INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

A major pathway for endocytosis involves the coat protein clathrin, comprised of heavy and light chains (HC and LC) and a large number of accessory factors. These proteins facilitate cargo capture, assembly of the clathrin lattice, membrane invagination, pinching off, and then uncoating of the vesicle once detached from the membrane. Many of these accessory proteins are held together by the cooperative interaction of different binding domains and appear to form a web beneath the cell surface (D'Hondt et al., 2000; Brodsky et al., 2001), although the role of the majority of these components and how they are regulated is still not completely understood.

Saccharomyces cerevisiae also contains clathrin HC and LC (Chc1p and Clc1p); however, there is only a partial dependence on clathrin for uptake at the cell surface (Payne et al., 1988; Tan et al., 1993; Huang et al., 1997). On the other hand, the actin cytoskeleton appears to be essential for endocytosis in yeast, as genetic screens to identify factors involved in internalization have repeatedly uncovered proteins important for function of the cortical actin cytoskeleton (for reviews see Baggett and Wendland, 2001; D'Hondt et al., 2000). Interestingly, many of these factors, or proteins that interact with these factors, are related to molecules that function in clathrin-mediated endocytosis in animal cells. For example, cortical actin patch components Rvs161p and Rvs167p are related to amphiphysins, which bind to clathrin HC, the heterotetrameric adaptor AP-2, and other regulatory components involved in clathrin-coated vesicle (CCV) formation in animal cells (David et al., 1996; McPherson et al., 1996; Ramjaun and McPherson, 1998; Slepnev et al., 2000). Eps15-homology (EH) domain proteins, Pan1p, End3p, and Ede1p were identified in screens for endocytosis mutants in yeast and found to be important for cortical actin structures (Raths et al., 1993; Bénédetti et al., 1994; Tang and Cai, 1996; Tang et al., 1997; Wendland and Emr, 1998; Gagny et al., 2000). Animal Eps15 has an AP-2 binding region and interacts with NPF motifs via the EH domain in proteins like epsin, which in turn has clathrin and AP-2 binding domains (Benmerah et al., 1996; Di Fiore et al., 1997; Iannolo et al., 1997; Chen et al., 1998; Drake et al., 2000). Yeast Pan1p is in a complex with End3p and another cortical actin patch component, Sla1p and has recently been shown to activate Arp2/3 complexes (Tang et al., 2000; Duncan et al., 2001; Zeng et al., 2001). This Pan1p function may couple actin polymerization to the endocytic machinery. Interestingly, Pan1p binds to NPF motifs in the yeast epsin-related proteins, Ent1/2p, and AP180/CALM-like proteins, Yap180p's, which are associated with cortical patches but also directly bind to clathrin HC in yeast (Wendland and Emr, 1998; Wendland et al., 1999). AP180/CALM promote the assembly of clathrin lattices in vitro and bind AP-2, and functional studies indicate that these proteins are also important components of the endocytic machinery in animal cells (McMahon, 1999; Traub et al., 1999).

A role for cortical actin in endocytosis in animal cells has been less clear, because actin-depolymerizing drugs, e.g., latrunculin A (Lat-A), disrupt clathrin-mediated endocytosis in some cell types or under some growth conditions but not in others (Fujimoto et al., 2000; Qualmann et al., 2000; Apodaca, 2001). Also, in some polarized epithelial cells such drug treatments inhibit internalization at the apical surface but not the basolateral membrane (Apodaca, 2001). Although actin may not be absolutely required for endocytosis in animal cells, other recent results support an important association of the actin cytoskeleton with clathrin-coated pits. Treatment of cells with Lat-A increases the lateral mobility of clathrin-coated pits at the cell surface, suggesting that cytoskeletal elements may localize the endocytic machinery to domains of the plasma membrane (Gaidarov et al., 1999; Bennett et al., 2001). In addition, a few proteins that associate with both the actin cytoskeleton and clathrin-coated pit components have been identified. Two of these factors, mAbp1p and HIP1/HIP1R, were first identified in yeast as actin-associated proteins Abp1 or Sla2p (also known as End4p) or in screens for endocytic mutants (Drubin et al., 1990; Holtzman et al., 1993; Raths et al., 1993). mAbp1 has an actin-binding region and a SH3 domain that also interacts with dynamin, and overexpression of the SH3 domain inhibits receptor-mediated endocytosis in animal cells (Kessels et al., 2001).

HIP1 was identified as a huntingtin interacting protein and is closely related to HIP1R (Kalchman et al., 1997; Wanker et al., 1997; Seki et al., 1998; Engqvist-Goldstein et al., 1999). These proteins, as well as Sla2p, contain an epsin N-terminal homology (ENTH) domain found in epsin and AP180/CALM, which may help to recruit these proteins to the plasma membrane by binding cell surface phosphoinositides (Ford et al., 2001; Itoh et al., 2001). The C terminus of the yeast and mammalian Sla2p-related proteins is homologous to the talin-like F-actin binding module (McCann and Craig, 1997). The central region of HIP1/HIP1R, which includes a coiled-coiled segment, has recently been shown to bind directly to clathrin and facilitates association with clathrin-coated pits (Engqvist-Goldstein et al., 2001; Metzler et al., 2001; Mishra et al., 2001; Waelter et al., 2001; Legendre-Guillemin et al., 2002). Although yeast Sla2p binds actin, shows partial colocalization with cortical actin patches, and is important for cortical actin organization and endocytosis (Holtzman et al., 1993; Raths et al., 1993; Wesp et al., 1997), no connection to clathrin-mediated trafficking has been reported.

In previous studies we identified the SCD5 gene as a multicopy suppressor of clathrin HC-deficient yeast (Nelson and Lemmon, 1993; Nelson et al., 1996). Here we demonstrate that Scd5p is important for actin organization and endocytosis. This finding led us to examine actin localization in cells lacking clathrin. We find that similar to scd5 mutants, loss of clathrin function causes major defects in cortical actin organization, and overexpression of SCD5 partially suppresses this defect. Moreover, two-hybrid screening with the clathrin LC has identified Sla2p as an interacting protein. Scd5p also associates with Sla2p and Sla2p's cortical localization is dependent on both clathrin and Scd5p. These data suggest that clathrin and Scd5p may play a role in organizing or stabilizing the cortical actin network, which is important for endocytosis in yeast.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Strains, Media, and Genetic Methods

Strains used in this study are listed in Table 1. YEPD (2% yeast extract, 1% peptone, and 2% dextrose), YEP-GAL (2% yeast extract, 1% peptone, and 2% galactose), 5-fluoro-orotic acid (FOA) plates, and synthetic complete dropout media (e.g., complete-uracil [C-URA]) were prepared as described in Nelson and Lemmon (1993). Yeast mating, sporulation, and tetrad analysis were performed as described (Guthrie and Fink, 1991). Yeast were transformed using the method of Gietz et al. (1995).

