Determination of cell polarity in germinated spores and hyphal tips of the filamentous ascomycete Ashbya gossypii requires a rhoGAP homolog

The ability to establish cell polarity and induce polarized cell growth is fundamental for morphogenesis and development of unicellular and multicellular organisms. One of the best known examples of polar cell growth is provided by the hyphal growth of filamentous fungi. Fungal mycelia can be grown from spores. The first step in breaking the dormancy of a spore is the initiation of an isotropic non-polarized growth phase that generates a spherical germ cell. In germ cells switching from isotropic to polarized growth results in the formation of the first hyphae which continuously extend from their hyphal tips, proliferate by branching, and form radially spreading mycelia. Differing from filamentous fungi are unicellular yeast-like fungi in which polarized growth is restricted to a short phase of the cell cycle during bud emergence (Lew and Reed, 1993, 1995; Kron and Gow, 1995; Waddle et al., 1996; Karpova et al., 1998). Genetic approaches have been applied in the unicellular fungus S. cerevisiae to identify genes required in this process (Chant, 1994; Cid et al., 1995). Detailed analyses demonstrated that during bud emergence in S. cerevisiae a protein network integrates and performs the tasks of bud-site selection, polarity establishment, and bud growth (Chant and Stowers, 1995; Cid et al., 1995; Drubin and Nelson, 1996; Cabib et al., 1998; Madden and Snyder, 1998; Johnson, 1999; Chant, 1999). The central parts of this protein network are constituted of ras/rho-like GTPases (Pringle et al., 1995; Van Aelst and D’Souza-Schorey, 1997; Hall, 1998). Ras/rho-like GTPases act as molecular modulators. Their signaling function depends on their association with GTP whereas hydrolysis of GTP to GDP renders them inactive. Guanine nucleotide cycling of rho-proteins between the GTP- or GDP-bound states is regulated by other proteins: GEFs (guanine nucleotide exchange factors) stimulate the exchange of GDP for GTP and GAPs (GTPase activating proteins) enhance the intrinsic GTPase activity of ras/rho-like GTPases and thus turn off their signaling (Tanaka and Takai, 1998). Therefore, a core GTPase module consists of at least three proteins. Establishment of a polarized actin cytoskeleton in yeast resulting in bud formation is controlled by the CDC42 rho-module consisting of the rho-GTPase Cdc42p, its GEF Cdc24p and GAP Bem3p (Johnson, 1999; Toenjes et al., 1999). Growth at restrictive temperature of conditional mutants of either CDC42, CDC24, or BEM3 results in the formation of large unbudded multinucleate cells which exhibit a delocalized deposition of chitin and a uniform distribution of actin cortical patches (Sloat et al., 1981; Adams et al., 1990; Zheng et al., 1994). Bud growth and the actin cytoskeleton in S. cerevisiae are regulated by several other GTPase-proteins including RHO1, RHO2, RHO3, and RHO4 (Tanaka and Takai, 1998). For the Rho1 protein the regulator proteins have been identified as Rom1/2-GEF and Bem2-GAP (Ozaki et al., 1996; Peterson et al., 1994). Temperature sensitive mutants of ROM1 or RHO1 become arrested as small budded cells under restrictive conditions. Whereas mutants in BEM2 form large unbudded multinucleate cells at elevated temperatures (Bender and Pringle, 1991).

In sharp contrast to the detailed knowledge in S. cerevisiae, very little is known with respect to the molecular mechanisms that control the establishment of cell polarity and polarized cell growth in filamentous fungi. In Neurospora crassa a large number of colony morphology mutants have been isolated (Garnjobst and Tatum, 1967; Perkins et al., 1982). Two of those mutants have recently been characterized. Mutations in the protein kinase cot-1 and in mcb-1, which codes for a regulatory subunit of cAMP-dependent protein kinase (PKA), were found to affect hyphal elongation and hyphal growth polarity, respectively (Yarden et al., 1992; Bruno et al., 1996). In Aspergillus nidulans, screens of a temperature sensitive mutant strain collection have been performed (Harris et al., 1994). Thereby three classes of mutants were isolated: (i) hypercellular mutants (hypA-hypE; Kaminskj and Hamer, 1998); (ii) polarity defective mutants (podA/hypA, podB, podC, and podD; Harris et al., 1999); (iii) swollen mutants (swoA-swoH; Momany et al., 1999). These mutants appear to be involved in different aspects of polar growth and polarity establishment in A. nidulans. However, except for hypA, which does not show homology to any known gene, none of these A. nidulans mutants has been characterized on the molecular level. Thus, we are not able to compare the molecular mechanisms underlying the yeast-like growth mode and polarized hyphal growth of filamentous fungi.

