The light chain subunit is required for clathrin function in Saccharomyces cerevisiae.

Clathrin, a multimeric protein involved in intracellular protein trafficking, is composed of three heavy chains (Chc) and three light chains (Clc). Upon disruption (clc1Δ) of the single Clc-encoding gene (CLC1) in yeast, the steady state protein levels of Chc decreased 5-10-fold compared with wild type cells; consequently, phenotypes exhibited by clc1Δ cells may result indirectly from the loss of Chc as opposed to the absence of Clc. As an approach to directly examine Clc function, clc1Δ strains were generated that carry a multicopy plasmid containing the clathrin heavy chain gene (CHC1), resulting in levels of Chc 5-10-fold elevated over wild-type levels. As with deletion of CHC1, deletion of CLC1 results in defects in growth, receptor-mediated endocytosis, and maturation of the mating pheromone α-factor. However, elevated Chc expression in clc1Δ cells partially suppresses the growth and α-factor maturation defects displayed by clc1Δ cells alone. Biochemical analyses indicate that trimerization and assembly of Chc are perturbed in the absence of Clc, resulting in vesiculation defects. Our results demonstrate that the light chain subunit of clathrin is required for efficient Chc trimerization, proper formation of clathrin coats, and the generation of clathrin-coated vesicles.

Distinct compartments in eukaryotic cells are maintained through the selective transport of proteins carried out by small vesicular carriers. Generation of these vesicles involves assembly of proteinaceous coats on the cytoplasmic surface of donor compartment membranes, which leads to the budding of coated vesicles. Although a variety of proteins are transported through vesicular movement, a subset of specific trafficking events are mediated by clathrin-coated vesicles. These include the retention of resident Golgi membrane proteins, receptormediated endocytosis, and the sorting of lysosomal/vacuolar proteins from the secretory pathway to the lysosome/vacuole (1,2).
The clathrin molecule, or triskelion, is a hexamer composed of three heavy chain subunits (Chc) and three light chain subunits (Clc) (1). In mammals, there are two forms of light chain, LC a and LC b , which share 60% amino acid identity and appear to be randomly distributed in clathrin trimers. In yeast, there is only one form of light chain encoded by the CLC1 gene (3). Formation of a clathrin-coated vesicle is initiated by binding of clathrin-associated protein complexes (APs) 1 to the donor membrane. Triskelions associate with the APs and polymerize into polygonal lattice structures. Such clathrin-coated membrane segments, known as coated pits, are thought to collect specific cargo proteins through interactions between the cargo protein cytoplasmic domains and the AP complexes. The clathrin-coated pit then invaginates and pinches off to form a clathrin-coated vesicle carrying the selected cargo proteins. Formation of the vesicle may involve rearrangement of subunits assembled on the coated pit, or it may be driven by the polymerization of new clathrin subunits into a polyhedral cage. Once the vesicle has formed, the clathrin lattice is depolymerized into triskelions to allow fusion of the vesicle with the target organelle membrane. The resulting triskelions are then available for another round of vesicle formation.
Conformational differences in clathrin triskelions at each stage of the assembly and disassembly cycle can allow for specific homotypic and heterotypic associations that contribute to regulation of the cycle. This regulation is likely to occur at a number of levels to ensure the appropriate temporal and spacial sequence of molecular interactions necessary for selective protein transport by clathrin-coated vesicles. Clc exhibits several properties that suggest it may act as a regulatory subunit (4). In vitro, Clc can bind calmodulin, Hsc70, and calcium (5)(6)(7). LC b can be phosphorylated in vitro by casein kinase II and is phosphorylated at the same sites in vivo (8). Calcium binding and Hsc70 recognition by Clc have been proposed to play a role in depolymerization of clathrin coats (6), although this view has recently been challenged (9). The roles of calmodulin binding and phosphorylation have not been determined. Differences between LC a and LC b suggest that each may act in separate regulatory functions; i.e. LC b contains a phosphorylation site, while LC a does not. Furthermore, there are neuronal isoforms of Clc, generated through differential mRNA splicing, that are presupposed to be involved in more specialized secretory and endocytic functions (10).
In order to evaluate the role of Clc in regulating clathrin function in vivo, we have initiated a genetic approach in the yeast Saccharomyces cerevisiae. Yeast Clc has only 18% amino acid identity with mammalian light chains, in contrast to the 50% sequence identity between clathrin heavy chains (3). Despite the sequence divergence, yeast Clc shares many biochemical properties with mammalian Clc including stoichiometric association with Chc in triskelions, size, overall acidic composition, heat and acid stability, and the ability to bind calmodulin and calcium (3,4). The evolutionary conservation of these properties argues that Clc in yeast and mammals carries out similar functions. Deletion of CLC1 in yeast results in a slow growth phenotype similar to that observed in strains lacking the Chc gene (CHC1), suggesting that Clc is critical to clathrin function (3). Here we present phenotypic characterization of protein trafficking pathways as well as clathrin structure and distribution in cells carrying the CLC1 disruption (clc1⌬) and in clc1⌬ cells expressing elevated levels of Chc1p. Our results suggest that Clc plays a role in stabilizing clathrin trimers and in formation of clathrin coats and clathrin-coated vesicles.