Plasmids

Plasmids were propagated in Escherichia coli DH5. Important plasmids used in this study are listed in Table 2. A clone containing the scd5-338 truncation mutation under its own promoter was generated in two steps. A 4.6-kb BamHI-ClaI SCD5 fragment was cloned into pBluescript SK⁺ (Stratagene, La Jolla, CA). This plasmid was subjected to site directed mutagenesis by the dut ung method (Kunkel et al., 1987; McClary et al., 1989) to generate pAC8. The primer 5'- GAGGAAGAGTAATTTACTAAGAATTCAG-3' used for mutagenesis created two stop codons at positions 535 and 536 of the Scd5p coding sequence. The mutation was verified by DNA sequencing. pAC10 was made by cloning the scd5-338 BamHI-ClaI fragment from pAC8 into YIp5 (Botstein et al., 1979). pJSC3 was generated by gap repair and contains scd5-338 from pAC8 in pRS315 (Sikorski and Hieter, 1989). The GFP-SCD5 construct contains SCD5 with the coding sequence for green fluorescent protein (GFP) fused in frame between codons 80 and 81 at a unique SacI site. A PCR product for gap repair was made using primers 5'-CAGAAATAATCACTCAAATACAGCAGCTGATAATGCCACTAACGTGAGCTCTGGTGCTGGTGCTGGTGCTGC-3' and 5'-CAGAAAGTGTTCGTACCTCCCCATTAGGTGGATTATCCTTTGAGGAGAGCTCACTTTGTACAATTCATCCATACC-3' and a modified YGFP3 coding sequence as a template. The modified YGFP3 contains the coding sequences for an N-terminal 5 × GlyAla linker followed by yeast enhanced GFP-S65G, S72A (gift of C. Schaerer-Brodbeck). The PCR product was cotransformed into SL1462 with YEp24-SCD5, which was cut with SacI. The final CEN, LEU2, GFP-SCD5 plasmid, pKRH21, was made by gap repairing a 2.2-kb fragment from YEpGFP-SCD5 into pCC545 gapped with XbaI.

pKH19 was created by engineering a CLC1 PCR product with a 5' NcoI site at the CLC1 start codon and a PvuII site downstream of the stop codon using primers 5'-CCATGGCAGAGAAATTC-3' and 5'-GCGCAGCTGATCATTTAAGCACCG-3', respectively. This was transferred by subcloning into pAS2 (Harper et al., 1993) to generate pKH15. The CLC1 ORF in pKH15 was then subcloned as a NcoI-BamHI fragment into pACT2 (Harper et al., 1993) to yield pKH19. pKH24 was generated by cloning a 5.0-kb ClaI-SalI fragment containing CHC1 into the SmaI-XhoI sites of pACT2. This fused codons 655-1653 of CHC1 to the GAL4 activation domain (GAD) coding sequence. In pKH31, a BamHI fragment containing the CLC1 ORF was cloned into pGBDU-C3 (James et al., 1996) to generate a full-length CLC1 ORF fused in frame to the coding region of the GAL4 binding domain (GBD). pGAD-SLA2-(292-520) (p31-2) and pGAD-SLA2-(292-968) (p31-10) were isolated from a two-hybrid screen using a LC bait (pKH31). pKH47 was generated by moving SLA2-(292-520) from p31-2 as an EcoRI-PstI fragment into pGBDU-C2 (James et al., 1996). To generate pKRH20, a 0.7-kb SCD5 PCR product was made using primers 5'-CGGGATCCATGTCGTTTGATTGGCTTA-3'and 5'-CGAGATCTACACCTCCTCG-3'. This produced a BamHI site at the SCD5 start codon and a stop at codon 227 followed by a BglII site. This product was cloned into pGBDU-C1 (James et al., 1996) fusing SCD5 codons 1-227 to the GBD coding region. pNT1 was made by cloning a XbaI-SalI fragment containing the scd5-338 mutation into the XbaI-SalI sites of pKRH20. This resulted in a construct coding for residues 1-534 of Scd5p fused to the GBD (GBD-scd5-338).

scd5-338 Integration and END3 Gene Deletion

To integrate the scd5-338 allele at the SCD5 chromosomal locus, pAC10 (URA3, scd5-338) was linearized by BlpI and transformed into scd5::TRP1/SCD5 strains, selecting on C-URA. YIp5-scd5-338 could integrate adjacent to SCD5 or scd5::TRP1. Several integrants strains were plated on FOA to select for recombinants that lost the vector and adjacent sequence. FOA^R colonies were then screened for the loss of TRP1 by plating replica plating on C-TRP. The Trp candidates represented events where integration occurred adjacent to the null allele and recombination excised scd5::TRP1 retaining the scd5-338 allele. The scd5-338 mutation generates a new DdeI site. The integration of scd5-338 was confirmed by heteroduplex analysis (Cotton, 1992) and PCR analysis combined with restriction enzyme digestion using DdeI. Tetrads were dissected to generate haploid progeny (SL3519, SL3521, and SL4182). The scd5-338 segregants were temperature sensitive for growth and expressed the truncated protein, as expected.

To generate a null allele of END3 the complete ORF was replaced by HIS3-MX6 using a PCR-based one-step gene-replacement method. The disruption PCR fragment was created with primers 5'-AGTTAGTGGGTATTGGAAAGGCCGGTAAAGATAACAGGG-ATCTCTGAAAACGGATCCCCGGGTTAATTAA-3' and 5'-AACAAACAGTAAATATTACACATTCATGTACATAAAATTAA TTATCGGTGGAATTCGAGCTCGTTTAAAC-3' using pFA6a-HIS3MX6 (Wach, 1996) as the template. This PCR product was transformed into a scd5-338/SCD5 heterozygous diploid generating SL4117. Colony PCR was used to verify the disruption.

Two-hybrid Interaction Analysis

For the two-hybrid screen with CLC1 as a bait, a yeast genomic two-hybrid library (James et al., 1996) was transformed into SL3004. YPJ96-4A (James et al., 1996) was transformed with pKH31 (pGBD-CLC1). The library strain and YPJ96-4A carrying pKH31 were mated in liquid culture for 4 h and plated on YEPD overnight. The cells were scraped from the plate and replated on synthetic medium lacking leucine, adenine, and uracil (C-ADE-LEU-URA) to select for mated zygotes and to monitor expression of the GAL2-ADE2 reporter from YPJ96-4A. Colonies appeared after 3-5 d and were further screened for false positives by testing against an empty bait vector and nonrelevant baits. Plasmids that passed these tests were sequenced from both ends of the insert.

Scd5p two-hybrid interactions studies used the mating method described above to generate strains expressing GAD and GBD plasmids. For clathrin two-hybrid interaction studies, YPJ96-4A was directly transformed with both GAD and GBD plasmids. For analysis of two-hybrid interactions, log phase cultures were diluted to 5 × 10⁶ cells/ml, and equal volumes of cells were spotted on C-ADE-LEU-URA and C-LEU-URA plates. The plates were grown for 3-5 d at 30°C.

Other Assays

The alpha-factor endocytosis assay was performed as described previously (Dulic et al., 1991). Lucifer yellow (LY) uptake was carried out by methods described in Munn et al. (1995). The latrunculin-A (Lat-A) halo sensitivity assay was performed as described by Ayscough et al. (1997), comparing a scd5-338 strain (SL4301) to an isogenic wild-type control (SL4302).

Microscopy

Yeast cells were grown to early exponential phase in YEPD or YEP-GAL. For F-actin, DNA, and chitin staining, cells were fixed with 3.7% formaldehyde and then stained with Alexa-568-phalloidin (Molecular Probes, Eugene, OR), 4'6'-diamidino-2-phenylindole (DAPI; Sigma, St. Louis, MO), or calcofluor white (Sigma), respectively, as described in Adams and Pringle (1991) and Pringle (1991). For simultaneous localization of GFP-Scd5p and actin, cells were fixed for 10 min with 2.0% formaldehyde and washed three times with PBS. The cells were stained with Alexa-594-phalloidin (Molecular Probes) as described in Adams and Pringle (1991) and visualized immediately. For GFP-Abp1p and actin colocalization, cells were fixed and processed as described in Heil-Chapdelaine et al. (1998).