A. gossypii is a filamentous ascomycete that, based on Rdna sequences, is more closely related to S. cerevisiae than either N. crassa or A. nidulans (Wendland et al., 1999). We have chosen A. gossypii as a model to initiate molecular genetic studies on hyphal morphogenesis because this fungus allows the facile identification and manipulation of genes. Specifically (i) A. gossypii harbors an exceptionally small genome of 8.85 Mb coding for approximately 4500 genes (S. Steiner et al., unpublished). (ii) The identification of genes in A. gossypii can be facilitated based on the conserved gene order between A. gossypii and S. cerevisiae (Altmann-Jöhl and Philippsen, 1996). (iii) Due to the high homologous recombination efficiency in A. gossypii, one-step gene replacement can be done conveniently by PCR-based gene targeting (Steiner et al., 1995; Wendland et al., 2000). (iv) A. gossypii allows the extrachromosomal free replication of plasmids bearing an autonomous replicator (Wright and Philippsen, 1991). Stability of freely replicating plasmids can be increased under non-selective conditions by the introduction of centromeric-DNA of A. gossypii (J. Wendland et al., unpublished).

In this report, we present the first isolation, deletion, and phenotypic characterization of a rhoGAP gene in a filamentous fungus. We demonstrate the requirement of AgBEM2, which is a homolog of the S. cerevisiae BEM2 gene for the determination of cell polarity during germination and hyphal growth in A. gossypii. Our results imply that polarized hyphal growth of filamentous fungi relies, at least in part, on the same regulatory networks, i.e. rho-like GTPase modules, as, e.g. in the yeast S. cerevisiae.

The Ashbya gossypii wild-type strain (ATCC10895) and a derivative of it deleted for the AgLEU2 and AgTHR4 genes was used (Altmann-Jöhl and Philippsen, 1996; C. Mohr and P. Philippsen, unpublished). For selection of G418/geneticin resistant transformants geneticin (GIBCO BRL) was included in the complete medium (AFM: 20 g/l glucose, 10 g/l yeast extract, 10 g/l peptone, 1 g/l myo-inositol) at a final concentration of 200 μg/ml. For selection of LEU⁺ transformants minimal medium was used lacking leucine (AMM: 20 g/l glucose, 0.69 g/l CSM-LEU (BIO101, Inc.; Vista CA), 1.7 g/l yeast nitrogen base without amino acids and (NH₄)₂SO₄ (Difco), 0.3 g/l myo-inositol, 1.0 g/l L-asparagine). Mycelia were routinely grown at 30°C unless otherwise specified. The Escherichia coli strains DH5α and XL1-Blue were used as hosts for plasmids.

DNA- and colony hybridization, as well as standard recombinant DNA techniques were performed according to the methods of Sambrook et al. (1989). Colony hybridization was done to isolate the complete AgBEM2 gene from a YEp352-based plasmid library (Maeting et al., 1999). Sequencing was done on an ABI377 automated sequencer according to the manufacturer’s instructions. Sequence comparisons were done by BLAST searches against available databases (Altschul and Lipman, 1990). Profile searches were done at the ISREC profile-scan server (http://www.isrec.isb-sib.ch). To generate the pairwise and the dot-matrix comparisons of AgBem2p and ScBem2p the GCG-software package form the University of Wisconsin was used (Devereux et al., 1984).

For the amplification of the ScLEU2 gene pRS415 was used (Sikorski and Hieter, 1989), using the primers LEU2a-5′-GACAGATCTCTTAGCAACCATTATTTTTTTCCTCAAC-3′ and LEU2b-5′-ACGGGATCCTTATCACGTTGAGCCATTAGTATCAAT-TTG-3′ (restriction sites underlined). The PCR fragment containing the entire LEU2 gene was cloned into the BamHI-site of pFA100 (Thierry et al., 1990) as a BglII-BamHI-fragment generating pLEUMX1 and pLEUMX2 in both orientations. Plasmid pBEM2 was created by inserting a 1.7 kb SalI-fragment containing the ScLEU2 gene excised from pLEUMX1 into the unique XhoI site of pBEM2-A (see results). In-frame deletions were produced based on pBEM2. Plasmid pBem2Δ885-1250, carrying Agbem2Δ2, was created by deleting a SpeI restriction fragment from pBEM2, plasmid pBem2Δ211-1436, carrying Agbem2Δ3, by deleting a NruI/MscI fragment from pBEM2, and plasmid pBem2Δ1152-2071, carrying Agbem2Δ4, by deleting an AocI fragment from pBEM2. The resulting plasmids were checked by restriction digest and the newly created junctions were sequenced. In plasmid pBem2Δ1152-2071 a frameshift at the new junction was found altering the SauI restriction site ‘CCTNAGG’ into ‘CCTAGG’. This frameshift led to a stop codon after the addition of a few amino acids as would have also been the case in a non-frameshifted clone followed by the original AgBEM2 terminator.