EXPERIMENTAL PROCEDURES
Plasmids-Plasmid constructions were carried out using standard molecular biology techniques (11). An AhaII to BamHI fragment containing CLC1 was inserted into pBM743 to generate pgalCLCURA3 (GAL-CLC1). pRSCLC14 was created by inserting a 1.5-kilobase pair BamHI fragment containing CLC1 into the BamHI site of pRS314. pRS424CHCo/e and pRS425CHCo/e were created by inserting a 6-kilobase pair BamHI fragment containing CHC1 into the BamHI site of pRS424 or pRS425.
Yeast Strains and Media-Yeast strains used in this study are listed. Yeast growth, mating, sporulation, and tetrad analyses were conducted as described by Sherman et al. (12). DNA transformations were performed by the lithium acetate procedure (13).
Immunoblotting-Cellular lysates were prepared by growing cells in SDCAA-trp to midlogarithmic phase. 1.2 ϫ 10 8 cells were collected, washed once in water, disrupted by agitation with glass beads in 0.05 ml of 2% SDS, and heated for 3 min at 100°C. 0.25 ml of Laemmli sample buffer (LSB) (14) was added to lysates and heated at 100°C again for 3 min. Insoluble material was removed by centrifugation at 16,000 ϫ g for 10 min.
Immunoblotting was carried out essentially according to Burnette (15) with secondary antibodies coupled to ALP (Bio-Rad). Antibodies were visualized using color development for ALP (Bio-Rad). Monoclonal antibodies to yeast Chc1p (16) were a gift from S. K. Lemmon (Case Western Reserve University), and antibodies to carboxypeptidase Y (CPY) and ␣-factor were a gift from S. D. Emr (University of California, San Diego). Immunoblot signals were quantitated using a Molecular Dynamics densitometer (Cupertino, CA).
Radiolabeling and Immunoprecipitations-For metabolic labeling of ␣-factor and CPY, cells were grown to midlogarithmic phase in SDYE at 30°C. Cells were labeled at 30°C. Labeling and immunoprecipitation for ␣-factor was performed as described by Seeger and Payne (17) except that the labeling period was 30 min instead of 10 min. Preparation of extracellular and intracellular fractions, labeling, and immunoprecipitation for CPY were conducted as described by Seeger and Payne (18). Quantitation of the various forms of ␣-factor and CPY was carried out using a Molecular Dynamics PhosphorImager.
Endocytosis Assay-Endocytosis assays were carried out as described by Dulic et al. (19) and Tan et al. (20). Cells were grown in SDCAA complete or SDCAAϪtrp or Ϫleu according to the plasmidborne selectable marker and then diluted and grown to midlogarithmic phase in YPDϩade at 30°C. 35 S-labeled ␣-factor (prepared as by Tan et al. (20)) was added at 1-2 ϫ 10 5 cpm/10 9 cells and allowed to bind to cells on ice for 60 min. Cells were prewarmed at 30°C for 2 min before the addition of 5% glucose to stimulate internalization, and the incubation continued at 30°C for 2, 5, or 20 min.
Fractionation Procedures-Clathrin-coated vesicles were isolated from GPY1407 or GPY1408 as described in Phan et al. (21) with the following changes. Cells were grown to 8 ϫ 10 7 cells/ml in 1.5L of YPD. Buffer A (100 mM MES-NaOH, pH 6.5, 0.5 mM MgCl 2 , 1 mM EGTA, 0.2 mM dithiothreitol, and 1 mM sodium azide) contained 1 mM phenylmethylsulfonyl fluoride. The lysate, obtained by glass bead disruption, was treated with 2 mg/ml RNase at 30°C for 30 min. Chromatography of the 100,000 ϫ g membrane pellet was performed using a 1.5 ϫ 125-cm Sephacryl S-1000 column (Sigma). Fractions (2.5 ml each) enriched for clathrin-coated vesicles were identified by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) of samples from each fraction followed by staining with Coomassie Brilliant Blue.
In experiments involving size exclusion chromatography and differential centrifugation, cells were grown to midlogarithmic phase in YPDϩade, washed once in water, and incubated in 100 mM Tris-SO 4 , pH 9.5, and 1 mM dithiothreitol for 10 min. Cells were collected and resuspended in YP with 1 M sorbitol, 10 mM Tris-HCl, pH 7.5, and 0.1% dextrose. Oxalyticase (50 units/A 500 unit; Enzymogenetics, Corvallis, OR) was added, and cells were incubated at 30°C for 30 min. Spheroplasts were sedimented and washed once in 1.2 M sorbitol.
For size exclusion chromatography, spheroplasts were resuspended in Buffer A and lysed by agitation with glass beads. The lysate was subjected to centrifugation at 100,000 ϫ g for 20 min in a Beckman TLA 100.2 rotor to yield a supernatant fraction (cytosol) and a pellet fraction. The cytosol was collected, 1 ⁄10 volume of 75% glycerol was added, and the solution was applied to a Sepharose CL-4B column (1 ϫ 28 cm) (Sigma) equilibrated with Buffer A. 1-ml fractions were collected, precipitated with 10% trichloroacetic acid, resuspended in LSB, and subjected to SDS-PAGE and immunoblotting. The pellet fraction was resuspended in 0.5 M Tris-HCl, pH 7.5, and homogenized with five strokes using pestle B in a 1-ml Kontes glass homogenizer, incubated on ice for 60 min, and then subjected to centrifugation at 100,000 ϫ g as before. This supernatant was collected and subjected to Sepharose CL-4B chromatography as described for the cytosol fraction.