For immunofluorescence, cells were prepared using a methanol/acetone fixation as described previously (Pringle et al., 1991). Affinity-purified rabbit anti-C-terminal Sla2p antiserum (Yang et al., 1999) was used at 1:75 dilution, and guinea pig anti-actin antiserum (Mulholland et al., 1994) was used at 1:2000 dilution. For fluorescent visualization, FITC-conjugated goat anti-rabbit IgG and Alexa-594-conjugated goat anti-guinea pig IgG were used at 1:800 dilution.

Most of the microscopy was performed using a Zeiss Axioplan-2 microscope (Thornwood, NY) equipped with Nomarski differential interference contrast (DIC) optics, a plan-Neofluor 100× objective (numerical aperture 1.3), and a Hamamatsu C4742-95 cooled CCD camera (Bridgewater, NJ) using QED acquisition software. A LSM 410 confocal microscope equipped with a plan-Apochromat 100× objective (numerical aperture 1.4) was used for analysis of cells stained with Alexa-568-phalloidin in Figures 2 and 6. Images were manipulated in Adobe Photoshop (San Jose, CA).

For quantification of microscopy, 100-500 cells were counted for each condition. Cells were scored as having depolarized actin patches if there were 10 patches in the mother cell of cells with buds. Actin patches were scored "large" (or aggregates of actin) if they were 0.5 µm diameter. Normal patches seen in wild-type cells were typically 0.1-0.2 µm diameter. Large patches were usually delocalized to the mother cell, but each was counted as a single patch. G-actin "bars" were scored by their unusual stick-like or bar morphology and were only detected with anti-actin antibodies. Actin bar structures were never observed in wild-type cells.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Endocytosis Is Defective in a scd5 Truncation Mutant

SCD5 was originally isolated as a multicopy suppressor of the lethality of a clathrin HC-deficient strain (Nelson and Lemmon, 1993). Previously we studied a temperature-sensitive scd5-338 mutant, expressing a 338-amino acid COOH-terminal truncation of Scd5p (Nelson et al., 1996). This strain exhibits a partial post-Golgi secretion defect at the nonpermissive temperature but no obvious defects in sorting/retention of trans-Golgi proteins typical of clathrin mutants (Nelson et al., 1996). Because clathrin mutants also display slowed receptor-mediated endocytosis (Payne et al., 1988; Tan et al., 1993; Huang et al., 1997), we examined whether Scd5p function is required for internalization of -factor by its receptor, Ste2p (Figure 1A). At 25°C, internalization of ³⁵S-labeled -factor was nearly normal compared with an isogenic wild-type strain. Preshifting scd5-338 cells to 37°C for 15 min resulted in a rapid and complete block in internalization of -factor.

View larger version (40K): [in this window] [in a new window]   Figure 1.   Endocytosis is defective in scd5-338 cells at 37°C. (A) Receptor-mediated internalization of radiolabeled -factor. Cultures were grown at 25°C to mid log phase in YEPD, pelleted, and resuspended at 1 × 109 cells/ml in fresh YEPD. Cells were preincubated at 25 (, ) or 37°C (, ) for 15 min before addition of 35S-radiolabeled -factor. Samples were harvested at times indicated and processed for determination of internalized -factor. Strains are isogenic wild-type (SL3921; , ) and scd5-338 (SL3993; , ). (B) Lucifer yellow (LY) uptake. Wild-type (SL1462; a and c) and scd5-338 (SL3739; b and d) cells were incubated at 25°C (a and b) or preshifted to 37°C (c and d) for 15 min. Then LY was added, and incubation was continued for 60 min. Cells were photographed using DIC (to visualize the vacuole) and fluorescence microscopy. Bar, 5 µm.

We also examined fluid-phase endocytosis in the scd5 mutant by visualizing accumulation of the fluorescent dye lucifer yellow (LY) in the vacuole (Figure 1B). At 25°C to 35-55% of scd5-338 cells, depending on the isolate, showed vacuolar staining with LY. When cells were shifted to 37°C, uptake to the vacuole was completely blocked. Nearly 100% of wild-type cells internalized LY at either temperature. We conclude that Scd5p is required for both fluid-phase and receptor-mediated endocytosis.

Scd5p Is Important for Actin Organization

The endocytic defects of scd5-338 are much more severe than observed in clathrin mutants (Payne et al., 1988; Tan et al., 1993; Huang et al., 1997) but resemble the internalization phenotypes of a number of mutants with a perturbed cortical actin cytoskeleton (e.g., see Kübler and Riezman, 1993; Bénédetti et al., 1994; Munn et al., 1995; Wesp et al., 1997; Madania et al., 1999; Schaerer-Brodbeck and Riezman, 2000). Moreover, some of these mutants, e.g., sla2/end4, rvs161, rvs167, and act1, also display partial post-Golgi secretory defects similar to those of scd5-338 (Novick and Botstein, 1985; Nelson et al., 1996; Mulholland et al., 1997; Breton et al., 2001). Therefore, we examined actin organization in the scd5-338 mutant.

Two major actin filament-based (F-actin) structures are seen in yeast cells: actin cables and cortical actin patches (Adams and Pringle, 1984; Pruyne and Bretscher, 2000a). The cables, which are bundles of actin filaments, are responsible for polarized growth and delivery of secretory vesicles, cell wall material, organelles, and other proteins into the growing daughter cell. Cortical actin patches appear at the site of bud emergence and then localize primarily to the growing bud. Late in the cell cycle, patches first redistribute across mother and daughter cells. Finally, cables reorient toward the mother/daughter cell neck and patches concentrate at the bud neck where they are involved in septum formation and cytokinesis (See Figure 2, A and B).

View larger version (43K): [in this window] [in a new window]   Figure 2.   Actin organization is defective in scd5-338 cells. Wild-type (SL1528, A, B, E, and F) and scd5-338 (SL3920, C, D, G, and H) strains were grown on YEPD to log phase at 25°C. Cells were immediately harvested (A, C, E, and G) or shifted to 34°C (B, D, F, and H) for 3 h before fixation, staining, and microscopic visualization as described in MATERIALS AND METHODS (Bar, 5 µm). (A-D) Stained with Alexa-568-phalloidin to visualize filamentous actin; (E-H) indirect immunofluorescence with anti-actin antibodies. Examples of cells through the cell cycle are shown in A-D. (I) Bar graph showing percent of cells with depolarized patches, large patches, or actin bars from the same experiments as shown in A-H. Quantification was determined as described in MATERIALS AND METHODS.

To examine actin organization, wild-type and scd5-338 cells were grown at 25°C or shifted to 34°C for 3 h and stained with Alexa-568-phalloidin to label F-actin. Many scd5-338 mutant cells displayed aberrant patch organization, which was more severe at 34°C (Figure 2, C and D). Often these patches were depolarized to the mother cell, and some patches appeared larger than normal cortical actin structures (see Figure 2I for quantification of patch morphology). In addition, there were fewer or thinner cables in some scd5-338 cells, particularly in those that had larger patches, and in some cells the cables seemed to be misoriented.

There is very little nonpolymerized globular actin (G-actin) and it is not aggregated in normal yeast cells; however, G-actin aggregates or actin bars are observed in the presence of some mutations affecting the actin cytoskeleton (e.g., srv2, act1-2, or sla1; Novick and Botstein, 1985; Holtzman et al., 1993; Cope et al., 1999). Therefore we stained SCD5 and scd5-338 cells with anti-actin antibodies, which detect both F- and G-actin. The staining of wild-type cells resembled the phalloidin pattern (Figure 2, E and F). In contrast, many scd5-338 cells accumulated actin bars and other large aggregates of actin, presumably G-actin, that were not observed by phalloidin staining (examples shown in Figure 2, G and H, and quantification in I).

