INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The actin cytoskeleton is involved in many processes in eukaryotic cells, including the polarization of cell growth and cytokinesis. These many roles involve the interaction of actin with a wide variety of actin-binding proteins, including proteins that can cross-link actin filaments into isotropic gels or bundles. Many actin cross-linking proteins have been purified and studied in vitro, but their functions in vivo remain poorly understood (reviewed by Matsudaira, 1994a; Otto, 1994; Furukawa and Fechheimer, 1997; Ayscough, 1998; Bartles, 2000).

Among the actin cross-linking proteins, one of the best characterized is -actinin. It was first isolated from rabbit skeletal muscle (Ebashi and Ebashi, 1965) and has subsequently been identified in both muscle and nonmuscle cells from a variety of animals, in Acanthamoeba, and in Dictyostelium (Furukawa and Fechheimer, 1997; Critchley and Flood, 1999). -Actinin forms a homodimer of antiparallel polypeptides (Critchley and Flood, 1999; Djinovic-Carugo et al., 1999), with the actin-binding domain of each polypeptide close to the NH₂ terminus, followed by four spectrin-like repeats and two COOH-terminal EF-hand motifs. Skeletal muscle isoforms are localized to the Z-disk and are not regulated by Ca²⁺, whereas nonmuscle isoforms are localized to focal adhesions, stress fibers, and other structures and are typically regulated by Ca²⁺. The observation that -actinin also binds to the cytoplasmic domains of various integral-membrane proteins has suggested that it may help to link the actin cytoskeleton to the plasma membrane. In addition, the localization of -actinin to the cleavage furrows of chick embryos (Fujiwara et al., 1978), sea urchin eggs (Mabuchi et al., 1985), and cultured mammalian cells (Sanger et al., 1987) has suggested that it may have a role in cytokinesis. However, the nature of this role is not clear, and neither Dictyostelium -actinin mutants (Schleicher et al., 1988; Rivero et al., 1999) nor mammalian cells microinjected with antibodies against -actinin (Jockusch et al., 1991) displayed any obvious defect in cytokinesis.

Genetic studies have also provided only limited insights into other aspects of -actinin function in vivo. In Drosophila, a single gene on the X chromosome encodes both muscle and nonmuscle isoforms by alternative splicing (Fyrberg et al., 1990; Roulier et al., 1992). Alleles with point mutations that sharply reduce levels of muscle -actinin allow growth to adulthood but result in abnormalities of myofibrillar organization associated with muscle weakness or paralysis (Fyrberg et al., 1990; Roulier et al., 1992). The phenotypes are surprisingly mild given the myofibrils' lack of a major protein of the Z disk. Moreover, even alleles that deplete all isoforms have no obvious nonmuscle phenotype; flies hemizygous or homozygous for such alleles can complete embryogenesis and differentiation of muscle and do not die until the second day of larval growth (Fyrberg et al., 1990; Roulier et al., 1992). Similarly, in Dictyostelium, inactivation of a Ca²⁺-sensitive -actinin confers only subtle defects (Schleicher et al., 1988; Rivero et al., 1999), although the phenotypic effects are somewhat more pronounced (but still nonlethal) in double mutants defective both in this -actinin and in a second actin cross-linking protein (either ABP-120 or the 34-kDa actin-bundling protein) (Rivero et al., 1996, 1999). Taken together, the available data suggest that -actinin might be redundant or overlapping in function with other proteins, but the identities of the redundant proteins and the degree of redundancy remain unclear.

Another well characterized actin cross-linking protein is fimbrin, a monomeric protein with two tandem -actinin-type actin-binding motifs but no spectrin-like repeats; most fimbrins also have two EF-hand motifs at their NH₂ termini (reviewed by Matsudaira, 1994a; Otto, 1994; Ayscough, 1998; Correia and Matsudaira, 1999). Fimbrins have been identified in budding yeast, ciliates, slime molds, plants, and a variety of animals. The Saccharomyces cerevisiae fimbrin, Sac6p, localizes to actin cables and patches (Drubin et al., 1988; Doyle and Botstein, 1996). A sac6 null mutant has a mild phenotype on synthetic medium but has more pronounced abnormalities on rich medium (Adams et al., 1991, 1993). At 23°C, it is viable with a reduced growth rate (variable in different genetic backgrounds), and the cells are rounder than wild-type cells. At 37°C, the mutant is inviable, and many cells appear lysed. Moreover, the mutant cells display defective actin cables and aberrant, mislocalized actin patches (Adams et al., 1991). Dictyostelium fimbrin also localizes to certain actin-rich regions of the cell (Prassler et al., 1997), but a mutant lacking fimbrin has no detectable defect in cytokinesis and is capable of completing development (Prassler and Gerisch, personal communication). Thus, fimbrin function may also overlap with that of other actin-binding proteins. Three human fimbrins (also called plastins) have been identified that differ in their patterns of tissue-specific expression (reviewed by Matsudaira, 1994a). Two of these proteins can substitute for Sac6p in vivo, indicating that at least some fimbrin functions have been highly conserved during evolution (Adams et al., 1995). It is not yet clear whether fimbrins function in cytokinesis. The sac6 mutant does not display an obvious cytokinesis defect, and, except in Tetrahymena (Watanabe et al., 1998, 2000), fimbrins have not been reported to concentrate in cleavage furrows.

The fission yeast Schizosaccharomyces pombe provides a simple, genetically tractable model system for studies of the function of the actin cytoskeleton in cellular morphogenesis and cytokinesis (Marks and Hyams, 1985; Nurse, 1994; Chang and Nurse, 1996; Gould and Simanis, 1997; Naqvi et al., 1999). During vegetative growth, the cylindrical S. pombe cell grows only at its poles. Immediately after cell division, only the preexisting pole (the "old end") grows; later, after "New End Take Off," both poles grow (Mitchison and Nurse, 1985). These growth patterns reflect the distribution of actin in the cell: actin cables generally run along the long axis of the cell, and cortical actin patches are concentrated at the growing pole or poles. During mitosis, growth at the cell poles ceases as the actin cytoskeleton reorganizes in preparation for division. A medial ring containing actin and a variety of associated proteins forms at the cell middle and eventually contracts during cytokinesis, and the actin patches also become concentrated at the cell middle, where they presumably contribute to the localized cell wall growth that forms the septum. During mating and sporulation, the actin cytoskeleton also undergoes rearrangements that suggest roles in various stages of the sexual cycle (Petersen et al., 1998). The rearrangements of the actin cytoskeleton during the life cycle are likely to involve actin cross-linking proteins, but little information on such proteins has been available. In this article, we describe the identification and characterization of both an -actinin-like protein (Ain1p) and a fimbrin (Fim1p) in S. pombe. These proteins appear to have overlapping functions during cytokinesis, and Fim1p appears to be involved also in the polarization of vegetative cells, in mating, and in sporulation.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Strains, Growth Conditions, and Genetic and Molecular Biology Methods