Complete deletion of the AgBEM2 open reading frame was achieved by using a PCR-based one-step gene replacement approach recently established in A. gossypii similar to the method used in S. cerevisiae (Wach et al., 1994; Wendland et al., 2000). In short, as a selectable marker gene we used the GEN3 module which consists of the kanamycin-resistance gene under the control of the S. cerevisiae TEF2 promoter and terminator sequences and confers resistance to G418/geneticin to A. gossypii transformants. In a PCR-step the module was amplified adding short target guide sequences that allow specific gene targeting in A. gossypii. For amplification BEM2-S1 (5′-ctacttgcgtactctttcgcgtgctcgtcagccaccgaacaacgcagGCTAGGGATAACA GGGTAAT-3′) and BEM2-S2 (5′-gattaaagaatgataaagaaccaaaaacaccac-gagcttgcataacaGGCATGCAAGCTTAGATCT-3′) primers (homology to the marker cassette in upper case letters and homology to the target locus in lower case letters) were used generating a PCR product that was used directly in transformation of either the A. gossypii wild-type strain or the A. gossypii leu2, thr4 strain. Preparation of A. gossypii mycelia for transformation was done as described previously (Maeting et al., 1999). After electroporation mycelia were resuspended in 1.0 ml AFM, plated on AFM plates, and incubated for 6 hours at 30°C to allow regeneration of the mycelia and expression of the resistance marker. Subsequently plates were first overlayed with 7.0 ml 0.5 % agarose containing 1.4 mg G418 and after an incubation period of 40 hours at 30°C a second overlay of 7.0 ml 0.5% agarose was applied containing 7.5 mg G418. Primary heterokaryotic transformants were isolated and via a sporulation step homokaryotic mutant Agbem2Δ1 strains were obtained according to a procedure described previously (Steiner et al., 1995). Correct deletion of AgBEM2 was shown by analytical PCR using the primers: G1: 5′-gctgagcctccccgcctag-3′; G2: 5′-gtttagtctgaccatctcatctg-3′; G3: 5′-tcgcagaccgataccaggatc-3′; G4: 5′-gcacatagtttcaaagcggcg-3′; I1: 5′-gtataagtacttggagaaaaag-3′; and I2: 5′-gcgagatcatcggtgaagtc-3′. Additionally DNA-hybridization was employed to verify locus specific gene targeting. Transformations of plasmid DNA were performed following the same transformation protocol. For selection of LEU⁺ transformants transformation mixtures were directly plated onto selective minimal medium plates.

For calcofluor (0.1 μg/ml) or DAPI staining (1.0 μg/ml) cells were grown either in liquid medium or on microscopy slides covered with thin layers of full medium solidified with 0.7% agarose. For calcofluor staining untreated germ cells or mycelia were directly incubated with the dye for up to 20 minutes at room temperature. To eliminate background fluorescence samples were washed with water. For DAPI staining, cells were fixed in 70% ethanol for two minutes, stained, and washed with water. For actin staining, cells were fixed in 3.7% formaldehyde containing 0.4% Triton X-100 (fixation for less than 30 minutes gave best results). Mycelia or germinated spores were collected by centrifugation and washed several times in PBS (9.4 g/l Na₂HPO₄, 1.8 g/l NaH₂PO₄, and 4.4 g/l NaCl) and resuspended in PBS containing 1-2 μl rhodamine-conjugated phalloidin (77 μM in methanol). After an incubation of 1-2 hours at room temperature samples were washed five times in PBS and finally resuspended in mounting medium (90% glycerol, 10% PBS including 0.1% p-phenylene-diamine). Microscopy was performed using a Zeiss Axioskop microscope with the appropriate filter combinations. For picture acquisition either a TE/CCD-1000PB back-illuminated cooled CCD camera (Princeton Instruments, Inc, Trenton, NJ, USA) or a VI-470 CCD camera (Optronics, Goleta, CA, USA) was used. With the former image acquisition and picture processing was done using Metamorph 3.51 software (Universal Imaging Corp., West Chester, PA, USA) and with the latter images were processed by the NIH-Image software package (http://rsb.info.nih.gov/nih-image).

