We show that Arf3p, a member of the ADP ribosylation family, is involved in the organization of actin cables and cortical patches in Saccharomyces cerevisiae. Profilin-deficient cells (pfy1Δ) have severe growth defects and lack actin cables. Overexpression of ARF3 restores actin cables and corrects growth defects in these cells. Cells deficient for the cortical patch proteins Las17p and Vrp1p have growth defects and a random cortical patch distribution. Overexpression of ARF3 in las17Δ and in vrp1Δ cells partially corrects growth defects and restores the polarized distribution of cortical patches. The N-terminal glycine, a myristoylation site in Arf3p, is necessary for its suppressor activity. arf3Δ cells show a random budding pattern. Overexpression of BNI1, GEA2 or SYP1, three genes involved in actin cytoskeleton formation, restores the normal axial budding pattern of arf3Δ cells. BUD6 is a polarity gene and GEA2 is involved in retrograde transport and the organization of the actin cytoskeleton. We have identified genetic interactions between ARF3 and BUD6, and between ARF3 and GEA2. Both double mutant strains have actin cytoskeleton defects. Our results support a role for ARF3 in cell polarity and the organization of the actin cytoskeleton.
The actin cytoskeleton is a complex structure found in some bacteria and in all eukaryotic cells (Gitai, 2005). The budding yeast Saccharomyces cerevisiae contains two types of polymerized actin structures, cables and patches (Kilmartin & Adams, 1984; Amberg, 1998; Yang & Pon, 2002). These structures are usually oriented towards regions of cell growth. For example, in cells with small buds, the cables extend from the cell cortex and neck of the bud into the mother cell. Actin cables, which contain linear actin filaments, are involved in polarized secretion and organelle segregation (Bretscher, 2003). Cortical patches are seen as numerous punctuate structures in developing buds and in the mother cell at later stages of the cell cycle. These globular structures are mobile, turn over rapidly and contain actin filaments with 70° side branches (Karpova et al., 1998; Huckaba et al., 2004; Young et al., 2004). Cortical patches function in the internalization step of endocytosis (Kaksonen et al., 2003, 2005; Duncan & Payne, 2005; Kim et al., 2006).
Recent evidence has shown that two systems of polymerization, containing largely nonoverlapping sets of proteins, are used to form these two classes of actin filaments (Pruyne & Bretscher, 2000b; Goode & Rodal, 2001). Actin cables are nucleated by the formin protein Bni1p and are decorated by tropomyosins, Tpm1p and Tpm2p (Liu & Bretscher, 1989; Drees et al., 1995; Sagot et al., 2002a). Actin monomers (Act1p), Bni1p, Bud6p, profilin (Pfy1p), Spa2p and signaling proteins are involved in the formation of actin cables. Evidence for the role of these proteins in cable formation has come from genetic studies, biochemical analysis and immunofluorescence microscopy. Briefly, cells lacking functional profilin or the formin proteins have cortical patches but few or no actin cables (Marcoux et al., 1998, 2000; Evangelista et al., 2002; Sagot et al., 2002a). The overexpression of a truncated BNI1 gene, coding for Bni1pFH1FH2, in a wild-type strain causes the formation of supernumerary convoluted actin cables (Pruyne et al., 2002; Sagot et al., 2002b; Zakrzewska et al., 2003). Long actin filaments are polymerized in vitro in a reaction containing actin monomers, Bni1p, Bud6p and Pfy1p (Pruyne et al., 2002; Sagot et al., 2002b; Moseley et al., 2004). Finally, protein–protein interaction studies provide evidence for a large protein complex, the polarisome, containing the proteins Bni1p, Bud6, Pea2p, Spa2p and Sph1p (Fujiwara et al., 1998; Sheu et al., 1998; Shih et al., 2005).
Numerous proteins are involved in the polymerization of the branched actin filaments in cortical patches. The Arp2/3 complex, which contains seven proteins including the actin-related proteins Arp2p and Arp3p, is essential for nucleation (Moreau et al., 1996; Winter et al., 1999). The Arp2/3 complex in vitro, however, has only weak nucleation activity. In budding yeast, Las17p and Abp1p are strong activators of the Arp2/3 complex while Pan1p, which is associated with the proteins Sla1p and End3p, is a weak activator. The class I myosins Myo3p and Myo5p and the yeast equivalent of the mammalian WASP-interacting protein, Vrp1p, also regulate the activity of the Arp2/3 complex, either negatively or positively (Lechler et al., 2000; Duncan et al., 2001; Goode et al., 2001; D'Agostino & Goode, 2005).