Latrunculin A (Lat-A) is an inhibitor of actin assembly, and many mutations affecting actin organization are hypersensitive to the drug (Ayscough et al., 1997). In addition, sla1 and end3 are slightly resistant to Lat-A (Ayscough et al., 1997). We found that scd5-338 also confers mild resistance to this inhibitor. In a Lat-A halo assay the scd5-338 mutant (SL4301) gave a diameter ratio of 0.8 ± 0.05 compared with an isogenic wild-type strain (SL4302; average of three experiments), although the observed effect could actually be stronger because scd5-338 grows slightly slower than wild type.

The scd5-338 mutant was examined for other phenotypes that are often associated with perturbation of the actin cytoskeleton. Mutations affecting actin can cause defects in bud site pattern selection and depolarized chitin deposition (Novick and Botstein, 1985; Yang et al., 1997). Also, multinucleated cells and aberrant morphologies, including multibudded cells, are sometimes seen because of the polarity defects in actin mutants (Novick and Botstein, 1985; Yang et al., 1997). To examine budding pattern, cells were stained with calcofluor white, which labels chitin in bud scars left on mother cells after cytokinesis (Pringle, 1991). In the scd5-338 cells chitin was delocalized both at 25°C and after shift to 34°C for 3 h (Figure 3, B and C, and 3, E and F). Also, the majority of scd5-338 diploids (>75%) displayed random bud scar placement at 25 and 34°C, whereas >93% of wild-type cells showed the normal bipolar budding pattern (Figure 3, A-C). Haploid scd5-338 cells also showed random budding (72% at 34°C), rather than the normal axial budding pattern (Figure 3, E-F). Haploid and diploid mutants were heterogeneous in size and shape but were generally larger and rounder than wild-type cells (Figures 1 and 3). Cell surface rims were observed in DIC profiles of scd5-338 mutant cells, indicating the mutation results in thickened cell walls (Figure 3). EM analysis also showed thickening of the mother cell wall, similar to other cortical actin mutants. In addition, mutant cells (17 and 33% of diploids at 25 and 34°C, respectively) were often multibudded, mostly appearing to have a small aborted bud on the mother cell (see Figure 3I). Similar numbers were observed in haploids. The concentrated patches of F-actin sometimes found in mother cells (for example, see Figure 2C) were often positioned at these sites of aborted budding. A significant number of scd5-338 cells were multinucleated (up to 14% at 34°C), compared with <1% seen in wild-type cells (Figure 3, G and H). From these data we conclude that scd5-338 causes cell polarity defects typical of many mutations affecting the actin cytoskeleton.

View larger version (74K): [in this window] [in a new window]   Figure 3.   scd5-338 cells exhibit cell polarity defects. Cells were grown in YEPD to log phase at 25°C or shifted to 34°C for 3 h before fixation and visualization. Top panels: (A-F) calcofluor white staining of chitin; (G-I) cells stained with DAPI to visualize DNA. Bottom panels: DIC images. Strains are as follows: wild-type diploid (SL1528) at 34°C (A and G); scd5-338 diploid (SL3920) at 25 (B and H) or 34°C (C and I); wild-type haploid (SL1462) at 34°C (D); scd5-338 haploid (SL3739) at 25 (E) or 34°C (F). Bar, 5 µm.

scd5-338 Exhibits Genetic Interactions with Mutations in Genes Encoding Other Proteins that Modulate the Actin Cytoskeleton and/or Endocytosis

Many mutations affecting actin cytoskeletal organization and/or endocytosis cause synergistic phenotypes when combined, and many of the normal gene products are physically associated or directly interact (Botstein et al., 1997; D'Hondt et al., 2000). Therefore, we examined genetic interactions between scd5-338 and mutations in genes important for these processes (Table 3). Haploids carrying mutations in genes to be tested were crossed to scd5-338 and subjected to segregation analysis. Then wild-type, single, and combination mutant progeny from tetrads were analyzed for synthetic growth phenotypes. We found sla2-41(end4-1), pan1-20, and end3 were synthetically lethal with scd5-338. abp1, and rvs161, and rvs167 showed strong negative synergistic growth defects with scd5-338, as the double mutants were inviable at 30°C and often grew very slowly at 25°C, whereas the single mutants grew relatively well or like wild-type cells at these temperatures. A weaker negative genetic interaction was observed with sac6, where the sac6 scd5-338 double mutant grew at 30°C but not at 32°C. These genetic interactions are likely to be specific because ent1, ent2, sla1, srv2, yap1801, yap1802, and yap1801 yap1802 had no effect when combined with scd5-338. In these cases, growth of the multiple mutants was similar to that of the scd5-338 mutant alone.

View this table: [in this window] [in a new window]   Table 3.  Genetic interactions between scd5-338 and mutations in other genes affecting endocytosis and/or actin cytoskeletal organization

The double mutants were also analyzed for enhanced defects in actin cytoskeleton organization at 25°C, compared with the single mutants, by staining with Alexa-568-phalloidin (summarized in Table 3). Generally the results paralleled the growth phenotypes. The mutations that exacerbated the growth of scd5-338 the most caused more severe actin defects. For example abp1 scd5-338 double mutants had fewer cables and fewer, but larger and depolarized, patches than scd5-338 alone, even although the actin cytoskeleton in abp1 single mutants was normal at 25°C (Drubin et al., 1990). In addition, disruptions in RVS161 and RVS167, which exhibit some defects in the actin cytoskeleton on their own (Bauer et al., 1993; Sivadon et al., 1995), resulted in the appearance of more and depolarized patches and fewer actin cables than the single mutants. sac6, which also perturbs actin (Adams et al., 1991), resulted in the appearance of fewer cables and a few larger patches in combination with scd5-338. This weaker synergistic effect on actin is consistent with the more limited effect of sac6 on growth of the scd5 mutant. The scd5-338 double mutants that contained sla1, ent1, ent2, yap1801, yap1802, or srv2 did not show an exaggerated actin phenotype. Instead, these double mutants resembled the scd5-338 mutant, despite the fact that some of these mutations (sla1 and srv2; Vojtek et al., 1991; Holtzman et al., 1993; Lila and Drubin, 1997) cause actin phenotypes on their own. The noninteracting mutations also did not exacerbate the LY internalization defect of scd5-338 at 25°C. We note that most of the mutations causing enhanced growth and actin cytoskeleton phenotypes with scd5-338 also had major endocytic defects on their own, so the LY analysis was uninformative. Overall this analysis indicates that mutations in several genes affecting actin organization and endocytosis also genetically interact with scd5-338, further supporting a role for Scd5p in actin function.

Scd5p Colocalizes with Cortical Actin

Because scd5-338 causes several phenotypes observed in other mutants with defects in cortical actin organization, we examined localization of Scd5p using a GFP-tagged protein in cells where this was the only source of the protein. The plasmid containing GFP-SCD5 (pKRH21) complements scd5 completely. We found that GFP-Scd5p colocalized with cortical actin patches visualized with Alexa-594-phalloidin. This was most easily observed in budding cells, where there was a striking polarized distribution of Scd5p to the bud (Figure 4). GFP-Scd5p could also be seen in mother cell actin patches. However, there was not complete overlap of Scd5p and cortical patches, as has been observed for other actin patch proteins like Sla2p (Yang et al., 1999). Overall, the localization results further support a role for Scd5p in cortical actin function.