The S. pombe strains used in this study are listed in Table 1 and/or described where appropriate below; all of them are congenic to strain 972 (Leupold, 1970). Standard growth media (YES rich medium and EMM minimal medium) and genetic methods (Moreno et al., 1991) were used except where noted. Some solid media contained 2.5 µg/ml phloxin B, which accumulates in dead cells (Moreno et al., 1991). SPA mating and sporulation medium contained 10 g/l glucose, 1 g/l KH₂PO₄, 1 ml/l 1000× vitamin stock solution (as used for EMM), and 30 g/l agar. For transformation, cells were grown to ~10⁷ cells/ml in YES medium and then transformed using a lithium-acetate method (Bähler et al., 1998b). For morphological observations and other experiments, cultures were grown overnight and maintained in mid-exponential phase (2 × 10⁶ to 10⁷ cells/ml) at the appropriate temperatures. For regulating the expression of genes under the control of the nmt1 promoter, inducing conditions were EMM medium, and repressing conditions were either EMM containing 5 µg/ml thiamine (all experiments except where noted) or YES medium. The nmt1 promoter begins to transcribe ~10 h after the removal of thiamine and reaches maximal steady-state levels ~6 h later (Maundrell, 1990). Growth rates were determined by measuring the OD₅₉₅ of mid-exponential cultures at two time points at least 2 h apart. Standard recombinant-DNA methods (Sambrook et al., 1989) were used except where noted. DNA was extracted from Escherichia coli and isolated from agarose gels using Qiagen kits (Qiagen, Santa Clarita, CA). DNA was sequenced by the UNC-CH Automated Sequencing Facility on a Model 373A DNA Sequencer with the Taq DyeDeoxy Terminator Cycle Sequencing kit (Applied Biosystems, Foster City, CA). Oligonucleotide primers were obtained from Integrated DNA Technologies (Coralville, IA).

Deletion and Tagging of ain1

ain1 was deleted and tagged in strain 972 by polymerase chain reaction (PCR)-based gene targeting (Bähler et al., 1998b) using kanMX6, which confers resistance to G418/Geneticin (Life Technologies, Gaithersburg, MD), as the selectable marker. For deletion, the forward primer contained sequences corresponding to nucleotides 163 to 83 relative to the start codon, and the reverse primer contained sequences corresponding to the complement of nucleotides 83 to 162 downstream of the stop codon. (The adjacent genes do not impinge closely on the region deleted.) Checking by PCR (Bähler et al., 1998b) identified two stable G418-resistant transformants that had sustained the desired integration at the ain1 locus. When these transformants were crossed to strain 975, all tetrads dissected produced four viable spores that segregated 2:2 for G418 resistance, indicating that the transforming DNA had integrated at a single site. One G418-resistant segregant (strain JW45) was chosen for further experiments. JW45, five other G418-resistant segregants, and the two original transformants all had the same phenotype under stress conditions (see RESULTS), suggesting that no suppressors existed in the original transformants.

To tag Ain1p at its COOH terminus with green fluorescent protein (GFP), the ain1-specific sequences (79 bp) of the forward primer ended just upstream of the stop codon, and the reverse primer was the same as that used for the deletion. Checking by PCR identified a G418-resistant transformant (strain JW46) that had sustained the desired integration event and segregated 2:2 for G418 resistance when crossed to strain 975. In addition, ain1⁺ was placed under the control of three versions of the thiamine-regulated nmt1 promoter (the wild-type 3nmt1, the attenuated 41nmt1, and the still weaker 81nmt1) (Basi et al., 1993; Forsburg, 1993), with or without an associated NH₂-terminal GFP tag (Bähler et al., 1998b). The ain1-specific sequences (70 bp) of the forward primer corresponded to nucleotides 163 to 94, and those (76 or 70 bp) of the reverse primers corresponded to the complement of the NH₂-terminal codons. Checking by PCR identified G418-resistant transformants (strains JW34, JW35, and JW49) that had sustained the desired integration events and segregated 2:2 for G418 resistance when crossed to strain 975. Strains JW46, JW34 (grown under repressing conditions), and JW35 (grown under inducing conditions) all formed normal colonies and had normal cell morphology on various media and at various temperatures, including EMM + 1 M KCl medium at 18°C, conditions under which ain1 deletion cells are severely defective (see RESULTS). In addition, although the ain1-1 fim1-1 double mutant was synthetically lethal at 25°C (see RESULTS), an ain1⁺-GFP fim1-1 strain was viable and resembled fim1-1 single-mutant cells in cell morphology. Thus, both Ain1p-GFP and GFP-Ain1p appear to provide normal Ain1p function.

Cloning, Deletion, and Tagging of fim1

fim1 sequences were amplified by PCR with degenerate primers that corresponded to two regions conserved among eight fimbrins known at the time (see Figure 2B). The forward primer (5'-CCNGAYACNATHGAYGARMG-3', where N = A, C, G, or T; Y = C or T; H = A, C, or T; R = A or G; M = A or C) encoded the peptide PDTIDER and was 768-fold degenerate, and the reverse primer (5'-ACRAANGCNARRTTNARYTT-3') was the complement of the sequence encoding the peptide KLNLAFV and was 2048-fold degenerate. Taq polymerase (Promega, Madison, WI) was used as recommended by the supplier, and DNA from an S. pombe genomic-DNA library (a gift from P. Young, Queens University, Ontario, Canada) constructed in the plasmid pWH5 (Wright et al., 1986) was used as template. After denaturing the template for 5 min at 94°C, 38 cycles of 94°C for 1 min, 50°C for 1.5 min, and 72°C for 1 min were performed followed by a 10-min extension at 72°C. A major PCR product of ~580 bp was obtained and cloned into the pGEM-T Easy vector (Promega), and sequencing of the cloned fragment revealed that it would encode a peptide of 192 amino acids that was 77 and 47% identical to the corresponding regions of S. cerevisiae Sac6p and human L-plastin, respectively. The cloned fragment was then used to screen the pWH5 library by colony hybridization. Plasmid pG16-2 was isolated and sequenced; it appeared to contain a complete fimbrin coding sequence (designated fim1) interrupted (at positions 235 to 310 and 1800 to 1853 relative to the A of the putative initiation codon) by two introns with conserved donor, splice-branch, and acceptor sequences. To confirm the presence of these introns, the absence of other introns within the fim1 coding region, and the identification of the putative start and stop codons, cDNA sequences were amplified by PCR using the Expand High Fidelity System (Boehringer Mannheim, Indianapolis, IN), one or the other of two cDNA libraries as template, and primers corresponding to fim1 genomic sequences or vector sequences. One library (kindly provided by Dr. C. Albright, Vanderbilt University, Nashville, TN) was constructed in vector pREP3X (Basi et al., 1993; Forsburg, 1993), and the other (Becker et al., 1991) was constructed in vector pDB20. The PCR products were cloned into the pGEM-T Easy vector and sequenced.