The small GTPase proteins with their regulatory GTPase-activating protein (GAP) and guanosine exchange factor (GEF) are important regulators of the actin cytoskeleton. Budding yeast contain six Rho proteins: Cdc42p and Rho1p–Rho5p, which all have roles in the organization of the actin cytoskeleton (Marcoux et al., 2000; Pruyne & Bretscher, 2000a; Lechler et al., 2001; Dong et al., 2003). Cdc42p is an activator of Bni1p and it also recruits two activators of the Arp2/3 complex, Las17p and Vrp1p, to the complex. Rho1p activates Pkc1p, which in turn regulates the polarisome (Fujiwara et al., 1998). Rho2p is a suppressor of the profilin-deficient phenotype (Marcoux et al., 2000). Rho3p and Rho4p share an essential function in Bni1p activation and Rho5p negatively regulates stress-induced actin depolarization (Schmitz et al., 2002).
Little is known about the role of another class of small GTPase proteins in the actin cytoskeleton, the ADP ribosylation family, which contains three members, Arf1p, Arf2p and Arf3p (Jackson & Casanova, 2000). Arf1p and Arf2p are partially redundant in function and are important in retrograde transport between the Golgi and the endoplasmic reticulum (Stearns et al., 1990; Roth, 1999; Springer et al., 1999; Yahara et al., 2001). Arf3p is not implicated in Golgi–endoplasmic reticulum transport (Lee et al., 1994; Huang et al., 2003). In the present work, we show that Arf3p plays a role in the organization of the actin cytoskeleton.
Materials and methods
Strains, media and transformations
Yeast strains and plasmids used in this study are listed in Tables 1 and 2. Cells were grown with rotary shaking at 30 or 37°C in YPD medium (1% yeast extract, 2% Bacto peptone and 2% glucose). The strains with a plasmid were grown in an appropriate synthetically defined medium (SD: 0.67% yeast nitrogen base without amino acids, 2% glucose and supplemented with appropriate autotrophic requirements). Caffeine sensitivity was tested in SD medium with 1.25 mg mL−1 caffeine. Transformations of the cells were performed by a modified lithium acetate procedure (Kaiser et al., 1994).
pRS426+ ARF3 with a mutation in N-terminal site (G2A)
pRS426+ ARF3 with a mutation in N-terminal site (N3A)
Multicopy plasmid+ GEA2
pRS426+ ARF3 and its promoter
Multicopy plasmid+ GEA1
The plasmid pRMP100 was constructed as follows. The ARF3 coding sequence was amplified by PCR from the wild-type (WT) strain BY4741 using the oligonucleotides ARF3 BamHI (BamHI Site is in bold characters) forward: 5′CGGGATCCGGTCTCATAACCCT TTCTTG and ARF3 EcoRI (EcoRI Site is in bold characters) reverse: 5′CGGAATTCGGTGTA TGCAGATTCAACACC. The PCR product was cloned into the BamHI and EcoRI sites of the pBluescript plasmid. The ARF3 gene was excised from this plasmid with SacI and KpnI and was cloned into the SacI and KpnI sites of the multicopy pRS426 plasmid. The plasmid p4753 was used to overexpress the Bni1p domains FH1 and FH2 (Bni1pFH1FH2) in the arf3Δ strain. The plasmids p88 and pGMS3 were used to overexpress the GEA1 and GEA2 genes respectively in arf3Δ.
To construct the plasmid pEML102, containing the SYP1 coding sequence, the SYP1 gene was excised from pBS- SYP1 (Marcoux et al., 2000) with SacI and KpnI and was cloned into the SacI and KpnI sites of the multicopy plasmid pRS426. The plasmid was cloned into arf3Δ cells. Cells were observed and photographed with a Leica fluorescence microscope. Images were captured by a CoolSNAP-Prool Monochrome camera using the IpEx32 program. Images were processed by Adobe Photoshop to enhance only brightness and contrast.
The ARF3 gene in pBluescript was used for site-directed mutagenesis. Two specific mutations were made in ARF3. The first was introduced into the second amino acid where a glycine was replaced by an alanine (G2A). The other mutation was in the third amino acid where an asparagine was substituted for an alanine (N3A). These mutations were made with the QuikChange™Site-Directed Mutagenesis kit (Stratagene). The mutagenic oligonucleotide primers are listed below and the mismatched bases are underlined:
The mutagenized genes with their promoters were cloned into the SacI/KpnI sites of the multicopy plasmid pRS426.