For differential centrifugation analysis, cells were lysed in Buffer A using 15 strokes with pestle B in a 7-or 2-ml Kontes glass homogenizer. Centrifugation was carried out at 1,500 ϫ g for 4 min (low speed spin). The resulting supernatant was subjected to centrifugation at 26,000 ϫ g for 30 min in a Sorvall type SS-34 rotor (medium speed spin). The supernatant was then subjected to centrifugation at 100,000 ϫ g for 20 min in a Beckman type TLA 100.2 rotor (high speed spin). Pellets were Chemical Cross-linking-Yeast clathrin was purified from GPY1118. Cells were grown to 8 ϫ 10 7 cells/ml in YPD and subjected to glass bead lysis in 50 mM Tris-HCl, pH 7.5, 1 mM dithiothreitol, 1 mM EDTA, and 1 mM phenylmethylsulfonyl fluoride. The lysate was precipitated with 35% ammonium sulfate. The resulting pellet was resuspended in the same buffer and subjected to chromatography on a 2.5 ϫ 100-cm Sepharose CL-4B column. Fractions (8 ml each) enriched for clathrin were identified by SDS-PAGE of samples from each fraction followed by staining with Coomassie Brilliant Blue.
Cross-linking and composite agarose gel electrophoresis were performed as described by Kirchhausen and Harrison (22) with the following modifications. Purified clathrin was dialyzed into phosphate buffer (100 mM phosphate, pH 7.5, 5 mM EDTA). Extracts were prepared from spheroplasts as described above for size exclusion experiments except that lysis was carried out in phosphate buffer. Phenylmethylsulfonyl fluoride was added to 1 mM, and then 0.1-ml aliquots were incubated in varying concentrations of dithiobis(succinimidyl propionate) (DSP) (Pierce) for 1 h on ice. Reactions were quenched with 0.01 ml of 1 M Tris-HCl, pH 7.5, and then 5 ϫ LSB without ␤-mercaptoethanol was added, and samples were heated to 100°C for 3 min.

Increased Expression of CHC1 Partially Suppresses the Growth Defect Caused by clc1⌬-
The half-life of Chc1p is reduced from greater than 2 h to approximately 1 h in clc1⌬ strains (23). Chc1p levels were examined by immunoblotting to determine if this instability results in reduced steady state levels of protein. In clc1⌬ cells, levels of Chc1p are approximately one-tenth the amount expressed by cells carrying a single copy of CLC1 (Fig. 1A, lane 1, compared with lanes 2 and 5). Consequently, defects in clc1⌬ strains may result indirectly from the reduction of Chc1p as opposed to the absence of Clc1p. To circumvent the coincident loss of Chc1p, we constructed a set of isogenic clc1⌬ strains that carry a plasmid vector with no insert (⌬), a single copy plasmid with CLC1 (WT), or a multi-copy plasmid with CHC1 (⌬MC CHC1). The ⌬MC CHC1 strain overexpresses Chc1p about 5-fold compared with the strain expressing single copy levels of Clc1p (Fig. 1A, lanes 2-4). The ⌬MC CHC1 strain thus allowed us to assess the phenotypic consequences of Clc1p deficiency under conditions where Chc1p is not a limiting factor for clathrin function.
Deletion of CLC1 causes a severe growth defect commensurate with that of chc1⌬ cells (3). To determine whether elevation of Chc1p levels influences the growth of cells lacking Clc1p, we compared the growth rates of ⌬MC CHC1 cells to wild-type and clc1⌬ cells. In these experiments, strains were used that also harbored a plasmid carrying a copy of CLC1 under control of the GAL1 promoter (GAL-CLC1). When grown on galactose-containing medium to maintain expression of GAL-CLC1, all three strains exhibited equivalent growth (Fig.  1C, Galactose). This result indicates that the elevated expression of Chc1p in the ⌬MC CHC1 strain does not affect cell growth if Clc1p is present. On dextrose-containing medium, where GAL-CLC1 expression is repressed, the poor growth of ⌬ cells became apparent (Fig. 1C, Dextrose). On this medium, ⌬MC CHC1 cells also did not achieve wild-type growth but exhibited improved growth relative to the ⌬ cells. ⌬MC CHC1 cells similarly exhibited intermediate growth rates when growth was monitored in liquid glucose-containing medium. The ⌬MC CHC1 cells doubled in number in 2.8 h, while the WT cells doubled in 1.8 h and the ⌬ cells doubled in 4.5 h. These observations reveal that Clc1p is important for cellular growth. Furthermore, the partial restoration of growth by increased levels of Chc1p in ⌬MC CHC1 cells suggests that Chc1p can act to some degree without Clc1p.
