Ras GTPase Activating Protein CoIra1 Is Involved in Infection-Related Morphogenesis by Regulating cAMP and MAPK Signaling Pathways through CoRas2 in Colletotrichum orbiculare

Colletotrichum orbiculare is the causative agent of anthracnose disease on cucurbitaceous plants. Several signaling pathways, including cAMP–PKA and mitogen-activating protein kinase (MAPK) pathways are involved in the infection-related morphogenesis and pathogenicity of C. orbiculare. However, upstream regulators of these pathways for this species remain unidentified. In this study, CoIRA1, encoding RAS GTPase activating protein, was identified by screening the Agrobacterium tumefaciens-mediated transformation (AtMT) mutant, which was defective in the pathogenesis of C. orbiculare. The coira1 disrupted mutant showed an abnormal infection-related morphogenesis and attenuated pathogenesis. In Saccharomyces cerevisiae, Ira1/2 inactivates Ras1/2, which activates adenylate cyclase, leading to the synthesis of cAMP. Increase in the intracellular cAMP levels in coira1 mutants and dominant active forms of CoRAS2 introduced transformants indicated that CoIra1 regulates intracellular cAMP levels through CoRas2. Moreover, the phenotypic analysis of transformants that express dominant active form CoRAS2 in the comekk1 mutant or a dominant active form CoMEKK1 in the coras2 mutant indicated that CoRas2 regulates the MAPK CoMekk1–Cmk1 signaling pathway. The CoRas2 localization pattern in vegetative hyphae of the coira1 mutant was similar to that of the wild-type, expressing a dominant active form of RFP–CoRAS2. Moreover, we demonstrated that bimolecular fluorescence complementation (BiFC) signals between CoIra1 and CoRas2 were detected in the plasma membrane of vegetative hyphae. Therefore, it is likely that CoIra1 negatively regulates CoRas2 in vegetative hyphae. Furthermore, cytological analysis of the localization of CoIraI and CoRas2 revealed the dynamic cellular localization of the proteins that leads to proper assembly of F-actin at appressorial pore required for successful penetration peg formation through the pore. Thus, our results indicated that CoIra1 is involved in infection-related morphogenesis and pathogenicity by proper regulation of cAMP and MAPK signaling pathways through CoRas2.