Staining of actin filaments was carried out according to a modified protocol for visualizing actin filaments (Zakrzewska et al., 2003). Cells were grown to exponential phase at 30 or 37°C. All of the cells were fixed at room temperature for 30 min by addition of 37% formaldehyde to 3.7% final concentration directly into the culture media. The cells were washed in phosphate-buffered saline (PBS) and stained with 0.3 μM AlexaFluor 488-conjugated phalloidin (Molecular Probes, Eugene, OR) for 1 h at room temperature in the dark. Cells were observed and photographed as previously described.
Calcofluor white staining
Staining of chitin with Calcofluor White was performed as described by Pringle (1991). Cells were observed and photographed. A minimum of 200 cells were counted to determine the budding pattern.
Construction of double mutant strains
The arf3Δ gea2Δstrain was obtained by sporulation of the heterozygous diploid strain arf3Δ ∷kan/gea2Δ ∷HIS3 (CJY049-11-4). The tetrads were dissected manually using a dissection microscope to isolate the arf3Δ gea2Δ double mutant haploid strain.
The arf3Δ bni1Δ, arf3Δ bud6Δ, arf3Δ drs2Δ, arf3Δ gea1Δ and arf3Δ lsb5Δ strains were obtained by a PCR method by deleting the ARF3 coding sequence in each of the single mutant strains. The uracil gene of the plasmid pRS316 was amplified with oligonucleotides containing the flanking sequences of the ARF3 gene – URA3 forward: 5′-ATA ATTGGGATTTAGAACGGAAAAAAGGAAAAGACAAGCT AATTGTAGAGCGTTTCGGTGATGAC and ARF3 URA3 reverse: 5′-GAAGGATGAACAATCGGTGT
ATGCAGATTCAACACCAATAAATGCAATGTTTCCTG ATGCGGTATTTTCTCCT (underlined sequences were used for homologous recombination). The PCR fragments were transformed separately into the single mutant strains to obtain the five double mutant strains. The genotypes of the double mutant strains were verified by PCR amplifications.
Overexpression of ARF3 restores actin cables in pfy1Δ cells
Cells lacking profilin (pfy1Δ) have severe defects in actin cytoskeleton organization and growth. These cells are large and round, grow slowly at 30°C and do not grow at 37°C or in the presence of 1.25 mg mL−1 caffeine (Marcoux et al., 2000). Actin cables are absent and the numerous cortical patches are randomly distributed in the bud and in the mother cell. During a genetic screen for suppressors of the profilin-deficient phenotype, we showed that overexpression of ARF3 in the YEp24 vector partially corrects the aberrant phenotype of pfy1Δ cells. The cells were normal in size, had cortical patches polarized in buds, but did not have visible actin cables (Zakrzewska et al., 2003).
We re-examined the correction of pfy1Δ cells by the ARF3 gene by cloning this gene into the smaller multicopy plasmid, pRS426. With this new construction, the pfy1Δ strain has a WT phenotype. The size of these cells is normal and all the cortical patches are polarized in the bud. The budding pattern is also corrected from a random distribution to an axial pattern (results not shown) and growth is restored at 37°C and in the presence of 1.25 mg mL−1 caffeine at 30°C (Fig. 1a). The most interesting observation is the presence of perfectly formed actin cables (Fig. 1b). These results show that overexpression of ARF3 restores formation of actin cables in a profilin-deficient cell. A similar result was obtained with the overexpression of the genes GEA1 and GEA2, which are GEFs for Arf proteins (Zakrzewska et al., 2003).
Suppression of the pfy1Δ mutant phenotype by ARF3 overexpression. (a) Suppression of the pfy1Δ caffeine sensitivity by ARF3. Wild type mutant cells (pfy1Δ) and mutant cells overexpressing ARF3 (pfy1Δ ARF3) were tested for caffeine sensitivity. A total of 1 × 105 cells and serial 10-fold dilutions were spotted on an SD URA− plate containing 1.25 mg mL−1 caffeine. The plate was incubated at 30° for 48 h. (b) Overexpression of ARF3 in pfy1Δ cells repolarizes the cortical actin patches and restores actin cables. Cells from exponential phase cultures, grown at 30°, were stained with Alexafluor488-conjugated phalloidin to visualize actin distribution. WT cells, mutant cells (pfy1Δ) and mutant cells overexpressing ARF3 (pfy1Δ ARF3) in multicopy plasmid pRS426 are shown. Actin cables are visible only in WT cells and mutant cells overexpressing ARF3.
