GCS1, an Arf Guanosine Triphosphatase-activating Protein in Saccharomyces cerevisiae, Is Required for Normal Actin Cytoskeletal Organization In Vivo and Stimulates Actin Polymerization In Vitro

Abstract

Recent cloning of a rat brain phosphatidylinositol 3,4,5-trisphosphate binding protein, centaurin α, identified a novel gene family based on homology to an amino-terminal zinc-binding domain. In Saccharomyces cerevisiae, the protein with the highest homology to centaurin α is Gcs1p, the product of the GCS1 gene. GCS1 was originally identified as a gene conditionally required for the reentry of cells into the cell cycle after stationary phase growth. Gcs1p was previously characterized as a guanosine triphosphatase-activating protein for the small guanosine triphosphatase Arf1, and gcs1 mutants displayed vesicle-trafficking defects. Here, we have shown that similar to centaurin α, recombinant Gcs1p bound phosphoinositide-based affinity resins with high affinity and specificity. A novel GCS1 disruption strain (gcs1Δ) exhibited morphological defects, as well as mislocalization of cortical actin patches. gcs1Δ was hypersensitive to the actin monomer-sequestering drug, latrunculin-B. Synthetic lethality was observed between null alleles of GCS1 and SLA2, the gene encoding a protein involved in stabilization of the actin cytoskeleton. In addition, synthetic growth defects were observed between null alleles of GCS1 and SAC6, the gene encoding the yeast fimbrin homologue. Recombinant Gcs1p bound to actin filaments, stimulated actin polymerization, and inhibited actin depolymerization in vitro. These data provide in vivo and in vitro evidence that Gcs1p interacts directly with the actin cytoskeleton in S. cerevisiae.

INTRODUCTION

Inositol lipids are involved in diverse pathways in eukaryotic cells, acting as membrane localization signals, working as cofactors for numerous enzymes, serving as substrates for the production of second messengers, and functioning as bona fide second messengers (for reviews see Lee and Rhee, 1995; De Camilli et al., 1996; Toker and Cantley, 1997). Recent interest has focused on the D-3 phosphoinositides: phosphatidylinositol (PtdIns) 3-phosphate (PtdIns(3)P), PtdIns 3,4-bisphosphate (PtdIns(3,4)P₂), PtdIns 3,5-bisphosphate (PtdIns(3,5)P₂), and PtdIns 3,4,5-trisphosphate (PtdIns(3,4,5)P₃), which are synthesized by constitutively active or receptor-stimulated phosphoinositide 3-kinases. Phosphoinositide 3-kinases are required for many fundamental cellular processes, including cell growth and survival, vesicular trafficking, and cytoskeletal organization (reviewed by Vanhaesebroeck et al., 1996; Toker and Cantley, 1997).

Regulation of these cellular processes is presumably mediated by the interaction of D-3 phosphoinositides with specific intracellular targets (Theibert et al., 1997). Numerous candidate targets for D-3 phosphoinositides have now been identified, including protein kinases and proteins involved in the regulation of vesicle trafficking and the actin cytoskeleton (see references in Toker and Cantley, 1997). Our laboratory has identified and cloned a rat brain PtdIns(3,4,5)P₃-binding protein, centaurin α, (Hammonds-Odie et al., 1996), and two related PtdIns(3,4,5)P₃-binding proteins were subsequently identified (Stricker et al., 1997; Tanaka et al., 1997). The deduced amino acid sequence predicts that centaurin α contains a pleckstrin homology (PH) domain and a putative zinc-binding domain (Hammonds-Odie et al., 1996). PH domains have been implicated in phosphoinositide binding in a variety of proteins (Gibson et al., 1994; Klarlund et al., 1997). Within the zinc-binding domain, centaurin α is similar to numerous proteins (Hammonds-Odie et al., 1996), including a rat liver Arf1 guanosine triphosphatase (GTPase)-activating protein (GAP) (Cukierman et al., 1995) and several yeast proteins (Ireland et al., 1994; Zhang et al., 1998).

In Saccharomyces cerevisiae, the protein with the highest degree of structural homology to centaurin α is Gcs1p. The GCS1 gene was originally identified as a cold-sensitive mutant that failed to resume logarithmic growth from stationary phase, a G₀ to G₁ progression (Ireland et al., 1994). Johnston and co-workers have shown that mutant gcs1 cells lose mitochondrial activity (Filipak et al., 1992) and exhibit vesicle trafficking defects at the nonpermissive 15°C temperature (Wang et al., 1996). Biochemically, Gcs1p displays Arf1p GAP activity (Poon et al., 1996) that has been localized to the zinc-binding domain (Antonny et al., 1997). Deletion of GCS1 in an arf1 null background results in a strong synthetic growth defect (Poon et al., 1996). In addition, overexpression of GCS1 or several related proteins, including GLO3 and SAT1, rescues an arf1 temperature-sensitive (t.s.) mutant (Zhang et al., 1998).

Arfs are members of the Ras GTPase superfamily that have been implicated in regulation of vesicle trafficking and the actin cytoskeleton in mammalian cells. Arfs have been shown to function in endoplasmic reticulum and Golgi transport, endocytosis, and exocytosis (Boman and Kahn, 1995). In vitro, mammalian Arfs are required for the recruitment of coat proteins in various vesicle-budding assays (Orci et al., 1993; Faundez et al., 1997) and stimulate phospholipase D activity (reviewed by Cockcroft, 1996). In mammalian cells, Arf6 is localized to the plasma membrane, and overexpression leads to alterations in the actin cytoskeleton (Radharkrishna et al., 1996; D’Souza-Schorey et al., 1997). Although several yeast Arf proteins have been characterized and implicated in the secretory pathway (Stearns et al., 1990; Lee et al., 1994), the mechanisms by which these Arfs function in vesicle trafficking and whether they are involved in regulation of the actin cytoskeleton in yeast are unresolved issues.

In addition to the conserved zinc-binding and PH domains, centaurin α and several centaurin homologues contain ankyrin repeats and an ezrin/radixin/moesin (ERM) homology domain, suggesting that they may interact with the actin cytoskeleton (Hammonds-Odie et al., 1996). To investigate whether this protein family may function in vivo via interactions with the actin cytoskeleton, we focused the current study on GCS1. In this report, we demonstrate that Gcs1p binds phosphoinositides, consistent with the presence of a PH domain. Next, we provide morphological, pharmacological, genetic, and biochemical evidence that support a role for Gcs1p in regulation of the actin cytoskeleton in S. cerevisiae.

