interaction between Gga2p and the endosomal syntaxin Pep12p. Thus, Ent3p has an additional function in Gga-dependent transport to the late endosome. Ent3p also facilitates the association between Gga2p and clathrin; however, Ent5p can partially substitute for this function. We conclude that the clathrin adaptors AP-1, Ent3p, Ent5p, and the Ggas cooperate in different ways to sort proteins between the TGN and the endosomes.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
In eukaryotic cells, the clathrin coat is involved in endocytosis as well as in the sorting and transport of proteins between the trans-Golgi network (TGN) and the endosomes. Clathrin heavy and light chain form the outer layer of the coat, and adaptor proteins mediate the interaction between clathrin, the membrane surface, and cargo proteins in the membrane, as well as with other accessory proteins (Owen et al., 2004). Several different adaptor proteins have been identified to date, many of which interact to facilitate clathrin-mediated endocytosis. In intracellular clathrin-dependent protein trafficking, however, the relationships between the different adaptors are less well understood (Traub, 2005).

Adaptor protein 1 (AP-1) is a tetrameric clathrin adaptor complex that localizes to the TGN and/or early endosomes in yeast and in mammalian cells. It has been implicated in forward transport from the TGN, in retrograde transport from the early endosome back to the TGN, or in retention of proteins at the TGN (Meyer et al., 2000; Valdivia et al., 2002; Ha et al., 2003; Waguri et al., 2003; Foote and Nothwehr, 2006). AP-1 binds to phosphatidylinositol-4-phosphate, a phosphoinositide that is enriched on these membranes (Wang et al., 2003). The small GTPase Arf-1 and specific signals on cargo proteins (Meyer et al., 2005), as well as a recently discovered mammalian protein, -BAR (Neubrand et al., 2005), are also involved in the recruitment of AP-1 to lipid membranes (Meyer et al., 2005). Two consensus AP-1 binding motifs on cytosolic tails of cargo proteins have been identified, and their direct role in binding to AP-1 has been demonstrated in a few mammalian proteins (Peden et al., 2001; Bonifacino and Traub, 2003; Janvier et al., 2003). The recognition of a third type of AP-1 binding motif requires the participation of another mammalian protein, PACS-1 (Crump et al., 2001). Finally, an unrelated AP-1 binding signal has recently been discovered in yeast (Foote and Nothwehr, 2006).

Another type of clathrin adaptor, the monomeric GGA protein, has been implicated in the transport of proteins from the TGN to the late endosome, although GGAs also interact and colocalize with some early endocytic proteins in mammalian cells. Two domains in the GGAs have a role in cargo binding: the VHS domain specifically recognizes a DXXLL signal that is present in many mammalian proteins and is not recognized by any other adaptors, whereas the GAT domain binds to ubiquitinated proteins (Bonifacino, 2004). In trafficking of mannose phosphate receptors, it has been suggested that the GGAs cooperate with AP-1, to which they convey cargo proteins after the initial recognition event (Doray et al., 2002). Two GGA proteins, Gga1p and Gga2p, exist in Saccharomyces cerevisiae and their functions are redundant. They are required for transport of proteins such as carboxypeptidase Y (CPY) and the syntaxin Pep12p to the late endosome, and for ubiquitin-dependent sorting of the amino acid permease Gap1p (Black and Pelham, 2000; Hirst et al., 2000; Costaguta et al., 2001; Hirst et al., 2001; Scott et al., 2004). However, only ubiquitin has been shown to be a direct binding partner of a yeast GGA (via the GAT domain; Scott et al., 2004). Furthermore, structural analyses suggest that the yeast GGA VHS domain may not be able to accommodate the DXXLL signal peptide (Misra et al., 2002).

Two novel phosphoinositide-binding proteins, Ent3p and Ent5p, were identified in a two-hybrid screen for accessory proteins that interact with AP-1 and the GGAs in yeast (Duncan et al., 2003). A similar AP-1/GGA-binding protein, epsinR, exists in mammalian cells (Kalthoff et al., 2002; Wasiak et al., 2002; Hirst et al., 2003; Mills et al., 2003). Ent3p and epsinR bind to lipids via their epsin amino-terminal homology domain, whereas Ent5p contains a similar AP180 N-terminal homology domain. All three proteins also contain clathrin binding motifs in their primary sequence (Legendre-Guillemin et al., 2004), and Ent5p and epsinR bind to clathrin in vitro, suggesting that they function as clathrin adaptors (Duncan et al., 2003; Mills et al., 2003). Furthermore, Ent3p binds to a potential cargo protein, the vacuolar soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) Vti1p, and epsinR binds to the mammalian orthologue Vti1b (Chidambaram et al., 2004; Hirst et al., 2004). Cells lacking both Ent3p and Ent5p display defects in trafficking of proteins between the TGN and the endosomes and the sorting of proteins into multivesicular bodies is also affected (Duncan et al., 2003; Eugster et al., 2004). In mammalian cells, epsinR has been implicated in retrograde sorting from the early endosome (Saint-Pol et al., 2004).