Overexpression of ARF3 corrects the cortical patch defects in las17Δ and vrp1Δ cells
Las17p and Vrp1p are two important proteins in the cortical patch formation pathway. Las17p, which is the yeast equivalent of human WASP protein, is the principal activator of the Arp2/3 complex (Lechler & Li, 1997; Li, 1997; Madania et al., 1999). In las17Δ cells, actin cables are normal but few cortical patches are present (Fig. 2a). In las17Δ cells overexpressing ARF3, the overall morphology of the actin cytoskeleton is normal and is similar to the actin cytoskeleton in WT cells (Fig. 1b). Notably, in the overexpressing cells numerous cortical patches are seen. In cells with small buds, the cortical patches are polarized and are primarily located in the bud (Fig. 2a). ARF3, however, is only a partial suppressor of the las17Δ phenotype. las17Δ cells overexpressing ARF3 are larger than WT cells, they do not grow in liquid culture at 37°C and they grow slowly at 30°C.
Overexpression of ARF3 corrects the phenotype of las17Δ and vrp1Δ strains. Cells from exponential phase cultures, grown at 30°, were stained with Alexafluor488-conjugated phalloidin to visualize actin distribution. (a) Mutant cells (las17Δ) and mutant cells overexpressing ARF3 in multicopy plasmid pRS426 (las17Δ ARF3) are shown. (b) Mutant cells (vrp1Δ) and mutant cells overexpressing ARF3 in multicopy plasmid pRS426 (vrp1Δ ARF3) are shown. (c) Correction of the growth defects of vrp1Δ cells by the overexpression of ARF3. WT cells, vrp1Δ ARF3 cells and vrp1Δ cells were tested for caffeine sensitivity. A total of 1 × 105 cells and serial 10-fold dilutions were spotted on an SD URA− plate containing 1.25 mg mL−1 caffeine. The plate was incubated at 30°C for 48 h.
Vrp1p, a proline-rich protein, is the yeast homologue of the human WASP-interacting protein (WIP). It binds actin monomers and also binds to the WH1 domain of Las17p (Vaduva et al., 1997; Evangelista et al., 2000). vrp1Δ cells are larger than normal, grow more slowly than WT cells at 30 and 37°C and are sensitive to 1.25 mg mL−1 caffeine (Donnelly et al., 1993). In most vrp1Δ cells, the cortical patches are distributed randomly in mother and daughter cells throughout the cell cycle. Importantly, no actin patches are seen in newly emerging buds, while actin cables appear normal (Fig. 2b). Overexpression of ARF3 in vrp1Δ cells partially restores the WT phenotype. The cells are near normal in size and have a normal distribution of cortical patches through the cell cycle at 30 and 37°C. Actin cables are clearly visible in the mother cell. The cells with small buds contain numerous cortical patches throughout the bud (Fig. 2b). The cells have a near normal growth rate on plates at 30 and 37°C and have normal growth in the presence of 1.25 mg mL−1 caffeine (Fig. 2c). ARF3, however, is only a partial suppressor of the vrp1Δ phenotype. vrp1Δ cells overexpressing ARF3 grow more slowly than WT cells in liquid culture at 37°C (results not shown).
These results show that the overexpression of ARF3 corrects cortical patch defects in las17Δ and vrp1Δ cells. Combined with the results on the formation of actin cables in pfy1Δ cells, these results demonstrate that the overexpression of ARF3 in certain mutant cells can activate the pathways to actin cable and cortical patch formation. It should be noted that the cortical patch defects are not suppressed in las17Δ and vrp1Δ cellsby the overexpression of GEA2, a gene that induces actin cable formation in pfy1Δ cells (results not shown).
The N-terminal glycine, a myristoylation site in Arf3p, is necessary for its suppressor activity
Rho-type GTPases have lipid modifications that are necessary for membrane anchoring. The glycine in the second position of the Arf3p sequence is N-myristolyated (Huang et al., 2003). To determine whether this modification is essential for the suppressor activity of Arf3p, site-directed mutagenesis was used to modify the binding site for myristoylation by exchanging a glycine for an alanine (G2A). As a control experiment, the third amino acid, an asparagine, was modified to an alanine (N3A).