Introduction
Colletotrichum orbiculare is the causative agent of cucumber anthracnose disease. The infection process is initiated by the recognition of an appropriate surface. A series of changes in the morphology upon recognizing the appropriate signals, including the formation of a specialized infection structure called appressorium, is important for the successful infection of the host plants.
Several signal-transduction related genes associated with these morphological changes have been characterized in C. orbiculare [1]. It also has been shown that the MAPK and cyclic AMP (cAMP) signaling pathways are linked to infection-related morphological changes in this fungus. A yeast MAPK kinase (MAPKK) kinase STE11 homolog, CoMEKK1, is involved in the appressorium development in C. orbiculare [2]. C. orbiculare mutant with the adenylate cyclase encoding gene cac1 shows defectiveness in conidial germination [3], however, no upstream regulators of cAMP and MAPK signaling pathways have been identified in C. orbiculare.
Ras is the prototypical member of the small GTP binding protein superfamily that plays a pivotal role in proliferation and differentiation in eukaryotic cells. In Magnaporthe oryzae, MoRas1 and MoRas2 interact with Mst50, which interacts directly with the mitogen-activated proteins Mst11 and Mst7 [4]. In Fusarium graminearum, FgRas2 is involved in hyphal growth and pathogenicity [5]. In C. trifolii, CtRas is involved in conidial germination and hyphal growth [6]. Ras acts as a molecular switch that exists both in the active (GTP-bound) and the inactive (GTPase-bound) states. Cycling between these two states is aided by the interaction of a GTPase activating protein and guaninenucleotide exchange factors. In Saccharomyces cerevisiae, the Ras activity is controlled positively by the guanine nucleotide exchange factor (GEF) Cdc25 and negatively by the GTPase activating proteins Ira1 and Ira2 (GAPs) [7]. In S. cerevisiae, the RAS GTPase activating proteins Ira1 and Ira2 inactivate Ras1 and Ras2, which in turn activate adenylate cyclase, promoting the synthesis of cAMP from ATP. Ira1 and Ira2 are important factors for controlling cAMP production. The cAMP signaling pathway plays a pivotal role in transducing environmental cues during cell development in Magnaporthe oryzae [8]. Moreover, the cAMP signaling pathway of C. orbiculare plays a critical role in regulating conidial germination and pathogenicity [9]. Adenylate cyclase gene MAC1 encodes a Ras-association domain [10], [11]. In terms of protein structure, the Ras protein putatively interacts with Mac1 in M. oryzae. It was shown that cyclase-associated protein Cap1 of M. oryzae directly interacts with Mac1 and plays a role in the activation of Mac1, which may function downstream of Ras [12]. In Ustilago maydis, RAS2 promotes filamentous growth through a MAP kinase cascade, however, the function of RAS2 in the cAMP signal transduction remains unknown [13]. Therefore, direct evidences for the involvement of the Ras protein in the cAMP signal transduction are not sufficient in phytopathogenic fungi.
Because Ras proteins transduce signals from the cell surface to various intracellular effectors, they are located and function only at the plasma membrane [14], [15]. Conversely, data provided by the yeast GFP fusion localization database [16] and other published data [17], [18] indicate that Ras2, Cyr1, Cdc25, and Ira proteins mainly localize at the internal membranes of the endoplasmic reticulum and mitochondrial membranes, but only marginally at the plasma membrane. Recently, it was reported that active Ras accumulates mainly in the plasma membrane and nucleus when the cells are grown on medium containing glucose, whereas it accumulates mainly in the mitochondria in glucosestarved cells [19]. Moreover, PKA activity causes Ira2 to move away from the mitochondria [19]. Therefore, it is considered that the behavior of Ras proteins and Ras GTPase activating proteins and their localization depends on various growth conditions. In plant pathogenic fungi, subcellular dynamics of Ras proteins and Ras GTPase activating proteins during infection and the location of this interaction remain unclear.
In this study, we identified a novel S. cerevisiae homolog gene CoIRA1, encoding the RAS GTPase-activating protein (RAS-GAP) of C. orbiculare and characterized the CoIRA1 function in relation to the activation of cAMP-PKA and MAPK signaling pathways. By phenotypic analysis of the coira1 mutant, cytological analysis of CoRas2 localization and BiFC assay, we showed that CoIra1 regulates these signaling pathways through CoRas2 as a negative regulator. Conclusively, we presented CoIra1 is involved in infection-related morphogenesis by regulating cAMP and MAPK signaling pathways through CoRas2 in C. orbiculare.

Fungal and bacterial strains
The 104-T (MAFF240422) C. orbiculare (Berk. & Mont.) Arx [syn. C. lagenarium (pass.); Ellis & Halst.] strain was used as the wild-type. All the C. orbiculare strains used in this study are listed in Table 1 and were cultured on PDA media (3.9% [w/v] PDA; Difco laboratories, Detroit) at 24uC. One shop TOP10 chemically competent E. coli cultured in Luria-Bartani media [20] at 37uC was used as a host for gene manipulation. When required, the supplement kanamycin was added to the medium at 50 mg/ml. A. tumefaciens C58C1 cultured in Luria-Bartani media at 28uC was used to transform C. orbiculare by AtMT.

Fungal transformation
For the fungal transformation, we used an AtMT protocol that was previously described [21] with slight modifications. The hygromycin-resistant transformants were selected on the PDA medium containing 100 mg/ml of hygromycin B (Wako Chemicals, Osaka, Japan), 50 mg/ml of cefotaxim (Wako Chemicals, Osaka, Japan), and 50 mg/ml of spectinomycin (Wako Chemicals, Osaka, Japan). The bialaphos-resistant transformants were selected on an SD medium containing 10 mg/mL of bialaphos (Meiji Seika Kaisha, Ltd., Tokyo, Japan), 100 mg/ml of cefotaxim, and 100 mg/ml of spectinomycin. The sulfonylurea-resistant transformants were selected on an SD medium containing 4 mg/ml of chlorimuron-ethyl (Chem Service West Chester, PA, USA.), 100 mg/ml of cefotaxim, and 100 mg/ml of spectinomycin.

Genomic DNA blot analysis
The total DNA from the mycelia of C. orbiculare was isolated, and a DNA blot analysis was performed using a previously described method [22]. DNA digestion, gel electrophoresis, labeling of probes, and hybridization were performed according to the manufacturer's instructions following standard methods [20]. DNA probes were labeled with DIG-dUTP using a BcaBESTTM DIG labeling kit (Takara Bio, Ohtsu, Japan). Hybridized DNA was detected with anti-Digoxygenin antibody Fab fragments conjugated to alkaline phosphatase (Roche Diagnostics, Tokyo, Japan). Light emission from the enzymatic dephosphorylation of the CDP-Star Detection Reagent (GE Healthcare, Tokyo, Japan) was detected using the Fujifilm LAS-1000 Plus Gel Documentation System (Fujifilm, Tokyo).

