The Sla2p Talin Domain Plays a Role in Endocytosis in Saccharomyces cerevisiae

Corresponding author: Beverly Wendland, Mudd Hall, Room 35, 3400 N. Charles St., The Johns Hopkins University, Baltimore, MD 21218., bwendland{at}jhu.edu (E-mail)

Communicating editor: T. STEARNS

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Clathrin-binding adaptors play critical roles for endocytosis in multicellular organisms, but their roles in budding yeast have remained unclear. To address this question, we created a quadruple mutant yeast strain lacking the genes encoding the candidate clathrin adaptors Yap1801p, Yap1802p, and Ent2p and containing a truncated version of Ent1p, Ent1CBMp, missing its clathrin-binding motif. This strain was viable and competent for endocytosis, suggesting the existence of other redundant adaptor-like factors. To identify these factors, we mutagenized the quadruple clathrin adaptor mutant strain and selected cells that were viable in the presence of full-length Ent1p, but inviable with only Ent1CBMp; these strains were named Rcb (requires clathrin binding). One mutant strain, rcb432, contained a mutation in SLA2 that resulted in lower levels of a truncated protein lacking the F-actin binding talin homology domain. Analyses of this sla2 mutant showed that the talin homology domain is required for endocytosis at elevated temperature, that SLA2 exhibits genetic interactions with both ENT1 and ENT2, and that the clathrin adaptors and Sla2p together regulate the actin cytoskeleton and revealed conditions under which Yap1801p and Yap1802p contribute to viability. Together, our data support the view that Sla2p is an adaptor that links actin to clathrin and endocytosis.

PLASMA membrane lipids and proteins, as well as contents of the extracellular space, are internalized by the formation of vesicles budding into the cytoplasm from the plasma membrane. This process, known as endocytosis, is vital to all cells for growth, signaling, and nutrient uptake. Although several endocytic pathways exist, the clathrin-dependent pathway is the best characterized. Clathrin is a heterohexameric coat protein that is believed to drive curvature of the membrane to form a vesicle during both endocytic and secretory pathway events (PEARSE 1976 ; UNGEWICKELL and BRANTON 1981 ). In the model system Saccharomyces cerevisiae, most strains can survive in the absence of clathrin (PAYNE and SCHEKMAN 1985 ; PAYNE et al. 1987 ); however, these strains are very unhealthy and exhibit reduced levels of endocytosis (PAYNE et al. 1988 ). Internalization in the absence of clathrin led to a debate about its role in endocytosis, but temperature-sensitive clathrin mutants exhibited a rapid onset of endocytosis defects, indicating that loss of clathrin function results in a primary endocytosis defect (TAN et al. 1993 ). This rapid but partial reduction in the internalization of specific receptor proteins when conditional clathrin mutants are shifted to nonpermissive conditions implies the presence of a clathrin-independent pathway that either shares cargo with or is capable of compensating for loss of the clathrin-dependent pathway (reviewed in BAGGETT and WENDLAND 2001 ). This view is consistent with studies in mammalian cells in which an acute block of the clathrin pathway caused an immediate compensatory upregulation of the clathrin-independent fluid phase uptake pathway (DAMKE et al. 1995 ). Recent studies have focused on clarifying the role of clathrin function in yeast endocytosis as well as identifying the machinery and adaptors acting in all pathways of yeast endocytosis.

Adaptor proteins, which select the protein cargo to be included in internalizing vesicles and also recruit endocytic machinery, have been the subject of many recent studies (reviewed in SANTOLINI et al. 1999 ). Receptor proteins are selected for endocytosis by sequences in their cytosolic tail that are either recognized by adaptor proteins directly (TAN et al. 1996 ; HOWARD et al. 2002 ) or modified by kinases and ubiquitin ligases to allow recognition (HICKE 1999 ; ROTH and DAVIS 2000 ; SHIH et al. 2000 ). It has been modeled that these adaptors, once bound to receptors, recruit proteins such as clathrin, as well as regulators of the actin cytoskeleton, to drive invagination of the plasma membrane, which culminates in vesicle scission (reviewed in SCHMID 1997 ). Many of the proteins involved in this process have been found by a combination of genetic selections, yeast two-hybrid screens, and sequence homology with known mammalian or yeast endocytic factors.

One conserved family of endocytosis proteins is the epsins, which are thought to be adaptors. They contain a conserved structural domain at their N terminus called ENTH (epsin N-terminal homology), which binds phosphatidylinositol-4,5-bisphosphate [PI(4,5)P₂; ITOH et al. 2001 ]. The yeast proteins Ent1p and Ent2p are homologs of mammalian epsin and bind PI(4,5)P₂ (AGUILAR et al. 2003 ) and are also suggested to act as adaptors (WENDLAND 2002 ). Ent1p and Ent2p contain identical C-terminal clathrin-binding motifs (CBMs), and Ent1p binds the terminal domain of clathrin through this motif (WENDLAND et al. 1999 ; AGUILAR et al. 2003 ). Deleting both ENT1 and ENT2 is lethal, indicating that the corresponding proteins play an essential role in yeast viability that may overlap with or be distinct from their endocytic function (WENDLAND et al. 1999 ). The yeast epsins also contain two ubiquitin interaction motifs (UIMs) that bind ubiquitin and are proposed to select ubiquitinated receptors for endocytosis (SHIH et al. 2002 ; AGUILAR et al. 2003 ).

Two other potential endocytic proteins, Yap1801p and Yap1802p, are highly homologous to the mammalian clathrin assembly factors CALM/AP180 (WENDLAND and EMR 1998 ; WENDLAND et al. 1999 ). These proteins share similar N-terminal ANTH domains that are structurally related to ENTH domains (FORD et al. 2002 ). Yap1801p and Yap1802p share additional similar domain structures with the yeast epsins, including the C-terminal CBM, but do not contain any UIMs. In contrast to the yeast epsins, deletion and mutational analyses have thus far not shown a function for Yap1801/2p in yeast cells (WENDLAND and EMR 1998 ; HUANG et al. 1999 ).

An important actin and endocytic protein, Sla2p, also contains an N-terminal structural domain similar to the ENTH domain. Unlike the four proteins described above, however, Sla2p was originally isolated as an actin-regulatory protein in a synthetic lethal with abp1 screen (HOLTZMAN et al. 1993 ). Alleles of SLA2 were also found in an early endocytosis screen (end4-1/sla2-41; RATHS et al. 1993 ) and in a screen for stabilization of the plasma membrane H⁺-ATPase Pma1p (mop2-1; NA et al. 1995 ). In addition to the ANTH domain at its N terminus, Sla2p contains a central coiled-coil region and a talin homology domain at its C terminus. The central coiled-coil region was recently shown by a yeast two-hybrid screen to bind the clathrin light chain and, to a lesser extent, the clathrin heavy chain (HENRY et al. 2002 ). The C-terminal talin homology domain binds F-actin in vitro (MCCANN and CRAIG 1997 , MCCANN and CRAIG 1999 ), but an in vivo function for this domain has yet to be shown. sla2 strains are viable, but are temperature sensitive for growth and have both endocytosis and actin cytoskeletal defects (WESP et al. 1997 ; YANG et al. 1999 ). In this study, we isolated an sla2 mutation that results in a truncation within the talin homology domain. This allele is temperature sensitive for endocytosis, but not for growth, indicating that the talin homology domain plays an important role in endocytosis at elevated temperatures. In addition, the genetic background used to isolate this allele, in which ENT1 is mutated and ENT2, YAP1801, and YAP1802 are deleted, allowed us to uncover a role for the yeast AP180 homologs Yap1801p and Yap1802p in yeast viability. Finally, we report a genetic interaction between SLA2 and either ENT1 or ENT2 that is important for viability. Our data support previous findings that Sla2p is important for both endocytosis and actin regulation and provide the first evidence for an in vivo role for the talin homology domain in endocytosis as well as an important genetic interaction between the yeast epsins and SLA2.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Media and materials:
Rich medium for yeast strain growth is standard yeast extract-peptone-dextrose (YPD). Synthetic medium contains yeast nitrogenous base plus dextrose and is supplemented with appropriate amino acids to maintain plasmid selection. 5-Fluoroorotic acid (5-FOA) was used at 750 µg/ml in synthetic medium. Bacterial strains were maintained on standard media containing 50 µg/ml carbenicillin (US Biologicals, Swampscott, MA), 30 µg/ml kanamycin, or 34 µg/ml chloramphenicol as appropriate to maintain plasmids. Materials were purchased from Fisher Scientific (Pittsburgh) or Sigma (St. Louis) unless otherwise noted.