Clc1p Is Required for Efficient ␣-Factor Maturation-Mutations in CHC1 result in defects in the proteolytic maturation of the mating pheromone ␣-factor (17,24). This defect is due to mislocalization of the resident TGN membrane endoprotease Kex2p, which normally initiates proteolytic processing of the ␣-factor precursor protein. In chc1⌬ cells, mislocalization of Kex2p to the cell surface results in secretion of the highly glycosylated precursor form of ␣-factor, which is easily distinguished from mature ␣-factor by SDS-PAGE (17,24). Consequently, the form of ␣-factor secreted by cells serves as a reliable indicator of Kex2p localization to the TGN. To determine whether loss of Clc1p affects Kex2p localization, ␣-factor maturation was monitored in the ⌬, WT, and ⌬MC CHC1 strains. Cells were metabolically labeled with [ 35 S]cysteine and [ 35 S]methionine, and ␣-factor was immunoprecipitated from the media. In the WT strain, ␣-factor maturation is efficient, with over 95% secreted as the mature protein (Fig. 2, lane 1). In contrast, the ⌬ strain secreted 80% of the ␣-factor as the highly glycosylated precursor (Fig. 2, lane 3). Elevated expression of Chc1p in ⌬MC CHC1 cells resulted in a slight but reproducible suppression of the clc1⌬ ␣-factor processing defect. This strain secreted 65% unprocessed ␣-factor (Fig. 2, lane 2). Overexpression of Chc1p in a wild-type (CHC1 CLC1) strain had no effect on ␣-factor processing (data not shown). The defective ␣-factor maturation in the ⌬MC CHC1 strain indicates that Clc1p is required for the efficient processing of ␣-factor, most likely by functioning in the localization of Kex2p. However, the partial suppression of the ␣-factor maturation defect in the ⌬MC CHC1 strain compared with the clc1⌬ strain parallels the relative growth defects in the two strains and supports the conclusion that Chc1p retains some activity in the absence of Clc1p.
Vacuolar Protein Sorting Is Efficient in clc1⌬ Cells-To determine the effect of CLC1 deletion on another transport event that occurs at the Golgi complex, sorting and transport to the vacuole of the soluble vacuolar protease CPY was examined. CPY is normally synthesized in a precursor form that is translocated into the ER, where it is subjected to signal sequence cleavage and core glycosylation. This produces a 67-kDa p1 form of the protein (25). This p1 form is then transported to the Golgi complex, where further glycosylation yields a 69-kDa p2 form (25). After sorting from the secretory pathway at the TGN and delivery to the vacuole, the p2 form is proteolytically processed to the 61-kDa mature protein (25). chc1-ts cells shifted to the nonpermissive temperature for short time periods fail to sort CPY at the TGN; thus, the p2 form of CPY is secreted from the cell (18). This observation suggests a role for Chc1p in vacuolar protein sorting. However, in chc1⌬ cells or in chc1-ts cells shifted to the nonpermissive temperature for long time periods, sorting of CPY occurs normally (18,26). It therefore appears that cells recover the ability to sort CPY following Chc inactivation, although the basis for this recuperation is not known.
Transport of CPY to the vacuole was monitored in the congenic clc1⌬ strains using a pulse-chase regimen. Cells were labeled with [ 35 S]cysteine and [ 35 S]methionine for 15 min at 30°C, and then excess unlabeled amino acids were added to quench labeling (pulse). Cells were then incubated for 0, 15, 30, or 45 min (chase), and CPY was immunoprecipitated from the extracellular (E) and intracellular (I) fractions at each time point (Fig. 3). The WT strain exhibits normal CPY sorting and transport with essentially complete maturation (86% mature, 4% p2 intracellular) within 15 min after commencing the chase (Fig. 3, WT). A slight fraction of CPY, 6%, was missorted and appeared in the medium in the p2 form at this time point. A small amount of mature CPY was also detected in the medium and most likely reflects a low level of cell lysis that occurs during preparation of extracellular fractions. For ⌬ cells, a subtle delay in processing of CPY was detected. p2 represented 16% of the total after the 15-min time point and 7% after the 45-min time point (Fig. 3, ⌬). Sorting was relatively efficient; only about 6% of the CPY was secreted as the p2 form. An increasing percentage of mature CPY, from 2% at the 0 time point to 12% at 45 min, was also found in the extracellular fraction. This is consistent with our observations that chc1⌬ and clc1⌬ cells are more prone to lysis. For ⌬MC CHC1 cells, there was little difference in the kinetics of processing or in the sorting of CPY in comparison with the ⌬ strain. Compared with the ⌬ cells, less mature CPY was found extracellularly in the ⌬MC CHC1 cells, probably because the ⌬MC CHC1 cells are less prone to lysis during preparation of the extracellular fractions. Based on these results, it is evident that CPY processing is only slightly perturbed in clc1⌬ cells and that elevated expression of Chc1p has little or no effect on CPY transport.

Clc1p Is Required for Efficient
Receptor-mediated Endocytosis-Loss of Chc function results in an immediate reduction in the rate of receptor-mediated internalization of the ␣-factor mating pheromone (20). In chc1⌬ cells or in chc1-ts cells at the nonpermissive temperature, radiolabeled ␣-factor uptake decreases 2-4-fold (20). The role of Clc1p in endocytosis was examined by measuring ␣-factor internalization by ⌬, ⌬MC CHC1, and WT strains. In the ⌬ strain, levels of ␣-factor uptake were diminished to levels comparable with that previously found for chc1⌬ cells, approximately 3-5-fold less than that of WT cells (Fig. 4 and data not shown). The ⌬MC CHC1 strain exhibited an identical defect in ␣-factor uptake. It should be noted that when Chc1p was overexpressed in the WT strain, ␣-factor internalization proceeded as it does in the wild-type strain (data not shown). Hence, overexpression of Chc1p itself does not interfere with normal ␣-factor internalization. Therefore, Clc1p is required for the efficient uptake of ␣-factor, and elevated expression of Chc1p does not provide detectable function in the absence of Clc1p.