The Ashbya gossypii BEM2 gene was initially identified in a screen of 650 end-sequenced plasmid clones. Two identical clones (pAG1060 and pAG1637) revealed sequence similarity to the 5′ end of S. cerevisae BEM2. Colony hybridization was performed to isolate the complete AgBEM2 gene (using a library kindly provided by K. P. Stahmann). Two overlapping clones (pBEM2-A and pBEM2-B) were isolated which together cover a region of 9.4 kb (Fig. 1). The complete double strand sequence was determined and the entire AgBEM2 gene including its promoter was located on pBEM2-A. Searches for other open reading frames and BLAST searches in the available databases revealed two complete and two partial genes on the 9.4 kb fragment. Interestingly, this four-gene cluster is conserved between A. gossypii and S. cerevisiae (Fig. 1). The degree of conservation among the A. gossypii and S. cerevisiae proteins ranges between 26% and 57% identity. AgBem2p is 44% identical and 66% similar to ScBem2p. The AgBEM2 open reading frame is predicted to encode a protein of 2071 aa whereas ScBem2p is a 2167 aa protein. Conservation between AgBem2p and ScBem2p can be found almost along their entire lengths (Fig. 2A). Only the amino-terminal regions of both proteins, which are rich in serine/threonine residues, are more divergent and account for most of the size difference (Fig. 2B).

A pairwise alignment of AgBem2p and ScBem2p revealed nine blocks of greater than 65% aa identity compared to the 44% overall aa identity (Fig. 2B). A profile search on both Bem2 proteins using the PROSITE Database assigned specific domains to three of these conserved blocks. The carboxy-teminal 250 aa residues of both Bem2 proteins show similarity to pleckstrin homology (PH)- and GTPase activating protein (GAP)-domains found in a large family of proteins (Lemmon et al., 1996; Lamarche and Hall, 1994); including e.g. S. cerevisiae Bem3p (Zheng et al., 1994), human Bcr (Heisterkamp et al., 1985; Hariharan and Adams, 1987; Lifshitz et al., 1988), and human rhoGAP (Lancaster et al., 1994). Sequence identity of the AgBem2p GAP-domain with the GAP-domains of human Bcr, ScBem3p and rhoGAP is 36%, 26%, and 23%, respectively (Fig. 3A). Alignment of these GAP-domains shows several highly conserved amino acids residues including an arginine residue at position 85 of the rhoGAP sequence which according to its resolved structure plays a critical role for GAP-funtion (Rittinger et al., 1997). ScBem2p, for example, has been shown to function specifically as a GTPase activating protein for ScRho1p (Peterson et al., 1994). Two other domains that were indicated by the profile search in both the A. gossypii and S. cerevisiae Bem2 proteins were a putative rasGEF-domain between the aa residues 451 and 598 and a LTE-domain between positions 989 and 1029 of the AgBem2p. Protein alignments of rasGEF domains of proteins from yeast, mouse and Drosophila suggested the presence of three ‘structurally conserved regions’, SCR1, SCR2, and SCR3, respectively (Boguski and McCormick, 1993). Alignment of the AgBem2p and ScBem2p rasGEF-domains with these SCRs showed recognizable agreements to the SCR2 whereas manual alignment of the AgBem2p sequence to SCR1 and SCR3 showed only moderate conservation (Fig. 3B). Since this putative rasGEF domain is located in a well conserved region between AgBem2p and ScBem2p suggests a functional role for this domain even if it does not correspond to the bona fide rasGEF consensus represented by the PROSITE annotation. For the LTE1/rasGRF-associated domain named after the S. cerevisiae LTE1 gene only a preliminary entry is available in PROSITE which does not implicate a specific function of this domain. Six additional blocks (I to VI) of high homology between AgBem2p and ScBem2p were found. Profile and blast searches with these short aa sequences revealed no motif and did not identify other proteins with similar sequences which could have indicated some functional relationship.