Yeast strains and plasmids:
Table 1 and Table 2 list yeast strains and plasmids used in this study, respectively. DNA deletions and manipulations were performed using standard techniques. JRY41 was derived by sporulation and tetrad dissection of the diploid DDY1194 (YANG et al. 1999 ), provided by the Drubin lab (UC Berkeley). JRY43 was created from JRY41 by transformation with a PCR fragment containing homologous untranslated regions of ENT2 surrounding HIS3 and selection for HIS3+ cells. Deletion of ENT2 was confirmed by PCR and Western analysis. Integration of ent1CBM at the ENT1 locus in ent1::HIS3 ent2::HIS3 + pENT2 cells was achieved using an integrating LEU2 vector, linearized by restriction digest downstream of the ent1CBM stop codon. To create the ent1CBM triple- strain, cells that had integrated ent1CBM were selected and confirmed by PCR and Western analysis and then crossed to ent2::HIS3 yap1801::HIS3 yap1802::LEU2 cells. The quadruple mutant ent1::HIS3::ent1CBM::LEU2 ent2::HIS3 yap1801::HIS3 yap1802::LEU2 was selected by tetrad analysis and Western blotting. BWY1413 was generated by four sequential backcrosses of SEY6210 to end4-1 (RATHS et al. 1993 ). Plasmids were created by standard techniques. pBW482 was generated by PCR of genomic DNA from JRY31, followed by digestion and subcloning into pBW384. pRSET-IC2-N237, containing a His₆-tagged fragment of dynein intermediate chain was provided by T. Schroer (KING et al. 2003 ) for use as a negative control for clathrin binding. To introduce plasmids, standard yeast transformations were performed.

Growth of yeast cells in serial dilution:
Yeast cells were grown in liquid culture to mid-log phase. Cells were then serially diluted to 0.25, 0.05, 0.01, and 0.002 OD₆₀₀/ml. The diluted cells were transferred to solid media using a pronged "frogger" or replicator tool and grown for 1–5 days at the temperatures shown.

Isolation of Rcb mutants and treatment of rcb432 and rcb243:
The strains JRY18 (MATa) and JRY19 (MAT) were mutated with ethyl methanesulfonate (EMS) to 30% viability. We screened a total of 25,000 colonies, and Rcb (requires clathrin binding) mutants were identified as dead colonies on 5-FOA media by replica plating. All colonies selected were retested by streaking as well as by plating serially diluted cells. rcb432 was backcrossed at least twice to JRY18 or JRY19 using standard techniques to determine linkage of a single mutation to the phenotypes. Outcrossing to SEY6210 was also performed to obtain the rcb432 mutation, sla2-432, alone and in the background of ent1 or ent2. Complementation cloning was performed using a CEN, TRP1 genomic library from C. CONNELLY and P. HIETER (unpublished data). To confirm linkage between sla2-432 and SLA2, we tagged the SLA2 locus with TRP1 in ent2::HIS3 cells and mated them to sla2-432 ent2::HIS3 cells (JRY31). Tetrad analysis of this mating showed 100% segregation of SLA2::TRP1 away from sla2-432 (identified by temperature-sensitive growth in the ent2 background). To sequence the sla2-432 allele, genomic DNA was isolated and PCR was used to amplify the SLA2 gene. Sequencing was performed in triplicate using overlapping primers and compared to wild-type SLA2 sequence obtained by the same methods.

rcb243 cells were found to be in the same complementation group as rcb432 by mating the two mutant strains and testing for inviability on 5-FOA media. However, Western blot analysis showed that rcb243 cells produce full-length Sla2p, and transformation with pSLA2 only partially complemented the Rcb phenotype of rcb243 cells. This could be due to a dominant effect of the sla2 allele found in rcb243 cells or to a mutation in a gene other than SLA2 that results in nonallelic noncomplementation.

Trichloro-acetic acid precipitated yeast cell extracts:
Yeast cells were grown to mid-log phase and equal numbers of cells were harvested by centrifugation. Cells were resuspended in 10% trichloro-acetic acid containing the protease inhibitor 4-(2-aminoethyl)benzenesulfonyl fluoride (AEBSF, 0.1 mM) and incubated for 20 min at 4°. Extracted proteins were pelleted by centrifugation and washed twice in acetone. Dried pellets were resuspended in Laemmli buffer with 0.1 mM AEBSF and subjected to standard glass bead disruption. Proteins were separated by SDS-PAGE and subjected to immunoblotting.

SDS-PAGE and immunoblotting:
Proteins were separated on 7.5–15% polyacrylamide minigels at 20–30 mA in running buffer (3 mM SDS, 25 mM Tris base, 192 mM glycine). Proteins were then transferred onto nitrocellulose membrane in Western transfer buffer (20% methanol, 0.0375% SDS, 48 mM Tris base, 39 mM glycine) for 90 min at 80 V, followed by blocking in 5% nonfat dried milk in TBST (0.25 M NaCl, 10 mM Tris, pH 7.5, 0.025% Tween-20). Blots were incubated with the given primary antibodies, washed in TBST, incubated with secondary antibodies conjugated to HRP obtained from Pierce (Rockford, IL) and used at 1:5000 for 30–45 min, and developed with SuperSignal West Pico chemiluminescent substrate (Pierce, Rockford, IL) for 5 min. Proteins were visualized on a Fluorchem 8000 chemiluminescence system (Alpha Innotech, San Leandro, CA). Quantification of band intensity was performed using the Fluorchem 2.0 software analysis system. Monoclonal anti-clathrin heavy chain antibody (skl1) was provided by S. Lemmon (Case Western Reserve University). Polyclonal anti-clathrin light chain serum and anti-Ste3p serum were the generous gift of G. Payne (UCLA), and polyclonal serum against the N terminus of Sla2p was provided by D. Drubin (YANG et al. 1999 ).

