The Small GTP-Binding Proteins AgRho2 and AgRho5 Regulate Tip-Branching, Maintenance of the Growth Axis and Actin-Ring-Integrity in the Filamentous Fungus Ashbya gossypii

GTPases of the Rho family are important molecular switches that regulate many basic cellular processes. The function of the Rho2 and Rho5 proteins from Saccharomyces cerevisiae and of their homologs in other species is poorly understood. Here, we report on the analysis of the AgRho2 and AgRho5 proteins of the filamentous fungus Ashbya gossypii. In contrast to S. cerevisiae mutants of both encoding genes displayed a strong morphological phenotype. The Agrho2 mutants showed defects in tip-branching, while Agrho5 mutants had a significantly decreased growth rate and failed to maintain their growth axis. In addition, the Agrho5 mutants had highly defective actin rings at septation sites. We also found that a deletion mutant of a putative GDP-GTP-exchange factor (GEF) that was homologous to a Rac-GEF from other species phenocopied the Agrho5 mutant, suggesting that both proteins act in the same pathway, but the AgRho5 protein has acquired functions that are fulfilled by Rac-proteins in other species.


Introduction
The family of small GTP-binding proteins of the Rho-type is composed of important regulators of the actin cytoskeleton in nearly all eukaryotic organisms. Because Rho-homologs are highly redundant in metazoan cells, the analysis of specific functions of single proteins is complicated. Therefore, model organisms with less Rho-GTPase redundancy are valuable tools for deciphering Rho-GTPase function and evolution. Often fungi are used for this purpose. A recent survey that is based on datasets containing many complete genome sequences suggests that generally six groups of paralogs exist in fungi [1]. Using the terminology from Saccharomyces cerevisiae these are Cdc42, Rac (Rho5, see below), Rho1, Rho2, Rho3 and Rho4. While some fungi contain additional Rho paralogs that seem to have been acquired by horizontal gene transfer (e.g. Aspergillus fumigatus and Neurospora crassa), some species within the above named groups carry additional copies which most probably stem from duplication events. This is, for example, the case for Rho1 in Ashbya gossypii (see below) and Schizzosaccharomyces pombe [1]. For nearly each Rho-group, at least one species carrying such duplications can be found [1]. There are also many species that lack members in one of the fungal Rho-family groups. A special case which should be specifically noted here is the Rac-homolog of S. cerevisiae. For a long time it was generally accepted that an Rac ortholog is not present in this yeast, but some functional analysis data, and more recent phylogenetic analyses, suggest that Rho5 might be a true Rac-ortholog in S. cerevisiae [1,2,3]. Therefore, within the currently available fungal genome datasets, Schizzosaccharomyces pombe seems to be the only fungal model organism that does not possess a Rac-homolog [1].
Analysis of Rho-GTPase function has been carried out in many fungal model organisms including N. crassa [4,5], Aspergillus nidulans [6,7], Ustilago maydis [8,9], Ashbya gossypii [10] and Schizzosaccharomyces pombe [11]. Even though studies in these organisms have significantly contributed to our understanding of Rho-GTPase function and especially functional diversification, the organism with the best characterized set of Rho-GTPases is still the baker's yeast S. cerevisiae, simply because the earliest attempts at studying Rho-proteins in a fungus were made in this organism [12] and by far the most publications on fungal Rho-GTPases are to be found for S. cerevisiae. Therefore, the function of many members of the Rho-family in S. cerevisiae is very well understood. This is the case for Cdc42, Rho1, Rho3 and Rho4. They play important roles as far as cell polarity, cytokinesis, cellular integrity and secretion are concerned, and, accordingly, mutants of these genes present clear phenotypes. However, the function of the other members of this protein family is less well characterized, often simply because no clear phenotype is associated with mutants of the corresponding genes.