MATERIALS AND METHODS

All chemicals, purchased from Sigma Chemical (St. Louis, MO) or Fisher Scientific (Pittsburgh, PA), were of the highest grade available unless otherwise indicated. ENHANCE was from Dupont-New England Nuclear (Boston, MA). Oxalyticase was from Enzogenetics (Corwallis, OR). Latrunculin-B (Lat-B) was from Calbiochem (San Diego, CA), and rhodamine phalloidin was from Molecular Probes (Eugene, OR).

Strains and Growth Medium

The genotypes of the strains used in this study are listed in Table 1. YPD, yeast minimal medium, presporulation medium, and sporulation medium have been described previously (Kaiser et al., 1994). To generate the gcs1Δ strain, the 5′- and 3′-regions of GCS1 were generated from yeast genomic DNA by PCR using the following sets of primers: 1) 5′-GGGAATTCTTATAAGCAGA TCTTTGGGGC-3′ (16A-1) and 5′-GGGGATCCCCATACGAAGAAGTTCCTCCGG-3′ (16A-2) and 2) 5′-GGGGATCCAGGCCGAGGACAAATGGGACG-3′ (16A-3) and 5′-GGGCATGCATG TCAATAA-GTAAGTGCCGC-3′ (16A-4), respectively (identity to the GCS1 gene is underlined). To generate pGCS-HIS3, the PCR products were cloned sequentially into the pTZ18 vector, after which the HIS3 gene was cloned into the BamHI site generated by the PCR primers.

Table 1.

Yeast strains used in this study

Strain	Genotype	Source
CTY182	Mat a, ura3-52 Δhis3-200, lys2-801AM	V. A. Bankaitis
CTY3	Mat α, ura3-52, Δhis3-200, Δtrp1, ade3	V. A. Bankaitis
YAT1	Mat α, ura3-52, Δhis3-200, Δtrp1	This study
YAT2	Mat a, ura3-52, Δhis3-200, Δtrp1, GCS1∷HIS3	This study
YAT3	Mat a, ura3-52, Δhis3-200, Δtrp1, GLO3∷HIS3	This study
W303-1a	Mat a, leu2-3,112, ura3-1, his3-11, trp1-1, ade2-1	G. C. Johnston
GWK9a	Mat a, leu2-3,112, ura3-1, his3-11, trp1-1, ade2-1, gcs1Δ∷URA3	G. C. Johnston
PPY147.28.2C	Mat α, leu2-3,112, ura3-1, his3-11, trp1-1, ade2-1, gcs1Δ∷URA3, (pPY805.25)	G. C. Johnston
PPY147.28.2D	Mat a, leu2-3,112, ura3-1, his3-11, trp1-1, ade2-1	G. C. Johnston
DDY216	Mat α; Δsac6∷URA3; lys2; trp1; his3; leu2; ura3	D. G. Drubin
DDY318	Mat α; Δsac6∷LEU2; lys2; his3; leu2; ura3	D. G. Drubin
DDY322	Mat α; Δabp1∷LEU2; Δhis3-200; leu2-3,112; ura3-52	D. G. Drubin
DDY340	Mat α; act1-104∷HIS3; Δhis3-200; leu2-3,112; ura3-52; can1-1; tub2-201	D. G. Drubin
DDY342	Mat α; act1-113∷HIS3; Δhis3-200; leu2-3,112; ura3-52; can1-1; tub2-201	D. G. Drubin
DDY343	Mat α; act1-115∷HIS3; Δhis3-200; leu2-3,112; ura3-52; ade2-101; can1-1; tub2-201	D. G. Drubin
DDY344	Mat α; act1-116∷HIS3; Δhis3-200; leu2-3,112; ura3-52; can1-1; tub2-201	D. G. Drubin
DDY351	Mat α; act1-129∷HIS3; Δhis3-200; leu2-3,112; ura3-52; can1-1; tub2-201	D. G. Drubin
DDY546	Mat α; Δsla2∷URA3; Δhis3-200; leu2-3,112; lys2-801am; ura3-52	D. G. Drubin
DDY950	Mat α; Δrvs167∷TRP1; lys2; trp1; leu2; ura3	D. G. Drubin

Gene disruptions were generated by digesting pGCS-HIS3 with EcoRI and SphI. Wild-type yeast was transformed by the lithium acetate method and plated onto selective (−histidine) media plates. Colonies were picked and screened by PCR using a primer whose sequence lies outside of the region disrupted: 5′-TCATGCTGACGACGTAC-3′ and 16A-4. Positive clones were backcrossed three times to an isogenic parental wild-type strain. Tetrads from a heterozygous GCS1/GCS1::HIS3 diploid strain were analyzed to determine whether the GCS1::HIS3 disruption segregated with mutant phenotypes (Lat-B sensitivity, NaCl sensitivity, and actin mislocalization). At least 20 tetrads were analyzed from each cross, and it was determined that the GCS1::HIS3 disruption segregated with the mutant phenotypes. One wild-type (YAT1) and one gcs1Δ (YAT2) spore were chosen and used throughout this study.

Growth Assays

To assess the reentry phenotype, cells were grown for 5 d and then diluted to early-log phase in fresh YPD and shifted to the indicated temperatures. The criteria for assessing stationary phase were consistent with those of Singer and co-workers (Drebot et al., 1987). Wild-type cells were assessed by light microscopy, and 90% of the cells were unbudded and reached maximal density at least 48 h earlier. Culture densities was measured by diluting cell aliquots into sonication buffer (PBS containing 1 mM EDTA and 1 mM EGTA) and sonicating for 10 s to disperse cell clumps (Pringle and Mor, 1975). Dilutions were performed such that the optical density measured 0.1–0.6 U as measured in a spectrophotometer at 595 nm.

Gcs1p Fusion Protein Purification

pQE-GCS1, a plasmid encoding for a His₆-Gcs1p fusion protein, was generated by isolating a full-length GCS1 PCR product using the following primers: 1) 5′-GGGGATCCATGTCAGATTGGAAAG-TGG-3′ and 2) 5′-GGGCATGCTTAGAAATCGTCCCATTTGTCC-3′ (underlined regions indicate identity to GCS1). The PCR product was gel purified and ligated in frame into the BamHI/SphI sites of the pQE-30 His tagged vector (Qiagen, Chatsworth, CA).