In addition to the two confirmed introns, a third possible intron is present at positions 120 to 80 (Figure 1); if these sequences were removed by splicing, the ATG at positions 125 to 123 would be in frame with the rest of the coding sequence and thus might be the actual initiation codon. However, several lines of evidence argue strongly against this possibility. First, although the 120 to 80 region has consensus intron donor and acceptor sequences (Figure 1), no commonly used branch site (Prabhala et al., 1992) is present. Second, removal of these sequences by splicing and initiation of translation at the 125 to 123 ATG would result in Fim1p's having 28 additional NH₂-terminal amino acids that would not correspond to the NH₂ termini of other known fimbrins (see Figure 2B). Third, three different cDNAs obtained from two different libraries (see above) all have 5' ends between positions 54 and 48 (Figure 1). Finally, when GFP-encoding sequences were inserted at the presumed actual start codon under control of the 3nmt1 promoter (strain JW113; see below), a GFP-fimbrin fusion protein was produced that showed nearly normal fimbrin localization (see RESULTS) under repressing conditions and caused a severe defect in cytokinesis (similar to that of 3nmt1-fim1⁺ cells; see Figure 7A) under inducing conditions. In contrast, when the same tagging was performed at the 125 to 123 ATG, no specific GFP localization was observed under either repressing or inducing conditions, and no overexpression phenotype was observed under inducing conditions.

View larger version (20K): [in this window] [in a new window]   Figure 1.   Identification of the fim1 start codon (see text for details). The putative start codon is underlined and in boldface; an upstream ATG (125 to 123) is in boldface; the possible intron donor and acceptor sequences are underlined; and the 5' ends of the three cDNAs are underscored by *.

View larger version (63K): [in this window] [in a new window]   Figure 2.   Sequences of S. pombe Ain1p and Fim1p. (A) Alignment of Ain1p (GenBank/EMBL/DDBJ accession number Z97208) and Drosophila muscle -actinin (DmActinin; accession number X51753). Proteins were aligned using the GCG GAP program with some modifications according to the results of a BLAST analysis (http://www.ncbi.nlm.nih.gov/BLAST/). Identical and similar (V/I/L, D/E, K/R, N/Q, S/T) amino acids are shaded in black and gray, respectively. Dots indicate gaps introduced to optimize the alignment; underscoring by ~ indicates the presumed actin-binding motif; and -//- indicates amino acids 370 to 599 of DmActinin, omitted here to optimize alignment of the NH2- and COOH-terminal regions of the proteins (see text). (B) Alignment of Fim1p (GenBank/EMBL/DDBJ accession number AF053722), S. cerevisiae Sac6p (SWISS-PROT accession number P32599), and the human fimbrin L-plastin (SWISS-PROT accession number P13796). Proteins were aligned using the GCG PileUp program. Amino acids identical or similar (see above) to those in Fim1p are shaded in black and gray, respectively. Two putative actin-binding motifs are underscored with ~. The amino acids of Sac6p and L-plastin underscored with * were used to design the degenerate PCR primers that allowed the cloning of fim1 (see MATERIALS AND METHODS).

fim1 was deleted and tagged as described for ain1 except that deletion was performed in the diploid strain JW9. For deletion, the forward primer contained sequences corresponding to nucleotides 116 to 17 relative to the start codon, whereas the reverse primer contained sequences corresponding to the complement of nucleotides 68 to 147 downstream of the stop codon. (The adjacent genes do not impinge closely on the region deleted.) Checking by PCR identified two G418-resistant transformants that had sustained the desired integration. Fifteen tetrads from each transformant yielded mostly three or four viable spores, and about half of the viable spores were G418-resistant (with 2:2 segregation in tetrads with four viable spores). One such segregant (strain JW144) was chosen for further experiments. To tag Fim1p at its COOH terminus with GFP or a triple hemagglutinin epitope (3HA), the forward primer contained 78 nucleotides corresponding to those just upstream of the fim1 stop codon, and the reverse primer was the same as that used for deletion. Checking by PCR and tetrad analysis revealed G418-resistant transformants (strains JW106 and JW109) that had sustained the desired integrations. The tagged proteins appeared to be at least largely functional as judged by two criteria. First, cells of strains JW106 and JW109 formed normal colonies and resembled wild-type cells in morphology during growth on EMM medium at 36°C or on EMM + 1 M KCl medium at temperatures from 18 to 36°C, conditions under which fim1 deletion cells had strong phenotypes (see RESULTS). Second, although the ain1-1 fim1-1 double mutant was synthetically lethal (see RESULTS), strains harboring ain1-1 and either one of the tagged fim1 genes were viable and had normal cell morphology. However, the ain1⁺ fim1⁺-GFP cells did display some morphological abnormalities during growth on rich medium at 36°C, suggesting that Fim1p-GFP was not completely normal in function.

To place fim1⁺ under control of the nmt1 promoters (with or without an NH₂-terminal GFP tag), the forward primer contained sequences corresponding to nucleotides 210 to 131 relative to the fim1 start codon, and the reverse primers contained 77 bp corresponding to the complement of the NH₂-terminal codons. Checking by PCR revealed G418-resistant transformants (strains JW110, JW111, and JW113) that had sustained the desired integrations and segregated 2:2 for G418 resistance.

Morphological Observations and Latrunculin Treatment

Cells were observed by fluorescence or differential-interference-contrast (DIC) microscopy using a Nikon Microphot SA microscope with a 60× Plan-apo objective. To visualize overall cell morphologies or GFP signals together with septa and DNA in the same cells, cells in growth medium were double-stained by adding 1/20 volume of a 1-mg/ml stock solution (in 50 mM sodium citrate, 100 mM sodium phosphate, pH 6.0) of Calcofluor (Sigma, St. Louis, MO) and 1/125 volume of a 1-mg/ml stock solution (in water) of bisBenzimide (Sigma), and then photographed immediately. Immunofluorescence of 3HA-tagged cells was performed as described previously (Bähler and Pringle, 1998), using the monoclonal HA11 primary antibody (Berkeley Antibody, Richmond, CA) and fluorescein isothiocyanate (FITC)-tagged goat anti-mouse-IgG secondary antibody (Jackson ImmunoResearch, Laboratories, West Grove, PA). Staining of F-actin with rhodamine-phalloidin (Molecular Probes, Eugene, OR) was performed using the method of Sawin and Nurse (1998) with two modifications. One-third volume of 16% EM-grade paraformaldehyde solution (Electron Microscopy Sciences, Ft. Washington, PA), preincubated to the same temperature as the culture, was added to a growing culture in EMM or YES medium, and fixation was continued for 15 min at the growth temperature. After staining with rhodamine-phalloidin, the cell suspension was mixed with an equal volume of 10 µg/ml bisBenzimide (in 0.1 M Na Pipes, pH 6.9, 1 mM EGTA, 1 mM MgSO₄) to visualize DNA. For double staining of Fim1p-3HA and actin, immunofluorescence was carried out essentially as described above, except that cells were in suspension in a microfuge tube rather than affixed to a slide. The cells were then stained with rhodamine-phalloidin (as described above) just before mounting and observation.

To determine the effects of actin depolymerization on the localization of Ain1p and Fim1p, cells were grown in EMM medium at 25°C to 5 × 10⁶ cells/ml, and then Latrunculin A (Lat-A; Molecular Probes) was added from a 20 mM stock solution in DMSO to a final concentration of 100 µM; a control culture received an equal volume of DMSO. Incubation was continued for 20 min at 25°C, and cells were double-stained with Calcofluor and bisBenzimide or fixed and stained with rhodamine-phalloidin or HA-specific antibodies (see above).