The open reading frame of AgBEM2 was completely deleted from the genome by one-step gene replacement using the recently established PCR-based gene targeting protocol for A. gossypii (see Materials and Methods). The deletion was constructed both in an A. gossypii wild-type and in an A. gossypii leu2, thr4 strain using the dominant selectable marker GEN3 that confers resistance to G418/geneticin (Fig. 4A). The hyphae of A. gossypii consist of multinucleate septate compartments. The primary transformation event generated a heterokaryon consisting of nuclei that carried either the wild-type AgBEM2 or the Agbem2Δ1 deletion (Fig. 4B). The heterokaryotic transformants were indistinguishable from wild type as far as hyphal and colony morphology were concerned but were able to grow on selective medium. In order to obtain homokaryotic mutant strains, in which all nuclei carried the mutant allele, spores (which are mononucleate and haploid) of the heterokaryotic strains were isolated and germinated under selective conditions (see Materials and Methods). Viable homokaryotic Agbem2Δ1 transformants could be obtained indicating that AgBEM2 is not essential. The constructed deletions were confirmed by analytical PCR (Fig. 4B). Additionally, DNA hybridization experiments on chromosomal DNA of the progenitor strains, as well as on the heterokaryotic and homokaryotic mutants indicated that the selectable marker module had integrated correctly only once into the genome at the BEM2 locus (data not shown).

Mycelial growth assays were performed comparing the radial growth rates of wild-type and Agbem2Δ1 strains at 20°C, 30°C, and 37°C (Fig. 5). Agbem2Δ1 shows a slow growth phenotype at all temperatures. The ability of Agbem2Δ1 mutant strains to grow above 33°C is in contrast to the Scbem2 strains which are inviable above this temperature (Kim et al., 1994; Peterson et al., 1994). Restoration of wild-type-like growth-rate was achieved in an A. gossypii bem2Δ1, leu2, thr4 strain by complementation with the freely replicating plasmid, pBEM2, containing the wild-type AgBEM2 and the ScLEU2 gene as a selection marker (Fig. 5). Freely replicating plasmids are stably maintained under selective conditions in A. gossypii but can be lost in the mycelium under non-selective conditions (Wright and Philippsen, 1991). Interestingly, pBEM2 is maintained in the mycelium of Agbem2Δ1 strains even without selective pressure suggesting that fast growing hyphae have become dependent on AgBEM2. Two lines of evidence indicated the free replication of the complementing plasmid. (i) Repeated rounds of spore isolation and re-growth experiments from non-selectively grown mycelia of Agbem2Δ1 transformants harboring the pBEM2 plasmid indicated differential plasmid distribution in the spores. Plasmid loss resulted in the reoccurrence of the slow growth phenotype that was observed in Agbem2Δ1 whereas fast-growing colonies had maintained pBEM2. (ii) Transformation of DNA obtained from the pBEM2 bearing A. gossypii strains into E. coli cells resulted in the formation of ampicillin resistant colonies.

In an attempt to identify subdomains within the 2071 aa of AgBem2p that are required for the complementation of Agbem2Δ1 we constructed internal deletions of AgBEM2 creating Agbem2Δ2, Agbem2Δ3, and Agbem2Δ4 based on pBEM2 (Fig. 6). Agbem2Δ2 fully complemented the Agbem2Δ1 strain whereas Agbem2Δ4 could not. Deletion of approximately 60% of the BEM2 open reading frame in Agbem2Δ3 was partially able to complement the BEM2 deletion. Complementation resulted in the restoration of elongated hyphal morphology, which led to increased rates of radial growth compared to the deletion strain transformed with a ScLEU2 control plasmid. Radial growth rates, however, were only 45% of that of wild type and microscopic inspection revealed an increased proportion of hyphae exhibiting an Agbem2Δ1-associated phenotype (data not shown). These results indicated that the carboxy-terminal part of AgBem2p containing the rhoGAP-domain is critical for its function. The partial complementation obtained with Agbem2Δ3 could be due to either the loss of function of the putative rasGEF-domain or a decrease in protein stability of AgBem2Δ3p since a large part of AgBEM2 (60%) had been removed. Similarly, it was shown in S. cerevisiae that the temperature sensitive growth defect of Scbem2 cells could be complemented by overexpressed carboxy-terminal fragments of ScBem2p containing the PH and GAP-domains (Peterson et al., 1994; Kim et al., 1994).

Germination of A. gossypii spores occurs as an isotropic growth phase that results in the formation of spherical germ cells. Wild-type spores germinate into germ cells that reach on average 8 μm in diameter (n=100; s.d.=0.5 μm; Fig. 7A). Seven to eight hours after inoculation of spores into rich medium at 30°C a switch from isotropic to polarized growth leads to the formation of the first hyphal tube (Fig. 7B). After formation of the first hypha a second hypha is produced opposite of the first resulting in a bipolar germ cell branching pattern (100/100 observed cases; Fig. 7C). From these initial hyphae a mycelium is generated first by lateral branching and later (16-18 hours post inoculum) by dichotomous tip branching (Fig. 7D,E).