We have previously shown that some mutations of syntenic homologs to the genes of Saccharomyces cerevisiae had a more severe phenotype in the close-relative species Ashbya gossypii [13]. This finding is due to the fast and strictly filamentous growth of this fungus, which depends on a large and very dynamic actin cytoskeleton. Therefore, investigations on homologs in A. gossypii often provide novel and additional information on gene function. All family members of the Rho-type GTPases from S. cerevisiae are also conserved in A. gossypii. Currently, AgCDC42, AgRHO3, AgRHO1a and AgRHO1b have been characterized in A. gossypii [14,15,16]. AgCdc42 is essential for the switch from isotrophic to filamentous growth during the germination of the spore and, most likely, for polarity at the hyphal tip, where the protein localizes [14,17]. The latter function has not been shown experimentally because deletion mutants of Agcdc42 fail to leave the isotropic growth phase after germination and consequently die [14]. The RHO1 gene has two homologs in A. gossypii. This duplication is not unique to A. gossypii, but is rather conserved in several species of the Saccharomycetaceae, including the putative ancestor of S. cerevisiae and A. gossypii that has been calculated from the genomes of several yeast species [18]. While mutants of Agrho1b show phenotypes that are very similar to Scrho1 mutants and suggest a function in cell wall biosynthesis and cellular integrity [14,15], the additional second copy of the RHO1 gene, AgRHO1a, which is present in the genome of A. gossypii, plays an important role in maintaining cellular integrity under high hyphal growth speeds [14,15]. However, the A. gossypii RHO1a gene carries a unique mutation that is most likely associated with the filamentous nature of A. gossypii and allows for a decoupling of AgRho1a and AgRho1b activation [15]. The Rho3 protein  functions in polarity maintenance. Mutants frequently lose polarity but eventually continue filamentous growth [14].
In contrast to the genes described above, the functions of the RHO2 and RHO5 genes are less understood. RHO2 and RHO5 have been only poorly characterized in S. cerevisiae. Rho2 is thought to have a partially overlapping function with Rho1 in actin cytoskeleton repolarization [19]. The role of Rho5 is also not clear. A null mutant has no obvious growth or cellular morphology phenotype. Instead, its role appeared to be in response to various stresses and some reports suggest a role for Rho5 in oxidative stress response, as well as in osmotic and cell wall stress [2,20,21]. In addition, the activation of both GTPases is still unclear because as yet no GDP-GTP-exchange factor has been identified [22].
Here, we report on the phenotypes of Agrho2 and Agrho5 mutants in Ashbya gossypii. We show that both of the GTPases have strong phenotypes that affect hyphal growth and morphology. Additionally, we demonstrate that the Agrho5 phenotype is copied by a mutant in a putative Agrho5 GDP-GTP exchange factor (GEF). The latter gene encodes a homolog to members of the DOCK-homology family, which are hypothesized to act as putative GEFs for Rac homologs in other fungal organisms [23], even though experimental proof for such an activity is still missing.

A. gossypii strains and growth conditions
The construction of all Ashbya gossypii strains was performed using PCR-based gene targeting, as described by Wendland et al. [24]. For the generation of targeting cassettes for deletions or gene fusions, we used template vectors from the pAGT series [25] or pGEN3 [24]. The transformation of A. gossypii with plasmids was performed according to the protocol by Wright et al. [26]. All strains used for this study are listed in Table 1, and the oligonucleotides used for strain construction are listed in Table 2. For the generation of the Agrho2*-strain, the ScLEU2 gene was integrated on a vector behind the Agrho2 G195C allele [13] using the primers Lok-Leu2-NS1 and Lok-Leu2-F2. The allele was excised with the ScLEU2-marker and the flanking sequences and was transformed into the Agrho2 deletion to replace the GEN3 deletion marker.
A. gossypii was cultured in Ashbya Full Medium (AFM, [27]) supplemented with or without 200 mg/ml geneticin (Sigma-Aldrich, Taufkirchen, Germany). To use the auxotrophic marker leu2, cells were cultured in synthetic minimal medium (ASC, [27]). To avoid auto-fluorescence of the medium, we also used synthetic medium for fluorescence microscopy.

Fluorescence microscopy
For fluorescence microscopy, we used the protocol described in Kemper et al. [31]. Actin and Calcofluor White stainings were performed as described earlier [32].