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Identification of an -Actinin Homologue and of a Fimbrin in S. pombe

The S. pombe genome project revealed an open reading frame (here designated ain1) on cosmid c15A10 of chromosome I that encodes an -actinin-like protein (Ain1p). ain1 contains two presumed introns (with consensus donor, splice-branch, and acceptor sequences) of 51 and 140 bp, and Ain1p is predicted to contain 621 amino acids. It is 32-35% identical to Dictyostelium -actinin (Noegel et al., 1987), Drosophila muscle -actinin (Fyrberg et al., 1990), and human nonmuscle -actinin-1 (SWISS-PROT accession number P12814; Millake et al., 1989) over its entire length and ~50% identical to these proteins over its NH₂-terminal 222 amino acids, which include a conserved actin-binding motif (Bresnick et al., 1990) of 27 amino acids (Figure 2A). Amino acids 225-343 of Ain1p are 27% identical to the first spectrin-like repeat (amino acids 248-366) of Drosophila muscle -actinin, and amino acids 344-450 of Ain1p are 21% identical to the fourth spectrin-like repeat (amino acids 603-715) of the Drosophila protein (Figure 2A). (These two regions of Ain1p are less similar to the second and third repeats of the Drosophila protein, which have thus been omitted from Figure 2A.) Thus, it seems that Ain1p has only two spectrin-like repeats, consistent with its shorter overall length compared with the other known -actinins. Amino acids 487-550 of Ain1p are 28% identical to the two EF-hand motifs (amino acids 752-821) of Drosophila muscle -actinin, but this region of Ain1p does not appear likely to contain a functional Ca²⁺-binding site (Strynadka and James, 1989). Ain1p and the Drosophila protein are also ~30% identical over their COOH-terminal ~70 amino acids. No other obvious motifs were found in Ain1p, except for one small region (amino acids 137-153) having 47% identity (65% similarity) to the phosphatidylinositol 4,5-bisphosphate-binding site in chicken muscle -actinin (Fukami et al., 1996).

The mild phenotype of ain1 deletion cells (see below) suggested that Ain1p might overlap in function with another actin-binding protein. Although no obvious candidate had been identified by the genome project, the presence of a fimbrin (Sac6p) in S. cerevisiae (Adams et al., 1991) suggested that S. pombe might also contain a fimbrin. Indeed, PCR using degenerate primers based on two regions conserved among known fimbrins yielded a product with strong similarity to other fimbrin genes (see MATERIALS AND METHODS). Use of this fragment to probe a genomic-DNA library yielded a plasmid containing a complete fimbrin coding sequence (here designated fim1) interrupted by two introns, as confirmed by the analysis of cDNA sequences (see MATERIALS AND METHODS). Sequences subsequently released by the genome project (cosmid c1778 on chromosome II) were identical within the fim1 coding region.

The predicted Fim1p sequence of 614 amino acids is very similar (64, 49, and 42% identical, respectively, over the full lengths of the proteins) to those of S. cerevisiae Sac6p (Adams et al., 1991), Dictyostelium fimbrin (Prassler et al., 1997), and human L-plastin (Lin et al., 1988, 1990) (Figure 2B). Like these related proteins, Fim1p has two apparent actin-binding motifs that are 41% identical to each other. The first of these (amino acids 205-231) is 56% identical to the putative actin-binding motif in Ain1p (Figure 2A), whereas the second (amino acids 467-493) is 41% identical to that in Ain1p. Like Sac6p (Adams et al., 1991), Fim1p does not appear to have EF-hand-like motifs that would be functional for Ca²⁺ binding (Strynadka and James, 1989).

Actin-dependent Localization of Ain1p to the Medial Ring and of Fim1p to the Medial Ring and Actin Patches

To localize Ain1p, we tagged the chromosomal ain1 locus with GFP-encoding sequences both at its 3' end (ain1⁺-GFP, under the ain1 promoter) and at its 5' end (41nmt1-GFP-ain1⁺ and 81nmt1-GFP-ain1⁺, regulated by the attenuated nmt1 promoters). When cells expressing Ain1p-GFP were examined, no signal was detected in interphase cells (Figure 3A, cell 1). In anaphase cells, Ain1p formed a ring in the middle of the cell (Figure 3A, cells 2-5). The time of appearance and localization of this ring were similar to those of other components of the medial ring (Gould and Simanis, 1997). The Ain1p ring then gradually contracted to a dot and disappeared as the septum formed; Ain1p was always at the leading edge of the centripetally growing septum (Figure 3A, cells 6-10). At no stage was Ain1p-GFP observed in the actin patches or cables. However, it seemed possible that Ain1p was actually present in these structures but that the Ain1p-GFP signals were too weak to detect. To test this possibility, we also examined the strains expressing GFP-Ain1p from the attenuated nmt1 promoters. When grown under repressing conditions, 41nmt1-GFP-ain1⁺ cells had a stronger GFP signal than did ain1⁺-GFP cells, but the Ain1p localization pattern was the same. When grown under inducing conditions, 81nmt1-GFP-ain1⁺ cells had an even stronger GFP signal. This signal was again localized primarily to the medial ring. However, when the cells were grown for 20 h under inducing conditions, Ain1p was also observed as a dot near the nucleus in some interphase cells (our unpublished results). Double staining to determine whether this dot corresponded to the spindle pole body was not possible because the GFP signal did not survive preparation of cells for immunofluorescence. Attempts to gain more information by tagging Ain1p with 3HA were unsuccessful (our unpublished results): either no signal was observed by immunofluorescence (COOH-terminal tag) or localization was observed only to the medial ring (NH₂-terminal tag under control of the nmt1 promoter). In summary, the localization of Ain1p suggested that it might be involved mainly in cytokinesis and/or septum formation.

View larger version (94K): [in this window] [in a new window]   Figure 3.   Localization of Ain1p and Fim1p during the cell cycle. In A and B, upper and lower panels show the same cells, and individual cells are numbered for reference in the text; in C, the two panels show the same cells. (A) Localization of Ain1p to the medial ring. Cells expressing Ain1p-GFP (strain JW46) were grown in EMM medium at 25°C and stained with Calcofluor and bisBenzimide to allow visualization of septa and DNA. (B and C) Localization of Fim1p to actin patches and the medial ring. Cells expressing Fim1p-3HA (strain JW106) were grown in EMM at 30°C, fixed, and double-stained either with HA-specific antibody and bisBenzimide (B) or with HA-specific antibody and rhodamine-phalloidin (C). Bar, 10 µm.