Dichotomous tip branching is usually not observed during the initial period of mycelium formation suggesting that a second switch is needed in A. gossypii cells to progress from unipolar hyphal tip growth to dichotomous tip growth which then occurs at regular intervals (Fig. 7E). At the edge of an A. gossypii colony almost all new tip cells were found to be produced by dichotomous tip branching compared to lateral branching of subapical hyphal cells.

In Agbem2Δ1 mutant strains the isotropic growth phase during spore germination was prolonged and produced enlarged germ cells with cell diameters of up to four times the size of wild-type germ cells. Interestingly, on the surfaces of Agbem2Δ1 germ cells protuberances were noted that might represent multiple randomly distributed initiation sites of polarized hyphal growth (Fig. 8A). This indicated the inability of Agbem2Δ1 germ cells to correctly initiate the switch from isotropic to polar growth. Eventually germ cells were either unable to produce hyphae and oversized cells lysed or Agbem2Δ1 germ cells switched to polarized growth. In contrast to wild-type germ cells, however, these Agbem2Δ1 cells were defective in generating the bipolar germ cell branching pattern (0/200 observations). Agbem2Δ1 germ cells were found to produce either only one or multiple growing hyphae (Fig. 8A,B). The dichotomous tip branching at later stages of mycelium development was not affected in Agbem2Δ1 nutants. However, the distance between two consecutive tip branches were found to be much shorter in Agbem2Δ1 hyphae than in wild-type hyphae (Figs 7E, 8C).

Fluorescence microscopy was used to analyze the distribution of chitin (and other β-linked polysaccharides) by calcofluor staining. In wild-type hyphae calcofluor decorated sites of septation that occur at regular intervals and showed that hyphal tips stain more brightly compared to subapical cell wall regions indicating areas of polarized cell growth (Fig. 9A). In Agbem2Δ1 hyphae staining at sites of septation was found to be comparably weak and in tip cells of Agbem2Δ1 hyphae calcoflour staining was most often uniformly distributed along the hyphal walls (Fig. 9B). This indicated a defect in polarized cell growth and a complete loss of cell polarity in those hyphal tip cells that had become spherical (Fig. 9B). The diameter of Agbem2Δ1 hyphae was up to three times larger than that of wild-type hyphae. We therefore asked whether the positioning of nuclei in Agbem2Δ1 hyphae was also different from that of the wild type. DAPI-fluorescence showed that wild-type nuclei are rather evenly distributed whereas nuclei in Agbem2Δ1 hyphae accumulated to large numbers in swollen areas of the hyphae (Fig. 9C,D).

Analysis of the polarity of the actin cytoskeleton in wild-type hyphae by staining with rhodamine-phalloidin showed that actin cortical patches were clustered in hyphal tips, indicating regions of active growth, and in ring like structures at sites of developing septa (Fig. 10A,B). Actin cables could be resolved in wild-type only after strongly overexposing actin patches but were invisible in Agbem2 mutants using this staining protocol. In Agbem2Δ1 hyphae actin-ring formation at sites of developing septa was found to occur (Fig. 10C). Actin cortical patches still localized to the hyphal apices, although the area of patch distribution was found to be much broader corresponding to the enlarged size of the tips of Agbem2Δ1 hyphae. Hyphae that had swollen at their tips during a period of isotropic growth, however, showed a uniform distribution of actin cortical patches (Fig. 10C). Thus, the defects in polarized cell surface growth and loss of cell polarity visualized by chitin staining were connected with a failure in maintaining the polarity of the actin cytoskeleton. Surprisingly, swollen Agbem2Δ1 hyphal tip compartments could establish new cell polarities resulting in the formation of multiple hyphal tips at randomly positioned sites (Fig. 10C). These ‘protuberances’ which had already been noted in Agbem2Δ1 germ cells were shown to contain accumulations of cortical actin patches and therefore corresponded to newly established sites of polarized growth (Fig. 10C). Initiation of multiple new cell polarities in Agbem2Δ1 mutants occurred at two different stages: (i) In Agbem2Δ1 germ cells after prolonged periods of isotropic growth (ii) in vegetative hyphae after loss of cell polarity. The observation that in germ cells not all newly established sites of growth continued to form hyphae might suggest the requirement of other factors whose supply could be limited.