Results
To gain an understanding of the function of the AgRho2 protein, we first constructed a deletion mutant by replacing the entire open-reading frame of the encoding gene by PCR-mediated gene targeting with a marker cassette. In addition, we made use of the fact that the GTPase structure and its regulation are highly conserved among Rho-type GTPases. This enabled the construction of GTP-locked mutants by the simple mutation of a conserved glutamine in the switch II region. We constructed such a mutant for AgRHO2 (Agrho2*) in vitro and integrated it at the original AgRHO2 locus in the genome. Figure 1A and B shows the macroscopic growth of the deletion and the GTP-locked rho2 mutants. While the deletion mutant grew slightly but significantly faster than the wild type, the Agrho2* mutant consistently grew slower. Interestingly a simple overexpression of the wild-type allele did not yield the same result, but grew rather like the wild type (Figure 1 C and D), suggesting that the activity of the protein is highly regulated. Although Agrho2* did not attain a similar growth speed as the deletion mutant, the microscopic growth of both mutants was highly similar. Figure 2A shows 3 time-points that are examples of typical defects of the Agrho2 deletion mutants, which we monitored by time-lapse video-microscopy (see Movie S1 for the complete process). Most of these defects are associated with untimed and irregular branching at the hyphal tip. Wild-type A. gossypii typically switches from a lateral branching pattern to a tip-branching pattern, which is the dichotomous division of one growing hyphae into two new hypha, at a growth speed above 80 mm/h [33]. A characteristic feature of the tipbranching process in wild-type strains is the broadening of the hyphal tip directly before tip-splitting occurs. Then, simultaneously, two novel branches begin growing at an approximately 45u angle. In Agrho2 deletion mutants, this tip-splitting is often disturbed, and branching at the tip proceeds at fast, as well as slow hyphal-tip growth speeds (Figure 2A). Actin staining with rhodamin phalloidine revealed that the broadening of the hyphal tip, which precedes tip-branching, occurred simultaneously with a loss of actin polarity at the growing tip ( Figure 2B). Despite the fact that the Agrho2* mutant grew more slowly than the deletion  mutant, the microscopic appearance was identical to that of the deletion phenotype. Figure 3A shows a typical example of defective tip-branching. The growth speed over the entire sequence shown in the figure is approximately 95 mm/h. During the first branching event that is visible in the sequence, only one side of the tip-branch continues to grow. The emergence of the second branch is delayed until a second tip-branch occurs at the first branch. Figure 3B shows that, like the Agrho2 deletion, the Agrho2* mutant shows a loss of actin polarity immediately before tip-branching. The tip-branch, which does not continue to grow, is completely free of actin.
The phenotypes associated with the loss and gain of the AgRho2 function suggest that the protein plays an important role in tip-branching. However, whether the loss of actin polarity at the hyphal tip during this process is directly influenced by AgRho2, or is an indirect side effect is not clear. To obtain further insight into AgRho2 during this process, we constructed amino-terminal fusions of the GFP-gene to AgRHO2 and integrated these fusions at the original gene locus. Figure 4A shows fluorescence images for the wild type and the strain carrying the Agrho2* allele. Surprisingly, the signal of the protein was not focused in the hyphal tip but was dispersed over the entire plasma-membrane for the wild type and for the GTP-locked allele. Additionally, no difference between straight growing tips and hyphal tips that had already undergone swelling, which occurs prior to tip-branching, was observed. The only accumulation that we found was associated with the Agrho2* alleles at mature septa. Because we recently found a connection between Rho-type GTPase signaling and spore formation in A. gossypii [32], we also searched for the localization of the GFP-fusions in sporangia. Figure 4B shows that the localization in sporangia was identical to hyphal localization. This was in contrast to many other proteins. The GFP-fusions of most proteins are not visible in sporangia compared to vegetatively growing hyphae. Therefore the visibility of GFP-AgRho2 in sporangia suggests a possible role for this protein in spore formation. However, even a close inspection of the Agrho2 deletion mutant for sporulation defects yielded no results. Figure 4C shows that even the average spore size of the mutant strain was identical to the wild-type spores.