These two Arf3p mutations were expressed in pfy1-111 arf3Δ cells. This strain was chosen as it does not contain a WT ARF3 gene and it has a defective actin organization at 37°C. We could not use the double mutant cells pfy1Δ arf3Δ for this experiment because the cells are inviable. At 37°C, the pfy1-111 arf3Δ cells have few actin cables, are larger than normal and have randomly distributed cortical patches (Fig. 3). The actin cytoskeleton and morphological abnormalities in this strain are suppressed by the overexpression of WT ARF3 or by ARF3N3A (data not shown). The cortical patches are polarized and numerous actin cables are visible. By contrast, overexpression of ARF3G2A does not suppress the phenotype of this strain; the cells are large with few cables and with a random distribution of cortical patches (Fig. 3). In addition, the overexpression of ARF3G2A does not correct the actin cytoskeleton defects or the growth defects in vrp1Δ cells (results not shown). These results show the importance of the glycine myristoylation site for the activity of the Arf3p protein.
The myristoylation site in the N-terminal glycine is necessary for the activity of Arf3p. Cells from exponential phase culture, grown at 37°, were stained with Alexafluor488-conjugated phalloidin to visualize actin distribution. WT cells, mutant cells (pfy1-111 arf3Δ) and mutant cells overexpressing ARF3G2A and ARF3N3A in multicopy plasmid pRS426 are shown. Visible actin cables are seen in WT and pfy1-111 arf3Δ ARF3N3A cells.
Arf3p requires functional formin proteins Bni1p and Bnr1p for actin cable formation
Budding yeast have two formin proteins, Bni1p and Bnr1p, that are necessary for actin cable formation. Neither Bni1p nor Bnr1p is essential for growth but cells lacking both proteins are nonviable. The double mutant bni1-FH2#1 bnr1Δ carrying a temperature-sensitive allele for BNI1 and a deleted BNR1 gene is viable with normal actin cytoskeleton at 25°C. However, at a nonpermissive temperature, the actin cytoskeleton of these cells is disorganized. The cortical patches are present but are distributed randomly in the bud and in the mother cell. Importantly, no actin cables are visible (Sagot et al., 2002a; Evangelista et al., 2003).
To test whether the formation of actin cables by the overexpression of ARF3 is dependent on the yeast formins, the double mutant bni1-FH2#1 bnr1Δ cell was transformed with ARF3 and the actin cytoskeleton was observed. Overexpression of ARF3 in these cells has no effect on the actin cytoskeleton: the cortical patches are still depolarized and no actin cables are visible (Fig. 4). These results show that Arf3p cannot replace the formin proteins for the production of actin cables.
Overexpression of ARF3 does not restore the organization of the actin cytoskeleton in bni1-FH2#1 bnr1Δmutant cells. Actin distribution was visualized with Alexafluor488-conjugated phalloidin in bni1-FH2#1 bnr1Δcells and bni1-FH2#1 bnr1Δcells overexpressing ARF3 in multicopy plasmid pRS426.
Overexpression of GEA2,SYP1 or a fragment coding for BnipFH1FH2 corrects the budding defect of arf3Δ cells
The budding pattern is axial in WT haploid cells, with the new bud emerging near the previous division site. arf3Δ cells have an abnormal budding pattern (Fig. 5). In these cells, bud formation is not axial but is distributed randomly over the cell surface. The phenomenon, however, is not seen in all cells. About 40% of arf3Δ mutant cells show random budding at 30°C. This proportion increases to 90% at 37°C.
Correction of the abnormal bud pattern in arf3Δ by overexpression of the truncated BNI1, coding for Bni1p (FH1FH2), GEA1,GEA2 and SYP1. WT cells, mutant cells (arf3Δ) and mutant cells overexpressing the truncated BNI1 in plasmid p4753, GEA1 in plasmid p88, GEA2 in plasmid pGMS3 or SYP1 in plasmid pEML102 were grown at 30°C, fixed and stained with Calcofluor White to visualize bud scars. In each experiment, a minimum of 200 cells with two bud scars or more were counted and analyzed. Mutant cells (arf3Δ) exhibit abnormal bud pattern in about 40% of cells. Overexpression of the truncated BNI1,GEA2 or SYP1 in mutant cells (arf3Δ) restores the normal axial budding pattern. Overexpression of GEA1 in arf3Δ does not suppress the random budding pattern. (a) Representative immunofluorescence-stained cells. (b) Graphic representation of the budding pattern in different strains.