Clathrin-binding assay:
His₆-IC2-N237 (negative control), Ent1p, Ent1CBMp, Ent2p, and Ent2CBMp (pRSET-IC2-N237, pBW247,241,493,494) were transformed into Rosetta(DE3) cells (Novagen, Madison, WI). Bacteria were grown at 37° overnight in Luria-Bertani medium plus 30 µg/ml kanamycin, 34 µg/ml chloramphenicol, and 10 mg/ml dextrose. These cells were diluted 1:15 from saturation into superbroth plus kanamycin and chloramphenicol and grown for 3 hr at 37°. Cells were cooled and induced with 1 mM isopropyl thiogalactoside (RPI, Mt. Prospect, IL) for 5 hr at 30°. Cells were washed in 50 mM glucose plus 10 mM EDTA, and cell pellets were frozen at -20° to initiate lysing. To prepare lysates, cell pellets were thawed and resuspended in lysis buffer (20 mM Tris, pH 7.5, 1 mM EDTA, 0.4 mg/ml lysozyme, 0.2 mM AEBSF) and incubated at room temperature for 20 min. Lysates were sonicated and cell debris was removed by centrifugation. KCl and Tween-20 were added to a concentration of 150 mM and 0.2%, respectively, to the final lysate.

Lysates were incubated with nickel agarose beads (QIAGEN, Valencia, CA) for 1–2 hr at 4° to allow binding. Protein-bound beads were washed several times with phosphate-buffered saline and then tested for approximately equal concentrations of protein bound by SDS-PAGE and standard Coomassie staining. To test His₆-tagged proteins for binding to yeast proteins, protein-bound beads were incubated with yeast cytosol (prepared by glass bead lysis and high-speed centrifugation as described in WENDLAND and EMR 1998 ) for 30–90 min at 4°. Beads were subjected to centrifugation and gentle washing. The initial supernatant was marked as unbound, and the final pellet was marked as the bound fraction. Bound and unbound proteins were denatured in Laemmli buffer and run on SDS-PAGE gels for immunoblotting.

Ste3p stabilization assay:
Yeast cells were grown in liquid culture to mid-log phase and then diluted to equal concentrations. Cells were preshifted to assay temperature for 5–15 min before treatment with cycloheximide to a final concentration of 5 µg/ml at time zero. Equal volumes of cells were removed at chase timepoints of 0, 10, 20, 40, 60, and 90 min into azide/fluoride stop solution (final concentration 10 mM NaF, 10 mM NaN₃) and maintained on ice. Cell extracts were prepared by resuspension of cell pellets in Laemmli urea sample buffer and standard glass bead disruption methods. Care was taken not to heat samples above 70° to prevent aggregation of Ste3p. Extracts were run on 10% acrylamide SDS-PAGE gels and immunoblotted with polyclonal anti-Ste3p serum (pretreated with Ste3p delete extract to reduce background).

Green fluorescent protein fusion proteins:
pSTE3-green fluorescent protein (GFP) was the gift of R. Piper (URBANOWSKI and PIPER 2001 ). Transformed cells were grown in selective media to early to mid-log phase at 26° and then shifted to 37° for 30 min. Cells were subjected to gentle (300 x g) centrifugation at 4° and resuspended in phosphate-buffered saline plus 2% glucose. Microscopy was performed on a Deltavision deconvolving fluorescence microscope under an FITC filter.

Staining of filamentous actin with rhodamine-phalloidin:
Cells were grown to early to mid-log phase at 26°. Cells were then either subjected to 90 min preshift to 37° or fixed immediately. Fixation was achieved in media supplemented with 4% formaldehyde and 0.1 M potassium phosphate, pH 6.5, for 1–2 hr. Cells were sometimes additionally fixed overnight in 4% formaldehyde in 0.1 M potassium phosphate, pH 6.5. After several washes in phosphate-buffered saline (PBS), cells were permeabilized in 0.02% Triton X-100 for 15 min. Cells were washed in PBS and labeled with 30–50 µl of 0.6 µM rhodamine-phalloidin (Molecular Probes, Eugene, OR) for 1 hr to overnight. Cells were washed and resuspended in DABCO (triethylene diamine) antifade and were viewed under TRITC filters on a Deltavision deconvolving fluorescence microscope. Images were deconvolved using Deltavision software to eliminate out-of-focus light.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Characterization of a C-terminal clathrin box motif:
The mammalian LLDLD clathrin box motif binds to the terminal domain of clathrin heavy chain (RAMJAUN and MCPHERSON 1998 ; DELL'ANGELICA 2001 ). Four ENTH/ANTH domain-containing proteins in S. cerevisiae contain a similar clathrin-binding motif LDStop (, bulky hydrophobic residue, Fig 1A), with the stop codon immediately following the four-amino-acid motif as a key element that mimics the last aspartic acid of the clathrin box motif. The Ent1p and Yap1801p CBMs were shown previously to be sufficient for binding to clathrin (WENDLAND et al. 1999 ), and we were interested in the role these four CBMs may play in the recruitment of clathrin to sites of endocytosis. To determine if this is the only clathrin-binding motif present in the yeast epsins Ent1p and Ent2p, we constructed truncated forms of these proteins in which the last four amino acids constituting the entire clathrin-binding motif were deleted. The truncated proteins are referred to as Ent1CBMp and Ent2CBMp. We constructed plasmids encoding His₆-tagged versions of Ent1p, Ent1CBMp, Ent2p, and Ent2CBMp, and overexpressed the proteins in Escherichia coli. Lysates containing proteins of the expected size were incubated with nickel agarose beads. Wild-type yeast extracts were then incubated with the immobilized His₆-protein beads and washed, and then bound and unbound fractions were separated by SDS-PAGE and analyzed by immunoblotting with antibodies against clathrin heavy and light chains (Fig 1B). The bound fraction was also probed with antibodies to the His₆-tag, which confirmed that approximately equal amounts of protein were present in each lane (not shown). The results showed that the clathrin heavy chain (Chc1p) bound both wild-type Ent1p and wild-type Ent2p (Fig 1B, lanes 2 and 4), while neither Ent1CBMp nor Ent2CBMp showed any binding (Fig 1B, lanes 3 and 5). Even though their C termini are identical, Ent2p appeared to bind more clathrin than did Ent1p. Both Ent1CBMp and Ent2CBMp showed some residual clathrin light chain (Clc1p) in the bound fraction, but quantification showed this to be <1% of wild-type levels for Ent1p, while we consistently observed 3% of wild-type levels for Ent2p. These data indicate that the C-terminal CBMs of Ent1p and Ent2p are both necessary and sufficient for clathrin heavy chain binding.

View larger version (26K): In this window In a new window Download PPT slide   Figure 1. An endocytic clathrin-binding motif. (A) Schematic of the endocytic proteins Ent1p, Ent2p, Yap1801p, and Yap1802p, highlighting the C-terminal clathrin-binding motif (LD[stop], , bulky hydrophobic residue). (B) Recombinant proteins were immobilized and then incubated with yeast cytosol. Supernatant (unbound) and pellet (bound) fractions were analyzed by Western blotting for clathrin heavy chain (top) and clathrin light chain (bottom). Left to right: His6-IC2-N237 (1), His6-Ent1p (2), His6-Ent1CBMp (3), His6-Ent2p (4), and His6-Ent2CBMp (5). (C) Serial dilutions (left to right: 0.25, 0.05, 0.01, and 0.002 OD600/ml) of wild-type (SEY6210 [empty vector]) and ent1CBM triple- [pENT1] (JRY19) cells were grown in liquid culture to mid-log phase and plated in equal volumes to rich media at 26° and 37° and to 5-FOA at 26°. (D) Quantification of Ste3p degradation in cycloheximide-treated cells at 37°. , wild type (SEY6210); •, ent1CBM triple- [pENT1] (JRY19); , ent1CBM triple- (JRY46).