Chc1p Is Able to Form Trimers in the Absence of Clc1p-Analysis of clathrin-mediated transport pathways indicates that the absence of Clc1p causes defective clathrin function even in the presence of excess Chc1p. To examine the basis for these defects, we employed the ⌬MC CHC1 strain to conduct biochemical studies of Chc1p, since the reduced levels of Chc1p in the ⌬ strain were difficult to detect. Trimerization of Chc1p in the absence of Clc1p was probed by chemical cross-linking of cytosol from the WT and ⌬MC CHC1 strains using the reversible cross-linker DSP. Kirchhausen and Harrison (22) have demonstrated trimer organization through chemical cross-linking of purified bovine brain clathrin triskelions. When a similar cross-linking procedure was performed on purified yeast clathrin, the same pattern of cross-linked species was observed after immunoblotting with anti-Chc1p antibody (Fig. 5A, lanes 1, 2,  and 8). Increasing the amount of DSP resulted in conversion of lower molecular weight species into the higher trimer form. These cross-linked species correspond to forms previously identified as Chc monomer (M 0 ), Chc monomer plus one light chain (M 1 ), Chc dimer forms with 0, 1, or 2 associated light chains (D 0 , D 1 , and D 2 ), and a Chc trimer form (T). The pattern produced by immunoblotting with antibodies directed against yeast Clc1p was the same as that with anti-Chc1p antibodies except for the absence of M 0 and D 0 (Fig. 5A, lane 6). To examine the trimerization status of Chc1p in the WT and ⌬MC CHC1 strains, cells were converted to spheroplasts and lysed by homogenization, and 100,000 ϫ g supernatant fractions (cytosol) were prepared. Cross-linked WT cytosol, when probed for Clc1p (Fig. 5A, lane 7) or Chc1p (Fig. 5A, lanes 3-5), revealed a banding pattern similar to purified triskelions. The bands are slightly more diffuse, which may be due to interactions with other cytosolic proteins or to less efficient crosslinking caused by quenching of the DSP by other extract components. When ⌬MC CHC1 cytosol was cross-linked, we observed a pattern consistent with trimer organization. The M 0 species cross-linked to two prominent higher molecular weight bands with increasing amounts of reagent (Fig. 5A, lanes  9 -12). These forms correspond to the D 0 and T sizes. As expected for a strain lacking Clc1p, the light chain-containing forms M 1 , D 1 , and D 2 were not detected. To determine if the banding pattern resulted from specific interactions of Chc1p in triskelions, cross-linking was performed on extracts from clc1⌬ strains that overexpress a truncated form of Chc1p lacking the trimerization domain ⌬MC CHC1(TriϪ). At DSP concentrations sufficient to convert most of the monomeric Chc1p in WT and ⌬MC CHC1 extracts to dimeric and trimeric species (Fig.  5A, lanes 5 and 12), almost all of the truncated Chc1p remained FIG. 4. Clc1p is required for efficient ␣-factor internalization. Uptake of radiolabeled ␣-factor was measured in ⌬ (squares), WT (triangles), and ⌬MC CHC1 (circles) cells. Radiolabeled ␣-factor was allowed to bind to cells at 4°C for 1 h. Cells were washed to remove unbound ␣-factor and then preshifted in the absence of glucose for 2 min at 30°C. Glucose was added to 5% to stimulate internalization (time 0), and the incubation continued at 30°C for 2, 5, or 20 min. WT and ⌬MC CHC1 cytosol panels, 100,000 ϫ g supernatant from WT or ⌬MC CHC1 cell extracts (cytosol) was fractionated by Sepharose CL-4B column chromatography. ⌬MC CHC1 tris ext panel, an extract prepared from ⌬MC CHC1 cells was sedimented at 100,000 ϫ g for 20 min, and the resulting pellet was extracted with 0.5 M Tris-HCl as described under "Experimental Procedures." The Tris-HCl-solubilized proteins were then applied to a Sepharose Cl-4B column. Selected fractions were precipitated with 10% trichloroacetic acid and analyzed by SDS-PAGE (6% acrylamide) and immunoblotting with monoclonal antibody to Chc1p. Column fraction numbers are indicated at the top of the panels. monomeric (M TϪ ), and no trimeric form was observed (Fig. 5A,  lane 14). A minor band (Fig. 5A, lane 14, asterisk) migrating more slowly than the monomer was detected, which could be due to the association of two Chc1p molecules through interactions along the arm domains similar to interactions that occur in assembled clathrin coats. The presence of a specific trimer band in the ⌬MC CHC1 strain provides strong evidence for the existence of Chc1p trimers in the absence of Clc1p.
