INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

In Saccharomyces cerevisiae, both fluid-phase and receptor-mediated endocytosis are dependent on actin (Kübler and Riezman, 1993). Insight into the mechanism and the requirements of initial endocytic events in yeast has come through the analysis of mutants defective for receptor-mediated endocytosis of the mating pheromone -factor and fluid-phase endocytosis of Lucifer yellow (LY). Identification of the genes affected in the mutants showed that actin (END7/ACT1) and genes that are required for proper actin cytoskeleton organization, such as END3, END4, END5, and END6 (Bénédetti et al., 1994; Munn et al., 1995) are involved in the internalization step of endocytosis. In addition, mutations affecting other proteins that regulate the integrity of the actin cytoskeleton such as fimbrin, calmodulin, and type I unconventional myosin also block endocytosis (Kübler and Riezman, 1993; Kübler et al., 1994; Geli and Riezman, 1996). Thus, many mutations that disturb the integrity of the actin cytoskeleton affect endocytosis.

In yeast, actin has been localized by indirect immunofluorescence to two distinct structures: actin cables that traverse the mother cell and cortical actin patches that are located at the sites of cell surface growth (Adams and Pringle, 1984). One of the proteins localized to the cortical actin patches is Abp1p, a protein with an SH3 domain (src-homology 3 domain; Drubin et al., 1988; Mulholland et al., 1994). Overexpression of ABP1 confers loss of polarized growth, delocalized assembly of cortical actin, and inviability at high temperatures (Drubin et al., 1988), whereas deletion of ABP1 does not produce a detectable phenotype. The mechanism of action on the actin cytoskeleton and the exact function of Abp1p, however, is unknown. Lila and Drubin (1997) demonstrated that Abp1p binds to the SH3 domain of Rvs167p. Rvs167p is homologous to amphiphysin, a protein involved in synaptic membrane recycling (David et al., 1996). Phenotypes of mutations in RVS167 and a related gene, RVS161/END6, are an inability to adapt to unfavorable growth conditions, altered actin organization, and defects in endocytosis (Bauer et al., 1993; Munn et al., 1995; Sivadon et al., 1995). In addition to binding to Rvs167p, Abp1p has been shown to bind, via its SH3 domain, to the proline-rich region of Srv2p in vitro and mediate the cortical actin cytoskeleton localization of Srv2p in vivo (Freeman et al., 1996; Lila and Drubin, 1997). Genetic studies indicate that Srv2p/CAP has at least two independent functions. First, it integrates RAS function through activation of the adenylate cyclase (Fedor-Chaiken et al., 1990; Field et al., 1990; Gerst et al., 1991). Second, Srv2p/CAP regulates the integrity of the actin cytoskeleton, possibly via actin monomer sequestration (Freeman et al., 1995).

To gain more insight into the function of ABP1, Holtzman et al. (1993) searched for genes that are required for viability in the absence of ABP1 and thereby identified SLA2, a nonessential gene that itself is required for proper organization of the actin cytoskeleton, growth at higher temperatures, and morphogenesis. Sla2p may promote actin polymerization (Li et al., 1995) although how Sla2p exerts this function or other function(s) is unknown.

We show herein that SLA2 is allelic to END4, a gene required for the internalization step of endocytosis (Raths et al., 1993). Amino acid sequence analysis of End4p suggests that it may contain domains that mediate protein-protein interactions. First, End4p contains a proline-rich region. SH3 domains mediate binding to components of the cell cortex (Drubin et al., 1990; Chenevert et al., 1992; Bauer et al., 1993; Holtzman et al., 1993) via proline-rich regions in target proteins (Ren et al., 1993; Freeman et al., 1996). Second, End4p contains a putative coiled-coil region, and coiled-coil domains have repeatedly been found to mediate protein-protein interactions (Crouzet et al., 1991; Bauer et al., 1993; Pollard et al., 1994). Interestingly, a putative leucine-zipper motif is part of the coiled-coil structure. Leucine zippers can serve to form multimeric structures through the interaction of amphipathic helices (Harbury et al., 1993). Third, the C terminus of End4p is homologous to the C terminus of talin (Holtzman et al., 1993). Talin regulates actin dynamics and it connects the actin cytoskeleton to the plasma membrane at adhesion plaques in mammalian cells (Burridge et al., 1988; Muguruma et al., 1990; Rees et al., 1990; Luna, 1991; Turner and Burridge, 1991). Talin and filamentous actin colocalize in macrophages specifically during receptor-mediated phagocytosis (Greenberg et al., 1990; Allen and Aderem, 1995), implicating a role for talin in this process.

We used a mutational analysis of END4 to correlate in vivo functions with regions of End4p and found that two regions of End4p participate in endocytosis. An N-terminal domain is essential for endocytosis. A central coiled-coil domain, required to form a complex, performs a redundant endocytic function with ABP1 and SRV2. We isolated a novel endocytosis mutant, srv2-14, that shows a recessive negative phenotype, and we propose that End4p may mediate endocytosis through the interaction with other actin cytoskeleton-associated proteins.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Media, Reagents, and Strains

Yeast cells were grown in YPUADT [1% yeast extract, 2% peptone (both Life Technologies, Paisley, Great Britain], 20 mg/l uracil, 20 mg/l adenine, and 20 mg/l tryptophan (E. Merck, Darmstadt, Germany; 2% glucose). Lyticase was prepared as described (Hicke et al., 1997). Chemicals were from Bio-Rad (Richmond, CA), Merck, Fluka (Buchs, Switzerland), or Sigma (St. Louis, MO). F1 and Emp47p antibodies were kindly provided by G. Schatz and S. Schröder, respectively (Biozentrum, Basel, Switzerland). DNA restriction enzymes were purchased either from Boehringer Mannheim (Mannheim, Germany) or New England Biolabs (Beverly, MA). Solid medium contained 2% Bactoagar (Difco, Detroit, MI). 5 fluoroorotic acid was used at 0.83 mg/ml in synthetic complete medium (Dulic et al., 1991). LY carbohydrazide was obtained from Fluka. Rhodamine-labeled phalloidin was obtained from Sigma. ³⁵S-labeled -factor was prepared and isolated as described (Dulic et al., 1991). Oligonucleotides were obtained from Microsynth (Balgach, Switzerland). DNA sequencing was performed with a United States Biochemical Sequenase II DNA sequencing kit and polymerase chain reaction (PCR) was carried out by following standard protocols using Pyrococcus furiosus (Pfu) polymerase. sla2 strains used for -factor assays were obtained by crossing sla2 mutant cells (kindly provided by Dr. D. Drubin, University of California, Berkeley) to either RH732 or RH449 and retaining temperature-sensitive bar1 segregants. A Mata srv2::HIS3 bar1 strain (RH4008) was obtained by crossing DDY952 (srv2::HIS3; D. Drubin) to RH3395 and selecting END4 srv2::HIS3 segregants. Other strains used are described in Table 1.