Phenotypes of ENT1, ENT2, YAP1801, and YAP1802 mutants:
Saccharomyces Genome Database (http://genome-www.stanford.edu/Saccharomyces) pattern match searches with the sequence [LI][LI][DE][LIM]-COOH identified the C-terminal CBM in Ent1p, Ent2p, Yap1801p, and Yap1802p, as well as untested CBMs in the vacuolar sorting protein Vps27p, which is involved in sorting at the late endosome, and the uncharacterized products of YML036w and YEL015w. It has been previously shown that the strain yap1801 yap1802 has no recognizable phenotypes (WENDLAND and EMR 1998 ), while the double deletion strain ent1 ent2 is inviable (WENDLAND et al. 1999 ). Although ent1 ent2 cell inviability is due solely to lack of the N-terminal ENTH domain (WENDLAND et al. 1999 ; our unpublished results), studies have shown possible functional roles for other regions of these proteins (WENDLAND and EMR 1998 ; WENDLAND et al. 1999 ; SHIH et al. 2002 ; AGUILAR et al. 2003 ). To investigate the role of the CBMs in the endocytic proteins Ent1p and Ent2p and the candidate endocytic proteins Yap1801p and Yap1802p, we created a quadruple mutant strain in which ENT1 was replaced with the truncated ent1CBM at the ENT1 locus, and ENT2, YAP1801, and YAP1802 were deleted (genotype abbreviated as ent1CBM triple-). We used ent1CBM rather than a larger deletion of ENT1 to avoid complications from loss of other functional regions of the yeast epsins. Ent1CBMp was produced at equivalent levels to the wild-type protein (not shown).

The quadruple mutant strain ent1CBM triple- was created in the presence of pENT1, a plasmid containing wild-type ENT1 and URA3 as its selectable marker. The requirement for these four CBMs for growth and viability was tested using the mutant strain ent1CBM triple- [pENT1] and plating on YPD and 5-FOA at 26° and 37° (Fig 1C). On 5-FOA, which forces cells to grow in the absence of pENT1, ent1CBM triple- cells grew similarly to wild-type cells transformed with empty vector. This suggests that the combined loss of the CBMs of these four endocytic proteins is not detrimental to growth.

Following the determination that ent1CBM triple- cells do not have growth defects at permissive or restrictive temperature, we investigated endocytosis in these cells. Wild-type, ent1CBM triple- [pENT1], and ent1CBM triple- cells were tested at 26° and 37° for uptake of the lipophilic dye FM4-64 to assess internalization rates and internal trafficking of bulk lipids (VIDA and EMR 1995 ). Internalization (as measured by amount of dye accumulated inside the cells) in the CBM-deleted strains was not significantly different from that in wild type (not shown). The normal vacuolar distribution of the dye in labeled cells was observed in all strains.

In addition to bulk endocytosis rates, we analyzed trafficking of the Ste3p receptor transmembrane protein. In MAT cells, Ste3p is constitutively endocytosed in the absence of mating pheromone and delivered to the vacuole for degradation with a t_1/2 15–20 min (ROTH et al. 1998 ). After treatment with cycloheximide, cells were chased from 0 to 90 min and analyzed by SDS-PAGE and quantitative immunoblotting with anti-Ste3p antibodies to determine the amount of Ste3p remaining. Wild-type, ent1CBM triple- [pENT1], and ent1CBM triple- cells were tested to evaluate rates of Ste3p degradation and results from at least three independent experiments were averaged (Fig 1D). Neither ent1CBM triple- [pENT1] nor ent1CBM triple- cells exhibited a kinetic defect in uptake and degradation of Ste3p by this assay. We also examined live cells for localization of two GFP fusion proteins, Ste3-GFP (URBANOWSKI and PIPER 2001 ) and Ste6-GFP (K. KELM, G. HUYER, J. HUANG and S. MICHAELIS, unpublished data; SHAW et al. 2001 ). Similar to Ste3p degradation, Ste3-GFP and Ste6-GFP showed no detectable endocytic defect in the CBM-deleted strains (not shown).

Rcb screen for additional endocytic factors:
Because the ent1CBM triple- cells had no detectable phenotype, we reasoned that additional endocytic factors may be functioning redundantly with these adaptor proteins in selecting receptors for endocytosis and recruitment of the required machinery. On the basis of this idea, we designed a screen to identify novel proteins that may be fulfilling this redundant function. We mutagenized ent1CBM triple- [pENT1] cells with EMS and allowed cells to recover on minimal media. Colonies that required wild-type ENT1 (encoded by the pENT1 plasmid) for viability were identified by replica plating mutagenized cells to 5-FOA at 26°. Twenty-six colonies that were consistently inviable on 5-FOA were designated rcb mutants, for their inability to grow in the absence of the four known adaptor protein CBMs. Of these, 65% were 5-FOA sensitive, even if wild-type ENT1 was supplied by an additional plasmid (that can be maintained on 5-FOA), and were discarded. The remaining nine mutants were mated to ent1CBM triple- [pENT1] cells and confirmed to contain recessive mutations. These mutants were then mated to one another and tested for complementation. Seven of the nine mutants were placed into one complementation group; however, transformation with plasmids containing ent1CBM allowed rescue, indicating that all seven mutants were likely to contain mutations in genomic ent1CBM that caused the Rcb phenotype. These seven were set aside. The remaining two Rcb mutant strains, rcb243 and rcb432, comprised a second complementation group and were selected for further study.

Characterization of rcb432, a temperature-sensitive Rcb mutant:
For the following studies, rcb243 and rcb432 were characterized in the background of ent1CBM triple- [pENT1], and their viability requirement for the Ent1p CBM prevented examination of any phenotypes in a completely CBM environment. For simplicity, the nonitalicized strain names rcb243 and rcb432 will be used to refer to cells with the rcb and ent1CBM triple- [pENT1] mutations. The two Rcb mutant strains were tested for temperature sensitivity at 37° and cold sensitivity at 14° by plating equal volumes of liquid cultures to rich medium (YPD) and incubating at several temperatures. Both the rcb243 and the rcb432 strains were inviable at 37°. rcb432 was selected for more detailed analysis, as rcb243 cells proved more difficult to work with (see MATERIALS AND METHODS). Transformation of rcb432 with single-copy plasmids containing ENT2, YAP1801, or YAP1802 or an additional copy of ENT1 confirmed that one of any of the four CBM-containing proteins was capable of rescuing viability on 5-FOA (Fig 2A, rows 3, 5, 7, and 8). Interestingly, only Ent2p rescued the temperature-sensitive growth at 37° (Fig 2A, row 5). Introducing a plasmid encoding an additional copy of Ent1CBMp did not restore growth on 5-FOA (Fig 2A, row 4), indicating that the rcb432 strain was not isolated due to a mutation in ent1CBM at the ENT1 locus. While together these data confirmed the Rcb phenotype, where any of the four CBM-containing proteins restored viability, supplementing rcb432 cells with a plasmid containing ent2CBM also partially rescued both inviability on 5-FOA at 26° and temperature-sensitive growth at 37° (Fig 2A, row 6). Thus, rcb432 is inviable in the absence of all of the four CBMs, unless both Ent1CBMp (genomic) and Ent2CBMp (episomal) are present.