Overnight cultures of pQE-GCS1 or pQE-40 (His₆-DHFR) were diluted 1:50 into 2 l of LB medium supplemented with 100 μg/ml ampicillin and grown uninduced for a further 8 h at 37°C. Cells were collected by centrifugation, and either native or denatured fusion proteins were batch purified using Ni⁺-NTA agarose beads. Denatured fusion protein was purified as described by the manufacturer and dialyzed for 12 h against 4 M urea, 0.1 M sodium phosphate, 0.01 M Tris, pH 8.0, followed by 12 h against 2 M urea, 0.01 M Tris, pH 8.0, and then twice more for 12 h against 0.01 M Tris, pH 8.0. Native protein was purified by lysing cells at 4°C for 90 min in 50 mM sodium phosphate, pH 7.8, 300 mM NaCl (buffer A) supplemented with 5 μg/ml lysozyme and 1% Triton X-100. Lysates were clarified by centrifugation and incubated with Ni⁺-NTA agarose beads in binding buffer (10 mM imidazole in buffer A) for at least 2 h at 4°C. Beads were washed extensively with binding buffer followed by stepwise washes with Triton X-100 to a final concentration of 0.1%. Protein was eluted 200 mM imidazole, 0.1% Triton X-100 buffer A and dialyzed twice for 12 h against 10 mM Tris, pH 7.4, 0.1% Triton X-100 final concentration. Protein concentrations were determined by both SDS-PAGE and Coomassie Protein Assay (Pierce Chemical). To determine whether His₆-Gcs1p fusion protein was biologically active, we assessed GAP activity by performing Arf1 GAP assays with recombinant Arf1p as described previously (Poon et al., 1996) and detected GAP activity similar to that reported (our unpublished results).

Polyclonal Antibody Production

Anti-Gcs1p antisera were prepared by immunization with the Ni²⁺-NTA–purified denatured His₆-Gcs1p fusion protein. The antigen was injected by Southern Biotechnology Associates (Birmingham, AL) into a rabbit using a standard immunization protocol.

Yeast Cell Extract Preparation and Immunoblotting

Yeast whole-cell extracts were prepared as described by Kaiser et al. (1994). Briefly, midlogarithmic cells were resuspended in SDS-PAGE sample buffer and boiled for 3 min. Glass beads were added, and the cells were vortexed vigorously for 2 min. Samples were boiled a second time for 3 min and were separated by SDS-PAGE. After transfer to nitrocellulose membranes, the lysates were immunoblotted using a 1:10,000 dilution of the anti-Gcs1p antisera.

Phosphoinositide-Binding Assays

Samples (300 μl) of Ni²⁺-NTA–purified fusion protein were incubated with 100 μl of a 1:1 slurry of Affigel-conjugated aminopropyl-inositol(1,3,4,5)P₄ (aminopropyl-InsP₄) (Hammonds-Odie et al., 1996) for 1 h at 4°C in binding buffer (10 mM Tris, pH 7.4, 50 mM NaCl). The beads were pelleted, the flow-through collected, and the beads were then washed in 1 ml of binding buffer. Protein was eluted from the resin by incubating the beads with 150 μl of SDS-PAGE sample buffer. Samples were separated by 10% SDS-PAGE, transferred to nitrocellulose, and immunoblotted using the anti–RGS-His₆ antibody (Qiagen). The presence of two additional bands in the “eluate,” which were more intense than the total fraction, suggests that His₆-Gcs1p underwent some degradation during the assay. To determine the affinity of His₆-Gcs1p to various phosphoinositides, competition binding assays were performed with the addition of phosphoinositides from a 10× stock to the binding reaction. Phosphoinositides with dipalmitoyl groups, PtdIns(3,5)P₂, and PtdIns(3,4,5)P₃, synthesized as described previously (Chen et al., 1996, 1998; Gu and Prestwich, 1996; Prestwich, 1996; Peng and Prestwich, 1998), were a generous gift from Echelon Research Laboratories (Salt Lake City, UT). PtdIns(3)P and PtdIns(3,4,5)P₃ were from Matraya (Pleasant Gap, PA). PtdIns, PtdIns(4)P, and PtdIns(4,5)P₂ were from Sigma Chemical. Blots were quantified using a Bio-Rad (Richmond, CA) densitometer.

Photolabeling was performed essentially as previously described (Hammonds-Odie et al., 1996) using fusion protein eluted from the aminopropyl-InsP₄ column with 1.5 M NaCl. Briefly, 100 ng of purified fusion protein in 10 mM Tris, pH 7.4, 1 mM EDTA were incubated with 110 nCi of [³H](3-[4-benzoyldihydrocinnamidyl]-propyl)-inositol tetrakisphosphate ([³H]BZDC-Ins(1,3,4,5)P₄) photoprobe (30 Ci/mM) in a final volume of 50 μl. Displacement was determined using the indicated concentrations of unlabeled phosphoinositides. Mixtures were exposed to 360 nm UV light for 1 h on ice, and the reactions were terminated with the addition of SDS sample buffer. Reactions were separated by 10% SDS-PAGE, and the gels were fixed and prepared for fluorography using the Enhance system. Autoradiographs were analyzed using a Bio-Rad densitometer.

Fluorescence Microscopy

Cells were grown overnight at 30°C and then diluted into YPD to early-log phase (0.2 OD/ml). After 6 h (mid-log phase), formaldehyde was added directly to the culture (3% final concentration) and incubated for 30 min at room temperature. Cells were resuspended in sonication buffer + 3% formaldehyde, sonicated for 10 s to disperse cell clumps, and incubated overnight at room temperature. Cells were washed three times in sonication buffer.

The actin cytoskeleton was visualized with rhodamine phalloidin (Molecular Probes) as previously described (Pringle et al., 1989). Cells were visualized with a 100× objective using a Leica DMRB microscope. All fields were photographed for equivalent exposure times at 400 ASA using Ilford δ 400 black and white film with a Leica Wild-MPS52 camera equipped with a 10× multiplier tube. Prints were developed to optimize visualization of the actin patches and cables.

To quantify the number of actin patches, the mother and bud cells of at least 200 randomly selected budding cells were counted by focusing up and down through the cell. Small buds were identified to be no more than 30% the size of the mother cell, and large buds were those larger than 30% of the mother cell. A standard two-tailed Student’s t test was employed to determine whether there was a significant change in the distribution of the number of patches between wild-type and gcs1Δ cells.

Halo Assays

Sensitivity to Lat-B was performed essentially as previously described (Ayscough et al., 1997). Briefly, 10 μl of midlogarithmic growth phase cells were added to 2 ml of 2× YPD or the appropriate selective minimal media, after which 2 ml of 1% agar were added to the cells, and the mixture was poured onto the surface of YPD or selective minimal media plates. Lat-B was diluted into DMSO and 4 μl of vehicle or the indicated concentration of Lat-B were pipetted onto a 6-mm filter disk (Scientific Specialties Group, Mt. Holly, PA), which was placed onto the top agar. Plates were placed at the indicated temperatures for 24 (YPD) to 72 (minimal media) h. Relative sensitivity was calculated as described previously (Reneke et al., 1988).