Sepharose CL-4B size exclusion chromatography was also performed on extracts from the strains to evaluate the trimerization status of Chc1p. In cytosol prepared from WT cells, Chc1p peaked at fractions 12 and 14, where purified triskelions elute (Fig. 5B, WT, and data not shown). The Chc1p elution pattern from cytosol of ⌬MC CHC1 was distinct, with Chc1p peaking in fractions 16 and 18 (Fig. 5B, ⌬MC CHC1 cytosol). Fractions 16 and 18 correspond to the elution position of monomeric Chc1p, established by chromatography of the truncated form of Chc1p lacking the trimerization domain. However, the Chc1p from ⌬MC CHC1 was also detected in earlier fractions (fractions 10 -14), suggesting the presence of multimers. These findings, combined with the results from the crosslinking experiments, suggest that Chc1p can associate into trimers in the absence of Clc1p but that the association is inefficient or unstable when compared with Clc1p-containing trimers.
Differential Fractionation of Chc1p Is Altered in the Absence of Clc1p-We carried out differential centrifugation of WT and ⌬MC CHC1 strains as an approach to determine if the trimers formed in ⌬MC CHC1 cells are capable of assembly. Cells were converted to spheroplasts and lysed by homogenization. Extracts were then subjected to a low speed spin (1,500 ϫ g for 5 min) to remove unbroken cells. Centrifugation of the low speed supernatant at 26,000 ϫ g for 30 min yielded a medium speed pellet (MSP) and a medium speed supernatant. The medium speed supernatant was further fractionated into a high speed pellet (HSP) and high speed supernatant by centrifugation at 100,000 ϫ g for 20 min. When Chc1p distribution was compared in the differential centrifugation fractions from ⌬MC CHC1 and WT extracts, a striking increase in the percentage of Chc1p fractionating in each of the ⌬MC CHC1 pellet fractions was observed. At both the medium and high speed steps, more than 50% of Chc1p from ⌬MC CHC1 extracts sedimented, resulting in only 22% of the total Chc1p present in the 100,000 ϫ g soluble fraction (Fig. 6B, lane 6, and Table II). In comparison, 78% of Chc1p from WT extracts remained soluble after the centrifugation regimen ( Fig. 6A and Table II). These alterations were specific to Chc1p rather than a change in the overall fractionation properties of the extracts, since there were no large differences found in the distribution of a cytosolic marker, glucose-6-phosphatase dehydrogenase, and a vacuolar membrane marker, alkaline phosphatase (Table II). A characteristic property of clathrin coats is that they can be depolymerized by incubation in 0.5 M Tris-HCl (27, 28). We applied this treatment to each pellet fraction from ⌬MC CHC1 extracts to eliminate the possibility that Chc1p was forming insoluble aggregates in the absence of Clc1p. When the HSP was treated with 0.5 M Tris-HCl, Chc1p was completely solubilized, while 80% of Chc1p was released from the MSP. Tris-HCl treatment completely releases Chc1p from each of the MSP and HSP fractions of WT extracts. Thus, while there may be some formation of insoluble Chc1p aggregates in the absence of Clc1p, nearly all of the Chc1p found in the MSP and HSP fractions from ⌬MC CHC1 extracts displays the same Tris-HCl sensitivity as properly assembled clathrin coats. To analyze the Trissolubilized Chc1p by size exclusion chromatography, a 100,000 ϫ g pellet fraction from ⌬MC CHC1 cells was extracted with 0.5 M Tris-HCl. Sepharose CL-4B chromatography of Tris-HCl-solubilized Chc1p showed that the Chc1p was distributed as both monomeric and multimeric forms with a peak at fraction 12 where WT trimers elute (Fig. 5B, ⌬MC CHC1 tris ext).
To determine if the increase in sedimentable Chc1p was due to the elevated expression levels instead of the absence of Clc1p, extracts from clc1⌬ cells overexpressing both Chc1p and Clc1p (⌬MC CHC1/CLC1) 10 -25-fold were fractionated (Table  II). An increase in the amount of Chc1p observed in the pellet fractions was apparent (42% of the total), although not to the extent found in the ⌬MC CHC1 extracts (78% of the total). In particular, only 20% of the Chc1p in the low speed supernatant was sedimented by the medium speed spin of the ⌬MC CHC1/ CLC1 strain extract, compared with 54% of the Chc1p in the ⌬MC CHC1 low speed supernatant (Table II, MSP column).
Taken together, the fractionation properties of Chc1p in the ⌬MC CHC1 extracts suggest that assembly of Chc1p into coat structures is abnormal in the absence of Clc1p.
Chc1p Does Not Form Coated Vesicles in the Absence of Clc1p in Vivo-The ability of Chc1p to polymerize in the absence of Clc1p raised the possibility that coated vesicles could form in the ⌬MC CHC1 cells. We investigated this possibility by subjecting HSP fractions from WT and ⌬MC CHC1 cells to chromatography through Sephacryl S-1000. This procedure has been the standard approach used to obtain enriched preparations of clathrin-coated vesicles from yeast (28,29). Fractions obtained from Sephacryl S-1000 column chromatography of the WT HSP were probed with antibodies directed against Chc1p, Aps1p, and Kex2p. Aps1p is a component of a AP complex related to the mammalian Golgi-localized AP-1 complex (21).
FIG. 6. Differential fractionation of Chc1p is altered in ⌬MC CHC1 cells. WT and ⌬MC CHC1 cells were lysed and subjected to differential centrifugation as described under "Experimental Procedures." ⌬MC CHC1 samples were diluted 1:10 compared with WT, and then each fraction was analyzed by SDS-PAGE (6% acrylamide) and immunoblotted with monoclonal antibodies to Chc1p.