View larger version (73K): In this window In a new window Download PPT slide   Figure 2. rcb432 is temperature sensitive for growth and endocytosis. (A) Serial dilutions (left to right: 0.25 and 0.05 OD600/ml) of ent1CBM triple- [pENT1] (JRY19) and rcb432 (sla2-432 ent1CBM triple- [pENT1], JRY25) cells were plated in equal volumes to rich media at 26° and 37° and to 5-FOA at 26°. rcb432 was prepared similarly after transformation with the indicated plasmids (rows 3–8: pBW633, pBW387, pBW634, pBW388, pBW635, and pBW636). (B) ent1CBM triple- [pENT1] (JRY19), rcb432 (JRY25), and rcb432 [pSLA2] (JRY25 + pBW384) cells were prepared as in A. (C) Quantification of Ste3p degradation in cycloheximide-treated cells at 37°. , wild type (SEY6210); , ent1CBM triple- [pENT1] (JRY19); •, rcb432 (JRY25); , rcb432 [pSLA2] (JRY25 + pBW384). (D) Ste3-GFP localization in mid-log wild-type (SEY6210) and rcb432 (sla2-432 ent1CBM triple- [pENT1(TRP1)], JRY27) cells transformed with a single-copy STE3-GFP plasmid. Bar, 5 µm.

Because the yeast epsins and AP180s are implicated in endocytosis, we tested rcb432 cells for endocytic defects. Bulk lipid uptake by the FM4-64 assay described above at 26° and 37° was not significantly different in rcb432, wild-type, or parental strains (not shown). Ste3p degradation was also measured as described above after cycloheximide treatment at 26° and 37°. Ste3p degradation in rcb432 cells was not significantly different from wild-type or parental strains at 26° (not shown), whereas no degradation occurred at all over the 90-min chase time in rcb432 cells at 37° (Fig 2C). To determine where trafficking of Ste3p is blocked, we observed localization of the fusion protein Ste3-GFP (URBANOWSKI and PIPER 2001 ) after a 30-min shift to 37° (Fig 2D). rcb432 cells accumulated Ste3-GFP at the cell surface, in contrast to wild-type localization at endosomal and vacuolar structures, indicating that the block in Ste3p degradation is at the internalization step rather than farther downstream. Thus, in contrast to bulk lipid uptake, internalization of the plasma membrane receptor Ste3p is blocked at the nonpermissive temperature of 37° in rcb432 cells.

Complementation cloning of rcb432 shows that it is an allele of SLA2:
We confirmed that a single recessive chromosomal mutation is responsible for the Rcb phenotype and temperature-sensitive growth by mating rcb432 to its parent strain ent1CBM triple- [pENT1] and observing 2:2 cosegregation of both phenotypes. Complementation cloning with a single-copy genomic library (TRP1) was then performed to identify the additional gene mutated in rcb432 cells. Cells were transformed and tested for rescue of inviability at the restrictive temperature, 37°. The temperature-resistant colonies were then tested for viability on 5-FOA. From these cells, we isolated two independent library fragments capable of rescuing both phenotypes that each contained SLA2; one also contained APG2. Because SLA2 is known to be involved in both endocytosis and actin cytoskeleton regulation, we subcloned SLA2 into a new vector (pSLA2) and found that SLA2 alone rescued both temperature-sensitive growth and death on 5-FOA (Fig 2B). In addition, transformation of rcb432 cells with pSLA2 also rescued the endocytic defect as measured by Ste3p degradation at 37° (Fig 2C).

To confirm that rcb432 cells indeed harbored a mutation in SLA2 and to identify the specific mutation, we performed linkage analysis followed by sequencing of the entire SLA2 gene in both wild-type and rcb432 strains (see MATERIALS AND METHODS). The sla2 allele from rcb432 differed from wild-type SLA2 in only one nucleotide (C2374T). This results in a nonsense mutation from glutamine to a stop codon, Q792stop, within the talin homology domain (Fig 3A). This rcb mutation, sla2^792stop, was named sla2-432. For comparison, we also sequenced the end4-1/sla2-41 allele that was isolated and described by RATHS et al. 1993 . Sequencing sla2-41 revealed a nonsense mutation as well, due to a single nucleotide change, C2007T. This results in a glutamine to a stop codon, Q503stop, within the central coiled-coil region (Fig 3A).

View larger version (31K): In this window In a new window Download PPT slide   Figure 3. sla2-432 alone exhibits a strong endocytic defect and is temperature sensitive for growth in ent1 or ent2 backgrounds. (A) Schematic of Sla2p. Arrows show the location of nonsense mutations on the basis of sequencing analyses in the indicated mutants. Bracket marks region deleted in Sla2768-968p. (B) Serial dilutions (left to right: 0.25, 0.05, 0.01, and 0.002 OD600/ml) of wild-type (SEY6210), ent1 (BWY1201), ent2 (BWY1203), sla2-432 (JRY33), sla2-432 ent1 (JRY51), and sla2-432 ent2 (JRY50) cells were plated in equal volumes to rich media at 26° and 37°. (C) Quantification of Ste3p degradation in cycloheximide-treated cells at 37°. , wild type (SEY6210); •, sla2-432 ent2 (JRY31); , sla2-432 (JRY33); , sla2-432 ent2 [pSLA2] (JRY31 + pBW384); , sla2-432 ent2 [psla2-432] (JRY31 + pBW482). Error bars indicate standard deviation of at least two independent experiments.

Endocytosis is defective in sla2-432 cells:
We mated the original rcb432 strain (sla2-432 ent1CBM triple- [pENT1]) to wild type to obtain cells containing only the sla2-432 allele for further characterization. We also crossed sla2-432 into ent1 and ent2 backgrounds. Growth at permissive (26°) and restrictive (37°) temperatures showed that sla2-432 alone is viable at 37°, but when combined with either ent1 or ent2, the cells became temperature sensitive for growth (Fig 3B). To determine the functional domains of Ent1p and Ent2p that are required for viability at 37°, we transformed sla2-432 ent1 and sla2-432 ent2 cells with several mutated or truncated ENT1 or ENT2 constructs and evaluated growth at restrictive temperature. As summarized in Table 3, we found that rescue of temperature sensitivity of either sla2-432 ent1 or sla2-432 ent2 required the addition of the majority of the corresponding Ent1p or Ent2p, including the ENTH domain and the UIMs. In contrast, the CBM was dispensable. This indicated that the sla2-432 allele exhibited genetic interactions with either ENT1 or ENT2, resulting in temperature sensitivity when the majority or all of either gene is absent.

We next tested endocytosis in both sla2-432 and sla2-432 ent2 cells by monitoring Ste3p degradation. As shown in Fig 3C, the temperature-sensitive endocytosis defect was present in sla2-432 cells despite a lack of temperature-sensitive growth. In addition, sla2-432 cells expressing STE3-GFP constructs showed plasma membrane accumulation of Ste3p (not shown) that was similar to rcb432 cells in Fig 2D, indicating the endocytosis block was still at the internalization step. This indicated that the two temperature-sensitive phenotypes were independent of each other and, particularly, that the endocytic defect was independent of the presence or absence of Ent2p. This endocytosis defect was, however, rescued by the addition of wild-type Sla2p (not shown). This indicates that, while both the growth and endocytosis phenotypes are linked to the sla2-432 allele, temperature-sensitive growth is due to a genetic interaction with either ENT1 or ENT2, while the endocytic defect at 37° is due solely to the sla2-432 mutation itself.