Genetic Interactions

All procedures were essentially as described by Sherman et al., (1986). Haploid mat α gcs1Δ yeast were crossed with haploid mat α yeast containing either mutant alleles of the ACT1 gene or deletions in the SLA2, SAC6, ABP1, or RVS167 genes (Table 1). Heterozygous diploids were selected on synthetic plates, and two separate colonies from each cross were chosen for culture and subsequent sporulation. The tetrad spores were dissected onto YPD plates and grown at 26°C for 4–5 d. Auxotrophies and temperature sensitivities were tested by frogging onto synthetic media or YPD.

Actin Cosedimentation

To assess the binding of Gcs1p to filamentous actin (F-actin), an actin cosedimentation assay was performed, similar to that described by Yao et al. (1996). F-actin was incubated with His₆-Gcs1p (prespun at 270,000 × g for 10 min at room temperature) in a total volume of 25 μl for 30 min at room temperature. All buffers and His₆-Gcs1p were supplemented with 0.1 volume of 10× F-buffer, and the Triton X-100 final concentration was 0.06%, which did not effect actin polymerization. The reactions were centrifuged at 270,000 × g for 10 min at 25°C. Supernatants were removed and the pellets were resuspended in SDS-sample buffer, and proteins were separated by 10% SDS-PAGE. After transfer to nitrocellulose, the His₆-Gcs1p was detected using the anti-RGS-His₄ antibody (Qiagen).

Light Scattering Assays

Monomeric actin in G-buffer (5 mM Tris [pH 7.4], 0.2 mM CaCl₂, 20 μM ATP, 20 μM DTT) was prespun at 270,000 × g at 4°C for 1 h; 10× initiation buffer (1 M KCl, 20 mM MgCl₂, 5 mM ATP) was added to initiate polymerization in the absence or presence of His₆-Gcs1p, and light scattering was measured at a 90° angle in a PTI Deltascan Fluorescent Spectrophotometer (Piscataway, NJ) at 400 nm. Actin depolymerization was performed by diluting 2 μM F-actin 20-fold into G-buffer in the absence or presence of His₆-Gcs1p.

Pyrene Actin Polymerization Assays

Actin polymerization was performed as described by the manufacturer (Cytoskeleton, Denver, CO). Briefly, 5 μM final concentration of monomeric actin (1:10 pyrene labeled) was incubated on ice for 10 min with the indicated concentrations of His₆-Gcs1p or His₆-DHFR. Samples were then equilibrated 10 min in a fluorescent spectrophotometer (ISS, Champaign, IL), after which polymerization was induced by the addition of KCl, MgCl₂, and ATP.

RESULTS

Gcs1p Contains a Putative Zinc-Binding Domain, a PH Domain, and ERM Homology Domain

Centaurin α is a mammalian brain PtdIns(3,4,5)P₃-binding protein that has an N-terminal cysteine-rich putative zinc-binding domain, a C-terminal PH domain, and homology to the actin-binding domain of the ERM family of cytoskeletal proteins (Hammonds-Odie et al., 1996). Database comparisons indicated that GCS1 is the yeast gene that shares the highest degree of structural homology to centaurin α (Figure 1A). Gcs1p contains an amino-terminal CxxCx₁₆CxxC putative zinc-binding domain (Figure 1B), which is 53% identical and 73% similar to centaurin α. Gcs1p is also 56% identical and 71% similar to a recently cloned rat liver Arf GAP (Cukierman et al., 1995) in this region, which has been shown to contain the Arf GAP domain (Antonny et al., 1997). A number of other yeast proteins, such as Glo3p and Sat1p, are homologous to Gcs1p in this region (Zhang et al., 1998). In addition, Gcs1p contains a region homologous to an actin-binding domain in the ERM protein family (Figure 1C) (Turunen et al., 1994). The ERM proteins function to link the actin cytoskeleton to the plasma membrane by binding actin via the C terminus to plasma membrane proteins, such as CD44, via the N terminus (Tsukita et al., 1997). Furthermore, a PH domain consensus sequence is present in Gcs1p (Figure 1D). PH domains are functional motifs found in many signal-transduction and cytoskeletal proteins and have been shown to mediate phosphoinositide and protein interactions (Gibson et al., 1994; Lemmon et al., 1996; Klarlund et al., 1997). Numerous yeast proteins, including several that regulate small GTPases, such as BEM2, BEM3, and ROM2, contain PH domains. The PH domain in Gcs1p is also related to the PH domain found in centaurin α; however, GLO3 and the rat liver Arf GAP do not appear to contain this domain.

Figure 1.

Sequence comparison of GCS1. (A) Domain structure of Gcs1p. (B) Comparison of the CxxCx₁₆CxxC amino-terminal zinc-binding domain of GCS1 to rat brain centaurin α, the yeast proteins GLO3, GTS1, and SPS18, and the rat liver Arf GAP. (C) Comparison of Gcs1p with merlin and the ERM family of proteins: ezrin, radixin, and moesin. (D) Comparison of the PH domain from GCS1 to six yeast proteins that contain PH domains according to the Stanford Saccharomyces Genome Database (SGD): NUM1, BEM2, OSH1, BUD4, BEM3, and BOB1. Shaded amino acids represent conserved amino acids in a majority of the family members; capital letters are amino acids conserved between GCS1 and other proteins. Conserved amino acids belong to the same Dayhoff group (GPAST, MILV, KRH, NQED, FWY, C).

Gcs1p Binds Phosphoinositide-based Probes With High Affinity

The structural homology between centaurin α and GCS1, as well as the presence of a PH domain in Gcs1p, suggested that Gcs1p may bind phosphoinositides. The identification of phosphoinositide binding in centaurin α, the clathrin adaptor/assembly protein AP-2, and α-COP of the Golgi coatomer COPI complex was facilitated using phosphoinositide-based affinity probes such as aminopropyl-InsP₄ affigel (Hammonds-Odie et al., 1996; Prestwich, 1996; Chaudhary et al., 1998). The inositol polyphosphate head group is conjugated via an aminopropyl moiety to the matrix, which results in these probes having a higher hydrophobic character than the free inositol polyphosphate, and therefore presumably mimic the structure of a phosphoinositide (Hammonds-Odie et al., 1996). Purification of recombinant His₆-Gcs1p from bacterial lysates yielded a major protein band at 45 kDa and two proteolytic fragments, all of which were recognized by antibodies against the fusion tag. This purified His₆-Gcs1p specifically and efficiently interacted with the aminopropyl-InsP₄ resin (Figure 2A). Approximately 80% of the total His₆-Gcs1p bound to the resin, while ∼20% of the His₆-Gcs1p immunoreactivity was present in the “flow through” and “wash” fractions. Efficient recovery of the recombinant protein (“eluate”) from the resin was effected using high-ionic strength buffer conditions that had been previously established for the recovery of centaurin α from the same affinity resin. No detectable bacterial proteins interacted with the resin under the binding conditions used, and the recombinant Gcs1p comprised all of the protein in the eluate fraction.