AgRHO5 is another uncharacterized Rho-type GTPase in A. gossypii. To initiate the characterization of the function of this gene, we deleted the open-reading frame. Such deletion mutants had a significantly slower macroscopic growth rate, which reached only half the rate of the wild-type and suggested a strong effect on hyphal growth ( Figure 5A and B). Because the sequences of AgRHO5 and its homolog ScRHO5 share structural homology with Rac-genes [2] and because there is no obvious Rac-homolog in S. cerevisiae or A. gossypii, we speculated that the protein might have taken over some of the functions of Rac-homologs in other fungi. In the dimorphic fungus Candida albicans, the Guanine Nucleotide Exchange Factor (GEF) Dck1 is required for Rac1 activation during invasive filamentous growth. We identified a homolog to DCK1, AER102W, in the genome of A. gossypii and speculated that this might be a potential regulator for AgRho5 in A. gossypii. To test this hypothesis, we also deleted the gene AER102W by direct gene targeting. As shown in Figure 5A and B, the effect of the deletion on macroscopic growth speed was similar to that of the Agrho5 deletion. In addition, both deletion strains were more sensitive to the actin destabilizing drug Latrunculin A than the wild type. The similarities between the two deletion mutants became even more evident when comparing the microscopic growth defects. Figure 6A shows that both mutations led to defects during spore germination. In A. gossypii, after an initial isotropic growth phase, a germ-tube is formed at an angle that is close to 90u to the spore needle axis. We determined the deviation of the first germ-tube for the wild type and for both deletion mutants. This deviation was only 14.6u for the wild type, while it was 25u for the Agrho5 and 22.4u for the dck1 deletion. This polarity defect was maintained during formation of the second germ-tube ( Figure 6B). While in the wild-type germlings the second germ-tube was formed at the opposite side to the first germ-tube, it was formed on the same side as the first germ-tube in both mutant strains. A strong polarity defect was also observed in the mature hyphae. When grown on solid medium, wild-type hyphae that were not branching grew straight. This was not the case for the mutant strains. The hyphae of both mutant strains had a wavy appearance and did not maintain a consistent direction of growth. Because actin is a good marker for polarity in A. gossypii, we investigated the Agrho5 and the Agdck1 deletions for actin defects. Figure 7A shows the typical hallmarks of a wild-type A. gossypii hypha stained with rhodamin-phalloidine: a concentration of actin patches at the hyphal tip and actin rings at the sides of septation. The insert in the fluorescent image shows a 3Dreconstruction of such a wild-type ring. These are usually flat, continuous and consist only of a single ring. Figure 7B and 7C show examples of defective actin rings associated with the Agrho5 and the Agdck1 deletions. Again, the defects in both strains were similar: the 3D reconstructions showed that actin ring structures were often discontinuous, having either multiple rings or rings that were split on one side to form two rings. Often, these rings were not closed or seemed to have lost contact with the membrane. The general appearance of the rings suggested that initial formation of the rings was successful, but the integrity of the ring was lost at some point during ring formation. These actin ring defects suggest that a defect in septum formation may have also occurred. However, staining with Calcofluor White for mature chitincontaining septa showed that the mutant and the wild-type strains were capable of forming mature septa ( Figure 7D).
Although the defect in actin ring formation was strong, it does not explain the polarity defects associated with the deletion of the two genes. However, the nature of the defects that were observed at the actin rings suggests an instability of actin structures that might also be present at the hyphal tip, where it could lead to an instable growth axis. To investigate this, we visualized actin using an actin-binding domain fused to GFP. This construct was similar to the construct that is marketed as Lifeact but utilizes the actinbinding site of the AgABP140 homolog. To monitor the stability of actin at the hyphal tip, we followed a growing hyphae using fluorescence microscopy over a time course of 240 minutes and determined the differences between the two halves of the hyphae. Figure 8A shows that between two time-points, dramatic changes occurred, and the actin signal was often stronger on one side of the hyphae than the other, which is a possible explanation for the wavy growth pattern of the hyphae (See Movie S2 for the complete movie).