Cloning of END4 and END14 (SRV2)

The END4 gene was cloned by complementation of the temperature-sensitive growth phenotype of an end4-1 strain (RH1597) using a yeast CEN-based genomic library (Rose et al., 1987). The end4-1 complementing plasmid (pSR2) contained a 16-kb Sau3A-BamHI insert. The complementing fragment of pSR2 was further defined and ligated as a 4.4-kb XbaI-SpeI fragment into linearized YCplac33 at the XbaI site to give plasmid pGP6 (Parraga, personal communication). The insert in pGP6 contains only one complete open reading frame (ORF). This plasmid complements the temperature-sensitive phenotype of RH1597 and restores both growth and endocytosis at 37°C in end4 mutants. Overlapping fragments of pGP6 were cloned into pBSK and used for sequencing. To confirm that the cloned fragment contained the authentic END4 locus, the chromosomal locus was tagged with URA3 by cloning the complementing fragment into an URA3 carrying integration vector followed, by linearizing this plasmid at the XhoI site within the putative END4 ORF and integrating this construct in the wild-type RH449 strain. Uracil prototrophs were crossed to a temperature-sensitive end4-1 strain (RH1807) and diploids were sporulated and tetrads were dissected as described. We observed tight linkage between uracil prototrophy and temperature-resistant growth, demonstrating that the cloned locus in pGP6 indeed corresponds to the authentic END4 locus. An ~4.2-kb BamHI-HindIII END4-containing fragment was ligated into high-copy-number vector YEplac181 to overexpress END4.

The END14 gene was cloned by complementation of the temperature-sensitive growth phenotype of an end14-1 strain (RH3993) using a yeast CEN-based genomic library (Munn et al., 1995). The end14-1 complementing fragment was found to reside within a 3.3-kb BamHI-HindIII fragment, which was cloned into YCplac111 to give pTA2 and cloned into YIplac128 to give pTA3. DNA sequencing revealed that SRV2 is the only ORF on this fragment. pTA3 was linearized using the unique restriction site StuI within the ORF of SRV2 and then transformed into wild-type RH977 strain to mark this locus with LEU2. Leucine prototrophs were then crossed to temperature-sensitive end14-1 strains and diploids were sporulated and dissected. In 48 dissected tetrads, leucine prototrophs were always temperature-resistant for growth showing that end14-1 is allelic to SRV2.

Deletion of END4 and END14 (SRV2)

All END4-containing plasmids were derived from pGP6. Plasmid pWA4 was constructed by ligating the ~4.2-kb BamHI-HindIII fragment from pGP6 into pBSK linearized with BamHI/HindIII. The complete END4 ORF including 135 nucleotides upstream and 65 nucleotides downstream were replaced by a 1.2-kb BamHI fragment containing the HIS3 ORF. This did not affect adjacent ORFs (Yeast Genome Project, final release of the complete genome on April 24, 1996. http://speedy.mips.biochem.mpg.de/mips/yeast). The resulting plasmid pWA9 was digested with EcoRI and the HIS3-containing 2.4-kb fragment was transformed into RH2966 for integration/replacement of END4, generating RH2974. Transformants were selected on minimal medium lacking histidine. Diploids were transformed with pGP6 to increase end4 spore viability, sporulated, and dissected as described. Deletion of only one copy of END4 was confirmed by tetrad analysis. Correct integration replacement of END4 was confirmed by the lack of End4p in total protein extracts of histidine prototrophic segregants of RH2974.

END14 was deleted by integration replacement. A URA3-containing PCR fragment was generated using pRS416 (Sikorski and Hieter, 1989) as a template with primers having homology to the 5 and 3 sequences of END14 (Lila and Drubin, 1997). Correct integration in diploid transformants prototrophic for uracil was confirmed by PCR analysis (our unpublished results).

Construction of Mutant end4 Alleles

All mutant end4 alleles were cloned into the yeast integration vector pRS404 carrying TRP1 as a selectable marker (kindly provided by Dr. P. Phillippsen, Biocenter, Basel, Switzerland). In all cases the HindIII site in the polylinker was eliminated either by inserting mutant end4 alleles or by specific elimination of the site (generating pRS404) prior to insertion of mutant end4 alleles. After cloning of mutant end4 alleles into this integration vector, the only HindIII site left was within the TRP1 sequence. To delete the End4p talin-like C-terminal domain, two complementary oligonucleotides encoding three consecutive stop codons were hybridized and blunt-end ligated into the END4 coding sequence at the HincII restriction site. This introduces in-frame stop codons just after amino acid 766. The resulting allele was named end4talin. To facilitate generation of end4 deletion constructs, we introduced an artificial StuI site into END4. This results in two changes in the amino acid sequence of End4p (A642V/I643L). Introduction of the unique StuI site was confirmed by DNA sequencing, and we showed that the introduced changes in the protein sequence of End4p does not interfere with END4 function (see RESULTS). An ~4.2-kb END4 BamHI-HindIII fragment containing the StuI site was cloned into pRS404 at the SpeI site to give plasmid pWA27. Subsequent end4 alleles were derived from pWA27 as follows: the ~800-bp END4 XhoI-StuI fragment of pWA27 was replaced with PCR-generated XhoI-StuI fragments that lacked various domains of the wild-type fragment. Alleles were named as follows (deleted amino acids in parentheses): end4coil1 (376-501), end4coil2 (495-573), end4coil1,2 (376-573), and end4Gln (376-440). The proline-rich sequence in END4 (amino acids [aa] 286-301) was deleted using a two-step PCR generating an ~670-bp KpnI-XhoI fragment. This fragment was used to replace the KpnI-XhoI fragment of END4 to generate end4286-301, which is referred to as end4Pro. The N-terminal deletion alleles were constructed as follows: a one-step PCR yielded the KpnI-XhoI fragments containing end4114-284 and end4318-373, respectively, which were used to replace the KpnI-XhoI fragment of END4.

In-frame deletions of the END4 gene created by PCR were confirmed by sequencing the relevant regions. All newly generated end4 alleles, and the wild-type END4 gene, were subcloned into TRP1-marked integration vectors. The resulting constructs, linearized with HindIII, were integrated at the trp1 locus of diploid RH2974. Integrants were selected by growth at 24°C on synthetic medium lacking tryptophan, restreaked for single colonies, then sporulated, and dissected as described below. Segregants prototrophic for both histidine and tryptophan were selected. Integration at trp1 was verified by crossing these segregants to RH2635. All segregants of the resulting diploids were prototrophic for tryptophan, confirming integration of the mutant end4 alleles at the trp1 locus. Expression of mutant End4p and concomitant lack of wild-type End4p was confirmed by immunodetection of End4p in total protein extracts prepared from His+ Trp+ segregants.