We chose the filamentous ascomycete A. gossypii as a model to investigate polarized hyphal growth on the molecular level due to its very small genome and the facile applicability of powerful molecular genetic tools. During an initial sequencing screen, we tried to identify A. gossypii homologs of genes known to be part of rho-GTPase modules in other organisms. In this study we have presented the characterization of an A. gossypii rhoGAP homolog. We could show, that a member of a rho-GTPase module is involved in the maintenance of hyphal growth and cell polarity. Several lines of evidence suggested that the rhoGAP homolog isolated from A. gossypii is the functional homolog of the S. cerevisiae BEM2 gene. (i) Both proteins show sequence similarity along their entire length. (ii) The function of both proteins depends on their carboxy termini carrying PH- and GAP-domains. (iii) The AgBEM2-locus displays synteny to the S. cerevisiae BEM2-locus. The order of at least four genes including their transcriptional orientation has been conserved. Another example for the conservation of a four-gene cluster has been reported for the A. gossypii and S. cerevisiae THR4-loci (Altmann-Jöhl and Philippsen, 1996). (iv) Phenotypic similarities were observed in the A. gossypii and S. cerevisiae mutant strains concerning heterogeneity in cell size, cell lysis, depolarized growth, and defects in the organization of the actin cytoskeleton. Preliminary experiments indicated that the AgBEM2 gene does not complement a S. cerevisiae BEM2 deletion. This might be due to the inability of the AgBEM2 promoter to function in S. cerevisiae. Analysis of the inter-ORF regions upstream of the AgBEM2 and ScBEM2 open reading frames, which is only 262 bp in A. gossypii and therefore 215 bp shorter than in S. cerevisiae revealed no conserved promoter elements. Biochemical studies have shown that ScBem2p shows GAP-activity towards ScRho1p but not towards ScCdc42p (Zheng et al., 1993, 1994; Peterson et al., 1994). Similar experiments with AgBem2p could delineate its function in A. gossypii.

The bipolar branching pattern of A. gossypii wild type was abolished in AgBEM2 mutants. In A. nidulans temperature-sensitive mutant strains interfering with a similar branching pattern have been isolated recently (Harris et al., 1999; Momany et al., 1999). Germination and growth of Agbem2Δ1 strains was not restricted at elevated temperatures, although at all temperatures prolonged periods of isotropic growth led to heterogeneity in the sizes of germ cells as well as to increased germ cell lysis. In contrast, Scbem2 cells become arrested as large unbudded multinucleate cells at temperatures above 33°C (Adams et al., 1990; Peterson et al., 1994; Kim et al., 1994). In S. cerevisiae heterogeneity in cell-size as well as a cell lysis phenotype has been described previously for the LYT3 gene of S. cerevisiae which was later found to be allelic to BEM2 (Cid et al., 1994, 1998).

Polarity defects in Agbem2 mutants eventually caused completely isotropic growth resulting in the deposition of chitin over the entire cell surface. Depolarized growth generated large balloon-shaped tip cells. In A. gossypii as well as in other filamentous fungi cortical actin is concentrated at active sites of growth and secretion, i.e. in hyphal apices and at sites of developing septa (Harris et al., 1994; Salo et al., 1989; Runeberg et al., 1986). In Agbem2Δ1 hyphae actin patches were depolarized as could be shown by their uniform distribution. A similar phenotype has been described for Scbem2 strains and also for A. nidulans and Saprolegnia ferax wild-type after treatment with cytochalasin A or latrunculin B, respectively, which both disrupt cortical actin organization (Wang and Bretscher, 1995; Torralba et al., 1998; Gupta and Heath, 1997). In S. cerevisiae genetic interactions of BEM2 with a number of cytoskeletal genes have been shown. Among these are synthetic lethality of BEM2 with alleles of TPM1 (tropomyosin), MYO1 (myosin II), MYO2 (myosin V), ACT1 (actin) and SAC6 (fimbrin; Wang and Bretscher, 1995). With respect to the polarity of the actin cytoskeleton it was shown that tropomyosin is a stabilizing component of actin cables. Conditionally defective tropomyosins lead to a disassembly of actin cables under restrictive conditions which results in a depolarization of actin cortical patches and isotropic growth in S. cerevisiae cells (Pryne et al., 1998). Therefore a genetic interaction between BEM2 and TPM1 provides a route via which a Rho1-GTPase module could modulate the actin cytoskeleton (Pryne et al., 1998). An alternative route was provided by showing that ScRho1p interacts with ScBni1p in a GTP-dependent manner (Kohno et al., 1996). ScBni1p interacts directly with actin and ScBNI1 is synthetic lethal with ScRHO1 (Kohno et al., 1996; Evangelista et al., 1997). Overproduction of N-terminally truncated ScBNI1 (452-1953) interestingly leads to depolarized growth resulting in large round cells (Evangelista et al., 1997) as does overproduction of ScRho1p (Peterson et al., 1994; Madaule et al., 1994) resembling Scbem2 mutant cells. It will be of interest to examine the role of AgBEM2 in the morphogenetic network of A. gossypii.