Construction of the CoIRA1 gene replacement vector and CoIRA1 complementation vector
To replace the CoIRA1 gene with the hygromycin-resistance gene, we constructed a CoIRA1 gene-replacement vector pBIG4MRBrev-coira1. We first amplified the upstream region of the CoIRA1 gene, the hygromycin-resistance gene, and the downstream region of the CoIRA1 gene by PCR using the primer pairs CoIRA1F1A-CoIRA1R2D, CoIRA1hphF1C-CoIR-A1hphR1D, and CoIRA1F2C-CoIRA1R1B, respectively. The primers used are listed in Table S1. Next, the pBIG4MRBrev-coira1 vector was constructed using the GeneART seamless cloning and assembly kit (Life Technologies, Carlsbad, California USA) with the amplified product and the A. tumefaciens binary vector pBIG4MRBrev. This vector, which contains a bialaphos resistance gene, was used as the gene replacement plasmid.
To perform a complementation assay of the coira1 mutant, we constructed the CoIRA1 complementation vector pBIG4MR-Brev-CoIRA1. We first amplified the upstream region of the CoIRA1 gene, the middle region of the CoIRA1 gene, and the downstream region of the CoIRA1 gene by PCR using the primer pairs CoIRA1F3A-CoIRA1R4D, CoIRA1F5D-CoIRA1R5C, and CoIRA1F4C-CoIRA1R4D, respectively. Next, the pBIG4MRBrev-CoIRA1 vector was constructed with the amplified product, and the A. tumefaciens binary vector pBIG4MRBrev was constructed using the GeneART seamless cloning and assembly kit. This vector, which contains a bialaphos resistance gene, was used as the gene replacement plasmid.

Construction of dominant active and negative CoRAS1 vectors
To construct a dominant active form of the CoRAS1 vector, we amplified a 1.4-kb fragment containing the upstream region of the CoRAS1 gene, a 2.0-kb fragment containing the downstream region of the CoRAS1 gene, and the linearized pBIG4MRBrev vector by PCR using the primer pairs CoRAS1F1A-CoR-AS1 G17V R1B, CoRAS1 G17V F1A-CoRAS1R1B, and CoRAS1p-BIF1A-R1B. Next, the pBIG4MRBrev-CoRAS1 G17V vector was constructed with the amplified product using the GeneART seamless cloning and assembly kit. To construct a dominant negative form of the CoRAS1 vector, we amplified a 1.4-kb fragment containing the upstream region of the CoRAS1 gene, a 2.0-kb fragment containing the downstream region of the CoRAS1 gene, and the linearized pBIG4MRBrev vector by PCR using the primer pairs RAS1 S22N F1A-RAS1R1B, RAS1F1A-RAS1 S22N R1B, and CoRAS1pBIF1A-R1B, respectively. Next, the pBIG4MRBrev-CoRAS1 S22N vector was constructed with the amplified product using the GeneART seamless cloning and assembly kit.

Construction of the CoRAS2 gene replacement, complementation, dominant active or negative vectors
To replace the CoRAS2 gene for a hygromycin-resistance gene, we constructed the CoRAS2 gene-replacement vector pBIG4MR-Brev-coras2. We first amplified the upstream region of the CoRAS2 gene, the hygromycin-resistance gene, the downstream region of the CoRAS2 gene, and the linearized pBIG4MRBrev vector by PCR using the primer pairs CoRAS2F1A-CoR-AS2R2D, CoRAS2hphF1C-CoRAS2hphR1D, CoRAS2F2C-CoRAS2R1B, and CoRAS2pBIF1-R1, respectively. Next, the pBIG4MRBrev-coras2 vector was constructed with the amplified product using the GeneART seamless cloning and assembly kit.
To perform a complementation assay for the coras2 mutant, we constructed the CoRAS2 complementation vector pBIG4MR-Brev-CoRAS2. We amplified a 3.3-kb fragment containing the CoRAS2 gene and the linearized pBIG4MRBrev vector by PCR using the primer pairs CoRAS2F1A-R1B and CoRAS2pBIF1A-R1B, respectively. Next, the pBIG4MRBrev-CoIRA1 vector was constructed with the amplified product using the GeneART seamless cloning and assembly kit.
To construct a dominant active form of the CoRAS2 vector, we amplified a 1.4-kb fragment containing the upstream region of the CoRAS2 gene, a 1.9-kb fragment containing the downstream region of the CoRAS2 gene, and the linearized pBIG4MRBrev vector by PCR using the primer pairs CoRAS2F1A-CoR-AS2 Q65L R1B, CoRAS2 Q65L F1A-CoRAS2R1B, and CoRAS2p-BIF1A-R1B, respectively. Next, the pBIG4MRBrev-CoR-AS2 Q65L vector was constructed with the amplified product using the GeneART seamless cloning and assembly kit. To construct a dominant negative form of the CoRas2 vector, we amplified a 1.2-kb fragment containing the upstream region of the CoRAS2 gene, a 2.2-kb fragment containing the downstream region of the CoRAS2 gene, and the linearized pBIG4MRBrev vector by PCR using the primer pairs CoRAS2F1A-CoRAS2-G19A R1B, CoRAS2 G19A F1A-CoRAS2R1B, and CoRAS2p-BIF1A-R1B, respectively. Next, the pBIG4MRBrev-CoR-AS2 G19A vector was constructed with the amplified product using the GeneART seamless cloning and assembly kit.