Endocytosis Assays

Pheromone-uptake assays were performed using a continuous presence protocol as described earlier (Dulic et al., 1991). Cultures were grown at 24°C to 0.5-1.5 × 10⁷ cells/ml, and the cells were harvested by centrifugation at 3000 rpm for 5 min in a GLC (General Laboratory Centrifuge, Sorvall, Lausanne, Switzerland) in 50-ml Falcon tubes. Cells were then resuspended in YPUADT at 2-3 × 10⁸ cells/ml and incubated at the respective growth temperature for 5 min on a rotary aquashaker (Kühner, Basel, Switzerland). Then ³⁵S-labeled -factor was added and samples were taken at the indicated time points after -factor addition. Cells were diluted 1:100 into ice-cold pH 1 (50 mM sodium citrate) or pH 6 (50 mM potassium phosphate) buffer. The pH 6 samples were filtered immediately onto GF/C filters (Whatman, Maidstone, England) on a vacuum pump, whereas the pH 1 samples were kept on ice for 20 min prior to filtering. Filters were washed with ice-cold buffer, transferred to 20-ml screw-cap polyethylene scintillation vials (Packard, Meriden, CT), and dried at 80°C for 15 min before adding 5 ml of emulsifier-safe scintillation cocktail (Packard). The radioactivity associated with each cell sample was measured in a -counter (1900 TR, Packard). The percentage of internalized -factor was calculated as follows: 100 × cpm(pH 1)/cpm(pH 6). Endocytic rates were determined as the slope of the internalization curves within the range of 5 to 15 min, using the linear regression curve fitting option on a HP48SX calculator, thus expressing percentage of internalized ligand per minute.

LY accumulation was assayed as described in Munn et al. (1995).

Generation of Antibodies, Protein Extracts, and Immunoblot Analysis

Antibodies against End4p were raised in female New Zealand White rabbits using the extreme N-terminal peptide (SRIDSDLQKALKK) and the extreme C-terminal peptide (KRLGEIRRHAYYNQDDD) as described (Harlow and Lane, 1988). Protein extracts were prepared as described (Horvath and Riezman, 1994). Protein extracts dissolved in sample buffer were separated on denaturing SDS-PAGE minigels (Hoefer Scientific Instruments, San Francisco, CA) and transferred to nitrocellulose filters as described (Schimmöller et al., 1995). Molecular weight markers were purchased from Bio-Rad. Nitrocellulose membranes were incubated for 30 min at room temperature (RT) in blocking buffer [0.1 M NaCl, 80 mM Na₂HPO₄, 20 mM NaH₂PO₄, 2% (wt/vol) nonfat-dry milk, 0.1% Nonidet P-40]. Antibodies were added in fresh buffer and incubation was continued either at RT for 1.5 h or at 4°C overnight; filters were subsequently washed three times with blocking buffer and three times with PBS (0.1 M NaCl, 80 mM Na₂HPO₄, 20 mM NaH₂PO₄). Then, goat anti-rabbit IgG coupled to horseradish peroxidase (Sigma) was added at a dilution of 1:8000 and incubation was continued for 1 h at RT before filters were washed as described above. Antibody conjugates were visualized using the ECL detection kit (Amersham) and Kodak XAR-5 film. Quantifications were carried out using densitometric scanning (Molecular Dynamics, Sunnyvale, CA).

Preparation of Spheroplasts and Cell Fractionation

Cultures were grown to 2-5 × 10⁷ cells/ml and then the cells were harvested in GSA bottles by centrifugation at 5000 rpm for 5 min (Sorvall, RC2-B). The cells were resuspended in 0.1 M Tris sulfate (pH 9.4), and 28 mM 2-mercaptoethanol, transferred to 50-ml Falcon tubes (Becton Dickinson, Franklin Lakes, NJ), and incubated at RT with slow shaking for 20 min (100 ml of buffer per 2-10 × 10¹⁰ cells). Cells were pelleted in a Sorvall GLC at 3000 rpm for 5 min, resuspended in 20 ml of 2% yeast extract, 1% peptone, 10 mM Tris (pH 8), and 0.7 M sorbitol containing lyticase (0.01-0.02 mg/ml) and incubated at 30°C on a rotary aquashaker at 100 rpm until >90% of the cells lysed upon osmotic shock. To check for lysis, an aliquot was diluted in 20 mM NaCl, 10 mM Tris, pH 8, and 5 mM MgCl₂ and observed under the microscope. Yeast cells were fractionated according to a modified version of the method of Pryer et al. (1993). Spheroplasts were washed once in 0.1 M NaCl, 10 mM Tris, pH 8, 5 mM MgCl₂, and 0.7 M sorbitol, resuspended in 2 ml of lysis buffer (0.1 M NaCl, 20 mM 2-(N-morpholino)ethane-sulfonic acid (MES) pH 6.5, 5 mM MgCl₂) containing leupeptin (1 µg/ml), aprotinin (1 µg/ml), and pepstatin (1 µg/ml) to 2-5 × 10⁹ cells/ml, and lysed on ice by douncing with 36 strokes in a 5-ml Dounce homogenizer (Wheaton, Dietikon, Switzerland). The lysate was centrifuged at 1000 × g for 5 min at 4°C in a tabletop centrifuge (Heraeus, Sepatech, Switzerland) to remove unbroken cells and large cell debris. Then, equal volumes (0.1 ml) of supernatant and reagents in lysis buffer were mixed to give final reagent concentrations of 1% Triton X-100, 0.5 M NaCl, 2.5 M urea, or 0.1 M Na₂CO₃ (pH 11.5), and incubated for 1 h on ice. Samples were centrifuged for 1 h at 55,000 rpm in a Beckman TLA100.3 rotor. Pellets and supernatants were denatured in SDS sample buffer and equivalent amounts were electrophoresed on SDS gels and analyzed by immunoblotting as described.

According to k-value calculations, one half of the total amount of a particle of about 9.1 S will sediment in 1 h of centrifugation at 55,000 rpm in a total volume of 0.2 ml in a TLA100.3 rotor (Beckman). If the particle has a sedimentation value of 8.3 S (estimation for End4coil1p homodimers is around 210 kDa) about 45% would be recovered under the same centrifugation conditions in the pellet fraction.