To localize Fim1p, we tagged the chromosomal fim1 locus at its 3' end with sequences encoding 3HA or GFP, so that the tagged proteins would be expressed from the fim1 promoter. When cells expressing Fim1p-3HA were examined by immunofluorescence, protein localization was detected throughout the cell cycle. In interphase cells and cells early in mitosis, Fim1p-3HA was found in patches that were concentrated at one (smaller cells; Figure 3B, cell 1) or both (larger cells; Figure 3B, cells 2-5) ends of the cells. Double staining revealed that Fim1p and actin were colocalized in these patches (Figure 3C). In early anaphase, Fim1p-3HA was still detectable in patches near the cell poles, but a medial ring-like structure also appeared (Figure 3B, cells 6-8). Later in anaphase, the medial ring became more prominent, and few patches (typically concentrated near the medial ring) were observed (Figure 3B, cells 9 and 10). Thus, the localization of Fim1p-3HA closely paralleled that of actin (Marks and Hyams, 1985; see also Figures 4B and 6), except that Fim1p-3HA was not observed in association with actin cables and that the Fim1p ring did not appear to contract at the time of septum formation (Figure 3B, cells 9 and 10; see DISCUSSION). Examination of cells expressing Fim1p-GFP revealed patches at the cell tips and division site like those seen in the cells expressing Fim1p-3HA (except that the patches at the division site were more pronounced during septum formation). However, the medial-ring-like structure was not clearly or consistently visualized with Fim1p-GFP, perhaps because it was obscured by the prominent and rapidly moving (our unpublished results) patches or because Fim1p-GFP was not fully normal in function (see MATERIALS AND METHODS).

View larger version (101K): [in this window] [in a new window]   Figure 4.   Actin dependence of Ain1p and Fim1p localization. Cells expressing Ain1p-GFP (strain JW46; A-D) or Fim1p-GFP (strain JW109; E and F) were treated with 100 µM Lat-A (added from a stock solution in DMSO; see MATERIALS AND METHODS) (A and C-F) or an equal volume of DMSO alone (B) at 25°C for 20 min, and then stained for actin (A and B) or examined for the GFP-fusion proteins (C and E) after staining with Calcofluor and bisBenzimide to allow visualization of septa and DNA in the same cells (D and F). Bar, 10 µm.

To determine whether the localizations of Ain1p and Fim1p depend on F-actin, we treated ain1⁺-GFP and fim1⁺-GFP cells with the inhibitor Lat-A, which specifically and efficiently disrupts actin filaments in S. pombe (Ayscough et al., 1997; Bähler and Pringle, 1998; Figure 4, A and B). Although treatment with DMSO alone had no obvious effect on the localization of Ain1p-GFP or Fim1p-GFP, localization of Ain1p-GFP (Figure 4, C and D) and of Fim1p-GFP (Figure 4, E and F) was completely lost in cells treated for 20 min with Lat-A in DMSO. Similar results (including a loss of medial-ring localization) were obtained with cells expressing Fim1p-3HA (strain JW106). Thus, the localizations of both Ain1p and Fim1p appear to depend on F-actin.

Involvement of Ain1p in Cell Division

To investigate the function of Ain1p, we constructed a strain (JW45) carrying a complete deletion of the ain1 coding sequences (ain1-1). The ain1-1 cells grew at a normal rate on YES or EMM medium at temperatures from 18 to 36°C and displayed no obvious defects in overall cell morphology or in actin organization (Figure 5A; and our unpublished results). However, under the stress conditions of low temperature and high salt (EMM + 1 M KCl medium at 18°C), ain1-1 cells grew much more slowly than wild type; moreover, although wild-type cells maintained a normal morphology under these conditions (Figure 5B), ain1-1 cells displayed pronounced abnormalities in cytokinesis and septation. In particular, some cells completed mitosis without forming septa or showing other indications of cytokinesis, some cells formed partial septa, some cells had double or triple septa, and some cells formed misplaced septa (Figure 5C). As expected for a defect affecting cytokinesis and septation but not other aspects of cell cycle progression, cells with multiple nuclei accumulated in the ain1-1 strain (but not in the wild-type strain) during growth under the stress conditions (Table 2).

View larger version (85K): [in this window] [in a new window]   Figure 5.   Morphological abnormalities of ain1 deletion and fim1 deletion cells. (A-C) Cells were double-stained with Calcofluor and bisBenzimide to visualize septa and DNA. (D-F) Cells were observed directly in culture medium by DIC microscopy. (A) ain1-1 cells (strain JW45) grown in EMM medium at 25°C. (B and C) Wild-type strain 972 (B) and strain JW45 (C) grown in EMM + 1 M KCl medium at 18°C for 40 h. (D) fim1-1 cells (strain JW142) grown in EMM at 25°C. (E and F) Strains 972 (E) and JW142 (F) grown for 5 h (20 h for the cells in the inset) after a shift from 25 to 36°C in EMM. Bar, 10 µm.

The abnormalities in cytokinesis and septation in the ain1 deletion cells under stress conditions might result from abnormalities in actin organization. Indeed, although wild-type cells grown under the stress conditions displayed normal actin organization (Figure 6A, left panels), and ain1-1 interphase cells displayed a seemingly normal organization of actin patches (Figure 6A, right panels, cell 1), the ain1-1 cells did not develop normal medial-ring structures. Instead, early in mitosis, some cells formed aberrant, thin, cable-like structures in the cell middle when most of the actin patches were still at the cell tips (Figure 6A, right panels, cell 2; the cable-like structures were more apparent under the microscope than they are in the printed micrograph). In anaphase cells, the majority of the actin patches accumulated in a broad region around the cell middle (Figure 6A, right panels, cells 3 and 4), but distinct actin rings were rarely seen. In ~1% of anaphase cells, very thin ring-like structures that were difficult to photograph were seen at the cell middles. Thus, Ain1p appears to be involved in formation of the actin ring, at least under stress conditions.

View larger version (84K): [in this window] [in a new window]   Figure 6.   Abnormal actin organization in ain1 deletion and fim1 deletion cells. (A) Failure of actin-ring formation in ain1-1 cells under stress conditions. Strains 972 (left panels) and JW45 (right panels) growing in YES medium at 30°C were shifted to EMM + 1 M KCl medium at 18°C for 20 h, fixed, and double-stained with rhodamine-phalloidin (upper panels) and bisBenzimide (lower panels). (B) Mislocalization and disorganization of actin patches in fim1-1 cells. Strains 972 (left panels) and JW142 (right panels) growing in EMM medium at 25°C were shifted to 36°C for 5 h, fixed, and double-stained with rhodamine-phalloidin (upper panels) and bisBenzimide (lower panels). Bar, 10 µm.

A role for Ain1p in actin-ring formation, cytokinesis, and septation is also supported by the results of overexpression studies. To overexpress Ain1p, the promoter of the chromosomal ain1 locus was replaced by the high-expression 3nmt1 promoter. When 3nmt1-ain1⁺ cells (strain JW49) were grown under repressing conditions, most cells looked normal, although ~5% had abnormal or double septa. After a shift to inducing conditions, cells with two or more nuclei accumulated in the population (Figure 7A, left panels); by 19 h of induction, ~60% of cells had two or more nuclei. Although most cells still formed septa, the septa were usually misoriented and/or disorganized, and some cells had double septa (Figure 7A, left panels). The Ain1p-overexpressing cells did not have other obvious morphological defects, which suggested that actin organization might be normal except for defects in actin-ring assembly. Indeed, rhodamine-phalloidin staining revealed an apparently normal organization and localization of actin patches in interphase cells (Figure 7B, left panels, arrowhead). In contrast, although most anaphase cells displayed actin-ring structures, these were generally misplaced and/or disorganized in ways that paralleled the abnormalities in septum organization (Figure 7B, left panels).