View larger version (45K): [in this window] [in a new window]   Figure 4.   Scd5p localizes with cortical actin. SL4300 (scd5/scd5 + pKRH21 [CEN, LEU2, GFP-SCD5]) was grown at 30°C to log phase. Cells were prepared for visualization of F-actin (left) and GFP-Scd5p (right) as described in MATERIALS AND METHODS. Arrows indicate examples of mother cell actin patches that also show GFP-Scd5p fluorescence.

Clathrin Deficiency Causes Defects in Cortical Actin Organization

Because Scd5p is important for endocytosis and the gene was isolated by its ability to rescue inviable strains of clathrin HC- (Chc1p) deficient yeast (Nelson and Lemmon, 1993; Nelson et al., 1996), we considered the possibility that this interaction is associated with endocytic defects seen in viable clathrin mutants. To investigate this, we transformed a viable clathrin HC null strain with YEp24, YEp24-SCD5, or a CHC1 complementing plasmid (YCp50-CHC1) and tested for suppression of the chc1 growth and receptor-mediated endocytosis defects. SCD5 overexpression suppressed the temperature sensitivity of the viable chc1 strain quite well at 34°C, although not at 37°C (Figure 5A). However, the slowed -factor internalization seen in chc1 mutants was minimally affected by overexpression of SCD5 (Figure 5B), even at 30°C, a permissive growth temperature.

View larger version (40K): [in this window] [in a new window]   Figure 5.   Overexpression of SCD5 partially suppresses the temperature-sensitive growth phenotype, but not the endocytic defect of chc1 scd1-v cells. (A) A chc1 scd1-v strain (SL3593) transformed with YEp24, YEp24-SCD5, or YCp50-CHC1 was streaked onto YEPD and incubated for 3 d at 30, 34, or 37°C. (B) Cells were grown to log phase in selective medium (C-URA). Radiolabeled -factor was prebound to cells at 4°C in YEPD. Cells were pelleted, and then endocytosis was initiated by resuspending in prewarmed YEPD at 30°C. Samples were taken at the indicated times and cells were processed to determine percentage of -factor internalized. Strains are chc1 scd1-v (SL3593) transformed with YEp24 (), YEp24-SCD5 () or YCp50-CHC1 ().

Because scd5-338 causes actin defects, we next examined whether clathrin deficiency affects actin organization. We made use of a strain expressing CHC1 under control of the repressible GAL1 promoter as its only source of clathrin HC. GAL1::CHC1 cells grown in galactose medium showed the normal pattern of actin localization by phalloidin staining (Figure 6A). Shifting to glucose for 15 h, which completely depletes Chc1p (Nelson and Lemmon, 1993), caused accumulation of larger actin patches. These large patches were usually delocalized to the mother cell during bud growth and were observed in 70% of the cells (Figure 6, B and I). The clathrin-depleted cells also displayed fewer actin patches at the mother-daughter neck during cytokinesis, when compared with cells expressing clathrin HC (Figure 6B). Furthermore, 35% of HC-depleted cells contained actin bars, whereas none of the cells grown on galactose contained G-actin bars (Figure 6, E, G, and I). Similar actin phenotypes were seen in chc1 and clathrin LC-deficient (clc1) strains.

View larger version (40K): [in this window] [in a new window]   Figure 6.   Overexpression of SCD5 partially suppresses the actin organization defect of clathrin-deficient yeast. A GAL1:: CHC1 strain (SL554) was transformed with YEp24 (A, B, E, and G) or YEp24-SCD5 (C, D, F, and H). Precultures were grown on selective C-URA + 2% galactose. Then cells were washed and resuspended in medium containing galactose (A, C, E, and F) or glucose to deplete clathrin HC (B, D, G and H) for 15 h at 30°C. Cells were then fixed and stained with Alexa-568 phalloidin (A-D) or anti-actin antibodies (E-H). Bar, 5 µm. (I) Graph showing the percentage of cells from the above cultures with large patches or actin bars.

We next tested whether overexpression of SCD5 could suppress the actin organization defect of Chc1p-depleted cells. On galactose medium GAL1:CHC1 cells transformed with YEp-SCD5 had normal actin organization (Figure 6, C and F), although during cytokinesis more depolarized patches were observed in the mother and daughter cells (see Figure 6C, 2nd row, 3rd cell). However, in total, <10% of cells with YEpSCD5 had depolarized patches when grown on galactose, and no large patches or bars were observed (Figure 6F). On glucose medium the number of Chc cells with large patches was significantly reduced by overexpression of SCD5 (Figure 6, D and I). Thus, extra Scd5p prevented aggregation of cortical actin. However, these normal size patches were still delocalized in the majority (83%) of the clathrin-deficient cells (Figure 6D). The actin bar defect of the Chc cells was also partially suppressed by increased Scd5p (Figure 6, H and I). In addition, SCD5 overexpression partially suppressed the enlarged size and morphological defects observed in chc1 and clc1 mutants.

Genetic Interactions of scd5-338 and chc1 Mutations

Because clathrin-deficient yeast have actin phenotypes and overexpression of SCD5 partially rescues this defect, we examined whether mutations in CHC1 and SCD5 cause synthetic phenotypes. We made use of a chc1-ts mutant (chc1-521), which exhibits TGN sorting and endocytic defects at 37°C, but is normal at 25°C (Seeger and Payne, 1992a, 1992b; Tan et al., 1993). The chc1-ts mutant was crossed to a scd5-338 strain, and haploid progeny from tetrad dissection were analyzed. We found that although both the chc1-ts and scd5-338 spore progeny grew up to 32 or 34°C, the chc1-ts scd5-338 double mutants grew very poorly at 25°C and were completely inviable at 30°C or higher (Figure 7A). Also, fluid-phase endocytosis of LY in the double mutant was almost completely blocked at 25°C (only 7% of cells had vacuolar staining), although 98% of the chc1-ts and 39% of the scd5-338 single mutants internalized the dye at this temperature (Figure 7B). Furthermore, although the chc1-ts mutant showed no actin phenotype at 25°C, it exacerbated the scd5-338 cortical actin defect causing appearance of more and larger actin patches than were observed in the scd5 mutant alone (Figure 7C). There were no large patches in the chc1-ts mutant and <10% of scd5-338 mutant had large patches, whereas 50% of the double mutant cells contained large patches of F-actin.

View larger version (86K): [in this window] [in a new window]   Figure 7.   Synthetic phenotypes are caused by combining scd5-338 and chc1-521. (A) Growth of scd5-338 chc1-521 double mutant. SL4182 (scd5-338) was crossed to SL2726 (chc1-521) and subjected to tetrad analysis. Resulting spore clones were streaked onto YEPD plates and incubated for 3 d at 25, 32, and 37°C. An example of a tetratype tetrad is shown. (B) Lucifer yellow uptake defect in scd5-338 chc1-521 mutant. Cells were grown to log phase in liquid YEPD at 25°C and incubated for 60 min with LY at 25°C. Top panels: LY fluorescence; bottom panels: location of vacuoles revealed by DIC imaging. Strains are as follows: (a) scd5-338, SL3740; (b) chc1-521, SL2726; and (c) scd5-338 chc1-521, SL4304. Bar, 5 µm. (C) Actin phenotype is exacerbated by combining scd5-338 with chc1-521. Cells were grown to log phase at 25°C in liquid YEPD, stained with Alexa-568-phalloidin, and photographed as described in MATERIALS AND METHODS. Strains are the same as shown in B.