Low Sla2 protein levels result in the temperature-sensitive growth of sla2-41 and sla2-432 alleles:
Sequencing the sla2-432 and sla2-41 alleles revealed nonsense mutations that should encode truncated proteins. To confirm the presence of the expected truncated proteins in sla2-432 and sla2-41 cells, whole-cell yeast extracts of equal numbers of cells were prepared from wild type, sla2-432, sla2-41, and a previously characterized SLA2 truncation mutant, sla2768-968, that encodes a protein only 24 amino acids shorter than Sla2-432p (Fig 3A; YANG et al. 1999 ). Proteins extracted from cells grown at 26° or shifted for 30 min to 37° were analyzed by SDS-PAGE and immunoblotted with polyclonal antisera against the N-terminal portion of Sla2p (Fig 4A). Anti-Sla2p reactive bands of the expected sizes were observed for each mutant strain, but the expression level and/or stability appeared quite variable between mutants. sla2-41 is known to be temperature sensitive for growth, with severe endocytic defects only at the restrictive temperature of 37° (RATHS et al. 1993 ; WESP et al. 1997 ). As shown (Fig 4A), after 30 min at 37°, no truncated sla2 protein was detected in sla2-41 cells, thus explaining the previously described irreversibility of this allele without new protein synthesis. In addition, Sla2-41p levels are already lower than those of wild-type Sla2p even at 26°. In contrast, Sla2-432p appeared relatively stable after shift to the restrictive temperature, but protein levels at both 26° and 37° were significantly lower than those in wild type. sla2768-968 cells produced a protein of a size similar to that in sla2-432, but Sla2768-968p levels were more similar to those in wild type.

View larger version (31K): In this window In a new window Download PPT slide   Figure 4. The talin homology domain of Sla2p plays a key role in endocytosis, but not in temperature-sensitive growth. (A) Extracts of cells at 26° or after 30 min at restrictive (37°) temperature were prepared and analyzed by Western blotting with anti-Sla2p antibodies. (B) Quantification of Ste3p degradation in cycloheximide-treated cells at 37°. , wild type (DDY131); •, sla2768-968 (JRY41); and , sla2768-968 ent2 (JRY43). Error bars indicate standard deviation of at least three independent experiments. (C) Serial dilutions (left to right: 0.25 and 0.05 OD600/ml) of wild type (SEY6210), sla2-432 ent2 (JRY31), sla2-432 ent2 [psla2-432] (JRY31 + pBW482), and sla2-432 ent2 [pSLA2] (JRY31 + pBW384) were plated in equal volumes to rich media at 26° and 37°.

The truncated sla2768-968 allele was engineered to delete the entire talin homology domain and was previously shown to rescue the temperature-sensitive growth, endocytosis, and actin defects of sla2 cells (YANG et al. 1999 ). When Sla2768-968p, missing the talin domain, replaced wild-type Sla2p, uptake of the fluid-phase dye Lucifer yellow was normal at both low (25°) and elevated (37°) temperatures (YANG et al. 1999 ). To compare the phenotypes of cells with this truncation to our sla2-432 allele, we constructed a sla2768-968 ent2 strain and tested it for growth and endocytosis. Unlike the temperature-sensitive sla2-432 ent2 strain, we found that growth of sla2768-968 ent2 was similar to that of wild type at 37° (not shown). However, Ste3p degradation assays showed that both sla2768-968 and sla2768-968 ent2 cells exhibited a significant endocytic defect at 37° (Fig 4B). While this result is in contrast to the normal endocytic rates previously reported with talin domain deletion proteins (using Lucifer yellow assays), sla2-432 also exhibited normal bulk endocytic flow at low and high temperatures (as measured by FM4-64 uptake). We interpret these data to suggest that general internalization rates are not affected in either sla2-432 or sla2768-968, but that the endocytosis of specific receptor proteins is selectively affected at high temperatures.

sla2 cells are temperature sensitive in an otherwise wild-type background, while sla2-432 (but not sla2768-968) cells are temperature sensitive for growth only in an ent1 or ent2 background. Therefore, we hypothesized that the decreased protein levels of Sla2-432p in combination with ent1 or ent2 might be the cause of the temperature-sensitive growth phenotype. To differentiate between the contributions of low protein levels vs. the loss of the talin homology domain to the temperature-sensitive growth and endocytic defects in sla2-432 cells, we increased the amount of Sla2-432p and tested for growth and endocytosis. Higher protein levels were achieved by introducing a single-copy plasmid encoding the sla2-432 allele into sla2-432 ent2 cells, which partially reversed inviability at 37° (Fig 4C). Quantitative immunoblotting with anti-Sla2p antibodies showed that sla2-432 ent2 cells expressed the truncated protein at 40% of wild-type protein levels, while sla2-432 ent2 [psla2-432] cells expressed the truncated protein at 60% of wild-type protein levels (not shown). Thus, an additional 50% of the truncated protein is able to rescue temperature-sensitive growth, indicating that the reduced level of Sla2-432p is likely to be the major cause of the temperature-sensitive growth phenotype when ENT2 is deleted. In contrast, endocytosis as measured by Ste3p degradation did not benefit from higher levels of Sla2-432p (Fig 3C). This indicates a segregation of two major phenotypes of the sla2-432 mutation. The temperature-sensitive endocytosis defect appears to be due mainly to loss of the talin domain, while temperature-sensitive growth in the absence of Ent1p or Ent2p appears to be mainly a function of decreased SLA2 protein levels.

sla2-432 cells exhibit actin defects that are exacerbated by loss of the 4 CBM-containing proteins:
A dynamic actin cytoskeleton is required for endocytosis in yeast cells, and numerous links between organization and regulation of the actin cytoskeleton have been shown for both yeast and mammals (reviewed in QUALMANN et al. 2000 ; SCHAFER 2002 ). In wild-type yeast cells, F-actin exists in two distinct pools, cortical actin patches and actin cables. Cortical actin patches localize at the plasma membrane to areas of growth, and actin cables organize along the axis of growth (Fig 5A). In many endocytosis mutants, this polarized pattern is disrupted and actin patches are sometimes found to be larger and fewer, with a "chunky" appearance (see discussion in MUNN et al. 1995 ).

View larger version (64K): In this window In a new window Download PPT slide   Figure 5. sla2-432 has a distinct actin defect that is more pronounced in the ent1CBM triple- background and is rescued by wild-type SLA2. (A) Wild type (SEY6210), (B) ent1CBM triple- [pENT1] (JRY19), (C) rcb432 (sla2-432 ent1CBM triple- [pENT1], JRY25), (D) sla2-432 ent2 (JRY31), (E) sla2-432 (JRY33), and (F) sla2-432 ent1CBM triple- [pENT1, pSLA2] (rcb432/JRY25 + pBW384) were stained with rhodamine-phalloidin to label the filamentous actin cytoskeleton. Left, rhodamine-labeled F-actin; middle, 4',6-diamidino-2-phenylindole-stained DNA; and right, differential interference contrast whole-cell images. F-actin images were deconvolved to eliminate interference from out-of-focus light. Bar, 5 µm.