Construction of the CoIRA1-VENUS vector
To construct the pBITEF-VENUS vector, we amplified a 0.9kb fragment containing the TEF promoter and the upstream region of the GFP gene, a 0.6-kb fragment containing the downstream region of the GFP gene, and the linearized pBIG4MRBrev vector by PCR using the primer pairs TEFF1-VENUSR1, VENUSF1-glyGFPR1, and pBIG4VENUSF1-R1, respectively. Next, the pBIG4MRBrev-TEF-VENUS vector was constructed with the amplified product using the GeneART seamless cloning and assembly kit. To construct the CoIRA1-VENUS vector, we amplified the VENUS gene, the hygromycinresistance gene, and the linearized pBIG4MRBrev-CoIRA1 vector by PCR using the primer pairs glyGFPF1-GFPR1, VENUShphF1-VENUSR1, and pBICoIRA1-VENUSF1-CoIR-ApBIcomR1-GFP, respectively. Next, the pBIG4MRBrev-CoIRA1-VENUS-hph vector was constructed with the amplified product using the GeneART seamless cloning and assembly kit.
Construction of the RFP fused CoRAS2 and the RFP fused CoRAS2 Q65L vectors To construct the RFP fused CoRAS2 vector, we amplified the RFP gene and linearized pBIG4MRBrev-CoRAS2 vector by PCR using the primer pairs RFPF1-glyRFPR1 and CoRAS2p-BIF1-R1, respectively. Next, the pBIG4MRBrev-RFP-CoRAS2 vector was constructed with the amplified product using the GeneART seamless cloning and assembly kit. To construct the RFP-fused CoRAS2 Q65L vector, we amplified the RFP gene and linearized pBIG4MRBrev-CoRAS2 Q65L vector using the primer pairs RFPF1-glyRFPR1 and CoRAS2pBIF1-R1, respectively. Next, the pBIG4MRBrev-RFP-CoRAS2 Q65L vector was constructed with the amplified product using the GeneART seamless cloning and assembly kit.
Construction of the VENUS-N (1-158 aa) fused CoIRA1 and VENUS-C (159-238 aa) fused CoRAS2 vectors for BiFC assays To construct the VENUS-N fused CoIRA1 vector, we amplified the VENUS-N fragment, the hygromycin-resistance gene, and the linearized pBIG4MRBrev-CoIRA1 vector by PCR using the primer pairs glyGFPF1-aGFPR1, VENUShphF1-VENUSR1, and pBICoIRA1-VENUSF1-CoIRApBIcomR1-GFP respectively. Next, the pBIG4MRBrev-CoIRA1-nVENUS-hph vector was constructed with the amplified product using the GeneART seamless cloning and assembly kit. To construct the VENUS-C fused CoRAS2 vector, we amplified the VENUS-C fragment and linearized pBIG4MRBrev-CoRAS2 vector by PCR using the primer pair BGFPF2-glyGFPR1 and CoRAS2PBIcompF3-c-CoRAS2PBIcompR3, respectively. Next, the pBIG4MRBrev-cVENUS-CoRAS2 vector was constructed with the amplified product using the GeneART seamless cloning and assembly kit.