The overexpression of ARF3,BNI1,GEA2 or SYP1 can restore a normal cytoskeleton in pfy1Δ cells (Marcoux et al., 2000; Zakrzewska et al., 2003). We therefore determined whether the overexpression of BNI1, GEA2 or SYP1 can correct the budding defects of arf3Δ cells. As the overexpression of an intact BNI1 gene is lethal, we used a truncated BNI1 gene containing the FHI/FH2 domains (Bni1pFH1FH2) or intact GEA2 and SYP1 genes. We also tested the effect of the overexpression of the genes GEA1,MID2,RHO2,ROM2 and WSC1, which were previously identified as suppressors of the profilin-deficient phenotype (Marcoux et al., 1998, 2000; Zakrzewska et al., 2003).
Only the overexpression of the truncated BNI1,SYP1 or GEA2 genes corrects the random budding of arf3Δ cells. At 30°C, only about 5% of the overexpressing cells show a random budding pattern compared with 40% of the arf3Δ cells. The vast majority of the cells overexpressing any of the three genes have the WT axial budding pattern (Fig. 5). These results indicate a genetic interaction between Arf3p and the proteins Bni1p, Gea2p and Syp1p. They also demonstrate that these three proteins can function in the absence of Arf3p. Interestingly, the overexpression of GEA1 has no effect on the random budding pattern of the arf3Δ strain (Fig. 5). Although Gea1p and Gea2p have partially overlapping functions in Golgi to endoplasmic transport, they are not equivalent in the suppression of the budding defect of the arf3Δ strain.
ARF3 shows genetic interactions with BUD6 and GEA2
No genetic lethal interactions are known for the combination of the ARF3 deletion and a second deletion. To test for possible genetic interactions that affect the organization of the actin cytoskeleton, we constructed five double mutant strains carrying the ARF3 deletion and another deletion. These genes were chosen on the basis of their coding for proteins involved in actin cable formation (BNI1,BUD6,GEA1,GEA2), and in possible protein–protein interactions with Arf3p (LSB5) (Sagot et al., 2002a; Evangelista et al., 2003; Zakrzewska et al., 2003; Costa et al., 2005; Moseley & Goode, 2005). We observed the actin cytoskeleton in strains carrying a single and a double deletion at 30 and 37°C.
The actin cytoskeleton and morphology are normal in all the single deletion strains at both temperatures, except for arf3Δ and bud6Δ strains (Fig. 6). arf3Δ cells at 30°C have a WT morphology and normal organization of the actin cytoskeleton. At 37°C, c. 20% of the cells have two buds but the actin cytoskeleton is normal in all cells with one bud (Fig. 6a). bud6Δ cells have a normal patch distribution at 30°C, but the actin cables are thin and more difficult to detect than in a WT strain. At 37°C, this strain shows a slight depolarization of actin patches but most of the patches are found in the bud in cells with small buds (Fig. 6b).
ARF3 interacts genetically with BUD6 and GEA2. Cells from exponential phase cultures, grown at 30 or 37°C, were stained with Alexafluor488-conjugated phalloidin to visualize actin distribution. (a) The actin cytoskeleton is normal in arf3Δ cells at 30 and at 37°C. (b) The bud6Δ cells have a normal actin cytoskeleton at 30°C. (c) The cortical patches in arf3Δ bud6Δ cells are depolarized at 30 and 37°C. (d) The actin cytoskeleton is normal in gea2Δ cells at 30 and 37°C. (e) The cortical patches are partially depolarized in the arf3Δ gea2Δ double mutant cells at 30°C and are completely depolarized at 37°C. Few actin cables are visible in arf3Δ gea2Δ cells at 37°C. (f) Growth defects of arf3Δ bud6Δ cells. WT, arf3Δ, bud6Δ and arf3Δ bud6Δ cells were tested for caffeine sensitivity. A total of 1 × 105 cells and serial 10-fold dilutions were spotted on an SD URA− plate containing 1.25 mg mL−1 caffeine. The plate was incubated at 30°C for 48 h.
Genetic interactions are only seen between ARF3,BUD6 and GEA2. The actin cytoskeleton is modified in arf3Δ bud6Δ cells. At 30°C, the cortical patches are partially depolarized. In some cells, no patches are seen at the tip of the bud and in all cells, numerous cortical patches are found in the mother cell. The actin cables are thinner and less numerous than those seen in arf3Δ or in bud6Δ cells (Fig. 6c). The presence of two buds in 50% of the arf3Δ bud6Δ cells growing at 37°C is a distinguishing feature. In these double budded cells, cables were faint and the cortical patches were depolarized (results not shown). The double mutant arf3Δ bud6Δ strain grows more slowly than WT and single mutant cells at 25, 30 and 37°C and on 1.25 mg mL−1 caffeine. The results for growth on caffeine at 30°C are shown (Fig. 6f). These results are a clear indication of a genetic interaction between ARF3 and BUD6.