In addition to a phenotypic characterization, we also constructed a mCherry-Rho5 fusion and expressed it in the deletion strain. Vector expression complemented the macroscopic defect on growth speed ( Figure 9B), but it failed to complement the wavy growth of hyphae that was observed using the microscope ( Figure 9A), suggesting that the mCherry interferes with at least some aspects of protein function. Nevertheless, the observed localization pattern was consistent with the phenotypes described above. A weak signal was found at the hyphal membrane and tip, and stronger signals were associated with the septa.
As a control, we also complemented the mutant with an AgRHO5 allele that was not fused to a fluorescent protein. Such a strain grew macroscopically identical to the wild type ( Figure 9D). Microscopy observation revealed that most of the hyphae grew straight, only some hyphal tips still displayed the wavy phenotype of the deletion ( Figure 9C). This is typical for complementation by a plasmid, because copy number in different hyphae is not identical.
To test further genetic interactions between the genes we observed here, we also constructed double mutants of AgRHO5 with AgDCK1 and AgRHO2. For the double deletion mutant in which both, Agrho5 and Agdck1 were deleted, we were unable to get a viable homokaryon from more than 200 spores tested, strongly supporting lethality of the double deletion. The effect of the combination of DAgrho2 with DAgrho5 is shown in Figure 10. Interestingly, the double deletion displays the same phenotype as the DAgrho5 deletion on its own. This is the case for the macroscopic growth ( Figure 10A) as well as for the microscopic appearance of the hyphae which display the same wavy growth as the DAgrho5 ( Figure 10B). Also the tip branching defect of the DAgrho2 strain is not visible in the double deletion.

Discussion
In this study, we performed an analysis of two uncharacterized Rho-type GTPases of the filamentous fungus A. gossypii. Mutants of the homologs of both genes in S. cerevisiae possess no known morphological phenotypes; therefore, it is remarkable that both mutants displayed a strong morphological defect.
Agrho2 mutants mainly displayed a tip-branching phenotype. Often the hyphal tip did not split synchronously and formed multiple branches. It is interesting that the deletion and GTPlocked mutant look similar under the microscope. This suggests that the mutation that we introduced led to a loss of protein function. However, the growth speed of the mycelia was significantly different between the two mutants. This suggests that the tip-branching defect and growth speed of the mutants are not directly linked. A loss of function of GTP-locked versions of Rhotype GTPases is usually observed if the process requires the cycling of the GTPases between the GTP and GDP-state [34]. This appears to be the case for the tip-branching defect. In contrast, the differences observed in the growth speed were consistent with a loss or gain of Rho2-protein function: the faster growth of the deletion mutant and a slower growth of the GTP-locked allele compared to the wild type suggested an inhibitory role of active AgRho2 on growth speed.
A similar, asynchronous tip-branching defect which we observed for the Agrho2 mutants, has been previously described for the Agpxl1 mutant [35]. AgPxl1 is a member of the paxillin family, which are scaffold proteins that are involved in the recruitment of diverse regulatory and structural proteins. They play a central role in the coordination of Rho-type GTPases in many organisms. We have recently shown that AgPxl1 interacts with the two AgRho1-homologs of A. gossypii to balance their activity during spore formation [32]. The high sequence identity between AgRHO1a, AgRHO1b and AgRHO2 (46% and 50% identity) along with the observed similar phenotypes, suggests that, in vegetatively growing cells, AgPxl1 could be involved in the regulation of AgRho2 at the hyphal tip. This is in agreement with the fact that a GFP-fusion of AgPxl1 localized to the hyphal tip and to septa in vegetatively growing hyphae. However, our attempts to prove such an interaction by two-hybrid experiments and by co-immunoprecipitation with purified proteins were both unsuccessful. Further work will be required to ascertain whether there is either no connection, or only an indirect link between the highly similar defects of Agrho2 and Agpxl1 mutants. Interestingly, AgRho2 appears to not play a role in spore formation, although its GFP-fusion localized to sporangia. Spore formation was normal, and we could not observe an effect on spore length as we did for the mutants of Agrho1 or Agrho1b. Therefore, AgRho2 appears to have a regulatory role at the hyphal tip that is essential for proper tip-branching. Together with other tip-branching mutants, such as the Agpxl1 deletion [35] or an activated allele of the formin protein AgBni1, that display premature tip-branching [13], the Agrho2 mutant provides further evidence that tip-branching is a highly regulated process.