Intracellular cAMP measurements
Mycelia were collected from three-day old PSY liquid cultures and frozen in liquid nitrogen. All samples were lyophilized for 24 h and weighed. For every 10 mg of mycelia, 200 ml of ice-cold 6% trichloroacetic acid was added. The precipitate was removed by centrifugation at 20006g for 15 min at 4uC, the supernatant was transferred to a new tube, and the TCA was extracted four times with five volumes of water-saturated ether. The concentration of cAMP was determined using the cAMP Biotrak Enzyme immunoassay system (GE Health Life Science, UK) according to the manufacturer's instructions.

Western blot
The total protein was isolated from vegetative hyphae using a previously described method [23]. The protein separated on SDS-PAGE gels was transferred onto a PVDF membrane using an Xcell SureLock Mini-Cell. The phosphorylation activation of Maf1 and Cmk1 MAPK kinase was detected using a PhosphoPlus p44/42 MAP kinase antibody kit (Cell Signaling Technology). Alkaline Phosphatase-conjugated secondary antibody and light emission from the enzymatic dephosphorylation of the CDP-Star Detection Reagent (GE Healthcare, Tokyo, Japan) was detected using the Fujifilm LAS-1000 Plus Gel Documentation System (Fujifilm, Tokyo). Anti-actin antibodies (Wako, Japan) were used at a 1:1000 dilution for Western blot analysis.

Pathogenicity tests
An inoculation assay on cucumber cotyledons (Cucumis sativus L. ''Suyo'') was performed as described by Tsuji et al. (1997) [24]. The conidia of C. orbiculare were obtained from seven-day old cultures, and drops of 10-ml conidial suspension (1610 5 conidia per ml) were added on the surface of cucumber cotyledons at different locations. To assess invasive growth ability, 10-ml drops of spore suspension were added on wounded sites that were created by scratching the leaf surface with a sterile toothpick. After inoculation, the cotyledons were incubated at 24uC for seven days.

Light microscopy
For appressorium formation and penetration assays in vitro, conidia were harvested from seven-day old PDA cultures and suspended in distilled water. The conidial suspension, adjusted to 1610 5 conidia per ml, was placed on a multiwell glass slide (eightwell multi-test slide; ICN Biomedicals, Aurora, OH, U.S.A.) and incubated in humid boxes at 24uC. Germlings were observed by a Nikon ECLIPSE E600 microscope with differential interference contrast optics (Nikon, Tokyo, Japan). To observe the formation of infectious hyphae in cucumber leaves, the conidial suspension was inoculated on the abaxial surface of cucumber cotyledons and incubated at 24uC for three days. Then, the inoculation site was cut off and stained with 0.1% lactophenol-aniline blue. VENUS and RFP fluorescence was observed by a Carl ZEISS Axio Imager M2 microscope (Zeiss, Gottingen, Germany) with 470 and 595 nm of excitation wavelength, respectively.

Results
Identification of an IRA1/2 homolog, CoIRA1, in C. orbiculare CoIRA1 was first identified as a mutant gene in the Agrobacterium tumefaciens-mediated transformation (AtMT) T-DNA mutant AA4510 of C. orbiculare, which shows an attenuated pathogenicity on cucumber leaves. DNA flanking regions, adjacent to the inserted plasmid, were isolated from the mutant AA4510 by thermal asymmetrical interlaced-polymerase chain reaction (TAIL-PCR) and the amplified products were subsequently sequenced. The TAIL-PCR result showed that the T-DNA fragment was inserted probable 10-bp downstream from the translational origin of the gene ENH81573, based on C. orbiculare genomic information [25]. This gene putatively encodes a 2255amino acid protein with a predicted RAS GTPase-activating protein (RASGAP) domain ( Figure S1A). We named this gene CoIRA1.
A blast search of the CoIRA1 fungal genome homologs in nonredundant protein database of the National Center for Biotechnology Information indicated that the derived amino acid sequence from this gene is significantly similar to that of the IRA1/2 homolog in S. cerevisiae and filamentous fungi, including C. graminicola, Neurospora crassa, and M. oryzae ( Figure S1B).