A genetic interaction is also detected between ARF3 and GEA2. At 30°C, the double mutant arf3Δ gea2Δ cells are somewhat larger and rounder than WT cells. The cortical patches are partially disorganized in these mutant cells, with the presence of cortical patches in the mother cell in all cell cycle stages. Actin cables are present, often forming highly convoluted structures not aligned along the bud mother cell axis. At 37°C, these double mutant cells are even larger (Fig. 6e). Cortical patches show a random distribution in all cells. The actin cables are either not visible or are seen as short disorganized fragments. These results demonstrate genetic interactions between ARF3 and BUD6 and between ARF3 and GEA2. Interestingly, the double mutant cells arf3Δ gea1Δ have a WT appearance at 30 and 37°C and a normal distribution of actin cables and patches (results not shown). No genetic interaction is therefore observed between ARF3 and GEA1.
The ARF family in budding yeast is composed of three members, Arf1p, Arf2p and Arf3p. Arf1p and Arf2p are 96% identical and have overlapping functions. A deletion of either gene is viable, but the deletion of both genes is lethal. The activated GTP-associated forms of both proteins bind to membranes and are involved in retrograde transport from the Golgi to the endoplasmic reticulum (Roth, 1999; Springer et al., 1999; Jackson & Casanova, 2000; Yahara et al., 2001). Arf3p is only 54% identical to Arf1p and Arf2p. The arf3Δ strain is viable as are the arf3Δ arf1Δ and arf3Δ arf2Δ double mutant strains. Arf3p is not involved in vesicular transport (Lee et al., 1994). The protein is partially localized to the heavy membrane fraction by biochemical analysis and to the cell periphery and buds by green fluorescent protein immunofluorescence (Huang et al., 2003). The polarized location of Arf3p is independent of the actin cytoskeleton but is dependent on an N-terminal myristoylation (Huang et al., 2003). The myristoylation is also necessary for the actin cytoskeleton activity of Arf3p (this work). arf3Δ cells have a random budding pattern but have a normal actin cytoskeleton and normal fluid-phase endocytosis as measured by the uptake of fluorescent dyes. Arf3p has been recently shown to interact physically with Lsb5p (Costa et al., 2005). Lsb5p locates to the cortical membrane region by an actin-independent but Arf3p-dependent mechanism. Lsb5p also interacts with Sla1p and Las17p, two cortical patch proteins important for endocytosis. The deletion of LSB5 in a strain lacking the actin-associated protein gene YSC84 leads to a severe phenotype of a disorganized actin cytoskeleton and the almost complete loss of fluid-phase endocytosis (Dewar et al., 2002).
Here we show that Arf3p is involved in the organization of the actin cytoskeleton. Six independent lines of evidence support this conclusion. First, the overexpression of ARF3 in a pRS426 vector restores visible cables in a profilin-deficient mutant, pfy1Δ. The resulting cells resemble WT cells in all aspects. Second, Las17p is an important activator of the Arp2/3 complex, a necessary element for cortical patch formation. Mutant las17Δ cells have clearly visible actin cables but contain few or no cortical patches. The overexpression of ARF3 in these cells leads to a normal amount and distribution of cortical patches. Third, vrp1Δ cells have few cortical patches in developing buds and have a general delocalization of cortical patches in other cell cycle stages. The overexpression of ARF3 in these cells restores the WT distribution of cortical patches. Fourth, the formin protein Bni1p is a key element in the formation of actin filaments. The overexpression of BNI1 corrects the random budding defects of arf3Δ cells. Fifth, the GEF protein Gea2p restores actin cables in pfy1Δ cells, indicating that it is involved in actin cable formation. The overexpression of GEA2 restores the normal budding pattern to arf3Δ cells, and ARF3 and GEA2 show a genetic interaction. The double mutant arf3Δ gea2Δ has few actin cables and has a random distribution of cortical patches. Finally, Bud6p is involved in actin cable formation. This protein binds actin monomers and also binds to the formin protein Bni1p in the polarisome (Moseley & Goode, 2005). The arf3Δ and the bud6Δ mutants individually have either a normal or near normal actin cytoskeleton organization. The arf3Δ bud6Δ double mutant has a severely compromised actin cytoskeleton. Cells have either only a few short or no visible actin cables. The cortical patches in most cells are delocalized. The double mutant cells also show growth defects.