To gain a better understanding of the intracellular trafficking steps mediated by clathrin, we have examined the transport of a cargo protein, the chitin synthase Chs3p, in S. cerevisiae. Chs3p is stored in an intracellular reservoir, from which it is transported to the cell surface in a cell cycle–dependent manner with the help of a cytosolic complex that contains Chs6p (Chuang and Schekman, 1996; Ziman et al., 1998; Sanchatjate and Schekman, 2006). Chs3p proved particularly useful in discerning the role of the yeast AP-1, because an inefficient intracellular retention of Chs3p was the first easily observable phenotype associated with mutations in AP-1 subunits. Mutations in the VPS genes, which are involved in trafficking to the late endosome and the vacuole (Bowers and Stevens, 2005), did not affect Chs3p localization. It was therefore proposed that AP-1 mediates continuous recycling of Chs3p between the TGN and the early endosome (Valdivia et al., 2002). Other proteins, such as the protease Kex2p, also follow this recycling route (Ha et al., 2003; Foote and Nothwehr, 2006).

In this study, we address the role of Ent3p and Ent5p in the trafficking of Chs3p, and we compare their function to the function of the other intracellular clathrin adaptors, AP-1 and the GGAs. We show that like AP-1, but independently of it, Ent5p directly binds to Chs3p. Furthermore, we find that Ent3p is important for the function of the GGAs and may mediate the binding of Gga2p to cargo proteins.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Strain and Plasmid Construction
Strains (Table 1) were constructed either by tetrad dissection of sporulated diploid strains, by integration of disruption cassettes replacing the whole coding region, or by integrations of insertions at the COOH-terminal codon (Longtine et al., 1998). All allelic replacements or insertions were confirmed by polymerase chain reaction (PCR). chs6::HIS3, gga1::TRP1, and gga2::LEU2 disruption cassettes were generated by PCR amplification from total DNA isolated from JCY479, RSY2364, and RSY1715, respectively. The plasmid pFA6::NATMX was used to generate disruption cassettes containing the NATMX marker (Tong et al., 2004). Resistance to calcofluor (CF) was assessed by growth on YPD agar plates supplemented with 100 µg/ml Fluorescent Brightener 28 (Sigma-Aldrich, St. Louis, MO). The pCHS3-GFP expression plasmid was generated by subcloning CHS3-GFP from a plasmid generated by Valdivia et al. (2002) into pRS315. The plasmid pFA6a-mRFP was used to generate the COOH-terminal integration cassettes ENT3::mRFP and ENT5::mRFP.

Subcellular Fractionations
Total cell lysates were fractionated on 5-ml 20–60% (wt/wt) continuous sucrose gradients as described previously (Kaiser et al., 2002). Continuous gradients were generated by diffusion for 2 h at room temperature (Kolling and Hollenberg, 1994). Gradients were centrifuged for 17 h at 4°C at 100,000 x g. Fractions (16) were collected manually from the top of a gradient and analyzed by SDS-polyacrylamide gel electrophoresis (PAGE) and immunoblotting with antibodies against Chs3p, Sec61p, Vph1p, Pma1p (F. Portillo, Universidad Autonoma de Madrid, Spain), Anp1p (S. Munro, Medical Research Council, Cambridge, United Kingdom), Tlg1p (H. Pelham, Medical Research Council, Cambridge, United Kingdom), and Pep12p (Invitrogen, Carlsbad, CA). Protein bands were quantified as described below.

Tandem Affinity Purification (TAP) from Yeast
A tandem affinity purification tag (Rigaut et al., 1999) composed of an S tag (Kim and Raines, 1993), a TEV protease cleavage site, and a ZZ tag (minimal protein A binding domain) was amplified from pKW804 (a gift from K. Weis, University of California, Berkeley, Berkeley, CA). The tag was integrated at the COOH termini of APS1, ENT3, ENT5, or GGA2 in DDY1810 to generate TSY127, ACY134, ACY133, or ACY182, respectively. For TAP, cells were grown to OD₆₀₀ 0.6–0.8, washed with H₂O, and incubated 10 min in prespheroplasting buffer (100 mM Tris, pH 9.4, and 40 mM -mercaptoethanol), followed by 30-min incubation in spheroplasting buffer (20 mM HEPES, pH 7.4, 0.7 M sorbitol, 0.5X YPD, and lyticase at 30 U/10⁷ cells) at 30°C. Spheroplasts were washed once in cross-linking buffer (20 mM HEPES, pH 7.4, 0.7 M sorbitol, and 100 mM KOAc) and resuspended in 0.02 ml of cross-linking buffer/10⁷ cells. Cross-linking was conducted with 5 mM dithiobis[succinimidylpropionate] (DSP) (Pierce Chemical, Rockford, IL) for 30 min at room temperature, and the reaction was quenched for 15 min in 100 mM Tris, pH 7.4. Cells were then resuspended in the same volume of ice-cold lysis buffer A (50 mM HEPES, pH 7.4, 350 mM NaCl, and 1% Triton X-100) with 2 mM phenylmethylsulfonyl fluoride and 2X protease inhibitors, and they were lysed with 15 strokes in a Dounce homogenizer. All subsequent steps were conducted at 4°C. Lysates were incubated for 30 min to extract membrane proteins, and the extracts were centrifuged at 3000 rpm for 6 min and at 13.5 krpm in an SS-34 rotor for 20 min. The supernatant fraction was passed over a 2-ml Sepharose CL-6B column (Sigma-Aldrich), and the flow-through was incubated for 3 h with immunoglobulin G (IgG)-Sepharose (GE Healthcare, Little Chalfont, Buckinghamshire, United Kingdom). Beads were washed with lysis buffer A and incubated overnight in a small volume of buffer containing 80 U of AcTev protease (Invitrogen). The supernatant was then removed from the IgG-Sepharose and incubated with 60 µl of 50% S-protein agarose (Novagen, Madison, WI) for 3 h. S-protein agarose beads were washed with lysis buffer, and 2X SDS protein sample buffer containing 100 mM dithiothreitol was added to elute the bound proteins and cleave the cross-linker DSP. Samples were analyzed by SDS-PAGE, followed by Sypro red staining or immunoblotting with antibodies against Chs3p, Clc1p (affinity-purified using glutathione S-transferase-Clc1p as described previously; Harsay and Schekman, 2002), Chc1p (S. Lemmon, University of Miami), Pep12p (Invitrogen), Apl2p and Gga2p (G. Payne, UCLA, Los Angeles, CA), Vps27p (T. Stevens, University of Oregon, Eugene, OR), and dsRed (Clontech, Mountain View, CA). S-protein-horseradish peroxidase (HRP) (Novagen) was used for detection of purified TAP-tagged proteins. Sypro-stained gels and nitrocellulose membranes were scanned on a Typhoon 9400 Imager (GE Healthcare). Protein bands were quantified using ImageQuant software (GE Healthcare).