Sucrose Velocity Gradients

Wild-type RH732 cells were grown at 24°C to 2-5 × 10⁷ cells/ml and 4-10 × 10¹⁰ cells were converted to spheroplasts as described above. Spheroplasts were washed once in 150 mM potassium acetate, 20 mM HEPES (pH 6.8), and 5 mM magnesium acetate containing 0.7 M sorbitol for osmotic support and resuspended in ice-cold lysis buffer (20 mM NaCl, 10 mM Tris-Cl, pH 8, 5 mM MgCl₂) containing leupeptin (1 µg/ml), aprotinin (1 µg/ml), and pepstatin (1 µg/ml) to 2-5 × 10⁹ cells/ml. Phenylmethylsulfonyl fluoride (100 mM stock in isopropanol) was added to a final concentration of 1 mM, and spheroplasts were lysed by 48 strokes on ice in a 5-ml Dounce homogenizer (Wheaton). The lysate was centrifuged at 1000 × g for 1 min in a tabletop centrifuge (Heraeus) to remove unbroken cells. Then, the supernatant was further centrifuged at 1000 × g for 3 min to give P1 and S1. S1 (560 µl) was mixed with 140 µl of 150 mM potassium acetate, 20 mM HEPES (pH 6.8) and 5 mM magnesium acetate containing 20% Triton X-100 and incubated for 20 min at RT before loading the entire volume onto preformed sucrose gradients. Centrifugation was carried out for 18 to 19 h at 4°C in a Centrikon TST41.14 rotor at 130,000 × g (32,000 rpm, Kontron TGA-65 ultracentrifuge, Zurich, Switzerland) and was terminated without brakes. The gradients were prepared as follows: 1.875-ml volumes of 30%, 25%, 21%, 17%, 13%, 9% sucrose (wt/vol) in 150 mM potassium acetate, 20 mM HEPES (pH 6.8), and 5 mM magnesium acetate containing 1% Triton X-100 were layered on top of each other in 13-ml plastic tubes (Beckman ultra-clear tubes, 14 × 89 mm) and stored overnight at 4°C. Fractions of 320 µl each were collected from the top of the gradient with a Gilson P1000 pipette. Pellets were resuspended in 1% SDS, and equal amounts were loaded onto SDS gels. Proteins were transferred to nitrocellulose filters and detected as described. Sucrose (as percent wt/vol) was determined by measuring refractive indices and reading values from a standard curve. Total protein was determined using the BCA kit (Pierce, Rockford, IL). Markers used were bovine serum albumin (66 kDa, Sigma), aldolase (147 kDa), -amylase (200 kDa, Sigma), catalase (232 kDa, Sigma), and apoferritin (443 kDa, Sigma). Each marker (0.7 mg) was loaded and fractions were collected as described. Calculations according to Griffith (1979) were carried out to determine the molecular weight of the End4p-containing complex (corrected for water at 20°C). Comparison of the End4p-containing complex with a standard curve derived from the markers yielded a molecular weight of the End4p complex of 222 ± 28 kDa (n = 5).

Two-Hybrid Analysis

For analysis of End4p protein interactions, the entire coding region of END4 was amplified by PCR using primers introducing BamHI sites 2 bp upstream of the initiator methionine and downstream of the last codon, respectively, using pGP6 as a template. This PCR product was digested with BamHI, gel purified, and ligated into BamHI-digested dephosphorylated vector pGAD424 (Fields and Sternglanz, 1994) to give plasmid pGCE, containing the gal4 DNA-activating domain fused to End4p. pGCE complements the growth defect of end4 mutants at 37°C (our unpublished results). Fusion of the gal4 transcription-activating domain to mutant end4 alleles was carried out using the same primers as for the full-length END4 with plasmids containing end4coil1 and end4coil2 alleles to give plasmids pGLE and pGSE, respectively. aa 318-590 of End4p fused to the gal4 transcription-activating domain was isolated during a screen of RPG1-interacting proteins (Palecek, unpublished observation). The BamHI fragment of pGCE containing the entire END4 sequence was cloned into the BamHI site of pBTM116 to give plasmid pLCE, containing an in-frame fusion of the lexA DNA-binding domain and End4p proteins. Yeast strain L40 containing two reporter genes (lexA-LacZ and lexA-HIS3) was transformed with pLCE and the plasmids containing the various DNA-activating domain fusion proteins. Histidine prototrophy was scored on plates containing 0.67% yeast nitrogen base and 2% glucose supplemented with 1 mM 3-amino-1,2,4-triazol after 3 d at 30°C. LacZ enzymic activity was measured as described in Miller (1972). -Galactosidase activities are expressed as units/milligram of protein.

Miscellaneous

Mating of haploid strains of yeast, sporulation of diploids, and tetrad analysis were performed as described in Sherman et al. (1974). Transformation of yeast with plasmid DNA and all DNA manipulations were done as described in Munn et al. (1995). Recovery of DNA fragments was performed as described (Sambrook et al., 1989) with DEAE membranes. The End4p protein sequence was analyzed using the University of Wisconsin GCG programs run on a VAX/VMS computer system. Proline at position 498 caused the structure prediction program to divide the region from aa 376 to aa 573 into two separate coiled-coil domains. Actin was stained with rhodamine-coupled phalloidin as described in Munn et al. (1995). Carboxypeptidase Y (CPY) metabolic labeling experiments were carried out as described in Mulholland et al., (1997) using a 5-min pulse prior to chase at the indicated time points after 15 min of preincubation at either 24°C or 37°C, respectively.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

END4 Is Essential for Endocytosis

end4 mutants were isolated independently in two separate screens for endocytosis mutants and in both cases the mutants were temperature-sensitive for growth (Raths et al., 1993; Munn et al., 1995). We cloned the wild-type END4 gene by phenotypic complementation of the temperature-sensitive growth defect of an end4-1 strain and found that END4 is identical to SLA2/MOP2. SLA2 was isolated because it is essential for viability in the absence of ABP1. Sla2p is required for growth at high temperature, viability with abp1, and polarization of the cortical actin cytoskeleton (Holtzman et al., 1993). MOP2 was isolated as a regulator of the abundance of the plasma membrane ATPase at the cell surface (Na et al., 1995). Henceforth we refer to SLA2/MOP2 as END4.

To create a null allele, END4 was disrupted in a diploid wild-type strain by integration replacement (Figure 1A). We noticed that end4 segregants were hardly recovered, suggesting that END4 may be essential for germination and/or growth in our strain background. Therefore, the heterozygous disrupted strain was transformed with a plasmid carrying the wild-type END4 gene on a centromeric plasmid (pGP6), sporulated, and dissected. end4 segregants carrying pGP6 were cured of the plasmid at 24°C. Deletion of END4 was confirmed by the lack of the gene product in plasmid-cured segregants prototrophic for histidine (Figure 1C). Growth of end4 strains at 37°C, but not at 24°C, depended on the presence of pGP6. Thus, deletion of END4 confers a temperature-sensitive growth phenotype (see Figure 3B) and it may interfere with germination. This may explain why Na et al. (1995) reported that MOP2 is essential for viability. In our background, a null allele results in vegetative growth phenotypes similar to the results described by Holtzman et al. (1993).