While AgRho2 appears to be required for branching, mainly at the tip at high-growth speeds, AgRho5 is required early in the development of A. gossypii. The Agrho5 deletion dramatically affected growth speed but not polarity. The hyphae appeared to be perfectly pointed and only the direction of polarity changed. These polarity changes already started at the germinating spores. The deletion of AgDCK1 was an exact phenocopy of the Agrho5 deletion, which suggests that both deletions function as activators of the same processes and supports the idea that RHO5 homologs are divergent Rac-homologs [2]. AgDCK1 is a homolog of the DOCK-family of proteins which, together with proteins of the socalled ELMO family, form a novel class of GTP-GDP exchange factors (GEFs). Even though direct biochemical proof still has to be provided, there is strong evidence to support the conclusion that the homolog of AgDCK1 in C. albicans functions as a putative GEF for the CaRac1 protein [23]. A Rac1 homolog is missing in A. gossypi and Rho5 homologs are absent in all filamentous fungi that possess a Rac homolog [3]; therefore, it seems likely that the AgRho5 protein has acquired some of the functions of the Rac protein and that the AgDck1 protein is the evolutionarily conserved GEF that regulates these functions. The processes that appear to be regulated by AgRho5 further support this hypothesis. Rac homologs in most fungi are responsible for hyphal growth, polarity maintenance and actin organization and have a varying degree of importance for the regulation of these processes [3]. Thus, the phenotypes observed in this study fit into this model. In addition, these results may contribute to a greater understanding of the potential effectors of the Rho5 protein. In S. cerevisiae, Rho5 was shown to bind to Ste50 and to be involved in the regulation of the osmotic stress response [20]. Because Ste50 and other components of this pathway, such as Cdc42, Ste20 and Ste11, are also involved in the regulation of filamentous growth, it is likely that a similar pathway may be responsible for the polarity defects observed in the A. gossypii rho5 deletion mutant. In contrast to S. cerevisiae, where Rho5 is involved in oxidative stress response [2], we were unable find any hints on an involvement of AgRho5 in response to oxidative stress (data not shown).
A characteristic feature of the phenotype of the Agrho5 and Agdck1 mutants, the instability of the actin structures at the hyphal tip and at septation sites and the sensitivity to Latrunculin A, may also point to proteins that may be involved in the affected processes. The homolog to the AgDck1 protein of S. cerevisiae physically interacts with Lsb3 (Las seventeen binding), which, in turn, binds to Las17 [36]. Mutants of the LAS17 homolog WAL1 (AgLAS17) in A. gossypii completely lack actin rings [37]. It is possible that all of these proteins act together in a pathway that regulates actin ring formation at future septation sites. The fact that the double deletion of Agrho5 and Agrho2 is lethal suggests further that both proteins signal into an additional branch independently, which is distinct from the signaling branch that is linked to the phenotype observed in the single deletion mutants. In addition, our result with an Agrho2/Agrho5 double deletion suggests that the Agrho5 deletion phenotype is dominant over the Agrho2 deletion phenotype. Given the fact that both phenotypes are related to hyphal growth, this is not surprising, as the effect of the Agrho5 deletion severely impairs hyphal growth, whereas the effects of an Agrho2 deletion are much less pronounced.
The strong phenotypes for the homologs of two Rho-family members that we have identified in this study -the mutants of which in S. cerevisiae only cause mild defects -emphasize the value of A. gossypii as a complementary, morphologically more complex model system to the yeast cells of S. cerevisiae.

Supporting Information
Movie S1 Supporting movie for Figure 2, showing the complete information that was used to generate the data shown in Figure 2.