For purification of Ent3-TAP and Ent5-TAP under more native conditions, the cross-linking step was omitted, and lysis buffer B (50 mM HEPES, pH 6.8, 350 mM K-acetate, 5 mM MgCl₂, 1 mM EDTA, 1 mM -mercaptoethanol, and 0.5% octyl glucoside) was used for lysis and subsequent purification steps.

For mass spectrometry analysis, SDS-PAGE gels were stained with colloidal blue Coomassie stain (Invitrogen), and protein bands were excised from the gel. Gel bands were digested with Trypsin Gold (Promega; Jimenez et al., 1998), and proteins were identified at the Howard Hughes Medical Institute mass spectrometry laboratory at the University of California, Berkeley, by peptide mass fingerprinting (using MSFit at http://prospector.ucsf.edu).

Metabolic Labeling and Immunoprecipitation
To analyze the processing of A-alkaline phosphatase (ALP) mutants, we transformed strains in which PHO8 (encoding wild-type ALP) was deleted with pSN100 (Nothwehr et al., 1993) or pSH46 (Foote and Nothwehr, 2006) encoding A(FA)-ALP or A(2–11, FA)-ALP, respectively. Cells were grown in SC-Ura medium to mid-logarithmic phase, harvested, labeled with [³⁵S]methionine/cysteine, and chased for different amounts of time as described previously (Seeger and Payne, 1992). Approximately 5 x 10⁶ cells were used for each time point. All experiments were performed at 30°C. A-ALP mutants were immunoprecipitated from cell lysates by using rabbit polyclonal antibodies against ALP (G. Payne). Radioactively labeled proteins were quantified from SDS-PAGE gels by using Typhoon 9400 Imager (GE Healthcare) and ImageQuant software (GE Healthcare). For calculation of the half-time of processing of A-ALP mutants, the log of the ratio of the precursor versus the mature form at each time point was plotted as a function of time, and the plots were analyzed by linear regression analysis. The precursor and mature forms of A(FA)-ALP and A(2–11, FA)-ALP migrated on SDS-PAGE in a manner consistent with their predicted sizes.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Ent3p and Ent5p Are Involved in Intracellular Trafficking of Chs3p
We sought a test of the role of proteins that interact with AP-1 in the intracellular retention of the chitin synthase Chs3p. Ent5p was shown to interact with the subunit of AP-1 and to colocalize with internal clathrin puncta in yeast (Duncan et al., 2003). Mutations in AP-1 subunits restore the transport of Chs3p to the cell surface in a chs6 background, in which the normal export route for Chs3p is blocked (Valdivia et al., 2002). We asked whether ent5 also suppressed the chs6 phenotype by constructing an ent5 chs6 strain and testing its sensitivity to the chitin-binding drug calcofluor (Figure 1A). The amount of chitin in the yeast cell wall depends on the activity and localization of Chs3p. We found that ent5 chs6 cells were as resistant to calcofluor as chs6 cells, indicating that the transport of Chs3p to the cell surface is blocked and that unlike the AP-1 subunits, Ent5p is not required for the intracellular retention of Chs3p. Ent3p is related to Ent5p, and it also colocalizes with intracellular clathrin, but it does not interact with AP-1 in vivo (Duncan et al., 2003). As expected, ent3 did not suppress the calcofluor resistance of chs6 cells. However, when mutations in ENT3 and ENT5 were combined with chs6, the calcofluor sensitivity was restored to wild-type levels. Abundant chitin rings could be observed when ent3 ent5 chs6 cells were stained with calcofluor; these cells could not be distinguished from wild-type cells (Figure 1B). The results presented in Figure 1 were obtained with strains constructed in the Euroscarf background (Brachmann et al., 1998), but they were reproduced with strains in the YPH499 background (Sikorski and Hieter, 1989; data not shown). These results suggest that transport of Chs3p to the cell surface is restored in ent3 ent5 chs6 cells and that both Ent3p and Ent5p may act in the AP-1 pathway but their functions may be redundant.

View larger version (83K): [in this window] [in a new window]   Figure 1. Simultaneous disruption of ENT3 and ENT5 suppresses the chs6 phenotype. (A) CF sensitivity of chs6 is restored to wild-type levels when ENT3 and ENT5 are mutated but not when either gene alone is mutated. (B) Staining of chitin with calcofluor.