View larger version (56K): [in this window] [in a new window]   Figure 1.   Deletion of END4 constitutively blocks endocytosis. (A) The complete END4 ORF on an AatII fragment was replaced with the prototrophic marker HIS3. The resulting EcoRI fragment was used to delete END4 by integration/replacement. B, BamHI; R, EcoRI; A, AatII; X, XhoI; H, HindIII. (B) Receptor-mediated internalization of -factor is blocked at both 24°C (solid symbols) and 37°C (open symbols) in RH3391 cells cured of pGP6 (end4, squares), whereas RH3393 cells (END4, circles) show no defect at either temperature. (C) End4p-specific antiserum generated against a C-terminal peptide specifically recognizes a single band with an apparent mobility of ~116 kDa that is absent in strain RH3391 cured of pGP6 (end4) extracts but more prominent in extracts prepared from strains expressing multiple copies of the gene (2µEND4). (D) In contrast to strain RH2887 (END4), strain RH3391 cured of pGP6 (end4) does not accumulate LY in the vacuole after incubation with the dye at 24°C for 1 h. Bar, 5 µm.

View larger version (53K): [in this window] [in a new window]   Figure 3.   N terminus of End4p is essential for actin organization and endocytosis. (A) Mutant end4 alleles devoid of the indicated regions (open space in the schematic representation) were generated as described in MATERIALS AND METHODS. Pro, proline-rich sequence; Q, glutamine-rich sequence; L, putative leucine-zipper; coil1 and coil2, putative coiled-coil regions 1 and 2, respectively. Proline at aa 498 divides the region from aa 376 to aa 573 into two separate coiled-coil domains (MATERIALS AND METHODS). The nomenclature of end4 alleles and the deleted amino acids are indicated. (B) END4 (section 1, RH3393), end4 (section 2, RH3212), end4N1 (section 3, RH3769), and end4coil1 (section 4, RH3395) strains were streaked onto YPUADT plates and incubated for 2 d at 37°C. Deletion of the N-terminal part of End4p but not deletion of the coiled-coil domain impairs growth at higher temperatures. (C) Receptor-mediated endocytosis of radiolabeled pheromone in END4 (RH3393) and end4N1 (RH3769) strains after a 5-min preincubation at either 24°C (solid symbols) or 37°C (open symbols). (D) END4 (RH3393) and end4N1 (RH3769) strains were grown at 24°C and processed for visualization of filamentous actin with rhodamine-conjugated phalloidin as described in MATERIALS AND METHODS. Bar, 5 µm.

The originally isolated end4 alleles showed temperature-sensitive endocytosis, partially defective at 24°C and completely defective at 37°C. To determine whether deletion of the END4 gene abolishes endocytosis, we assayed fluid-phase endocytosis of LY and receptor-mediated endocytosis of -factor in end4 mutants. Cells were grown at 24°C and radiolabeled -factor was added to cells after a 5-min incubation either at 24°C or 37°C. The percentage of internalized pheromone at each time point was determined by dividing internal (pH 1 resistant) counts by total cell-associated (pH 6 resistant) counts. Wild-type cells internalized -factor rapidly, irrespective of the temperature of incubation (Figure 1B). In contrast, disruption of the END4 gene resulted in a complete loss of pheromone internalization at both 24°C and 37°C (Figure 1B). Since end4 cells were also defective for LY accumulation in the vacuole at 24°C (Figure 1D), we conclude that END4 is essential for both fluid-phase and receptor-mediated endocytosis. We also found that the five previously isolated temperature-sensitive sla2 alleles (Holtzman et al., 1993) all showed a defect in -factor internalization at the restrictive, but not at the permissive, temperature for growth (our unpublished observations). Thus, the independently isolated end4 and sla2 alleles are true temperature-sensitive loss of function mutations.

The Coiled-Coil Domain of End4p Is Necessary for Protein Interaction

Antibodies generated against a C-terminal peptide of End4p recognized a single protein in yeast lysates with an apparent molecular weight of about 116 kDa as determined by denaturing SDS-PAGE (Figure 1C). This band was absent in extracts prepared from end4 strains but was much stronger in extracts from strains carrying the END4 gene on a high-copy-number plasmid. The fact that End4p migrates more slowly than predicted from its size (109 kDa) may indicate that the protein adopts a special conformation resistant to SDS denaturation or that the protein is posttranslationally modified. To determine whether End4p is soluble or membrane-associated, we incubated wild-type cell extracts with detergent, high salt, urea, or carbonate as described in Figure 2A. After a 1-h high-speed centrifugation, pellets and supernatants were analyzed by immunodetection for End4p, Emp47p, and F1 content. The integral membrane protein Emp47p (Schröder et al., 1995) was only solubilized by treatment of lysates with detergent, whereas a peripheral membrane protein, the F1 subunit of the mitochondrial ATP synthase, was also solubilized by treatment with high salt, urea, and carbonate (Figure 2A). End4p distributed almost equally between the soluble and particulate fraction and treatment of the lysate with detergent did not alter this distribution (Figure 2A). This demonstrates that End4p is not a transmembrane protein but partitions into the particulate fraction as a proteinaceous complex. Consistent with this idea, incubation of cell extracts with agents that disrupt protein-protein interactions solubilized End4p.

View larger version (32K): [in this window] [in a new window]   Figure 2.   Coiled-coil domain of End4p mediates complex formation. (A) Wild-type cell extracts were incubated on ice for 1 h in 1% Triton X-100 (lane T), 0.5 M NaCl (lane N), 2.5 M urea (lane U), or 0.1 M carbonate, pH 11.5 (lane C) or were mock treated with lysis buffer (lane B). After high-speed centrifugation, equivalent volumes of pellets (lanes P) and supernatants (lanes S) were analyzed for End4p and control proteins by immunodetection. (B) Wild-type cell lysates were fractionated on linear 9-30% sucrose gradients. Fractions were collected from the top (fraction 1) and End4p content was quantified by Western blotting. Arrowheads indicate the position of the peaks of marker proteins; numbers are given in kDa. (C) Wild-type END4 as well as mutant end4 alleles were fused in-frame to the gal4 transcription-activation domain. These fusion constructs were tested in the two-hybrid system for interaction with the wild-type END4 gene fused to the LexA DNA-binding domain. -Galactosidase activities of extracts prepared from yeast strain L40 harboring the various combinations of plasmids were measured and expressed as units/milligram of protein. Alternatively, growth on plates lacking histidine was scored to monitor interaction of End4p. (D) Extracts of end4376-501 (RH3395) and end4495-573 (RH3397) mutants were prepared in lysis buffer and processed as described in A. Mutant End4p was visualized in pellets (lanes P) and supernatants (lanes S) by immunodetection.

The fractionation results suggested that End4p may exist in two forms, one soluble (and not sedimentable) and the other part of a sedimentable complex. To address this question, we analyzed wild-type lysates using velocity centrifugation on linear sucrose gradients. This method allows to resolve (putative) multiple species of End4p. As shown in Figure 2B, End4p was recovered from the gradient as a single peak. The sedimentation coefficient of proteins comigrating with End4p is 10.6 ± 0.8S (Griffith, 1979), which corresponds to an apparent molecular weight of 220 ± 30 kDa, if a globular shape is assumed. As End4p is predicted to have a molecular weight of 109 kDa, the protein most likely exists in a complex, possibly as an End4p homodimer or associated with other protein(s). Because End4p has a predicted coiled-coil structure, it is possible that the protein adopts a nonglobular shape that would cause it to sediment more slowly on velocity gradients. In this case, 220 kDa would be an underestimate of the molecular weight of the complex. The apparent discrepancy between the results from the differential centrifugation and velocity sedimentation experiments can be easily explained. Fifty percent of a 9.1S complex would have sedimented in 1 h under our conditions for differential centrifugation (see MATERIALS AND METHODS).