Clathrin and Scd5p Interact with Sla2p by Two-hybrid Analysis

A yeast two-hybrid screen was carried out to identify clathrin LC interacting proteins (see MATERIALS AND METHODS). Thirteen positive clones were obtained. Two of the clones contained C-terminal fragments of Chc1p including the LC binding domain (Liu et al., 1995), as would be expected. The other 11 clones (representing five distinct isolates) contained various fragments of Sla2p coding sequences. Figure 8A shows two-hybrid interaction results with two of the LC interacting Sla2p clones. The smallest Sla2p prey coded for amino acids 292-520, which includes part of the predicted central coil-coiled domain as well as the glutamine-rich region, the leucine zipper, and a portion of the proline-rich region (Figure 8B; Wesp et al., 1997; Yang et al., 1999). A slightly smaller fragment (residues 292-501) also interacted with LC (Figure 8A). Interestingly, similar regions in the related mammalian proteins, HIP1 and HIP1R, were recently shown to interact with clathrin (Engqvist-Goldstein et al., 2001; Metzler et al., 2001; Mishra et al., 2001; Waelter et al., 2001). Chc1p-655-1653, which binds LC, also interacted with Sla2p-292-501. However, this association was somewhat weaker than the interaction of LC with HC or LC with the Sla2p coil domain, as the pGAD-CHC1-655-1653/pGBD-SLA2-292-501 transformant grew consistently more slowly on the C-adenine reporter plates. This suggests that the interaction of HC with Sla2p might be through LC. In addition, confirming previous studies (Wesp et al., 1997; Yang et al., 1999), we found that the coil-coiled domain of Sla2p can interact with itself (Figure 8A).

View larger version (49K): [in this window] [in a new window]   Figure 8.   Clathrin and Scd5p interact with Sla2p by two-hybrid analysis. (A) Yeast strain YPJ96-4 was transformed with GAL4 binding domain (GBD) bait plasmids pGBDU (empty), pKH19 (GBD-CLC1), or pKH47 (GBD-SLA2-292-501) in combination with GAL4 activation domain (GAD) prey plasmids pGAD (empty), pKH24 (GAD-CHC1-655-1653), pKH19 (GAD-CLC1), p31-10 (GAD-SLA2-292-968), or p31-2 (GAD-SLA2-292-520). Equal numbers of cells from transformants were spotted on complete synthetic medium lacking leucine and uracil (C-LEU-URA) and medium lacking adenine, leucine, and uracil (C-ADE-LEU-URA) and grown for 3 and 5 d. Interactions of baits and preys were scored for activation of the GAL2:: ADE2 reporter by growth on C-ADE-LEU-URA. (B) Model of Sla2p. Sla2p domains and region that associates with clathrin are indicated. (C) YPJ96-4 was transformed with GAL4 binding domain bait plasmids pGBDU (empty) or pNT1 (GBD-scd5-338). SL3004 was transformed with GAL activation domain (GAD) plasmids pGAD (empty), pKH24 (GAD-CHC1-655-1653), pKH19 (GAD-CLC1), p31-10 (GAD-SLA2-292-968). Bait and prey plasmids were combined by the mating method. Cells were spotted on plates as described in A and grown for 3 d.

Genetic interactions were also observed between clathrin mutations and sla2, supporting the two-hybrid physical association results. Crosses of chc1 mutants to sla2 mutants resulted in strong synergistic growth phenotypes in double mutant spore progeny (Figure 9). In fact, chc1 was synthetically lethal with sla2 (Figure 9A). Similar results were observed when clc1 was combined with sla2 mutations. Overall these results suggest that clathrin and Sla2p function in related pathways.

View larger version (52K): [in this window] [in a new window]   Figure 9.   Clathrin mutants display negative synergistic growth defects with sla2. (A) sla2 and chc1 are synthetically lethal. SL4162 (sla2) was crossed to SL13 (chc1 scd1-v YCp50-CHC1). The CHC1 plasmid was dropped, and the diploid was sporulated and dissected for tetrad analysis. An example of a tetratype tetrad is shown. (B) Tetrads were dissected from SL13 (chc1 scd1-v YCp50-CHC1) X SL4049 (sla2-41) that had lost YCp50-CHC1 or SL2726 (chc1-521) X SL4049. Overnight cultures of spore segregants were diluted to 107 cells/ml, and serial fourfold dilutions were made in YEPD. Equal volumes of cells were spotted on YEPD and plates were grown for 2 d at the indicated temperatures.

Because scd5-338 shows synthetic lethality with sla2-41 (end4-1) (Table 3) and strong genetic interactions with chc1-521 (Figure 7), we tested whether Scd5p interacts with clathrin or Sla2p by two-hybrid analysis. When full-length Scd5p was fused with the GAL4 DNA binding domain (GBD), the fusion protein activated reporter genes even in the absence of the GAL4 activation domain (GAD) preys. However, the GBD-Scd5-338 truncation only weakly autoactivated, allowing tests for interaction with Sla2p and clathrin preys. We found that Scd5-338p could not interact with full-length LC or Chc1p-655-1653 (Figure 8C). However, a two-hybrid interaction was observed with the longer Sla2p-292-968 clone (Figure 8C) but not the Sla2p 292-520 fragment. Thus, the N-terminal region of Scd5p appears to associate with Sla2p, but this interaction depends on sequences C-terminal to the minimal region that binds clathrin LC.

Clathrin and Scd5p Are Required for Normal Sla2p Localization

In wild-type cells, Sla2p partially colocalizes with cortical actin patches (Yang et al., 1999). This is most easily observed in unbudded and small budded cells where actin is highly polarized. Interestingly, not all actin patches stain with anti-Sla2p nor does every Sla2p patch have associated actin (Yang et al., 1999; Figure 10, A and B). Because Scd5p and Chc1p interact with Sla2p by two-hybrid analysis, we tested whether Sla2p localization is affected in scd5-338 or Chc1p-depleted cells. In scd5-338 cells grown at 25°C, Sla2p was partially delocalized to the cytoplasm, but there was still some residual cortical staining of Sla2p (Figure 10, C and D). This remaining cortical Sla2p was much less polarized than in wild-type cells, and less was found in actin patches (Figure 10, C, D, and M). After a 3-h shift to 34°C, nearly all of the Sla2p was cytoplasmic, although faint cortical rim staining was observed in some scd5-338 cells (Figure 10F). However, these small patches at the cell periphery rarely colocalized with actin (Figure 10M). When GAL1::CHC1 cells were depleted of clathrin by growth on glucose, Sla2p was completely cytoplasmic in most cells (Figure 10, H and M), whereas cells grown on galactose showed the normal Sla2p staining pattern.

View larger version (80K): [in this window] [in a new window]   Figure 10.   Sla2p is mislocalized in scd5-338 and clathrin-depleted cells. Wild-type (SL1528; A and B) and scd5-338 (SL3920; C-F) strains were grown on YEPD to log phase at 25°C (C and D) or shifted to 34°C (A, B, E, F) for 3 h. The GAL1:: CHC1 strain (SL554; G and H) was grown to log phase in YEP-GAL and then shifted to YEPD for 15 h at 30°C to deplete clathrin HC. In I and J scd5-338 (SL3920) + pJC4 [pGFP-ABP1] was grown in C-URA and shifted to 34°C for 3 h. In K and L the GAL1: CHC1 strain (SL554) + pJC4 [pGFP-ABP1] was grown in C-URA+GAL and then shifted to glucose medium for 15 h to deplete clathrin HC. Cells were fixed, stained with anti-actin (A, C, E, and G) and anti-Sla2p antibodies (B, D, F, and H); or in cells for GFP-Abp1p localization (J and L) actin was visualized with Alexa-594 phalloidin (I and K). (M) Quantification of the percent of cells with Sla2p or Abp1p showing colocalization with actin patches. Note that the GAL1::CHC1 strain grown on galactose gave similar results as wild-type cells. Bar, 5 µm.