View larger version (32K): [in this window] [in a new window]   Figure 2. Intracellular distribution of Chs3p in ent3 ent5 cells. (A) Wild-type cells were fractionated on continuous sucrose gradients as described in Materials and Methods. The distribution of Chs3p and of the marker proteins Pma1p (plasma membrane), Vph1p (vacuole), Anp1p (early Golgi), Tlg1p (TGN and endosomes), and Sec61p (ER) was assessed by immunoblot analysis. (B) The distribution of Chs3p on sucrose gradients from wild type, chs6, and chs6 apl2. Apl2p is the -subunit of AP-1 in yeast. Insets show the distribution of Tlg1p and Pma1p in these strains. (C) The distribution profile of Chs3p in ent3 ent5 and ent3 ent5 chs6 cells is different from in wild-type cells, whereas its distribution profile in ent5 chs6 cells is the same as in chs6 cells. Insets show the distribution of Tlg1p and Pma1p in these strains.

View larger version (107K): [in this window] [in a new window]   Figure 3. Chs3-GFP localization in ent3 ent5 cells. Chs3-GFP can be visualized at the bud-neck (asterisk) in wild-type, ent3 ent5, and ent3 ent5 chs6 cells but not in ap-1 chs6 cells.

View larger version (66K): [in this window] [in a new window]   Figure 4. Mutations in ENT3 and ENT5 do not affect the binding of AP-1 to Chs3p, Gga2p or clathrin. The gene encoding the Aps1p subunit of AP-1 was tagged chromosomally at the C terminus in wild-type or ent3 ent5 cells. Cells were converted to spheroplasts and incubated with 5 mM cross-linker DSP before lysis. Aps1-TAP was purified by TAP on an IgG-Sepharose column followed by an S-protein agarose column. As a negative control, Aps1-TAP was purified from apl4 cells where the subunit is missing and the AP-1 complex cannot assemble. The same amount of cells (1500 OD600 units) was used in each purification. DSP was then cleaved with a reducing agent, and 1/80,000 of each lysate (IgG-load) and one half of each final purified sample (S-elution) were loaded on an SDS-polyacrylamide gel. After electrophoresis, the gels were stained for total protein with Sypro red (A). The subunits of AP-1 are marked, and an asterisk marks the position of the band that corresponds to Chc1p. (B) The same samples were blotted with antibodies against Chs3p, Gga2p, and clathrin subunits Chc1p and Clc1p. An asterisk denotes residual Aps1-TAP that cross-reacts with the anti-Clc1p antibody. (C) Quantification of the protein bands in S-elution fractions. Aps1p and Apm1p were quantified from Sypro-stained gels, and the other proteins were quantified from immunoblots. The amount of a protein in each strain is expressed as a percentage of the signal in the wild-type strain that was analyzed at the same time. The values presented are the average values from three independent experiments.

Ent3p and Ent5p Both Bind to Clathrin and Cargo Proteins In Vivo
Both Ent3p and Ent5p bind to membranes in a phosphoinositide-specific manner, and Ent5p also binds to clathrin in vivo (Duncan et al., 2003; Eugster et al., 2004). In addition, Ent3p has been shown to interact with a vacuolar SNARE, Vti1p (Chidambaram et al., 2004). These results suggest that Ent3p and Ent5p may act as bona fide clathrin cargo adaptors. To examine whether either protein might directly interact with Chs3p, we used TAP-tagged forms of Ent3p or Ent5p. Both tagged proteins are functional, as confirmed by calcofluor sensitivity assays and their correct localization in the cell (data not shown). Tagged Ent3p and Ent5p could be efficiently purified from yeast, but the amount of purified Ent3p was reduced compared with Ent5p, probably due to a four- to fivefold lower expression level of Ent3p (Figure 5A; data not shown). To identify all stable binding partners of Ent3p and Ent5p, we stained the final purified samples on an SDS gel with Coomassie blue, excised the protein bands, and analyzed them by mass spectrometry. Interestingly, clathrin heavy chain (Chc1p) was identified in both purified samples. In addition, the heat-shock proteins Ssa1p and Ssa2p, common contaminants in protein purifications, were identified in our samples. Other Coomassie-stained bands were Ent3p in Ent3p purification, Ent5p in Ent5p purification, their degradation products, and IgG. The unidentified species of 95 kDa (marked with asterisks) was not detected in subsequent purifications.

View larger version (53K): [in this window] [in a new window]   Figure 5. Purification of Ent3p and Ent5p from wild-type and mutant cells. ENT3 and ENT5 were chromosomally tagged at their COOH termini with the S-Tev-ZZ tag. TAPs were performed as described in Materials and Methods. (A) Ent3p and Ent5p were purified from ACY134 or ACY133 cells (1200 OD600 units), respectively. Cells were converted to spheroplasts and split into two aliquots for purifications with or without DSP. The final purified samples were incubated with reducing agent, and one third was loaded on an SDS-polyacrylamide gel. The gel was stained with colloidal Coomassie blue; protein bands were excised from the gel and analyzed by mass spectrometry. The proteins that were identified are noted in the image. (B) Ent5p was purified from spheroplasts prepared from wild-type cells (ACY133), which were incubated with 5 mM DSP or with dimethyl sulfoxide alone. An aliquot (1/60,000) of each lysate (IgG-Load) and one third of each final purified sample (S-elution) were analyzed by immunoblotting for the presence of the indicated proteins. (C) Ent5p was purified from wild-type, apl4, ent3, and gga1 gga2 cells (2000 OD600 units) treated with 5 mM DSP. Samples were analyzed by immunoblotting. S-protein-HRP conjugate was used for the detection of purified Ent5p in 1/50 of each S-elution sample. (D) Ent3p was purified from wild-type, ent5, and gga1 gga2 cells (3000 OD600 units) treated with 5 mM DSP. The final purified samples (one half of each sample; S-elution) as well as 1/90,000 of each lysate (IgG-Load) were analyzed by immunoblotting with antibodies against Chc1p, Gga2p, Pep12p, and Clc1p. Ent3p in the "S-elution" samples was detected by Sypro stain.