The middle part of End4p (aa 376-573) has the potential to form a coiled-coil structure (MATERIALS AND METHODS), and it has been reported previously that coiled-coil domains mediate protein-protein interactions (Crouzet et al., 1991; Bauer et al., 1993; Pollard et al., 1994). In light of our results from the fractionation and sedimentation analysis, we tested whether the coiled-coil domain of End4p can mediate End4p-End4p interaction using the two-hybrid system (Fields and Sternglanz, 1994). We therefore created fusion proteins of full-length End4p (or mutant end4 alleles) with the gal4 transcription-activating domain and tested for interaction with full-length End4p. We found that full-length End4p interacts with itself in the two-hybrid system (Figure 2C) and this interaction depends on the presence of the coiled-coil domain of End4p (aa 376-573). Moreover, this coiled-coil domain itself is sufficient to mediate interaction with End4p in the two-hybrid system. An End4p fragment consisting of aa 318-376 did not interact with full-length End4p in the two-hybrid system (Palecek, unpublished observation). Thus, it is likely that the coiled-coil domain of End4p mediates partitioning of End4p into a sedimentable complex. Consistent with this, sedimentation of End4p upon high-speed centrifugation depends on the presence of the coiled-coil domain (Figure 2, A and D). These results are consistent with the idea that End4p homodimer formation via its coiled-coil domain may be a prerequisite for End4p sedimentation. Alternatively, the coiled-coil domain may mediate interaction with other protein(s) than End4p and dimerization may not be required for association with the sedimentable complex.

The N-terminal Domain of End4p Is Essential for Endocytosis and Actin Organization

Analysis of the End4p amino acid sequence suggests that several regions could be important for the function of the protein (Holtzman et al., 1993; see MATERIALS AND METHODS). These include sequences rich in prolines, glutamines, and the putative coiled-coil domain. In addition, the C terminus of End4p has homology to talin (Holtzman et al., 1993), a protein that connects the actin cytoskeleton to membranes in mammalian cells (Rees et al., 1990). To investigate possible in vivo functions mediated by those sequences, we generated end4 mutants carrying deletions of the above-mentioned regions. In particular, we wanted to know whether the endocytic function of END4 can be separated from its function in actin cytoskeleton organization and from its growth requirement in an abp1 background. A schematic overview of the end4 deletion alleles is given in Figure 3A. Mutant end4 alleles were integrated (expressed from the wild-type promoter) in an end4 background and we then determined whether they could rescue the multiple defects of the end4 null mutant. All mutant end4xy gene products were stably expressed at levels comparable to wild-type End4p, and we found that all mutant end4 alleles could fully rescue the low spore viability of end4 segregants (our unpublished results). However, end4N1 mutants carrying a deletion in the N-terminal part of End4p could not grow at 37°C, whereas other end4 mutants did not display any growth defect (Figure 3B; our unpublished observations). Thus, except for the N-terminal region, large parts of End4p including the coiled-coil domain and the talin-like region can be deleted without affecting growth.

All of the previously tested end4 alleles displayed a temperature-dependent endocytosis defect (Raths et al., 1993; Munn et al., 1995; see above) and end4 mutants are defective for endocytosis even at their permissive growth temperature (Figure 1). To determine whether any of the deleted regions of End4p are required for endocytosis, we measured internalization of radiolabeled pheromone. The endocytic rates were expressed as a percentage of the wild-type internalization rate between 5 and 15 min after ligand addition. Strikingly, end4N1 mutants are defective for internalization of pheromone even at the permissive growth temperature of 24°C (Figure 3C), whereas all other end4 mutants internalized -factor with nearly wild-type kinetics at 24°C (Table 2). As for the growth requirement, large parts of End4p including the coiled-coil domain and the talin-like region can be deleted without affecting endocytosis.

END4/SLA2 is required for the organization and polarization of the actin cytoskeleton (Holtzman et al., 1993). Because deletion of END4 confers both a loss of actin organization and an endocytosis defect, we tested whether any of the newly generated end4 alleles had an actin defect. Mutants were grown at 24°C and the actin cytoskeleton was visualized with rhodamine-conjugated phalloidin. Except for end4N1, none of the end4 mutants had an obvious actin defect at 24°C (Figure 3D; our unpublished observations). In end4N1 mutants, many actin patches were found in both mother and daughter cells, irrespective of the stage of the cell cycle, whereas in wild-type strains actin was polarized to actively growing regions in small buds. However, actin cables in end4N1 were similarly organized as in wild-type strains. Note that end4N1 mutants are larger and more spherical than wild-type counterparts. Thus, our results demonstrate that only the N-terminal domain of End4p is required for growth, actin organization, and endocytosis and that the coiled-coil region, the talin-like domain, the proline-rich region, and the Gln-rich region are apparently not essential for those processes.

Functional Overlap of the Coiled-Coil Region of End4p with ABP1 and SRV2 in Endocytosis

Because end4 abp1 double mutants are inviable (Holtzman et al., 1993), we tested whether any of the newly generated end4 deletion alleles caused inviability of an abp1 strain. Finding viable and lethal combinations of end4xy abp1 double mutants could help to correlate in vivo functions with different parts of End4p. We therefore crossed the end4 deletion mutants to abp1 strains and analyzed double mutant segregants for growth on rich medium at 24°C and 37°C. We found that end4N1 abp1 and end4Gln abp1 double mutants were inviable. Examination of those dissection plates showed that all predicted double mutants ceased growth after 2-10 cell divisions. All other combinations of end4xy abp1 double mutants were viable at 24°C (Table 3). However, in contrast to each single mutant, end4coil1 abp1 and end4coil1,2 abp1 double mutants were unable to grow at 37°C, whereas deletion of the proline-rich region, the talin-like domain, another domain in the N terminus, and deletion of the second coiled-coil domain of End4p in an abp1 background does not confer inviability at 37°C. To determine whether the coiled-coil region of End4p is required for organization of the actin cytoskeleton in addition to growth at high temperature in an abp1 background, we looked at the actin cytoskeleton in end4coil1 abp1 mutants at their permissive growth temperature of 24°C. In wild-type cells and in end4coil1 and abp1 single mutants, actin cables traversing the mother cell and cortical actin concentrated in daughter cells could be seen (Figure 4A). In marked contrast, end4coil1 abp1 double mutants have a strong defect in actin cytoskeleton organization and cells are larger and rounder than wild-type cells. Notably, the actin defect is similar to the one observed in end4N1 mutants, though in end4coil1 abp1 mutants, the loss of polarization of actin patches is not as strong as in end4N1 mutants. Thus, in the absence of ABP1, deletion of the first coiled-coil domain of End4p results in a disorganized actin cytoskeleton at 24°C and impaired growth at 37°C. Because a defective actin cytoskeleton paralleled the endocytosis defect in all end4 mutants examined so far, we tested whether this was also true for the end4coil abp1 double mutants by measuring endocytosis at 24°C. We calculated the amount of pheromone internalized per minute in the linear range of 5 to 15 min after pheromone addition and expressed this as a percentage of the wild-type endocytic rate. Both end4coil1 and abp1 single mutants internalized -factor with wild-type kinetics (Figure 4B; Kübler and Riezman, 1993). Strikingly, deletion of the first coiled-coil domain of End4p in an abp1 background completely abolished endocytosis even at the permissive growth temperature of 24°C. As expected, the same effect was observed when both coiled-coil domains of End4p were deleted in the absence of ABP1, whereas combination of end4Pro, end4N2, end4coil2, and end4talin with abp1 did not significantly reduce endocytosis below the levels of the respective single mutants at any temperature (our unpublished observations). This strongly suggests that the coiled-coil domain of End4p mediates a function in endocytosis that is redundant with the function(s) of ABP1.