Figure 2.

His₆-Gcs1p binding to PtdIns(3,4,5)P₃ analogs. (A) Ni²⁺-NTA agarose-purified His₆-Gcs1p (total) was incubated with the Affigel-aminopropyl-InsP₄ resin. The flow through was collected (flow through), and the resin was washed (wash) and eluted with 2× SDS-sample buffer (eluate). Fractions were separated by SDS-PAGE and subjected to immunoblot analysis using an anti-RGS-His₄ antibody. (B) His₆-Gcs1p, eluted from the Affigel-aminopropyl-InsP₄ resin with 1.5 M NaCl buffer, was dialyzed, and then incubated with 110 nCi [³H]BZDC-Ins(1,3,4,5)P₄ photolabel, in the absence (total) or presence of 10 μM unlabeled phosphoinositide shown. Proteins were separated by 10% SDS-PAGE, and the gels were fixed, dried, and fluorographed. (C) Competition of His₆-Gcs1p binding to the Affigel-aminopropyl-InsP₄ resin by including in the binding assay increasing concentrations of unlabeled phosphoinositides. The resin was washed, and bound protein was eluted with sample buffer and separated by SDS-PAGE. Gels were transferred to nitrocellulose, and the His₆-Gcs1p was detected with the anti-RGS-His₄ antibody. Blots were densitized, and the results are presented as a percentage of total binding in the absence of unlabeled phosphoinositide; each value represents at least three independent determinations.

His₆-Gcs1p was also efficiently photoaffinity labeled with a [[³H]]BZDC-Ins(1,3,4,5)P₄ photoprobe (Figure 2B). The labeling was displaced by addition of 10 μM PtdIns(3,4,5P)₃ (a lipid containing the Ins(1,3,4,5)P₄ head group of the photoprobe) or PtdIns(4,5)P₂ to the binding reaction. In yeast, the phosphoinositides identified to date are PtdIns, PtdIns(3)P, PtdIns(4)P, PtdIns(3,5)P₂, and PtdIns(4,5)P₂ (Dove et al., 1997). To characterize the binding specificity of His₆-Gcs1p, increasing concentrations of these physiologically relevant phosphoinositides were added to the aminopropyl-InsP₄ resin–binding assays. His₆-Gcs1p was displaced from the affinity resin by increasing concentrations of phosphoinositides. PtdIns(3,5)P₂ was the most potent displacer, followed by PtdIns, PtdIns(4,5)P₂, and PtdIns(3)P (Figure 2C). The IC₅₀ for displacement by PtdIns(3,5)P₂ was approximately 7 μM, similar to the IC₅₀ obtained with photoaffinity labeling (our unpublished data). These data demonstrate that Gcs1p binds phosphoinositide analogues with high affinity and specificity.

gcs1Δ Deletion Mutant Strains Display Mutant Growth and Morphological Phenotypes

A gcs1 deletion mutant strain (gcs1Δ), in which the majority of the GCS1 gene was deleted, was generated by replacing nucleotides 224-1021 with the HIS3 gene cassette (Figure 3A). PCR analysis indicated that the deletion construct was properly targeted to the GCS1 locus. By immunoblot analysis, using anti-Gcs1p antisera, no detectable Gcs1p protein was present in the gcs1Δ strain while the wild-type stain expressed Gcs1p during vegetative growth (Figure 3B). Similar to the reentry phenotype originally described for gcs1, this gcs1Δ strain also did not progress at 15°C from stationary phase to logarithmic growth. Although normal for growth at the 30°C permissive temperature, vegetative cells displayed several morphological defects (see below), suggesting that GCS1 may function not only in the transition from stationary phase, but also during vegetative growth. To determine whether gcs1Δ had any other growth phenotypes, cells were grown in the presence of elevated sorbitol or NaCl concentrations. gcs1Δ grew normally in YPD medium with high sorbitol concentrations (≤1.4 M), but grew slowly at 26°C, or not at all at 30 or 37°C on YPD medium containing 0.9 M NaCl (Figure 3C). In addition, gcs1Δ was unable to grow in YPD with 40 mM NaF (our unpublished data), a previously reported gcs1 mutant phenotype (Poon et al., 1996). These data demonstrate that Gcs1p is required for growth in various stress conditions at the permissive temperature.

Figure 3.

Disruption of GCS1. (A) GCS1 was disrupted by replacing nucleotides 224-1021 with the HIS3 gene by homologous recombination. (B) Genomic DNA isolated from wild-type and gcs1Δ was screened by PCR (left). Whole-cell lysates were prepared from midlogarithmic wild-type and gcs1Δ cells grown at 30°C were separated by SDS-PAGE and transferred to nitrocellulose. Gcs1p was detected by immunoblot analysis using antisera generated against full-length His₆-Gcs1p fusion protein (right). (C) Serially diluted cell suspensions (10 μl) were spotted onto YPD and YPD plates supplemented with either 1.4 M sorbitol or 0.9 M NaCl. The plates were incubated for 48–72 h at the indicated temperatures.

Previous reports show that gcs1 mutant strains exhibit endosomal and exocytic trafficking defects at the 15°C nonpermissive temperature, consistent with its reported role as an Arf1 GAP (Poon et al., 1996; Wang et al., 1996). At the permissive temperature, maturation of carboxypeptidase Y (a resident vacuolar enzyme) and secretion of invertase in gcs1Δ were normal (our unpublished results), indicating that Gcs1p is not essential for protein trafficking to the vacuole or the plasma membrane. In ultrastructural analysis by electron microscopy, no accumulation of intracellular membranous structures, such as 50-nm vesicles or collapsed Golgi, was observed in gcs1Δ cells. However, in contrast to wild-type cells that contain one to two large vacuoles, gcs1Δ cells exhibited numerous (>4) small membrane-bound vacuolar-like structures, suggesting that Gcs1p is required for normal vacuolar morphology (our unpublished results). The presence of aberrant vacuolar structures is a pleiotropic phenotype observed in strains with mutations in genes required for maintenance of the actin cytoskeleton and/or various vesicle-trafficking pathways.