In contrast to the results obtained with Ent5p, no Chs3p was detected in the sample of purified Ent3-TAP by immunoblot. This was not due to less efficient cross-linking of Ent3p because clathrin and Gga2p copurified with Ent3p just as efficiently as with Ent5p (Figure 5D). Additionally, the late-endosomal t-SNARE Pep12p was very abundant in the purified Ent3p sample, indicating an efficient association of Ent3p with Pep12p in the membrane. We conclude that both Ent3p and Ent5p bind to proteins in the membrane, but their binding specificities differ, and only Ent5p may associate with Chs3p.

Ent3p does not interact with clathrin in vitro (Duncan et al., 2003), although it associates with clathrin in vivo (Friant et al., 2003). We therefore tested whether a third protein, Ent5p or Gga1p/Gga2p, might facilitate the cross-linking between Ent3-TAP and clathrin. We purified Ent3p from ent5 or gga1 gga2 cells (Figure 5D). Under both conditions, Ent3p was cross-linked to clathrin as efficiently as in wild-type cells, suggesting that like Ent5p, Ent3p may directly bind to clathrin in vivo. However, Ent3p binds less efficiently to clathrin than Ent5p. Clathrin copurified with Ent5-TAP, but not with Ent3-TAP, even in the absence of cross-linker when a milder lysis buffer was used during the purification (data not shown). In contrast to clathrin, the amount of Pep12p that copurified with Ent3p from gga1 gga2 cells may have decreased (Figure 5D).

It was suggested that Ent3p and Ent5p also interact with Vps27p and are involved in the formation of multivesicular bodies (Eugster et al., 2004); however, by immunoblotting, we did not observe a significant amount of Vps27p copurifying with Ent3-TAP or Ent5-TAP. We also did not observe Ent3p and Ent5p associating with each other when we purified Ent3-TAP from cells expressing Ent5-monomeric red fluorescent protein (mRFP) and Ent5-TAP from cells expressing Ent3-mRFP (data not shown).

Our finding that Ent5p but not Ent3p binds to Chs3p presents an obvious question: if the functions of Ent3p and Ent5p in Chs3p trafficking are not redundant, why does ent5 by itself not suppress the chs6 trafficking defect? To address this discrepancy, we examined other possible differences in the functions of Ent3p and Ent5p.

Ent3p, but Not Ent5p, Facilitates the Function of Gga2p
Because Ent3p and Ent5p interact with Gga2p (Duncan et al., 2003), we examined the role of these interactions in the context of Chs3p traffic. We crossed ent3 chs6 or ent5 chs6 cells with a gga2 strain and tested haploid mutants from these crosses for calcofluor sensitivity. We observed a genetic interaction between ENT3 and GGA2 but not between ENT5 and GGA2 (Figure 6A). The triple mutant ent3 gga2 chs6 displayed an increased sensitivity to calcofluor compared with the chs6 cells, suggesting that some Chs3p trafficked to the cell surface in this mutant. However, this was not the case in ent5 gga2 chs6 or in gga2 chs6 cells. A similar but less pronounced difference in calcofluor sensitivity was observed between ent3 gga1 chs6 and ent5 gga1 chs6 cells, whereas both ent3 and ent5 displayed some synthetic growth phenotype when combined with gga1 gga2, and thus the calcofluor sensitivity of such mutants could not be determined (data not shown). In contrast to our previous findings (Valdivia et al., 2002), gga1 gga2 chs6 triple mutants were also calcofluor sensitive (Figure 6B). The genetic interaction between ENT3 and GGA2 points to a role of Ent3p in Gga-mediated trafficking, and we suggest that the Gga proteins also contribute to the intracellular retention of Chs3p.

View larger version (82K): [in this window] [in a new window]   Figure 6. ENT3 displays synthetic interaction with GGA2. (A) The CF resistance of chs6 cells is partially suppressed by the combined ent3 gga2 mutations, whereas ent5 gga2 has no effect. (B) Combined gga1 gga2 mutations also suppress chs6.