The Arf GTPase family of proteins requires the action of guanosine nucleotide exchange factors for their activity. The yeast proteins Gea1p, Gea2p, Sec7p and Syt1p were identified as exchange factors for Arf1p. The proteins Gea1p and Gea2p, which are found in a membrane-bound and soluble form, are involved in retrograde transport from the Golgi to the endoplasmic reticulum. These two proteins are also involved in the organization of the actin cytoskeleton. The overexpression of GEA1 or GEA2 in pfy1Δ cells restores the polarized distribution of cortical patches and produces visible actin cables (Zakrzewska et al., 2003). Several lines of evidence indicate that these two genes are not completely redundant for their activity in vesicular transport and actin cytoskeleton organization. Gea1p and Gea2p have only partially overlapping cellular locations. Gea1p localizes to a large number of small foci distributed throughout the cell while Gea2p is localized to bright foci within the cytoplasm, consistent with Golgi or endosomal components. The arf1Δ gea1Δ double mutant is viable while the arf1Δ gea2Δ double mutant is nonviable (Spang et al., 2001). In the present work we show that the arf3Δ gea1Δ strain has a normal actin cytoskeleton while the arf3Δ gea2Δ strain has fewer actin cables and depolarized cortical patches. In addition, the overexpression of GEA2 but not GEA1 corrects the random budding defect of the arf3Δ strain. The genetic interaction between ARF3 and GEA2 suggests that Arf3p and Gea2p functionally interact in yeast cells.
The identity of the GEF protein for Arf3p is not known. It could be Gea1p, Gea2p, another GEF or some combination of these proteins. Our immunofluorence studies show that the proper location of Arf3p is independent of the presence of Gea1p and Gea2p (results not shown). Our genetic interactions suggest, however, that Gea2p is not the unique GEF for Arf3p. The arf3Δ gea2Δ strain has a severely depolarized actin cytoskeleton while each of the single mutations has a normal cytoskeleton, demonstrating that the effects of the arf3Δ and gea2Δ mutations are additive. In addition, the overexpression of GEA2 corrects the random budding defect of the arf3Δ strain, showing that excess Gea2p can bypass the requirement for Arf3p. These results suggest that if Gea2p is an activator of Arf3p, it must also have other protein targets. On the other hand, no genetic interaction was observed between ARF3 and GEA1, and the overexpression of GEA1 does not suppress the random budding of the arf3Δ strain. These results are compatible with a possible role for Gea1p as a GEF for Arf3p, but further experiments are necessary to verify this possibility.
Syp1p was identified as a multicopy suppressor of the actin cytoskeleton defects in pfy1Δ cells. This protein is found in regions of active growth throughout the cell cycle: the presumptive bud site, growing buds and the septum in dividing cells (Marcoux et al., 2000). The exact role for this protein is not known. In the present work, we identified SYP1 as a multicopy suppressor of the budding defects of the arf3Δ strain. This shows that Syp1p can function in the absence of Arf3p and is further evidence that Syp1p is involved in cell polarity.
The polarisome is a large protein complex involved in the formation of actin cables. It contains the proteins Bni1p, Bud6, Pea2p, Spa2p and Sph1p. The proteins Pfy1p, Gea1p and Gea2p are also involved in cable formation. Although none of the proteins is a known binding partner for Arf3p, we have identified genetic interactions between ARF3 and BUD6 and ARF3 and PFY1 (Zakrzewska et al., 2003; this work). Finally, the overexpression of ARF3 is able to bypass the cell's need for Pfy1p, Las17p and Vrp1p. Arf3p, a Rho-type GTPase, is one of the few budding yeast proteins implicated in actin cables and actin cortical patch formation. This suggests that Arf3p has protein targets in both pathways.
We thank Manon Valcourt for technical help. This work was supported by a grant from the Natural Sciences an Engineering Research Council of Canada (NSERC). A.A.L. and M.P.P. were supported by fellowships from the Centre de Recherche sur la Structure, la Fonction et l'Ingénierie des Protéines (CREFSIP). M.P.P. was also supported by a NSERC fellowship. We thank Charles Boone for the p4753 plasmid and Catherine Jackson for the plasmids p88 and pGMS3.
(2001) The ADP ribosylation factor-nucleotide exchange factors Gea1p and Gea2p have overlapping, but not redundant functions in retrograde transport from the Golgi to the endoplasmic reticulum. Mol Biol Cell 12: 1035–1045.