Gcs1p Is Required For Normal Actin Cytoskeleton Distribution

Examination of gcs1Δ cells by phase microscopy indicated that many of the mutant cells were larger than the wild-type strain, were multibudded, and/or displayed elongated bud neck structures (Figure 4C). These morphological characteristics, as well as sensitivity to high-ionic strength medium and vesicle-trafficking defects, are phenotypes that are frequently observed in strains with mutations in genes required for organization of the actin cytoskeleton (Drubin et al., 1993; Ayscough and Drubin, 1996). In S. cerevisiae, the actin cytoskeleton shows a characteristic polarization as the cell progresses through the cell cycle. In unbudded cells, actin is concentrated in cortical actin patches at points juxtaposed to the membrane, from which the next daughter cell emerges. As a bud emerges, the cortical actin patches accumulate in the bud, followed by the emergence of actin cables, which are oriented along an axis between the mother and daughter cells. Strains harboring mutations in genes associated with the actin cytoskeleton can display changes in actin cables and/or cortical actin patch number and polarization (Welch et al., 1994).

Figure 4.

Actin cytoskeleton distribution of log-phase wild-type andgcs1Δ cells. Wild-type (A and B) and gcs1Δ (C and D) cells grown at 30°C were fixed, and stained with rhodamine-conjugated phalloidin as described in MATERIALS AND METHODS. Cells were viewed with Nomarski optics (left) and fluorescence rhodamine filter (right) with a 100× oil immersion objective. Representative fields are shown; note the increase in the number of actin patches in gcs1Δ. Focal planes were chosen to maximize the number of actin patches observed.

To assess the actin cytoskeleton, F-actin was examined in midlogarithmic wild-type and gcs1Δ cells grown at 30°C by staining fixed cells with rhodamine phalloidin (Figure 4). Staining in wild-type cells was consistent with the actin distributions described above. In contrast, an increased number of actin patches in the mother cell were present in the gcs1Δ strain. In addition, although actin cables were evident in gcs1Δ cells, the cables often appeared misaligned. To determine whether the apparent increase in the number of actin patches in gcs1Δ was statistically significant, we quantified and compared the number of patches in mother and daughter cells of the wild-type and mutant strains. Whereas only 25% of budding wild-type mother cells had >3 actin patches, ∼80% of the gcs1Δ budding cells had >3 actin patches (p < 0.001). The number of actin patches in the daughter cell was similar between the wild-type and gcs1Δ cells. This phenotype did not appear to be the result of a cell cycle-progression defect since both small and large budded gcs1Δ cells displayed an increase in the number of actin patches in the mother cell (our unpublished results). Together, these data suggest that Gcs1p is required for the normal organization of the actin cytoskeleton.

GCS1 Is Required In Vivo to Stabilize the Actin Cytoskeleton

Yeast strains carrying mutations in genes encoding proteins implicated in regulation of the actin cytoskeleton also exhibit altered sensitivity to reagents that disrupt actin polymerization or depolymerization. Lat-B, a cell-permeant marine toxin binds to monomeric G-actin and inhibits actin polymerization (Coue et al., 1987). Hypersensitivity to latrunculin, which is indicative of an unstable filamentous actin network, has been reported for mutant alleles in the actin-binding proteins sla2, cap2, srv2 (adenylyl cyclase-associated protein), and sac6 (a yeast fimbrin homologue) and in specific actin alleles, such as act1–111, act1–108, and act1–136 (Ayscough et al., 1997). To investigate the sensitivity of gcs1Δ cells to Lat-B, a halo sensitivity assay was employed (Ayscough et al., 1997). Compared with the parental wild-type strain, gcs1Δ cells were ∼2.5 times more sensitive to Lat-B (Figure 5A). A gcs1 null strain (gcs1–6) generated in an unrelated background strain (Ireland, et al., 1994) displayed a similar hypersensitivity to Lat-B. To verify that the increased sensitivity was the direct result of the loss of Gcs1p, halo assays were performed with both a wild-type and gcs1 null strain bearing a loss-of-function gcs1^ts plasmid. At the permissive 30°C temperature, both strains were equally sensitive to Lat-B, whereas at 37°C the null gcs1 strain harboring the gcs1^ts plasmid was ∼1.5 times more sensitive to Lat-B (Figure 5B). These data show that gcs1Δ cells are hypersensitive to Lat-B, a phenotype that points to a role for Gcs1p in stabilizing the actin cytoskeleton.

Figure 5.

Sensitivity of gcs1 mutants to Lat-B. (A) Halo assays were used to determine the sensitivity of wild-type (left) and gcs1Δ (right) cells to the indicated concentration of Lat-B. (B) Sensitivity of wild-type (left) and gcs1–6 (right) cells expressing the gcs1^ts plasmid at either 30 or 37°C.

Genetic Interactions between Mutants in GCS1, SLA2, and SAC6, Genes That Encode Actin-associated Proteins

The above results suggest that Gcs1p is important for organization of the actin cytoskeleton. To further establish an in vivo role for Gcs1p in regulation of the actin cytoskeleton, we tested for genetic interactions between gcs1Δ and 1) deletions of proteins known to be associated with the actin cytoskeleton in yeast or 2) mutants in the single-yeast actin gene, ACT1. Gene knockouts and actin mutants with a variety of phenotypes and genetic interactions were chosen. For example, mutants in the SLA2 gene are defective in actin polarization, endocytosis, and are temperature sensitive. (Ireland et al., 1994), whereas mutants in the ABP1 gene, encoding the actin-binding protein Abp1p, behave in a manner similar to wild-type cells (Drubin et al., 1993). Similarly, the act1–104 allele is not t.s. and has no reported synthetic genetic interactions with actin-associated proteins, whereas the act1–129 allele is temperature sensitive and is synthetic lethal in combination with sac6Δ, abp1Δ, and sla2Δ (Holtzman et al., 1994).

Haploid progeny that were determined to harbor both gcs1Δ and sla2Δ were inviable, indicating a synthetic lethality (Figure 6). In addition, gcs1Δ mutants, when combined with a deletion of the SAC6 gene (encoding fimbrin, an actin cross-linking protein) (Adams et al., 1991), demonstrated an inability to grow at 20 and 34°C. No synthetic effects were observed when the gcs1Δ mutation was combined with null mutants in the ABP1 or RVS167 genes or with any of the five ACT1 mutant alleles tested (Wertman et al., 1992). This suggests that deletion of the GCS1 gene results in defects in actin cytoskeleton regulation that are mild in an otherwise wild-type background, but severe in combination with certain mutations that themselves disrupt actin organization, specifically sla2Δ (Wertman et al., 1992) and sac6Δ (Adams et al., 1991). These genetic data are consistent with Gcs1p interacting in vivo with the actin cytoskeleton in S. cerevisiae.

Figure 6.