View larger version (55K): [in this window] [in a new window]   Figure 7. The function of Gga2p depends on the presence of Ent3p. GGA2 was tagged chromosomally at the COOH terminus with the S-Tev-ZZ tag in wild-type, ent3, ent5, ent3 ent5, apl4, and ent3 apl4 cells. Gga2p was purified from each strain (1600 OD600 units) by TAPs as described in Materials and Methods. The cross-linker DSP at 5 mM was used in all purifications. (A) Representative purified sample (S-eluate) from indicated strain backgrounds was analyzed by SDS-PAGE and stained for total protein with Sypro stain. In addition to Gga2p, clathrin heavy chain (Chc1p) could be detected in the samples. (B) The same samples as in A as well as the starting lysates (IgG-Load) were analyzed by immunoblotting with antibodies against clathrin light chain (Clc1p) and Pep12p. (C) Quantification of the amount of Pep12p and Clc1p that copurify with Gga2-TAP expressed in different strains. Protein bands from immunoblots were quantified and are expressed as a percentage of the amount purified from wild-type cells for each protein. IgG-Load represents the starting lysates, and S-elution is the final purified sample. The n value below each strain is the number of independent experiments that were quantified. Wild-type cells were analyzed in each experiment. (D) Distribution profile of Pep12p in WT, ent3, ent5, and ent3 ent5 cells. Total cell membranes were fractionated on continuous sucrose gradients as described in Figure 2. Fractions were probed for Pep12p by immunoblotting.

Interplay between the AP-1, Ent3p/Ent5p, and Gga Pathways
We have shown that Ent3p, and to a lesser extent Ent5p, function in the Gga transport pathway, but they also direct the trafficking of Chs3p in a similar manner as AP-1. To better understand the relationship between these adaptors, we analyzed how mutations in them affect the processing of a model TGN protein, A(FA)-ALP. A(FA)-ALP consists of the transmembrane and the luminal domains of the vacuolar ALP, and the cytosolic domain of the TGN protein Ste13p containing a mutation in the aromatic amino acid motif FXFXD (Nothwehr et al., 1993; Bryant and Stevens, 1997). Like Chs3p, A(FA)-ALP requires AP-1 for its efficient retention at the TGN (Ha et al., 2003; Foote and Nothwehr, 2006). In wild-type cells, A(FA)-ALP is delivered to the vacuole, where it is proteolytically cleaved with a half-life of 60 min (Nothwehr et al., 1993; Foote and Nothwehr, 2006; Figure 8, A and B). The processing time is reduced to 45 min in ap-1 mutant cells (Ha et al., 2003). In ent3 ent5 cells, the half-life of A(FA)-ALP was even more significantly reduced, to 27 min. Deletion of ENT3 alone caused a small but reproducible reduction in the processing time, whereas ent5 had no statistically significant effect (Figure 8, A and B).

View larger version (40K): [in this window] [in a new window]   Figure 8. Mutations in ENT3 and ENT5 stimulate the transport of the TGN protein A(FA)-ALP to the vacuole. (A and B) Strain RSY3341 (wild type), RSY3342 (ent3), RSY3343 (ent5), and RSY3344 (ent3 ent5) were transformed with plasmids expressing A(FA)-ALP (pSN100) or A(2-11, FA)-ALP (pSH46). Cells were metabolically labeled for 10 min and subjected to chase for the indicated time (in minutes) at 30°C. A-ALP mutants were immunoprecipitated from cell lysates and analyzed by SDS-PAGE. Processing of A(FA)-ALP is shown in A, and precursor (p) and mature (m) form of A(FA)-ALP are indicated. Half-times (t1/2) of processing were indicated for each strain from the four time points and are plotted in B for A(FA)-ALP (), and for A(2-11, FA)-ALP (). A(2-11, FA)-ALP does not interact with AP-1; therefore, its processing is equivalent to the processing of A(FA)-ALP in ap-1 cells. The n value under each bar indicates the number of times that each experiment was repeated. (C) Processing of A(FA)-ALP (), and of A(2-11, FA)-ALP () in strains ACY193 (wild type), ACY194 (ent3), ACY195 (gga1 gga2), and ACY196 (ent3 gga1 gga2), was analyzed as shown in A, and the t1/2 values of processing were calculated. Each experiment was repeated two times.

Mutations in Gga1p and Gga2p also slightly increase the rate of processing of A(FA)-ALP (Foote and Nothwehr, 2006), although this difference is less consistent (Ha et al., 2003). We have shown that Ent3p has a role in Gga-mediated protein transport. We therefore asked whether the effect of ent3 on the processing of A(FA)-ALP is due to a block in the Gga pathway by comparing the processing of A(FA)-ALP in ent3, gga1 gga2, and ent3 gga1 gga2 cells (Figure 8C; note that the strains used in Figure 8C are different from the strains used in Figure 8, A and B, but the results are similar). Mutations in GGA1 and GGA2 increased the rate of A(FA)-ALP processing at a comparable or slightly higher level than ent3, and there was no added effect of ent3 in the gga1 gga2 background, suggesting that the effect of ent3 on the processing of A(FA)-ALP may indeed be due to a role of Ent3p in the Gga pathway, independent of its role along with Ent5p in the AP-1-parallel pathway. In contrast to ent3 and ent5, gga1 gga2 do not affect the processing of A(2-11, FA)-ALP (Figure 8C), and in ent3 gga1 gga2 cells A(2-11, FA)-ALP was processed at the same rate as in ent3 cells (data not shown). This result suggests that the Ggas function downstream of AP-1 in the case of this cargo protein (Figure 8C and Figure 9).