We also localized another cortical actin patch protein, Abp1p (Drubin et al., 1988), which was tagged with GFP. Interestingly, in the majority of scd5-338 cells grown at either 25 or 34°C, as well as cells depleted of clathrin, GFP-Abp1p was still associated with actin patches (Figure 10, I-M). Similar results were obtained with Rvs167p. Thus, loss of Scd5p or clathrin function specifically affects Sla2p and does not cause a general effect on all cortical patch proteins.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Scd5p Plays a Role in Actin Organization and Endocytosis

In this study we found that Scd5p is crucial for normal actin organization and endocytosis. Both fluid-phase and receptor-mediated endocytosis were completely blocked in the scd5-338 mutant at the nonpermissive temperature, which is a phenotype caused by many mutations that affect the cortical actin cytoskeleton. In these cells cortical actin patches were often larger or depolarized to the mother cell, and a major accumulation of G-actin in bar structures was observed, suggesting a defect in actin assembly or destabilization of F-actin. Other phenotypes typical of mutants with actin defects were observed in scd5-338 (Pruyne and Bretscher, 2000a, 2000b), including delocalized chitin deposition, bud site selection defects, thickened cell walls, and the presence of multibudded and multinucleated cells. Supporting the phenotypic analysis, Scd5p colocalized with actin patches in the cell.

In our previous work on SCD5 we showed that the scd5-338 mutation caused a partial post-Golgi secretory defect at the nonpermissive temperature (Nelson et al., 1996). Based on the evidence presented here, this secretory defect could be due to the effect of loss of Scd5p function on the actin cytoskeleton and/or endocytosis. Several other actin patch mutants display similar partial secretory defects or accumulation of secretory vesicles, including act1-1, myo3 myo5, sla2, rvs161, and rvs167 (Novick and Botstein, 1985; Goodson et al., 1996; Mulholland et al., 1997; Breton et al., 2001). Budding continues and secretory vesicles still polarize to growth sites in many of these patch mutants, including scd5-338 (K. Henry, unpublished results). It has been suggested that a defect in endocytosis may affect recycling of components (e.g., v-SNAREs) required for assembly of mature fusion-competent secretory vesicles (Pruyne and Bretscher, 2000a). However, we noted that the scd5 mutant had cables that were often misoriented within the cell, so the secretion phenotypes could be due to a partial effect on polarization of actin cables. This polarization defect might be caused indirectly by effects on cortical actin patches, which may contribute to anchoring of cables in the bud. Alternatively endocytic defects might affect localization of important membrane proteins that provide cues needed for cable polarization. A more complex function for Scd5p is suggested by the scd5-338 budding pattern defect. Unlike most cortical patch mutants, which show defects only in the diploid bipolar budding pattern (Yang et al., 1997; Pruyne and Bretscher, 2000a), scd5-338 was defective in both haploid and diploid specific budding. Also SCD5 is an essential gene, whereas many cortical actin patch components that affect endocytosis are not. This suggests that Scd5p could play a broader role in actin organization and cell polarization or may function in other cellular processes unrelated to actin.

scd5-338 Shows Genetic Interactions with Mutations in Genes Encoding Other Cortical Actin Patch Components

We found that the scd5-338 mutation caused more severe growth and actin organization phenotypes when combined with a number of mutations in cortical actin patch components, supporting the idea that Scd5p is important for cortical actin function. Interestingly, scd5-338 was synthetically lethal with pan1-20 and end3 but phenotypes were not exacerbated with sla1, even although Pan1p, End3p, and Sla1p exist in a complex (Tang et al., 2000; Zeng et al., 2001). This difference could be explained by the fact that pan1 and end3 mutants have severe endocytic phenotypes (Bénédetti et al., 1994; Tang et al., 1997; Duncan et al., 2001), whereas sla1 is only modestly affected (Ayscough et al., 1999; K. Henry, unpublished results). Because interaction of End3p with Pan1p is not dependent on the presence of Sla1p (Tang et al., 2000), perhaps Pan1p and End3p association are most critical for function of this complex. Also, END3, but not SLA1, overexpression rescues the temperature sensitivity of pan1-4, and overexpression of SLA1 prevents END3 overexpression from rescuing the pan1 mutant at 37°C (Tang et al., 2000). Thus, Sla1p may act antagonistically to End3p, which could explain the difference in genetic interactions with scd5-338.

scd5-338 had a strong genetic interaction with abp1 and rvs167, but not srv2. Interestingly both Srv2p (adenylyl cyclase-associated protein) and Rvs167p interact with Abp1p (Freeman et al., 1996; Lila and Drubin, 1997; Drees et al., 2001; Fazi et al., 2001). However, rvs167, but not srv2, shows a pattern of genetic interactions similar to abp1 when combined with other mutations in genes encoding actin cytoskeleton components (Lila and Drubin, 1997). Thus, the results with scd5-338 further support the idea that Abp1p and Rvs167p functions are closely related.

Sla2p Localization Is Dependent on Scd5p

Scd5p is also physically associated with components of the cortical actin cytoskeleton. In addition to showing overlapping localization with cortical actin patches, Scd5p interacts with Sla2p by two-hybrid analysis. In preliminary studies, we find that it also interacts with Rvs167p, but not Rvs161p, by two-hybrid screening (J. Chang, unpublished results). Thus far we have not been able to demonstrate association of Sla2p and Scd5p by coprecipitation, but their interaction may be transient or only occur in the context of the plasma membrane or when assembled with cortical actin structures. However, based on the synthetic lethality caused by combining sla2-41 (end4-1) with scd5-338, we believe the physical interactions detected by two-hybrid analysis are functionally important. As further support for this, we found that Sla2p dissociated from cortical structures and accumulated in the cytoplasm in scd5-338 cells. The delocalization of Sla2p seen in scd5-338 appears to be specific, because other actin patch-associated components, Abp1p and Rvs167p, localize in cortical sites relatively normally in these cells. Also, scd5-338 cells still have phalloidin-staining actin patch-like structures, indicating the effect of scd5-338 on Sla2p localization is not merely caused by actin depolymerization.

It is interesting to note that although Scd5-338p associates with Sla2p by two-hybrid interaction, the same truncation mutation causes delocalization of Sla2p. Possibly the C-terminal region of Scd5p is important for its localization with cortical actin, and perturbation of Scd5p localization thereby affects Sla2p's association with actin patches.

Scd5p has also been shown to interact with Glc7p (Tu et al., 1996; Uetz et al., 2000; Venturi et al., 2000; JiSuk Chang, unpublished results), which is yeast PP1, a broad specificity Ser/Thr phosphatase with many cellular functions (Stark, 1996). PP1's activity is restricted toward physiological substrates in vivo by different regulatory or targeting subunits (Stark, 1996), so it is possible that Scd5p targets PP1 to dephosphorylate specific actin-associated components. It could play a major role in regulating assembly, disassembly, or specific organization of actin-associated factors needed for endocytosis. Supporting this idea, genetic interactions with scd5-338 were observed with mutations in a number of actin-associated proteins that are known to be phosphorylated. One of these proteins is Sla2p, which interacts with Ark1p, an actin regulating kinase (ARK; Cope et al., 1999). Pan1p and Sla1p are phosphorylated by Prk1p (another ARK), as well as Ark1p (Zeng and Cai, 1999; Zeng et al., 2001). Recent evidence indicates that Prk1p phosphorylation disrupts association of Pan1p with Sla1p and End3p (Zeng et al., 2001). Also, in ark1 prk1 double mutants cortical F-actin patches, and many patch components, including Sla2p, are found in large aggregates (Cope et al., 1999). Because we find that Sla2p dissociates from the cortex in the scd5-338 mutant, dephosphorylation of Sla2p or other patch components may be important for assembly in actin patches, and Scd5p/PP1 may regulate this.