View larger version (62K): [in this window] [in a new window]   Figure 4.   Abp1p and Srv2p have cryptic endocytic functions. (A) Visualization of filamentous actin with rhodamine-conjugated phalloidin of END4 (RH3393), abp1 (RH3297), end4coil1 (RH3395), and end4coil1 abp1 (RH3407) cells grown at 24°C. (B) Internalization rates expressed as a percentage of wild-type (RH3393) in end4coil1 (RH3395), abp1 (RH3297), and srv2 (RH4008) single mutants and in end4coil1 abp1 (RH3407) or end4coil1 srv2 (RH4009) double mutants transformed with the indicated plasmids, respectively. p[abp1SH3] encodes a mutant allele of ABP1 with a deleted SH3 domain; p[SRV2-1] encodes wild-type SRV2 and p[SRV2-5] encodes a mutant allele of SRV2 that lacks both binding sites for the SH3 domain of Abp1p. All plasmids are centromere-based and described in Lila and Drubin (1997).

Using a previously described screen for endocytosis mutants (Munn and Riezman, 1994), we isolated a novel mutant, end14-1, that displays a temperature-sensitive growth phenotype. The endocytic defect of end14-1 was tightly linked to impaired growth at 37°C through several back crosses, thus a single mutation caused both phenotypes. We used a CEN library to clone the wild-type END14 gene by phenotypic complementation of the temperature-sensitive growth defect of an end14-1 strain. Integrative mapping demonstrated that the cloned DNA encoded the authentic END14 gene (MATERIALS AND METHODS). Mapping of the complementing ORF and sequence analysis revealed that END14 is allelic to SRV2. From now on we refer to end14-1 as srv2-14. At the permissive growth temperature of 24°C, srv2-14 mutants are already quite defective for pheromone internalization and endocytosis of both -factor and LY is completely blocked at 37°C (Figure 5, A and B). The srv2-14 mutation therefore impairs both receptor-mediated and fluid-phase endocytosis. Surprisingly, deletion of SRV2 did not confer an endocytosis defect (Figure 5A). This observation suggests that srv2-14 is not a simple loss-of-function allele but acts in a recessive negative manner. To determine whether srv2-14 mutants display a general defect in membrane transport, we followed the maturation of soluble vacuolar CPY. CPY is sequentially processed to its mature form because it is transported from the endoplasmic reticulum to the Golgi and finally to the vacuole. As shown in Figure 5C, CPY matures with wild-type kinetics in srv2-14 mutants, even at the restrictive growth temperature of 37°C. These results show that srv2-14 mutants are impaired in the internalization step of endocytosis but the early secretory pathway and the Golgi to vacuole transport steps remain unaffected.

View larger version (41K): [in this window] [in a new window]   Figure 5.   srv2-14, but not srv2, mutations confer an endocytic defect. (A) Endocytosis of radiolabeled pheromone in SRV2 (RH978) and srv2-14 (RH3993) strains was measured at 24°C (solid symbols) or 37°C (open symbols) after a 15-min (SRV2, srv2-14) preincubation at the indicated temperature. Endocytosis in srv2 (RH4008, triangles) and SRV2 (RH3393, circles) was measured after a 5-min preincubation at 24°C. (B) Accumulation of LY in SRV2 (RH978) and srv2-14 (RH3993) mutants grown at 24°C and preincubated for 15 min at 37°C prior to addition of LY for 1 h. (C) Maturation of radiolabeled CPY was measured in SRV2 (RH978) and srv2-14 (RH3993) mutants grown at 24°C and preincubated for 15 min at 37°C prior to starting a pulse-chase experiment as described in MATERIALS AND METHODS. p1, endoplasmic reticulum form of CPY; p2, Golgi-modified form of CPY; m, mature vacuolar form of CPY.

Recently, Lila and Drubin (1997) showed a functional interaction of the SH3 domain of Abp1p and a proline-rich sequence of Srv2p. We therefore tested whether the SH3 domain of Abp1p, and thus the binding of Abp1p to Srv2p, is required for the redundant endocytic function of Abp1p. As shown in Figure 4B, deletion of the SH3 domain of Abp1p in an end4coil1 background results in a complete loss of endocytosis at 24°C. This is consistent with the finding that Abp1p is required for endocytosis in the absence of the coiled-coil domain of End4p and it also demonstrates that the endocytic function of Abp1p depends on its SH3 domain. If binding of Abp1p via its SH3 domain to Srv2p were required for endocytosis in the end4coil1 mutant, one would expect that the absence of Srv2p or mutations in Srv2p that abolish binding of Srv2p to the SH3 domain of Abp1p would mimic the end4coil1 abp1SH3 endocytosis defect if combined with an end4coil1 mutation. We therefore measured endocytosis of -factor in strains carrying both the end4coil1 mutation and a deletion of SRV2 or a deletion of the Abp1p-binding sites of Srv2p (SRV2-5; Lila and Drubin, 1997). Even though strains lacking Srv2p (such as strains lacking Abp1p) had no measurable endocytosis defect (Figures 4B and 5A), endocytosis in end4coil1 srv2 double mutants was completely blocked, very similar to the end4coil1 abp1 double mutant. Surprisingly, end4coil1 SRV2-5 double mutants internalized pheromone almost as efficiently as wild-type strains (Figure 4B). This suggests that binding of Srv2p to Abp1p, and therefore the cortical localization of Srv2p (Lila and Drubin, 1997), is most likely not required for the cryptic endocytic function of Abp1p or Srv2p. Thus, these results show that the coiled-coil domain of End4p mediates an endocytic function that is redundant with the endocytic function(s) of Abp1p and Srv2p.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

END4 has previously been shown to be required for the process of endocytosis. We used a mutational analysis of END4 to correlate in vivo functions with regions of End4p and we found that two parts of End4p participate in endocytosis. One of them resides in the N-terminal region and is essential for endocytosis and actin organization. The other contains a coiled-coil domain of End4p that is required for partitioning of End4p into a proteinaceous complex. In addition, this domain mediates an endocytic function that is redundant with function(s) of ABP1 and SRV2. On the basis of biochemical, functional, and genetic data, we propose that END4 may control endocytosis through the interaction with other actin-associated proteins (Figure 6).