TABLE II
Distribution of Chc1p is altered in ⌬MC CHC1 cells WT and ⌬MC CHC1 cells were lysed and subjected to differential centrifugation as described under "Experimental Procedures." Each fraction was analyzed by SDS-PAGE and immunoblotted with polyclonal antibodies to glucose-6-phosphate dehydrogenase (G6PD) or ALP or with monoclonal antibodies to Chclp. Values for glucose-6-phosphate dehydrogenase and ALP represent the average from two experiments. Values for Chc1p were collected from three experiments, except those for ⌬MC CHC1/CLC1, which represent one experiment. Protein levels were determined by densitometry. Numbers represent the relative distribution of each protein in the supernatant (S) and pellet fractions (P) of each centrifugation step (MS, medium speed (26,000 ϫ g for 30 min); HS, high speed (100,000 ϫ g for 20 min); T, total]. The low speed spin (1,500 ϫ g for 5 min) was not included in the quantitation because it contained a significant level of unbroken cells as determined from the distribution of the cytosolic marker glucose-6-phosphate dehydrogenase. We note that a 3-fold increase in Chc1p was detected in the low speed pellet from ⌬MC CHC1 extracts compared with WT extracts. As expected, Aps1 co-fractionated with clathrin-coated vesicles, which peaked in fractions 42-48 (Fig. 7A). A portion of Kex2p also coeluted with Chc1p and Aps1p, consistent with a role for clathrin coats in the localization of Kex2p to the TGN as reported previously (21). Electron microscopy of samples negatively stained with uranyl acetate confirmed the presence of clathrin-coated vesicles in these fractions (data not shown). When the HSP prepared from ⌬MC CHC1 cells was subjected to Sephacryl S-1000 chromatography, a dramatic shift in the elution profile of Chc1p was observed (Fig. 7B). Chc1p was not enriched in fractions expected to contain coated vesicles; instead, it separated into two peaks, one at fractions 26 -28, which may be either Chc1p associated with large membranes or very large assemblies of Chc1p, and a pool at fractions 48 -52, which eluted after the position of WT coated vesicles. The Chc1p in fractions 48 -52 could still be sedimented by centrifugation at 100,000 ϫ g, indicating that it does not represent free Chc1p released during resuspension of the the HSP or during chromatography. Thus, this population of Chc1p appears to be assembled in a form that elutes as a complex that is smaller than the clathrin coats surrounding vesicles in WT cells.
The elution profiles of Aps1p and Kex2p from the ⌬MC CHC1 HSP were also shifted with respect to the position of WT clathrin-coated vesicles (Fig. 7B). The perturbation of Aps1p and Kex2p fractionation properties in cells with aberrant clathrin provides further evidence that these proteins are normally associated with clathrin coats. Both Aps1p and Kex2p were enriched in fractions 26 -28, most likely indicative of association with larger membrane structures. Kex2p also showed a smaller peak in fractions 48 -52, and a slight increase in the abundance of Aps1p was also observed in these fractions. The coincidence of the elution profiles of Aps1p and Kex2p with Chc1p raised the possibility that these proteins are able to associate into coat structures. Electron microscopic examination of material from fractions 26, 42, 48, and 53 showed clear vesicular structures were present; however, it did not reveal the presence of lattice-like structures characteristic of conventional clathrin coats. Thus, in cells lacking Clc1p, vesicles with conventional clathrin coats were not detected. DISCUSSION Because of the instability of Chc1p in clc1⌬ cells, phenotypes in a clc1⌬ strain cannot be attributed solely to the loss of Clc1p. We have used a clc1⌬ strain expressing elevated levels of Chc1p to investigate the effect of the absence of Clc1p on clathrin-mediated processes within the cell. Our results reveal that the clathrin-mediated transport processes of receptor-mediated endocytosis and TGN membrane protein localization are compromised by the absence of Clc1p. Biochemical analyses indicate that there are defects in Chc1p trimerization, coat assembly, and coated vesicle formation. However, elevated expression of Chc1p partly suppresses clc1⌬ cell growth and ␣-factor maturation defects, arguing that Chc1p retains some degree of function in the absence of Clc1p. Consistent with the in vivo results, Chc1p in cell extracts from the ⌬MC CHC1 cells is capable of forming trimers and associating into larger, Trissensitive structures, which could represent assembled clathrin coats.