View larger version (97K): [in this window] [in a new window]   Figure 7.   Defects in cytokinesis, septation, and actin organization resulting from overexpression of Ain1p or Fim1p. Cells in which the expression of Ain1p or Fim1p was under the control of the 3nmt1 promoter were grown under repressing conditions and then shifted to inducing conditions (see MATERIALS AND METHODS) at 30°C. (A) Cells of strains JW49 (left panels) and JW110 (right panels) were double-stained with Calcofluor and bisBenzimide at 16 h (upper panels) or 19 h (lower panels). (B) Cells of strains JW49 (left panels) and JW110 (right panels) were fixed and double-stained with rhodamine-phalloidin (upper panels) and bisBenzimide (lower panels) at 16 h. The arrowhead indicates an interphase cell (see text). Bars, 10 µm.

Additional evidence for a role of Ain1p in actin-ring organization was obtained by analyzing genetic interactions between ain1-1 and other mutations known to affect medial-ring positioning and/or organization. ain1-1 showed strong synthetic growth and cytokinesis defects with act1-48 (actin), myo2-E1 (myosin II), cdc4-8 (myosin light chain-like protein), and cdc3-6 (profilin), and weaker synthetic defects with cdc8-110 (tropomyosin), rng2-D5 (IQGAP), cdc12-112 (a formin), mid1-F, and pom1-1 (Table 3; and our unpublished results). In contrast, no genetic interactions were detected between ain1-1 and mutations affecting the regulation of septation (cdc14-118 and cdc16-116), actin cortical-patch function (arp3-c1), or microtubule function (nda3-KM311) (see Table 1 for references), supporting the hypothesis that the function of Ain1p may be restricted to medial-ring assembly and positioning.

View this table: [in this window] [in a new window]   Table 3.  Genetic interactions of ain1-1 with other mutations affecting medial-ring positioning and/or organization

Involvement of Fim1p in Cell Polarization and Perhaps in Cell Division

To investigate the function of Fim1p, we constructed a strain (JW142) carrying a complete deletion of the fim1 coding sequences (fim1-1). Strain JW142 could grow on either YES or EMM medium at temperatures from 18 to 36°C, but it displayed a growth defect whose severity depended on the conditions. It grew at a rate similar to that of wild type in YES medium at 25°C but grew more slowly at higher temperatures and/or in EMM medium. On EMM at 36°C, it formed tiny colonies that were red when the plates contained phloxin, suggesting that many of the cells were dead, and it was unable to form colonies on EMM + 1 M KCl at temperatures from 18 to 36°C. The fim1-1 cells also displayed pronounced morphological defects in EMM medium; even at 25°C, ~30% of the cells (n = 492) were rounder than normal (Figure 5D), and 5% of the cells appeared dead (shrunken and phloxin red; some cells were clearly lysed). These defects were more pronounced at 36°C: although wild-type cells looked normal under these conditions (Figure 5E), by 5 h after a shift of fim1-1 cells from 25 to 36°C, ~16% of the cells (n = 488) appeared dead (among which about half were clearly lysed), and ~18% of the cells were clearly rounder than normal (Figure 5F). By 10 h at 36°C, ~21% of the fim1-1 cells (n = 419) appeared dead, and ~30% were rounder than normal. Beginning at ~10 h, and more frequently at later times, cells could be found that had apparently lost polarity completely and were perfectly round (Figure 5F, inset). At either 25 or 36°C, the fim1-1 strain displayed a modest increase in the number of cells with two nuclei (~45% compared with the ~28% characteristic of wild type), but only ~1% of the cells had more than two nuclei. These data suggest either that cell division was generally completed (although possibly after some delay) even in the morphologically aberrant cells and/or that the nuclear cycle was also delayed (perhaps by a checkpoint function). fim1-1 cells plated on EMM + 1 M KCl medium at 18, 25, or 36°C also did not display any obvious defect in cytokinesis or septation.

It seemed likely that the partial loss of growth polarization in fim1 deletion cells reflected defects in the organization of the actin cytoskeleton. Indeed, although wild-type cells showed normal actin distribution in EMM medium at 36°C (Figure 6B, left panels), fim1-1 cells did not display normal actin patches under these conditions. Instead, the patches appeared to be disorganized and mislocalized: they were not usually distinct, they sometimes appeared to fuse to form a cap-like structure at the cell end, and they also covered a larger area than just the cell tips (Figure 6B, right panels). The actin patches in the perfectly round cells were uniformly distributed in the cell cortex, indicating that the cells had completely lost polarity. Consistent with the absence of an obvious cytokinesis defect, actin rings appeared to form normally (Figure 6B, right panels). However, the actin patches did not relocalize efficiently to the cell middle; thus, when the actin ring had formed and begun to contract, considerable F-actin remained at the cell tip(s) (Figure 6B, right panels), which was never observed in wild-type cells. It was also striking that some cells at this stage appeared to have the actin patches concentrated at one pole of the cell; in contrast, wild-type cells always have patches concentrated at both poles before medial-ring formation. The fim1-1 cells still had actin cables, but these were difficult to photograph and thus are difficult to see in the printed micrographs.

The data described above suggested that Fim1p might be involved in polarization of cell growth and/or actin cortical-patch function, a hypothesis that was further supported by genetic evidence. In particular, fim1-1 was synthetically lethal with the cold-sensitive arp3-c1 mutation: no double mutant was recovered on YES medium at either 30 or 32°C (conditions permissive for both single mutants) from 17 tetrads dissected. The 15 predicted double-mutant segregants germinated and then died as 1-30 small, round cells; no elongated cells were found. Thus, it appeared that the double mutant was defective in polarized growth but not in cytokinesis.

However, a possible role for Fim1p in cytokinesis and/or septation was suggested by overexpression studies. To overexpress Fim1p, the promoter of the chromosomal fim1 locus was replaced by the 3nmt1 promoter. 3nmt1-fim1⁺ cells (strain JW110) were indistinguishable from wild-type cells under repressing conditions, but a pronounced defect in cytokinesis and septation was observed after a shift to inducing conditions. The number of cells with two or more nuclei increased dramatically (Figure 7A, right panels; by 19 h, 44% of the cells had two nuclei, 43% had four, and 8% had eight), and septum formation was grossly defective (Figure 7A). Some cells did not appear to deposit any septal material, others deposited small amounts of such material at normal or abnormal locations, and still others did form aberrant septa, but no cells formed normal-looking septa. In addition, these cells had severe defects in the organization of F-actin. No actin rings or concentrations of actin patches in the cell middles were detectable, although many patches were not at the cell tips (Figure 7B, right panels). Most of the actin patches at the cell tips were more brightly stained than those in wild-type cells (grown and stained at the same time), but their organization otherwise appeared normal (Figure 7B; cf. the wild-type cells in Figures 4B and 6). The actin cables also appeared normal but were difficult to photograph.