Sla2p is believed to play a key role in regulation of cortical actin patches and localizes to multiple cortical patches at the plasma membrane of growing buds with 80–90% colocalization to actin patches in small-budded cells (WESP et al. 1997 ). Mutations that disrupt Sla2p localization to cortical actin patches correlate with defects in actin cytoskeletal structure, including the formation of fewer, depolarized, and chunky patches (AYSCOUGH et al. 1999 ; YANG et al. 1999 ). To determine the effect of Sla2-432p on the actin cytoskeleton, we stained the original rcb432 isolate (sla2-432 ent1CBM triple- [pENT1]) as well as sla2-432 ent2 and sla2-432 cells with rhodamine-conjugated phalloidin, which labels filamentous actin. Cells were grown at 26° and either fixed immediately at 26° or shifted to 37° for 90 min, followed by fixation (Fig 5). At both permissive (26°) and nonpermissive (37°) temperatures, the original rcb432 cells had distinct, large, cortically located F-actin clumps in 60% of cells (Fig 5C). Polarization was also affected in these mutants, with the large, cortical clumps distributed in both mother cells and small buds. Similar to the rescue of endocytosis defects and temperature sensitivity, addition of wild-type Sla2p to rcb432 cells rescued most of the actin cytoskeletal defects (Fig 5F). In sla2-432 ent2 and sla2-432 cells, the actin defect was much less pronounced and was characterized by only 10–50% of the cells with smaller chunky or depolarized, but still cortical, F-actin patches (Fig 5D and Fig E). Thus, the CBM-containing proteins also contribute to actin organization since their presence partially suppresses the severe actin defects seen in rcb432 cells.

To determine whether the formation of these abnormally large cortical patches was due to decreased Sla2-432p levels or loss of the talin domain, we stained rcb432 cells alone, plus pSLA2, or plus psla2-432 at 26°. Using blind quantification, both rcb432 cells and rcb432 [psla2-432] cells contained an abnormal, large F-actin patch in 60% of cells, and 30–40% of the cells were depolarized. In contrast, rcb432 [pSLA2] cells contained only 30% of cells with an abnormal large patch and exhibited >80% polarization (Fig 5F). The lack of complete rescue in rcb432 [pSLA2] cells indicates that Sla2-432p has a partially dominant effect on actin organization. In addition, rescue of the rcb432 endocytosis defect by pSLA2 is also not complete (Fig 2C). The idea that Sla2-432p may have dominant effects is not surprising given reports that Sla2p is likely to homodimerize (WESP et al. 1997 ; YANG et al. 1999 ). Overall, these data indicate that the actin cytoskeletal defect in rcb432 cells is more likely due to loss of the talin domain than to decreased protein levels, which further links regulation of the actin cytoskeleton to the process of endocytosis.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The subject of clathrin-dependent internalization in yeast cells remains controversial in spite of clathrin deletion and temperature-sensitive mutants revealing a significant role for clathrin in endocytosis (reviewed in BAGGETT and WENDLAND 2001 ). Since the strain ent1CBM triple-, missing all four endocytic CBMs, failed to display any overt endocytic or growth phenotypes (Fig 1), we postulated the existence of other, redundant clathrin-binding proteins acting as adaptors in yeast. We designed the Rcb screen to isolate these factors, as well as clathrin-independent endocytic factors acting in parallel to the clathrin-dependent pathway. From this screen, we isolated rcb432, containing a mutation in the gene SLA2.

rcb432 cells (sla2-432 ent1CBM triple- [pENT1]) exhibited normal bulk lipid endocytosis (by FM4-64 uptake), but at 37° were defective for endocytosis of Ste3p, a constitutively internalized transmembrane receptor. Signals that trigger endocytosis on the cytosolic tail of Ste3p have been well characterized (ROTH and DAVIS 2000 ; ROTH et al. 1998 ) and may be recognized by clathrin-binding adaptor proteins such as Ent1p or Ent2p (WENDLAND et al. 1999 ; AGUILAR et al. 2003 ). That our Rcb screen yielded an SLA2 allele with a block in Ste3p endocytosis was particularly interesting given the recent discovery that Sla2p can bind the clathrin light chain through its central region (HENRY et al. 2002 ). In addition, the Sla2p homologs Hip1 and Hip12/Hip1R have been shown to bind clathrin, AP-2, and F-actin in mammalian cells and are capable of stimulating clathrin polymerization (ENGQVIST-GOLDSTEIN et al. 1999 , ENGQVIST-GOLDSTEIN et al. 2001 ; METZLER et al. 2001 ; LEGENDRE-GUILLEMIN et al. 2002 ). Together, these data suggest that Sla2p and its mammalian homologs may have clathrin adaptor-like functions that are redundant with other ENTH/ANTH domain-containing proteins. However, the fact that rcb432 cells are partially rescued by the presence of both truncated Ent1CBMp and Ent2CBMp, as well as the discovery that Ste3p endocytosis is blocked in sla2-432 cells, even in the presence of all four CBM-containing adaptor proteins (Fig 3C), indicates that Sla2p may also be playing a different or additional role to Ent1p, Ent2p, Yap1801p, or Yap1802p in the endocytosis of receptor proteins.

The talin homology domain plays an important role in endocytosis:
The sla2-432 allele has a nonsense mutation at the beginning of the talin homology domain or I/LWEQ module. On the basis of the structure of the I/LWEQ module, it is unlikely that the small N-terminal portion of the talin homology domain present in our sla2-432 allele can bind F-actin (MCCANN and CRAIG 1997 , MCCANN and CRAIG 1999 ). Both sla2-432 cells and sla2768-968 cells (missing the entire talin homology domain) grew at elevated temperatures, but exhibited defects in receptor endocytosis, indicating that the talin domain is required for receptor endocytosis at high temperature. Previous studies of SLA2 mutants lacking only the talin homology domain did not uncover endocytic defects; however, these experiments monitored either bulk fluid phase uptake (at 25° and 37°; YANG et al. 1999 ) or internalization of -factor by the Ste2p receptor (at 24°; WESP et al. 1997 ). Our data indicate that the role played by the talin homology domain in endocytosis is revealed only at elevated temperatures and may be restricted to affecting specific receptor proteins since bulk lipid uptake in sla2-432 cells was also normal at 37°. Consistent with this, WESP et al. 1997 tested numerous temperature sensitive for growth alleles of SLA2 for internalization of -factor at low and high temperatures and found endocytosis defects only at restrictive temperatures for most alleles. Even the strong sla2-41 allele exhibits only mild endocytosis defects at permissive temperature (WESP et al. 1997 ) despite loss of the entire second half of the protein, including half of the central coiled-coil region and the entire talin homology domain (Fig 3A and Fig 4A). On the basis of these data, as well as endocytosis defects in sla2 cells at both low and high temperatures, it is clear that Sla2p plays a key role in endocytosis. At elevated temperatures, where the heat-shock response results in reorganization of the actin cytoskeleton and rapid changes in the membrane's protein and lipid profile, the molecular function of Sla2p may be particularly important for viability. Under these conditions, even partial loss of Sla2p function may result in a strong endocytosis defect, particularly if trafficking of an Sla2p-dependent membrane protein is monitored.