The ␣-factor maturation defect in clc1⌬ cells argues that Clc1p contributes to clathrin function in the TGN localization of Kex2p. The prevailing model for TGN protein localization in yeast suggests that proper localization is maintained by transport of TGN proteins to endosomes followed by retrieval to the Golgi complex (2,30). Based on studies of chc1 mutants, clathrin has been proposed to act in the localization pathway by collecting proteins into clathrin-coated vesicles, which form from the TGN and are targeted to endosomes (2,17). In the absence of Chc1p function, Kex2p is no longer directed to the endosomal pathway but instead is mislocalized to the cell surface, resulting in decreased maturation of ␣-factor precursor (17,24). Our analyses of clc1⌬ strains adds support for this model of clathrin function in Kex2p localization. First, even with elevated Chc1p levels, clc1⌬ cells exhibit ␣-factor maturation defects diagnostic of Kex2p mislocalization. Second, the loss of clathrin-coated vesicles in ⌬MC CHC1 cells detected by Sephacryl S-1000 column chromatography was accompanied by a dramatic shift in the fractionation pattern of Kex2p compared with wild-type cells. This shift argues that Kex2p is associated with clathrin-coated membranes in wild-type cells. However, the increased expression of Chc1p in ⌬MC CHC1 cells partly ameliorates the ␣-factor maturation defect caused by clc1⌬, suggesting that Chc1p alone can function weakly in localization of Kex2p to the TGN. This localization could be due to a low level of Chc1p-coated vesicle formation. Compatible with this suggestion, coelution of Chc1p and Kex2p in fractions 48 -52 from the Sephacryl S-1000 column could reflect aberrantly or partially coated vesicles. Alternatively, even without being able to form vesicles, Chc1p coats assembled onto TGN membranes could still interact with Kex2p, thereby retarding transport from the TGN to the cell surface and raising the level of TGN Kex2p available to cleave the ␣-factor precursor.
In contrast to growth and pheromone maturation, receptormediated endocytosis of ␣-factor is not affected by the increased levels of Chc1p in clc1⌬ cells. This difference offers a potential insight into the suppressing activity of Chc1p in growth and TGN protein localization. The assay for ␣-factor internalization assesses the complete sequestration of surface-bound pheromone resulting from incorporation into endocytic vesicles. Thus, this assay presumably represents a direct measure of clathrin function in the genesis of a transport vesicle. The inability of high Chc1p levels to suppress the endocytosis defect in clc1⌬ cells suggests that Chc1p cannot function in membrane vesiculation without Clc1p. This observation, combined with the absence of conventional clathrin-coated vesicles in ⌬MC CHC1 cell extracts, leads us to favor the aforementioned possibility that the suppression of the pheromone maturation 100,000 ϫ g pellet fraction from WT or ⌬MC CHC1 cells was resuspended in Buffer A and chromatographed through Sephacryl S-1000. A portion of each fraction was trichloroacetic acid-precipitated, analyzed by SDS-PAGE (6% acrylamide), and immunoblotted with antibodies to Chc1p, Aps1p, or Kex2p. Column fraction numbers are indicated at the top of the panels. defects by increased Chc1p expression is due to assembly of static or frozen Chc1p coat structures incapable of forming vesicles rather than to low levels of Chc1p-coated vesicle formation. The partial suppression of TGN protein mislocalization might then contribute to improved growth rates.
In vitro studies of mammalian clathrin suggest that Chc does not require Clc subunits to trimerize. Trimer stability is not affected by removal of Clc from preformed triskelions (31), and an N-terminally truncated Chc forms trimers when expressed in E. coli in the absence of Clc (32). Our analysis of ⌬MC CHC1 cells also indicates that Chc1p is able to trimerize in the absence of Clc1p, but Clc-deficient triskelions appear to be unstable. The difference in Chc trimer stability likely reflects the different preparations of clathrin. The mammalian studies used preformed triskelions obtained from purified clathrincoated vesicles or used Chc trimers formed by overexpression in bacteria. In contrast, our experiments analyzed Chc1p expressed in a native context, albeit at somewhat elevated levels. Taken together, the analyses of mammalian and yeast clathrin suggest that Chc is capable of trimerizing without Clc but that Clc facilitates the formation of stable trimers.
Clc-deficient mammalian triskelions will assemble into coatlike lattices in vitro, but assembly does not display the dependence on calcium or clathrin AP complexes that wild-type triskelion assembly exhibits (32,33). The less stringent requirements for Clc-deficient triskelion assembly in vitro have fostered the proposal that Clc acts to prevent premature assembly of clathrin in vivo. The distribution of Chc1p in differential centrifugation fractions of ⌬MC CHC1 extracts conforms to this model. Compared with wild-type clathrin, a much higher percentage of Chc1p is associated with the pellet fractions, a result expected if Clc-deficient Chc1p assembled in an unregulated manner (33,34). We cannot completely exclude the possibility that the Chc1p in the pellets is nonspecifically aggregated rather than polymerized into cages. However, the sensitivity of the sedimented Chc1p to Tris extraction and the trimeric properties of the extracted Chc1p are characteristic of clathrin coats and support the possibility that a significant portion of Chc1p is polymerizing into coat-like structures. Even with this possible increase in polymerized clathrin, clathrin-coated vesicle formation in ⌬MC CHC1 cells is severely defective. This defect could be the consequence of nonproductive Chc1p assembly without Clc1p, or it could reveal a role for Clc1p in the conversion of planar clathrin lattices into closed polyhedral cages. Further experiments will be needed to distinguish between these possibilities. In either case, our data provide evidence that Clc1p is necessary in vivo to maintain Chc1p in a form capable of participating in transport vesicle biogenesis.
The studies presented here represent the first characterization of Clc function in clathrin-dependent protein transport in vivo. Clc1p is important for clathrin triskelion stability and appears to be required for formation of clathrin-coated vesicles at the plasma membrane and Golgi complex. These phenotypes now provide a foundation for a genetic characterization of specific Clc properties such as phosphorylation, calcium binding, and calmodulin binding.