View larger version (16K): [in this window] [in a new window]   Figure 9. Model of how clathrin-dependent pathways could contribute to the intracellular retention of Chs3p and A(FA)-ALP. AP-1, Ent5p, and Ent3p function in parallel to retain Chs3p and A(FA)-ALP at a compartment that may represent the TGN or the early endosome. Separate arrows indicated independent functions of these proteins and are not meant to represent different transport routes. Ent3p and Ent5p can partially substitute for each other, which is indicated by dashed arrows. In addition, Ent3p functions in the Gga pathway, which in A(FA)-ALP operates downstream of the AP-1 pathway. This may still be the same compartment, which has been reorganized by AP-1, Ent3p, and Ent5p.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Loss of Ent3p and Ent5p resulted in a pronounced accumulation of Chs3p at the cell surface (Figures 2 and 3). In a population of asynchronous ent3 ent5 cells, we found both by cell fractionation experiments and by microscopy that a larger fraction of total Chs3p may have been localized to the cell surface, compared with wild-type cells or ap-1 cells. Furthermore, the internal compartments populated by Chs3p in ent3 ent5 cells seemed different: smaller and more dispersed by microscopy and of a different buoyant density on sucrose gradients. Unlike ap-1 mutant cells (Yeung and Payne, 2001), ent3 ent5 cells display defects in trafficking of proteins to the late endosome and the vacuole. However, endocytosis is not affected in ent3 ent5 mutants (Duncan et al., 2003). The apparent retention of Chs3p at the cell surface may be due to its increased recycling from the early endosome or the TGN back to the cell surface.

Ent3p and Ent5p Have the Properties of Clathrin Adaptors
Clathrin adaptors mediate the interaction among clathrin, cargo proteins, and the lipid membrane. Ent3p and Ent5p contain phosophoinositide binding domains that can associate with the lipid membrane. Ent5p, but not Ent3p, binds to clathrin in vitro, probably via a type I clathrin binding box (Duncan et al., 2003). In vivo, however, Ent3p can also associate with clathrin (Friant et al., 2003; this study). This interaction is probably direct, because it is not affected by deletion of the other binding partners of Ent3p (Figure 5), but it may be weaker than the interaction between Ent5p and clathrin. A DLL clathrin binding motif was identified in the primary sequence of Ent3p (Legendre-Guillemin et al., 2004).

Ent3p but not Ent5p interacts with a vacuolar SNARE Vti1p (Chidambaram et al., 2004). We find that Ent5p may also bind to cargo proteins, because it associates with Chs3p and this interaction is independent of other known interacting partners of Ent5p. In addition, the endosomal t-SNARE Pep12p copurified with Ent3p and to a lesser extent with Ent5p (Figure 5). Thus, both Ent3p and Ent5p possess all the characteristics of clathrin adaptors. However, Ent3p and Ent5p also associate with other clathrin adaptors—Gga2p and AP-1 (Duncan et al., 2003; Figure 5). In addition, AP-1 and Gga2p also interact (Costaguta et al., 2001; Figure 4). We therefore asked whether these adaptors function independently or in cooperation with each other during clathrin-mediated protein sorting.

Ent3p Has an Important Role in the Gga Transport Pathway
In addition to the cargo binding differences noted above, ENT3 displayed a genetic interaction with GGA2, whereas ENT5 did not (Figure 6), and the loss of Ent3p but not of Ent5p affected the Gga2p cross-linking pattern. Most importantly, the interaction between Gga2p and Pep12p, an endosomal t-SNARE that must be transported from the TGN to function at the late endosome (Becherer et al., 1996), was lost in an ent3 background (Figure 7). Pep12p is rerouted to the cell surface when its transport from the TGN to the late endosome is blocked by mutations in GGA1 and GGA2 (Black and Pelham, 2000). We also observed a change in the intracellular localization of Pep12p in ent3 but not in ent5 cells (Figure 7D), and mutations in GGA1 and GGA2 had an effect similar to ent3 (data not shown). The mislocalization of Pep12p was even more pronounced in ent3 ent5 cells, but this may be due to defects in multiple pathways (Figure 7D). These data implicate Ent3p in Gga1p/Gga2p-mediated cargo sorting. Ent3p and Gga2p colocalize in the cells to a larger extent than do Ent5p and Gga2p (Costaguta et al., 2006). Conversely, the interaction between Ent3p and Pep12p may also be reduced in a gga1 gga2 background; thus, we suggest that Ent3p cooperates with Gga1p and Gga2p in sorting of at least a subset of cargo in the TGN-to-late endosome trafficking pathway.

In mammalian cells, two domains of the Gga proteins, the GAT domain and the VHS domain, have been shown to bind directly to signals in the cargo proteins, monoubiquitin and DXXLL, respectively; however, only the former interaction has been demonstrated in yeast (Bonifacino, 2004; Scott et al., 2004). The primary sequence of the yeast VHS domain differs somewhat from the mammalian sequence and may not be able to accommodate the VHS consensus signal peptide (Misra et al., 2002). An unrelated signal, FSDSPEF, in the cytosolic tail of Pep12p is important for the sorting of Pep12p to the late endosome, but a direct interaction between this signal and the Ggas has not been demonstrated (Black and Pelham, 2000). Perhaps this signal binds to Ent3p instead.