View larger version (27K): [in this window] [in a new window]   Figure 6.   Hypothetical model. Two parts of End4p participate in endocytosis: the N terminus, which is essential for endocytosis and actin organization, and the coiled-coil domain, which mediates an endocytic function redundant with Abp1p and Srv2p. We propose that this function involves binding to Rvs167p, a protein itself required for endocytosis. The shaded arrow indicates that the physical interaction of End4p and Rvs167p is based solely on two-hybrid data. The open arrow indicates that a recessive negative mutation but not a deletion of SRV2 results in a defect in endocytosis (see text for further explanations).

A deletion of the N-terminal domain of End4p completely abolished endocytosis. In addition, end4N1 mutants display a loss of actin organization. Moreover, an actin defect always paralleled an endocytic defect of end4 mutants, suggesting that an actin-related function of End4p is essential for endocytosis. However, it is unlikely that end4N1 is a true end4 null allele for the following reasons. First, the mutant allele fully complemented the germination defect of a null allele. Second, the mutant protein is stably expressed, and third, end4N1 mutants grow faster and reach higher density in stationary-phase cultures than end4 mutants (our unpublished observations). Perhaps the N-terminal part of End4p interacts with another protein(s) to mediate correct localization or maintenance of actin structures that may be a prerequisite for endocytosis. In support of this idea End4p stimulates polarized actin polymerization in vitro (Li et al., 1995) and the talin-like domain of End4p was shown to be sufficient for interaction with actin filaments (McCann and Craig, 1997). The fact that deletion of this talin-like domain of End4p has no impact on endocytosis and actin organization suggests that either this function is not required for endocytosis or that other protein(s) could substitute for this loss of function.

Deletion of a central coiled-coil domain of End4p results in failure to partition into a proteinaceous complex. Furthermore, the absence of this region creates a synthetic endocytic defect in the absence of Abp1p or Srv2p. Thus, deletion of the coiled-coil domain of End4p clearly results in different phenotypes than those that are associated with an end4N1 allele. Consistent with this, end4N1 abp1 double mutants are inviable whereas end4coil abp1 double mutants are viable. It is therefore possible that the coiled-coil domain is functionally different from the N-terminal region of End4p. On the basis of our results, we propose that END4, ABP1, and SRV2 have an overlapping endocytic function that is mediated through the coiled-coil domain of End4p. Consistent with this, deletion of the coiled-coil domain of End4p or deletion of ABP1 or SRV2 does not produce an obvious endocytic defect. However, combined deletions of the coiled-coil domain of End4p and deletion of ABP1 (or its SH3 domain) or deletion of SRV2 lead to a complete endocytosis defect, supporting the view of a common function of End4p, Abp1p, and Srv2p.

Abp1p (via its SH3 domain) binds to proline-rich sequences of Srv2p (Lila and Drubin, 1997) and both Abp1p or Srv2p are required for the redundant endocytic function with End4p. We therefore asked, whether an association of Srv2p with Abp1p was required for their endocytic function. The answer, however, is still ambiguous. Whereas deletion of the proline-rich sequences of Srv2p did not abrogate endocytosis in an end4coil1 mutant, deletion of the SH3 domain of Abp1p completely blocked endocytosis in an end4coil1 mutant. It could be that deletion of the SH3 domain of Abp1p is more deleterious to the physical interaction of Abp1p and Srv2p in vivo than deletion of the proline-rich sequences of Srv2p. Or as proposed (Lila and Drubin, 1997), the SH3 domain of Abp1p could bind other proteins in addition to Srv2p. It could be that one of those proteins is required for endocytosis in an end4coil1 mutant.

What could the common function of End4p, Abp1p, and Srv2p be? We speculate that it involves contacting Rvs167p. It has been shown that Abp1p binds to Rvs167p and that Abp1p provides a physical connection between Srv2p and Rvs167p. Interestingly, initial two-hybrid analysis indicates that the coiled-coil domain of End4p interacts with Rvs167p (Lombardi, unpublished observations). Consistent with this hypothesis that the common and essential function for endocytosis of the coiled-coil region of End4p, Abp1p, and Srv2p may be a physical connection to Rvs167p is the finding that Rvs167p is essential for endocytosis (Munn et al., 1995). Alternatively, Abp1p and other proteins may be required for their mutual localization to perform their function(s). Failure to properly localize Rvs167p, Rvs161p, and Srv2p in abp1 strains could be the reason why end4 abp1 double mutants are inviable. Consistent with this view, combined null alleles of end4 rvs167, end4 rvs161, and end4 srv2 double mutants are inviable, although Abp1p is present. It could therefore be that the synthetic effects observed with the End4p coiled-coil mutant and abp1 or srv2 reflect partial effects on two distinct functions, one mediated by End4p and the other mediated by Rvs167p.

SRV2 has been isolated genetically as a gene required for RAS-mediated activation of the adenylate cyclase pathway and biochemically as an adenylate cyclase associated protein (Fedor-Chaiken et al., 1990; Field et al., 1990). However, SRV2 has functions that are clearly not related to the RAS/adenylyl cyclase pathway and it was shown that Srv2p independently integrates RAS function and regulates the actin cytoskeleton organization (Field et al., 1990; Gerst et al., 1991). Most recently, Freeman et al. (1996) demonstrated that Srv2p binds to Abp1p independently of its cyclase association or actin-binding function. We report herein the isolation of a novel endocytosis mutant, srv2-14. It is likely that the recessive negative srv2-14 mutation interferes with endocytosis by disrupting an actin-related function because actin is known to be required for endocytosis and because mutations that either increase or decrease the activity of adenylate cyclase do not affect endocytosis (Zanolari and Riezman, unpublished observations). How could it do this? It could be that Srv2-14p can enter into a protein complex and modify its function. As Srv2p is known to bind Abp1p, perhaps it binds Abp1p in such a manner that it can no longer release Rvs167p. This would sequester Rvs167p, which is strictly required for endocytosis. In this case, one would have to postulate that the wild-type Srv2p would successfully compete with the mutant Srv2-14p for binding to Abp1p, otherwise the mutation would be dominant. A complete loss of function of SRV2 would not block endocytosis because the affinity of Abp1p for Rvs167p would not be increased. In the future it will be interesting to characterize this novel mutant srv2-14 protein and its interactions with other actin-associated proteins.