Synthetic Lethality and Cell Division Defect of an ain1 fim1 Double Mutant

To ask whether Ain1p and Fim1p overlap in function, ain1-1 (JW45) and fim1-1 (JW142) strains were crossed, and tetrads were dissected onto YES plates at 25°C. From 36 tetrads, all 35 predicted double mutants were inviable; 34 arrested as a single, highly elongated cell, and one arrested after two cell divisions. Thus, ain1-1 and fim1-1 were synthetically lethal under conditions in which both single mutants resembled wild type (see above).

The elongated morphology of the double-mutant cells suggested that they might be defective in cytokinesis and/or septation. To explore this possibility, an ain1-1 strain (JW53) was crossed to a strain (JW111) with the wild type fim1⁺ coding sequence under control of the attenuated 41nmt1 promoter. Strain JW111 itself resembled wild type under both inducing and repressing conditions. However, no ain1-1 41nmt1-fim1⁺ double mutant was recovered on YES plates (repressing conditions) from 19 tetrads dissected; the 22 predicted double mutants arrested as one to several hundred highly elongated cells with no septa or with aberrant septa. However, ain1-1 41nmt1-fim1⁺ double mutants were recovered by germinating spores on EMM (inducing conditions). The double mutants were indistinguishable from wild type under inducing conditions (Figure 8A) but had severe defects in cytokinesis and septation under repressing conditions and could form only tiny colonies at 25°C and no visible colonies at 36°C. The cell division defect became evident within 12 h after a shift to repressing conditions; by 16 h, 98% of the cells (n = 346) had two or four nuclei, while some cells had no septa and other cells had partial or misplaced septa (Figure 8, B and C). A few cells did have seemingly complete septa (which might have formed before Fim1p was depleted) but were delayed or defective in cell separation (Figure 8, B and C). By 24 h under repressing conditions, most cells had accumulated four or eight nuclei and begun to lyse. The defects in cytokinesis and septation were associated with severe defects in medial-ring assembly and/or positioning and in the concentration of actin patches at presumptive septation sites (Figure 8, D and E). The cells also displayed misplaced and aberrant cable-like structures (Figure 8, D and E) that were distinct from the weakly stained actin cables normally seen in the cytoplasm of wild-type cells. Thus, it appears that Ain1p and Fim1p play overlapping roles in the organization of actin for cytokinesis and septation.

View larger version (112K): [in this window] [in a new window]   Figure 8.   Cell division defect of an ain1 fim1 double mutant. ain1-1 41nmt1-fim1+ cells (strain JW164) growing in EMM medium without thiamine (inducing conditions) at 25°C (A) or 16 h after a shift to YES medium (repressing conditions, leading to depletion of Fim1p) at 25°C (B-E) were double-stained for septa and DNA without fixation (A and B), stained for septa alone without fixation (C), or double-stained for F-actin (D) and DNA (E) after fixation (see MATERIALS AND METHODS). Bar, 10 µm.

Involvement of Fim1p, but Apparently Not of Ain1p, in the Sexual Cycle

To investigate the possible functions of Ain1p and Fim1p during mating, sporulation, and spore germination, we examined the localizations of these proteins during these processes. When ain1⁺-GFP strains of opposite mating type (strains JW46 and JW52) were mated, no localized signal was detected at any stage of the sexual cycle. As expected from this result, mating ain1-1 strains of opposite mating type (strains JW45 and JW53) resulted in spore formation as efficient as that in wild type.

In contrast, Fim1p-GFP was detectable at all stages in patches resembling those seen in vegetative cells. During mating, most of these patches were localized to the extended tips of the conjugating cells (Figure 9A). Later, when most asci had begun to form spores, the Fim1p-GFP patches localized predominantly to the cortices of the developing spores (Figure 9, B and C), and this localization was maintained in mature asci (Figure 9, D and E). Before germination, the Fim1p-GFP patches did not appear polarized in the spores, but upon germination, the Fim1p-GFP patches were concentrated in the growing tips (Figure 9F). Throughout the sexual cycle, the localization of Fim1p was very similar to that of actin (Petersen et al., 1998). These data suggest that Fim1p might be involved both in growth polarization during mating and spore germination and in the formation of the spore plasma membrane and/or spore wall. Support for these possibilities was provided by the defective mating and sporulation of fim1 deletion strains. Fifty hours after crossing two fim1-1 strains, only ~16% of the cells had become asci (compared with ~51% in wild type), apparently reflecting at least in part a defect in mating. Moreover, <20% of the asci (n = 300) had four spores, while the others had zero to three spores (Figure 9G), and longer incubation did not change these numbers appreciably. The defect appeared to be in spore formation rather than in meiosis, because DNA staining revealed that most asci contained four nuclei. In contrast, >97% of the asci from a cross of wild-type strains 972 and 975 had four spores under the same conditions.

View larger version (83K): [in this window] [in a new window]   Figure 9.   (A-F) Localization of Fim1p-GFP during mating (A), sporulation (B and D), and spore germination (F). (C and E) DIC images of the cells shown in B and D, respectively. fim1+-GFP strains JW109 and JW117 were crossed on an SPA plate at 25°C, and samples taken after 12 h (A), 24 h (B and C), and 50 h (D and E) were examined after resuspension in liquid SPA medium. After 60 h, the mature asci were digested with Glusulase (NEN, Boston, MA) to release spores, which were then incubated in liquid EMM medium at 25°C and examined after 12 h (F). (G) Defective sporulation in fim1 deletion cells. fim1-1 strains JW144 and JW142 were crossed on an SPA plate at 25°C and examined by DIC after 50 h. Bar, 10 µm.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Evolution and Functional Relationships of Actin-bundling Proteins

-Actinins and fimbrins are members of a family of actin cross-linking proteins that share very similar actin-binding domains; this family also includes spectrin, dystrophin, ABP-120, filamin, and cortexillin (reviewed by Matsudaira, 1994a; Faix et al., 1996; Furukawa and Fechheimer, 1997; Bartles, 2000). As shown in this study, there are at least two members of this family in S. pombe. Fim1p appears to be a typical fimbrin, having strong similarity to other fimbrins throughout its sequence. In contrast, Ain1p is less similar in sequence to other known -actinins and is also considerably shorter, having just two spectrin-like repeats instead of the usual four. Because sequence alignments and phylogenetic trees suggest that spectrins evolved from -actinin by duplications of the spectrin-like repeats (Pascual et al., 1997), it seems plausible that other modern -actinins evolved from an ancestral Ain1p-like protein by duplication of its two repeats.