Genetic interactions between GCS1, actin-associated proteins, and actin mutants. gcs1Δ yeast were crossed with strains containing mutations in the single actin gene, ACT1, and with strains containing null mutants of a number of actin-associated proteins. (A) GCS1 and SLA2 are synthetic lethal. Shown are tetrads that were tetratype (T) or nonparental ditype (NPD), based on auxotrophies. In all cases, the haploid progeny predicted to be gcs1Δ, sla2Δ failed to grow at 25°C. (B) A table indicating the crosses made and the resulting temperature sensitivity of the double-mutant progeny. Note that certain of the mutants to which the gcs1Δ strain was crossed were already temperature sensitive. The only synthetic effect observed, other than with sla2Δ, was with sac6Δ. ^†, gcs1Δ, sac6Δ strains showed some variability in phenotype; some predicted double mutants failed to grow. The least severe temperature sensitivity is indicated. *, These crosses were made using GWK9A (gcs1Δ::URA3) provided by Dr. G.C. Johnston (see also Table 1).

Gcs1p Binds to F-Actin and Stimulates Actin Polymerization In Vitro

Homology with the actin-binding domain in the ERM protein family, in addition to the in vivo data described above, suggested that Gcs1p may interact directly with actin. To test for such interaction, an in vitro actin cosedimentation assay was performed by incubating purified recombinant His₆-tagged Gcs1p (Figure 7A) with polymerized F-actin (Figure 7B). In the absence of actin, the majority of His₆-Gcs1p remained in the supernatant. Addition of F-actin resulted in His₆-Gcs1p association with the F-actin– containing pellet. Addition of BSA, a protein that does not bind actin, to the cosedimentation reaction did not inhibit His₆-Gcs1p binding to F-actin, suggesting that the binding of His₆-Gcs1p to actin was specific. Moreover, a control His₆-DHFR fusion protein did not cosediment with F-actin, even at the highest F-actin concentrations tested (our unpublished data) demonstrating that actin did not bind the His₆ fusion tag. To characterize the binding of His₆-Gcs1p to F-actin, we incubated increasing concentrations of His₆-Gcs1p with F-actin (Figure 7, C and D). Binding was linear with increasing concentrations of His₆-Gcs1p and was saturable. Fifty percent maximal actin binding was observed with ∼75 nM His₆-Gcs1p, and at saturation the stoichiometry of His₆-Gcs1p:actin binding was 1:50.

Figure 7.

Gcs1p interacts with actin in vitro. (A) His₆-Gcs1p was purified using Ni²⁺-NTA resin. The purified protein was analyzed by Coomassie stain (lane 1) and by immunoblot analysis using an anti-RGS-His₄ antibody (lane 2). (B) Purified His₆-Gcs1p was incubated with 3 μM polymerized yeast F-actin in the presence or absence of BSA and centrifuged at 270,000 × g. The pellets (P) and supernatants (S) were separated by SDS-PAGE and subjected to immunoblot analysis using an anti-RGS-His₄ antibody. (C) Increasing concentrations (14–210 nM) of purified His₆-Gcs1p were incubated with 3 μM polymerized yeast F-actin and centrifuged at 270,000 × g. The pellets (P) and supernatants (S) were separated by SDS-PAGE and subjected to immunoblot analysis using an anti-RGS-His₄ antibody. (D) Quantification of immunoblot in panel C using a densitometer.

Numerous proteins that bind F-actin modulate actin polymerization dynamics. To examine whether Gcs1p regulates actin polymerization in vitro, a light scattering assay (Figure 8A) was employed. Two major stages of polymerization, nucleation and elongation, can be assessed with this assay as a time-dependent increase in light scattering. Addition of His₆-Gcs1p resulted in a dose-dependent stimulation of actin polymerization, leading to a decrease in the lag phase of polymerization, as well as an increase in the net polymerization rate and the steady-state level of actin polymerization. To test whether the increase in polymerization was a result of actin bundling, we also performed a polymerization assay using pyrene-labeled rabbit skeletal muscle actin. This assay measures actin polymerization as an increase in fluorescence intensity and is insensitive to the distribution of the filaments. His₆-Gcs1p stimulated actin polymerization in this assay with similar effects as seen in the light scattering assay (Figure 8B). The addition of His₆-DHFR did not stimulate actin polymerization, indicating that the effect was not due to the fusion tag or a bacterial contaminant.

Figure 8.

Time course for actin polymerization in the presence of His₆-Gcs1p. (A) Monomeric yeast G-actin (2 μM) was polymerized in the absence or presence of increasing concentrations of His₆-Gcs1p. Actin polymerization was monitored as an increase in light scattering as described in MATERIALS AND METHODS. Inset trace is a magnification of the first 105 s of the polymerization assay. In addition, 370 nM His₆-Gcs1p was incubated in the absence of actin to determine the intrinsic light scattering of His₆-Gcs1p. (B) Monomeric 1:10 pyrene-labeled rabbit muscle actin (5 μM): muscle actin was polymerized in the absence or presence of 0.2 μM His₆-Gcs1p or His₆-DHFR. Polymerization was monitored as an increase in fluorescence intensity as described in MATERIALS AND METHODS. Inset trace is a magnification of the first 300 s of the polymerization assay.

Shortening of the lag phase and enhanced rate of polymerization have been previously observed for proteins that nucleate actin polymerization. Another feature of several actin nucleating proteins is their ability to decrease the rate of actin depolymerization. Therefore, we examined whether addition of His₆-Gcs1p affected the rate of actin depolymerization after dilution of F-actin. Addition of His₆-Gcs1p inhibited actin depolymerization in a dose-dependent manner (Figure 9). Addition of heat-inactivated His₆-Gcs1p had no effect on actin polymerization or depolymerization (our unpublished data). These data show that Gcs1p can directly interact with actin, stimulate actin polymerization, and inhibit actin depolymerization in vitro. The ability of Gcs1p to both stimulate actin polymerization and block depolymerization suggests that it may function to stabilize actin filaments.

Figure 9.

Time course for actin depolymerization in the presence of His₆-Gcs1p. F-actin (2 μM) was diluted 30-fold into G-buffer containing increasing concentrations of His₆-Gcs1p. Actin depolymerization was monitored as an decrease in light scattering as described in MATERIALS AND METHODS.

DISCUSSION

In this report, we show that that Gcs1p is involved in regulating the actin cytoskeleton in S. cerevisiae. Five independent lines of evidence support this conclusion. First, Gcs1p contains a region of homology with the actin-binding domain of the ERM family of actin-binding proteins. Second, a gcs1Δ strain displayed mutant morphological and growth phenotypes, including mislocalized cortical actin patches, sensitivity to hyperionic conditions, and aberrant budding, that are consistent with defects in the actin cytoskeleton. Third, gcs1 mutant strains were hypersensitive to Lat-B, an actin monomer-sequestering drug. Fourth, synthetic growth defects were observed in strains in which gcs1Δ was combined with null mutations in either SLA2 or SAC6, whose gene products encode proteins implicated in the stabilization of the actin cytoskeleton. Fifth, Gcs1p bound F-actin, stimulated actin polymerization, and inhibited actin depolymerization in vitro. Furthermore, we have identified a candidate mechanism for regulation of Gcs1p: interaction with phosphoinositides.