Surprisingly, we found that growth of Agbem2Δ1 hyphal tips did not become arrested after an isotropic growth phase. Instead, repolarization of the actin cytoskeleton occurred generating multiple sites of activated polarized growth. Several observations suggested that multiple branch points were produced simultaneously from swollen hyphal tips: (i) In a few cases simultaneous outgrowth of new hyphal tips could be observed via time lapse video microscopy. (ii) Samples from logarithmically grown cultures contained hyphae with apical swellings from which already multiple elongated hyphae of comparable length had emerged suggesting that their growth had started at around the same time. (iii) Cyclic reoccurrence of rounds of polarity establishment, hyphal growth, and loss of cell polarity was observed in a few instances.

We could show that the carboxy terminus of AgBem2p containing a PH- and rhoGAP-domain is critical for its function. Additionally both, AgBem2p and ScBem2p, contain a domain with moderate homology to rasGEF domains as well as seven other conserved sequence elements of unknown function. The possession of both a putative rasGEF- and a rhoGAP-domain in AgBem2p and ScBem2p might provide a link between ras- and rhoGTPase modules. Such a link could be important to control the isotropic growth phase during germination of spores. It was shown in the zygomycete Phycomyces blakesleeanus that a RhoA homolog was expressed during germination and in S. cerevisiae that during spore germination signaling through the Ras protein pathway but not via Cdc28p is required (Ramirez-Ramirez et al., 1999; Herman and Rine, 1997). Ras proteins have been isolated from various ascomycetous fungi and were shown to be regulators of growth and morphology (Truesdell et al., 1999; Kana-uchi et al., 1997; Som and Kolaparthi, 1994).

The role of AgBem2p in the morphogenetic network might also suggest an explanation for the fact that at a given time only a portion of Agbem2Δ1 hyphae (<5%) displayed completely isotropic growth resulting in spherical swellings of hyphal tips. The amount of Rho1p-GTP increases upon stimulation of GDP-GTP exchange by Rho1p-GEF. Signaling via Rho1p-GTP in an Agbem2Δ1 background is not turned down in A. gossypii which might initially result in delocalized but still polar growth generating hyphae with an enlarged diameter compared to wild-type cells. Eventually, if signaling via Rho1p-GTP surpassed an overflow level, cell polarity would be lost. Interestingly, loss of cell poarity can be overcome in A. gossypii both at the germ cell stage and during vegetative growth. An important role in the process of re-establishment of polarity might be exerted by A. gossypii homologs of genes required for polarizing the actin cytoskeleton, e.g. the homologs of the S. cerevisiae genes CDC42 and CDC24. Therefore, we have recently isolated and started to characterize the AgCDC42 and AgCDC24 genes (J. Wendland et al., unpublished). Deletion of both genes generated similar phenotypes as have been described for the temperature-sensitive mutant strains of S. cerevisiae at the non-permissive temperature (Bender and Pringle, 1989; Adams et al., 1990). A. gossypii spores of heterokaryotic mutant strains germinated but then failed to initiate the switch from isotropic to polarized growth and became arrested as large, unbranched, multinucleate germ cells (J. Wendland et al., unpublished). This indicates a role of the AgCdc42p-GTPase module during establishment of polarity, which is taking place in a deregulated manner both in Agbem2Δ1 germ cells and swollen hyphal tips.

Our results suggest that in filamentous fungi rhoGTPase modules such as driven by the rho-GTPasesCDC42 and RHO1 are key regulators for polarized hyphal growth in a similar way as has been shown in yeast and animal cells. Further studies may therefore reveal the peculiarities of the different morphogenetic regulatory machineries that result in the generation of distinct growth forms.

We thank Natascha Springer for her work on AgBEM2 during an ESBS practical course, members of the Ashbya-group for helpful discussions, Peter Stahmann for generously providing an A. gossypii plasmid library, and Clarence Chan for providing strains and plasmids. We greatly appreciate the efforts of Dominic Hoepfner and Florian Schaerer who set up excellent hardware and software facilities used for fluorescence microscopy. This work was supported by grants from the University of Basel. The nucleotide sequence reported in this paper has been submitted to the DDBJ, GenBank/EMBL Data Bank with the accession number AF195007.