That endocytosis defects are present only at high temperature in sla2 strains missing the talin homology domain indicates a requirement for this domain in Sla2p function at restrictive temperatures or perhaps in the cell's response to the temperature shift. One possibility is that the talin homology domain helps to stabilize Sla2p at cortical actin patches during the heat-shock response where it helps to reorganize the actin cytoskeleton. Alternatively, the talin domain may be involved in regulating Sla2p function. Two small coiled-coil regions surrounding the Sla2p talin homology domain occlude F-actin binding by associating with one another (YANG et al. 1999 ), which may act as a regulatory switch in vivo. Perhaps this regulation or F-actin binding in general becomes even more important at elevated temperatures.

sla2-432 exhibits a genetic interaction with both ENT1 and ENT2:
Unlike the endocytic phenotype that is observed regardless of genetic background, we found that sla2-432 cells were inviable at 37° only if either ENT1 or ENT2 was deleted (Fig 3B). Further testing revealed that the sla2-432 allele encoded a truncated protein that was present at lower levels than wild-type Sla2p (Fig 4A). Since sla2768-968 ent2 cells were not temperature sensitive for growth, we asked if sla2-432 ent2 cells were inviable at 37° due to the combined loss of Ent2p and decreased Sla2 protein levels, rather than to the absence of the Sla2p talin homology domain. Increasing Sla2-432p levels by 50% in sla2-432 ent2 cells rescued inviability at 37° (Fig 4C), indicating that Sla2-432p can function in the role fulfilled by wild-type Sla2p in viability at restrictive temperature. Thus, decreased Sla2 protein levels, combined with loss of either ENT1 or ENT2, lead to a synthetic growth phenotype. By introducing ENT1 and ENT2 mutant plasmids into sla2-432 ent1 or sla2-432 ent2 cells, we identified the ENTH domains and UIMs of the yeast epsins as participants in the genetic interaction with sla2-432 (Table 3). However, if normal Sla2p levels are present, just a single ENTH domain from either Ent1p or Ent2p is sufficient for viability at 37° (our unpublished results). Taken to a greater extreme, loss of all of Sla2p results in death at 37° even when all of both Ent1p and Ent2p are present, as seen in the temperature-sensitive sla2 and sla2-41 cells. These data indicate that the temperature sensitivity in sla2 cells is also likely to be due to loss of the same essential pathway that requires the presence of either an ENT1 or ENT2 ENTH domain.

A novel role for the yeast AP180s:
Analysis of the sla2-432 allele in the original strain used for the Rcb screen also revealed a role in cell viability for the yeast AP180 homologs, Yap1801p and Yap1802p. Deletion analyses have not thus far provided any insight into a function for these genes (WENDLAND and EMR 1998 ; HUANG et al. 1999 ), but the similarity in domain structure between Yap1801p and Yap1802p and the yeast epsins, Ent1p and Ent2p, lends support to the idea that these four proteins may have a shared function. Inviability of the rcb432 strain (sla2-432 ent1CBM triple- [pENT1]) upon loss of wild-type ENT1 is rescued by reintroducing either Yap1801p or Yap1802p (Fig 2A, rows 7 and 8). This provides the first evidence that these proteins perform an important function in yeast, which may be in supporting the roles of Ent1p and Ent2p as cargo selectors and/or clathrin assembly factors. In addition, it is possible that Yap1801p and Yap1802p may recognize uncharacterized endocytic signals on certain receptor proteins. These four potential clathrin-binding adaptor proteins share similarities, but perhaps Ent1p and Ent2p are more promiscuous and, in yap1801 yap1802 cells, can also select the cargo normally recognized by the yeast AP180s.

Sla2p as a regulator of actin and endocytosis:
The data available indicate that Sla2p may regulate both endocytosis and actin cytoskeletal dynamics. The actin cytoskeleton of sla2 cells is depolarized and found in larger, aggregate-like patches rather than in the smaller, finer actin patches seen in wild-type cells (HOLTZMAN et al. 1993 ). Mutations in the endocytic proteins Scd5p or Sla1p decrease the amount of Sla2p in actin-positive cortical patches, which also increases the formation of larger, aggregated actin patches (AYSCOUGH et al. 1999 ; HENRY et al. 2002 ). Together, these data suggest that Sla2p plays a role in regulation of actin cytoskeletal dynamics at cortical actin patches, possibly by reorganizing and breaking up large polymers of actin and/or negatively regulating actin nucleators. Studies with the sla2-432 allele indicate that the talin homology domain is also required for this actin-organizing function, and this role is further revealed as increasing numbers of the potential clathrin-binding adaptor genes are deleted (Fig 5). In the original rcb432 strain (sla2-432 ent1CBM triple- [pENT1]), F-actin patches are disrupted to the extent that most cells have only one large cortical actin patch (Fig 5C). Sla2p exhibits interactions with several actin-binding and -regulatory proteins, and here we demonstrate interactions with clathrin adaptors as well. We propose that the loss of adaptors at sites of endocytosis, in combination with mutant sla2-432, leads to a more severe defect in actin because either communication between Sla2p and adaptors is compromised or the malfunctioning Sla2 protein may reduce the spatial and temporal coordination necessary for proper coated pit formation. The recruitment or regulation of other critical adaptor-binding components such as Pan1p or Ede1p may also be compromised. Whether endocytic cues are capable of affecting actin organization to such a degree is unclear, but our results indicate that the loss of clathrin adaptors does have a strong effect on normal regulation of F-actin in cortical patches.

Remaining questions and future considerations:
It remains to be discovered how Sla2p is regulated, how the clathrin adaptors coordinate with Sla2p or with other actin regulators, and the order in which these factors recognize and respond to cues from endocytic sites on the plasma membrane. Sla2p also interacts with the regulatory kinase Ark1p, as well as Scd5p, a protein whose function is unknown; these interactions may also be important for regulation of Sla2p (COPE et al. 1999 ; HENRY et al. 2002 ). The isolation of Sla2p from the Rcb screen, as well as its ability to bind clathrin, suggests that it may also function as an adaptor. Understanding the interactions between Sla2p, the clathrin adaptors, and the proteins that regulate the actin cytoskeleton will be required to develop a molecular understanding of how Sla2p functions in endocytosis and in recruiting and regulating the actin machinery as well.

	FOOTNOTES

¹ Present address: Center for Cancer Research, Massachusetts Institute of Technology, Cambridge, MA 02139.

	ACKNOWLEDGMENTS

We thank David Drubin, Susan Michaelis, Yidi Sun, and Wendland lab members for critical reading of the manuscript and helpful discussions. We also thank David Drubin, Sandra Lemmon, Susan Michaelis, Greg Payne, Robert Piper, and Trina Schroer for gifts of reagants, plasmids, and/or strains. This study was funded by a Burroughs Wellcome Fund New Investigator in the Pharmacological Sciences Award and National Institutes of Health GM-60979. J. Baggett was also funded by a National Science Foundation predoctoral fellowship.

Manuscript received May 1, 2003; Accepted for publication July 7, 2003.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

AGUILAR, R. C., H. A. WATSON, and B. WENDLAND, 2003  The yeast epsin Ent1 is recruited to membranes through multiple independent interactions. J. Biol. Chem. 278:10737-10743.[Abstract/Free Full Text]

AYSCOUGH, K. R., J. J. EBY, T. LILA, H. DEWAR, and K. G. KOZMINSKI et al., 1999  Sla1p is a functionally modular component of the yeast cortical actin cytoskeleton required for correct localization of both Rho1p-GTPase and Sla2p, a protein with talin homology. Mol. Biol. Cell 10:1061-1075.[Abstract/Free Full Text]

BAGGETT, J. J. and B. WENDLAND, 2001  Clathrin function in yeast endocytosis. Traffic 2:297-302.[Medline]

COPE, M. J., S. Y