Interestingly, we also found that Ent3p, and to a lesser extent Ent5p, facilitated the binding of Gga2p to clathrin. A lower level of clathrin copurified with Gga2p from ent3 cells (Figure 7). Even less clathrin associated with Gga2p purified from ent3 ent5 cells, whereas the ent5 mutation by itself had no effect. However, we showed that the binding of AP-1 to clathrin was not affected by ent3 ent5 (Figure 4); thus, the decreased interaction between Gga2p and clathrin in these cells is probably not due to a general mislocalization of clathrin. We propose that an additional role of Ent3p in the Gga trafficking pathway is to mediate the interaction between the Gga proteins and clathrin but that Ent5p can partially substitute for this function of Ent3p. The Ggas possess two clathrin-binding motifs that when mutated decrease the amount of clathrin binding in vitro. A residual interaction with clathrin depends on the GAE domain of Gga2p or Gga1p (Mullins and Bonifacino, 2001), which corresponds to the site bound by Ent3p and Ent5p. Nonetheless, the clathrin-binding function of Ent3p and Ent5p is probably not essential for the function of yeast Ggas, because processing of CPY and -factor are minimally affected by deletion of the GAE domain within Gga2p (Mullins and Bonifacino, 2001).

The Relationship among AP-1, Ent3p, and Ent5p Function
In contrast to Gga2p, the absence of Ent3p and Ent5p had no effect on the interaction of AP-1 with its cargo or clathrin (Figure 4), suggesting that they may not be required for the function of AP-1. We studied the effect of ent3 ent5 mutations on the trafficking of another model TGN protein, A(FA)-ALP (Nothwehr et al., 1993). A(FA)-ALP is slowly transported from the TGN to the late endosome and the vacuole, probably via the early endosome (Foote and Nothwehr, 2006). Mutations in AP-1 increase the rate of transport of A(FA)-ALP to the vacuole (Ha et al., 2003; Foote and Nothwehr, 2006), and we now show that ent3 ent5 has a similar but more pronounced effect on trafficking of this chimeric protein (Figure 8). A similar acceleration in A(FA)-ALP processing was observed when a lipid-remodeling enzyme, the polyinositide phosphatase Inp53p, was mutated (Ha et al., 2001). Interestingly, the effects of mutations in the AP-1–interacting motif of A(FA)-ALP and in ENT3 and ENT5 are additive, suggesting that these proteins function in parallel at the same intracellular compartment (Figure 9). Independent parallel functions of these proteins are also suggested by the fact that apl2 ent3 ent5 cells have a severe growth phenotype (Costaguta et al., 2006; our unpublished data), whereas ent3 ent5 cells do not.

Ent5p and AP-1, but not Ent3p, interact with Chs3p to a similar extent. However, mutations in AP-1 suppress the chs6-dependent block in Chs3p export, whereas ent5 does not. It is possible that Ent3p can to some extent substitute for the loss of Ent5p. Likewise, the effect of ent5, and to a lesser extent ent3, on the processing of A(FA)-ALP is only revealed when both mutations are present or when the interaction with AP-1 is also abrogated (Figure 8B), suggesting that there is some cooperativity among AP-1, Ent3p, and Ent5p. Similarly, ent3 or ent5 affect the processing of -factor only in an ap-1 background (Costaguta et al., 2006).

We attribute the small increase in the rate of processing of A(FA)-ALP in ent3 cells to the role of Ent3p in the Gga pathway, because a similar increase was observed in gga1 gga2 (or in gga1 gga2 ent3) cells (Foote and Nothwehr, 2006; Figure 8C). However, the gga mutations do not affect the processing of A(2-11, FA)-ALP, a mutant protein that cannot bind to AP-1, suggesting that the Gga proteins function downstream of AP-1 in the processing of A(FA)-ALP (Figure 9). In contrast, it was shown that in mammalian cells AP-1 acts downstream of the Ggas in the transport of mannose phosphate receptor to the lysosome (Doray et al., 2002).

It is possible that this discrepancy reflects differences between yeast and mammalian cells; the localization of AP-1 is also affected by the loss of Gga function in mammalian but not in yeast cells (Ghosh et al., 2003; Costaguta et al., 2006). Alternatively, it may reflect differences in how the clathrin adaptors sort the different cargo proteins (see below).

In contrast to the TGN resident protein A(FA)-ALP, whose vacuolar processing is accelerated in ent3 ent5 or in gga1 gga2 mutants, delivery of the vacuolar protein Cps1p from the TGN to the vacuole is delayed in these same mutants (Costaguta et al., 2001; Duncan et al., 2003). Cps1p trafficking, however, is not affected by mutations in AP-1 alone. Pep12p is another example of an AP-1–independent Gga cargo protein. Thus, the role of the Gga proteins, along with Ent3p, may be cargo-specific, and they may function in more than one pathway in yeast. In mammalian cells, the GGAs distribute more broadly than the TGN markers; some GGA protein is also detected at the early endosome (Puertollano et al., 2003). Furthermore, Ggas localize to an exaggerated endosomal compartment that is present in some (class E) vps mutants in yeast (Hirst et al., 2001; Boman et al., 2002), suggesting that their function is not limited to the TGN.

Our analysis of the in vivo physical interactions between the different internal clathrin adaptor proteins in yeast points to a more complex network of trafficking pathways than was previously envisioned, and it is in good agreement with a recent report by Costaguta et al. (2006). In mammalian cells, GGA and AP-1 colocalized on coated buds and vesicles in the area of the TGN, and extended tubules containing GGA1, AP-1, and epsinR have also been observed (Doray et al., 2002; Hirst et al., 2003; Puertollano et al., 2003). Likewise, there is a considerable colocalization of AP-1, Gga2p, Ent3p, and Ent5p in yeast (Costaguta et al., 2006). These ad