Clathrin Is Important for Organization of Actin in Yeast

In previous studies we identified SCD5 as a multicopy suppressor of clathrin-deficient yeast, but scd5-338 cells do not show the Golgi retention defects that lead to CPY sorting and alpha factor processing phenotypes seen in clathrin mutants (Nelson et al., 1996). Also, overexpression of SCD5 could not suppress these clathrin mutant phenotypes (Nelson and Lemmon, 1993). Here we report our discovery that clathrin mutants, like scd5-338 cells, have major cortical actin defects and accumulate significant nonfilamentous G-actin. SCD5 overexpression partially suppressed these actin defects, and combining scd5-338 with a chc1-ts mutation exacerbated growth, actin and endocytic phenotypes. Furthermore, clathrin mutations genetically interact with sla2 mutations, and Sla2p association with the cortex is dependent on the presence of clathrin. Previous studies are also consistent with a connection between clathrin and actin in yeast. Clathrin-deficient yeast become polyploid at high frequency, and aberrant nuclear divisions within the mother cell are observed (Lemmon and Jones, 1987; Lemmon et al., 1990). Like scd5-338 and other actin patch mutants, clathrin mutants are enlarged and round in appearance, they have delocalized chitin deposition, and their cell walls are thickened (Lemmon and Jones, 1987; Lemmon et al., 1990; S.K. Lemmon, unpublished results). Moreover, clathrin interacts with other cortical patch components, such as Ent1/2p (Wendland et al., 1999), in addition to Sla2p. Therefore, we suggest that clathrin plays a role in cortical actin function in yeast.

Clathrin Interacts with Sla2p

Clathrin is a trimeric molecule with three HC's joined at the C terminus and a LC bound to each leg in the hub region adjacent to the triskelion vertex. Each HC contains an N-terminal globular terminal domain (TD) that interacts with a number of proteins containing a signature clathrin binding motif ("clathrin box") related to LLDLD in amphiphysin (Dell'Angelica, 2001). This interaction is thought to facilitate the recruitment of clathrin and many of its associated factors to sites of clathrin-mediated transport. The yeast epsins (Ent1/2p) and AP180-like proteins (Yap1801p/Yap1802p) both bind to the Chc1p TD via this type of clathrin binding motif (Wendland and Emr, 1998; Wendland et al., 1999). Interestingly, we identified Sla2p in a two-hybrid screen with the clathrin LC. Although Sla2p has some sequences distantly related to the LLDLD motif, we found that the clathrin LC interacted with a portion of the Sla2p central coiled-coil domain, which does not contain potential clathrin TD-binding sequences. A fragment of clathrin HC including the LC binding region, but not the TD, also interacted with Sla2p, albeit more weakly than LC. In preliminary studies we find that LC interacts with Sla2p by two-hybrid analysis in a chc1 strain, indicating that this association is not dependent on HC (T. Newpher, unpublished results). We note that, because Sla2p dimerizes via the central coiled-coil domain (Wesp et al., 1997; Yang et al., 1999; this study), the two-hybrid coiled-coil product could associate with endogenous Sla2p. Thus, we cannot yet exclude the possibility that the LC interacts with another portion of Sla2p or that clathrin HC interacts with other Sla2p regions, possibly via the TD. However, consistent with an importance for the LC-Sla2p coiled-coil interaction, deletion of the Sla2p central coiled-coil domain causes defects in endocytosis and actin organization that resemble those of clathrin-deficient cells (Wesp et al., 1997; Yang et al., 1999).

Recent work has shown that the mammalian Sla2p-related proteins, HIP1 and HIP1R, localize to sites of clathrin-mediated endocytosis in vivo, are enriched in clathrin-coated vesicles, and bind directly to clathrin (Engqvist-Goldstein et al., 1999, 2001; Metzler et al., 2001; Mishra et al., 2001; Waelter et al., 2001). Additional studies indicate that that HIP1R may link the actin cytoskeleton to coated pits and suggest that cortical actin helps organize or isolate regions of active clathrin-mediated internalization (Gaidarov et al., 1999; Bennett et al., 2001; Engqvist-Goldstein et al., 2001). Thus, HIP1(R) may be important for this organization of coated-pits or may be actively involved in the endocytic process itself.

The binding of clathrin to HIP1 and HIP1R appear to be distinct. In HIP1, there are clathrin N-TD-binding sequences as well as AP-2 binding motifs (DPF and FXDXF) adjacent to the central coiled-coil region (Metzler et al., 2001; Mishra et al., 2001; Waelter et al., 2001). In HIP1R the binding to clathrin appears to occur principally via the coil-coiled region and involves a direct interaction with clathrin LC (Engqvist-Goldstein et al., 2001; Legendre-Guillemin et al., 2002), similar to our findings of Sla2p and yeast LC binding. In addition, overexpression of a hub fragment of clathrin, which contains the LC binding site and inhibits clathrin-mediated endocytosis, causes dissociation of HIP1R from coated pits in HeLa cells (Bennett et al., 2001). Similarly, we found that depletion of clathrin causes dissociation of Sla2p from the cell cortex. Overall this indicates there are close connections between clathrin, Sla2p-related proteins and the actin cytoskeleton in all eukaryotes.

Why then might clathrin deficiency in yeast only cause a partial defect in endocytosis? Clathrin may help collect specific membrane proteins, localize regulatory factors and or mark/domains for endocytosis, but not be absolutely required for generating force for internalization, which may be primarily an actin-mediated process in yeast. Clathrin may be diffusely distributed in the cortical actin network, which would explain why surface-coated pits have yet to be detected by microscopic methods. Possibly, classic surface-derived CCV do not even form in yeast, although some plasma membrane proteins have been detected in enriched CCV fractions (Pishvaee et al., 2000). Still, it has not been ruled out that these vesicles are of endosomal origin or on a recycling trajectory (Valdivia et al., 2002). Recent work also indicates that clathrin may be important for formation of a subclass of secretory vesicles, so CCV could contain newly synthesized plasma membrane proteins (Gurunathan et al., 2002; Harsay and Schekman, 2002). Nonetheless, our findings suggest that the endocytic phenotype of clathrin-deficient cells may relate to a role for clathrin in actin-associated processes. This effect may only be partial because clathrin is one of several factors that facilitate organization or stabilization of the cortical actin network. Interestingly, many homologues of mammalian clathrin associated factors that are involved in endocytosis in yeast (e.g., Ent1/2p [epsins], Pan1p [Eps15-related], and Rvs161 and Rvs167 [amphiphysins]), are also important for actin organization (Munn et al., 1995; Sivadon et al., 1995; Tang and Cai, 1996; Wendland et al., 1996; Balguerie et al., 1999; Wendland et al., 1999).

A final question concerns why SCD5 was isolated as a multicopy suppressor of clathrin-deficient lethality. Mutations in both clathrin and SCD5 show genetic interactions with sla2 and both clathrin and Scd5p are required for localization of Sla2p at the cell cortex. Our data suggest that Scd5p, as a potential targeting subunit for PP1, may promote the recruitment of proteins, like Sla2p, to the cortical actin network. Thus, overexpression of SCD5 could counter the destabilization of cortical actin components that might occur in the absence of clathrin. Further studies will be directed toward determining whether Scd5p plays a regulatory role in actin organization and endocytosis and determining what are its targets.