In addition to inhibiting depolymerization, Gcs1p shortened the lag phase and increased the net rate and steady-state level of actin polymerization. A protein that stimulates actin polymerization in a similar manner is the mammalian actin-binding protein talin (Kaufmann et al., 1991). Talin is a membrane-linking, F-actin–binding protein of the band 4.1 superfamily (McCann and Craig, 1997), proposed to function by promoting actin filament nucleation and elongation (Kaufmann et al., 1991). It is noteworthy that numerous actin-binding proteins, including talin, also bind phosphoinositides. Phosphoinositide binding has been shown to inhibit the interactions between actin and gelsolin or profilin (Janmey, 1994; Kandzari et al., 1996). Actin polymerization, however, can be modulated by several classes of actin-binding proteins, including those that sever, cap, nucleate, sequester, and bundle actin (Pollard and Cooper, 1986). Thus, although the in vitro biochemical data presented here suggest that Gcs1p may bind/cap the end of actin filaments, it is possible that Gcs1p stabilizes actin filaments by another mechanism. The precise manner by which Gcs1p interacts with actin awaits a detailed biochemical analysis.

Several yeast proteins have been identified and are proposed to function in modulating the actin cytoskeleton. These include Sla2p, a protein associated with cortical actin patches that is necessary for actin nucleation in vitro (Li et al., 1995), Cap2p, an actin-capping protein (Amatruda et al., 1992), Sac6p, a fimbrin homologue that bundles actin filaments (Adams et al., 1991), and Tpm1p, yeast tropomyosin (Liu and Bretscher, 1992). Strains harboring loss-of-function mutations in these genes are believed to have a destabilized actin cytoskeleton, which results in an increased microfilament-turnover rate, leading to hypersensitivity to the actin monomer-binding drug latrunculin (Ayscough et al., 1997). Hence, the gcs1Δ phenotypes, including hypersensitivity to Lat-B, are consistent with a role for Gcs1p in stabilizing the actin cytoskeleton.

gcs1Δ/sla2 and gcs1Δ/sac6 double mutants showed synthetic growth defects, whereas GCS1 did not interact genetically with ABP1, RVS167, or any of the actin alleles tested. In addition to stabilization of the actin cytoskeleton, Sla2p and Sac6p have been implicated as regulators of vesicle trafficking (Kubler and Riezman, 1993; Wesp et al., 1997). Gcs1p is also required for secretion and endocytosis at the nonpermissive temperature (Poon et al., 1996; Wang et al., 1996) and for normal vacuolar morphology at the permissive temperature. This raises the question of whether the genetic interactions result from combining defects in the actin cytoskeleton, in vesicle trafficking, or both. We currently do not know the nature of the genetic interactions between GCS1 and SLA2 or SAC6. However, these data are in agreement with a large body of evidence showing that the actin cytoskeleton is an important component of vesicle trafficking in yeast (Wendland et al., 1998). In addition to Sla2p and Sac6p, the myosin family of molecular motors, tropomyosin, and actin itself are required for vesicle trafficking (Novick and Botstein, 1985; Liu and Bretscher, 1992; Kubler and Riezman, 1993; Welch et al., 1994; Brown, 1997). Conversely, several of the endocytosis mutants (end 4 [sla2], end5 [vpr1], end6 [rvs161], end7 [act1], and end14 [srv2]) are allelic to proteins known to be directly associated with the actin cytoskeleton (Munn et al., 1995; Wesp et al., 1997). Moreover, late-acting secretory mutants such as sec1, sec3, sec6, and the Rab GTPases sec4 and ypt1 have a depolarized actin cytoskeleton (Segev and Botstein, 1987; Lillie and Brown, 1994; Haarer et al., 1996; Mulholland et al., 1997).

In addition to its role in actin cytoskeletal dynamics reported herein, Gcs1p has been shown to act as an Arf1 GAP in vitro and interacts with ARF1 in vivo (Poon et al., 1996). Arf1p is required for secretion in yeast (Stearns et al., 1990). Additional pathways that involve Arfs have been proposed based upon the finding that overexpression of GCS1 and other members of this gene family, including GLO3, SAT1, and SAT2, rescue a loss of function arf1–3^ts mutant, via a pathway that appears to be independent of the secretory function of Arf1p (Zhang et al., 1998). Our finding that Gcs1p is involved in regulation of the actin cytoskeleton, together with the data showing that Gcs1p interacts with Arf1, provide an intriguing possibility that Gcs1p may link the Arf and actin cytoskeletal pathways in yeast.

Gcs1p binds to phosphoinositide-based affinity probes, potentially through an identified PH domain. Although the physiological role of phosphoinositide binding has not been determined, the fact that deletion of the PH domain in Gcs1p yielded a phenotype similar to the null strain (Ireland et al., 1994) indicates that the PH domain is important for Gcs1p function in vivo. Based on its proposed function in other proteins (Toker and Cantley, 1997), phosphoinositide binding may act as a membrane localization signal and/or as a modulator of interactions with other proteins such as Arf or actin. Of the physiologically relevant yeast phosphoinositides tested, Gcs1p bound PtdIns(3,5)P₂ with the highest affinity. Present at low levels under normal growth, PtdIns(3,5)P₂ is synthesized rapidly upon shift to hyperosmotic conditions (Dove et al., 1997) and requires Vps34p, the PtdIns 3-kinase, and Fab1p, a PtdIns(3)P 5-kinase (Dove et al., 1997; Gary et al., 1998). Mutations in genes encoding proteins involved in phosphoinositide metabolism share similar phenotypes with gcs1 mutants. For example, mutations in VPS34, FAB1, PIK1, the PtdIns 4-kinase, and the PtdIns polyphosphate 5-phosphatase genes have mutant growth, actin cytoskeleton, and vesicle-trafficking phenotypes (Banta et al., 1988; Robinson et al., 1988; Garcia-Bustos et al., 1994; Yamamoto et al., 1995; Cutler et al., 1997; Srinivasan et al., 1997).

In summary, the data presented here suggest that Gcs1p is involved in the regulation of the actin cytoskeleton. Furthermore, the in vitro biochemical data suggest that Gcs1p can modulate actin dynamics directly, indicating a second functional pathway in addition to its Arf1 GAP activity. How specific interactions with Arf1p, actin, and phosphoinositides are integrated with Gcs1p function will provide important clues in understanding the role(s) of Gcs1p in cytoskeletal regulation in yeast and the functional activities of the potential GCS1 homologue, centaurin α, in mammalian brain.