The relatively short length of Ain1p may have important functional implications. The observation that -actinins and fimbrins bind to the same region of actin (Matsudaira, 1994b) suggests that they may have related but complementary functions. Indeed, it appears that these two proteins typically mediate the formation of distinct types of actin filament bundles. In particular, fimbrins, with their two closely spaced actin-binding domains, appear to form very tightly packed bundles, whereas -actinins, whose actin-binding domains are at the opposite ends of the long homodimers, appear to form looser bundles (Meyer and Aebi, 1990; Drenckhahn et al., 1991; Alberts et al., 1994; Höfer and Drenckhahn, 1996; Djinovic-Carugo et al., 1999; Bartles, 2000). In contrast to the bundles formed by typical -actinins, those formed by the putative Ain1p homodimers would presumably be more tightly packed and hence more similar to the bundles formed by fimbrins. This similarity may help to explain the apparent partial overlap in function between Ain1p and Fim1p during cytokinesis (see also below). Such functional overlap between -actinin and fimbrin does not appear to be typical: although these two proteins may both be involved in connecting actin filaments to the plasma membrane at sites of cell adhesions in some types of cells (Matsudaira, 1994a), we are not aware of other evidence for functional overlap between these proteins in cell types other than S. pombe.

In this regard, it is also interesting that S. cerevisiae contains no -actinin-like protein and has the fimbrin Sac6p as its only member of this protein family (Ayscough, 1998). Because the homology between Ain1p and other -actinins suggests strongly that the common ancestor of the fungi, protozoa, slime molds, and animals possessed an -actinin-like protein, this protein must have been lost during the evolution of S. cerevisiae. The contrast with S. pombe is particularly striking because the actin cytoskeletons of the two yeasts seem generally quite similar in organization and function (Marks and Hyams, 1985; Botstein et al., 1997; Gould and Simanis, 1997; Bähler and Peter, 2000; Pruyne and Bretscher, 2000). However, one significant difference is in the relative importance of the actomyosin contractile ring during cytokinesis. Although myosin II and the contractile ring are essential for cytokinesis in S. pombe (Bezanilla et al., 1997; Kitayama et al., 1997; May et al., 1997; Motegi et al., 1997), they are not essential in S. cerevisiae (Bi et al., 1998; Lippincott and Li, 1998). A possible explanation for this difference is that the primary role of the contractile ring is to guide septum deposition (Hales et al., 1999; Vallen et al., 2000) and that the need for such guidance is greater in forming the ~3-µm-diameter septum in S. pombe than in forming the ~1-µm-diameter septum in S. cerevisiae. In any case, it seems possible that an Ain1p-like protein was rendered dispensable in S. cerevisiae by the combination of functional redundancy with Sac6p and the nonessential role of the actomyosin ring in cytokinesis.

It is also important to note that although we found no evidence for a role of Ain1p other than in cytokinesis, Fim1p clearly also has a distinct role(s) in the organization of the actin cytoskeleton in interphase cells (see also below). However, this other Fim1p role is also nonessential. Although it remains possible that this nonessentiality reflects redundancy with an unidentified member of the -actinin/fimbrin family, it seems more likely (with the S. pombe genome sequence ~90% complete) that it reflects overlap in function with another type(s) of actin-binding protein, as appears to be the case also for Sac6p in S. cerevisiae (Adams et al., 1993). Partial overlap in function between -actinin and the structurally unrelated 34-kDa actin-bundling protein has also been observed in Dictyostelium (Rivero et al., 1999).

Roles of Ain1p and Fim1p in Cytokinesis

Several lines of evidence indicate that Ain1p functions specifically in the contractile medial ring during cytokinesis. First, Ain1p was detected exclusively in this structure. Second, ain1 deletion cells display cell division defects under stress conditions, much as do cells lacking the nonessential myosin II Myp2p/Myo3p (Bezanilla et al., 1997; Motegi et al., 1997). Third, cells overexpressing Ain1p have defects in cell division. Fourth, the ain1 deletion mutation shows negative genetic interactions with several other mutations affecting medial-ring formation and cytokinesis. In contrast, the putative role of Fim1p in cytokinesis is less obvious. In fim1 deletion cells, actin patches do not relocalize normally from the cell tip(s) to the cell center at the time of division, and some cells are lysed. In at least some cases, the lysis appeared to have occurred just after cell separation (our unpublished results), suggesting that it may result from defective cell wall or plasma membrane deposition at the new cell end. Moreover, at least some cells are delayed in cell division, as reflected by an increase in cells with two nuclei. These findings suggest that Fim1p might be involved in the targeted delivery of cell surface materials during cell division, much like its probable role during the polarized growth of interphase cells (see below). In addition, the localization of Fim1p to the medial ring and the cytokinesis defect of cells overexpressing Fim1p suggest that Fim1p may also be directly involved in medial-ring formation and/or function. This suggestion is strongly supported by the synthetic-lethal cytokinesis defect of cells lacking both Ain1p and Fim1p; this result also indicates that the roles of these two proteins in the medial ring are at least partly redundant.

What might be the role(s) of Ain1p and Fim1p in the medial ring? Medial-ring precursors start to accumulate at the cell center during the early stages of mitosis but only coalesce into a distinct ring during anaphase (Arai et al., 1998; Bähler et al., 1998a). Ain1p, Fim1p, and other actin-binding proteins such as Rng2p (Eng et al., 1998) may be involved in organizing the medial-ring precursors into a functional contractile ring (for review, see Bähler and Peter, 2000). Consistent with this hypothesis, the ring localization of both Ain1p and Fim1p depends on actin, and actin forms only aberrant cables and localizes to a broad region in ain1 fim1 double-mutant cells or in ain1 single-mutant cells under stress conditions. In the latter case, at least, the actin structures resemble those seen during the early transient stage of medial-ring formation in wild-type cells. In some cases, pairs of actin cross-linking proteins appear to function sequentially in assembling a final structure (Tilney et al., 1998; Bartles, 2000). However, Ain1p and Fim1p appear to assemble into the medial-ring structure at about the same time, consistent with the evidence for overlap in function.

In contrast, when the medial ring contracts, Ain1p stays at its leading edge while Fim1p appears to remain in an uncontracted ring. Although it is possible that the accumulation of Fim1p patches at the cell center obscures the detection of a contracting ring (as can happen when cells are stained for actin; see Balasubramanian et al., 1997), it is also possible that actin filaments are bundled too tightly by Fim1p to allow access by myosin, so that myosin must displace Fim1p as contraction begins. Such displacement has been observed with the Dictyostelium proteins in vitro (Prassler et al., 1997). Because -actinins can bind (directly or indirectly) to membrane proteins (Critchley and Flood, 1999), Ain1p may help to link the medial ring to the plasma membrane during contraction. However, the successful cytokinesis of ain1 deletion cells under normal conditions shows that this role must also be redundant with that of at least one other protein.

It is not yet known whether either -actinin or fimbrin functions in cytokinesis in other organisms. Although -actinins concentrate in the cleavage furrows of some animal cells and a fimbrin concentrates in the cleavage furrow of Tetrahymena (see INTRODUCTION), no strong evidence has been obtained to support direct roles for these proteins in cytokinesis in these cell types. Moreover, in Dictyostelium, neither -actinin nor fimbrin is detectably enriched in the cleavage furrow, and no substantial cytokinesis defect has been detected in cells that are lacking either protein (Schleicher et al., 1988; Pras