CoIRA1 is required for infection-related morphogenesis
To analyze the function of CoIRA1, the disruption vector pBIG4MRBrevcoira1 was designed to replace the wild-type CoIRA1 gene with the hygromycin phosphotransferase (hph) gene, by double crossover homologous recombination. Successful replacement of the targeted gene in the transformants was confirmed by genomic DNA blot analysis ( Figure S2). A single 3.4-kb band was detected in the wild type, whereas a single 5.6-kb band was detected in coira1 mutants, as expected from such a targeted gene replacement event ( Figure S2). In ectopic transformants, the 3.4-kb band and several additional bands were detected, indicating ectopic insertion events.
The hyphal growth of the coira1 mutant was similar to that of the wild-type, but the coira1 mutant formed a rather dark colony compared with the wild-type ( Figure S3). The coira1 mutant also showed reduced conidiation compared with the wild-type. To investigate infection-related morphogenesis in the coira1 mutant, we observed conidial germination and appressorium development on a glass slide, and infection hyphae development on cellulose membranes. In the wild-type, coira1 mutant, ectopic transformants, and CoIRA1 reintroduced transformants, approximately 80% of the conidia germinated and formed darkly melanized appressoria after 24 h, but the frequency of abnormal appressorium formation in the coira1 mutant was higher compared with the wild-type, as observed on glass slides ( Figures 1A and B). The appressorium turgor pressure is required to penetrate the plant surface mechanically during infection [26], therefore, we evaluated the appressorium turgor in the coira1 mutant using a cytorrhysis assay [27]. The proportion of the collapsed appressoria at each glycerol concentration in the coira1 mutant was similar to that of the wild-type ( Figure S4), indicating that CoIra1 is not involved in the generation of the appressorium turgor pressure. The wild-type, ectopic transformants, and CoIRA1 reintroduced transformants formed normal infection-hyphae that penetrated the cellulose membranes with high frequency. The coira1 mutant effectively formed infection hyphae as did the wild-type. Interestingly, the coira1 mutant hyphae had a bulbous shape, which was quite different from those of the wild-type (Figures 2A and B). These results indicated that CoIRA1 is involved in the progression of normal morphogenesis during infection.

CoIRA1 is involved in the pathogenicity of the host plant
To investigate whether the pathogenicity of coira1 mutants is attenuated, conidial suspensions were inoculated onto cucumber cotyledons. The pathogenicity of the coira1 mutant was found to be attenuated during the infection of the cucumber plants compared with that of the wild-type ( Figure 3A). Moreover, when conidial suspensions were applied directly on wounded sites, the coira1 mutant caused the formation of smaller lesions compared with the wild-type ( Figure S5). Microscopic observations of the infection process of the coira1 mutant showed that it formed normal darkly melanized appressoria and infection hyphae, however, its frequency of infection hyphae formation was lower than that of the wild-type ( Figures 3B and C). These data indicated that during the infection of the cucumber, the coira1 mutants had a defective infection-related morphogenesis. In C. orbiculare, we showed that the ssd1 mutant that are defective in proper fungal cell walls constitution failed to penetrate into host epidermal cells due to the increased defence reaction of the host plant with rapidly induced callose deposition at at the attempted penetration site from appressoria [28]. Therefore, to investigate whether the observed low frequency of the infection hyphae formation in the coira1 mutant was induced by a host defense mechanism, we monitored the callose deposition under the appressorium of the coira1 mutant on the cucumber cotyledons and found that the extent and frequency of the deposition was similar to that of the wild-type ( Figure S6A). Moreover, the coira1 mutant caused the formation of smaller lesions than the wild-type on heat-shocked cucumber cotyledons ( Figure S6B), which impaired the host defense responses, indicating that the induction of plant defense responses is not involved in the attenuated pathogenicity observed for coira1 mutants.

CoIRA1 is involved in cAMP signaling
In S. cerevisiae, Ira1/2 inactivates Ras1/2, which in turn activates adenylate cyclase, leading to the synthesis of cAMP from ATP [29]. We therefore examined whether the appressorium morphogenesis of the coira1 mutants is affected by exogenous cAMP signals. In the presence of 10 mM cAMP, the frequency of abnormal appressorium formation observed in the coira1 mutant was higher compared with that of the wild-type ( Figures 4A and  B). This data suggested that CoIra1 is involved in cAMP signaling pathway during the process of the appressorium formation. We identified two Ras genes and named CoRAS1 (ENH84705) and CoRAS2 (ENH80898). The comparative amino acid sequence analysis of Ras proteins showed that CoRAS1 was 90% identical to M. oryzae MoRAS1 and CoRAS2 was 80% identical to M. oryzae MoRAS2. To better understand relationship between CoIra1 and Ras proteins, we set up a hypothesis that the wild-type expressing a dominant active form CoRAS1 and CoRAS2 increases abnormal appressorium formation compared with the wild-type in the presence of excess cAMP, while the coira1 mutant expressing a dominant active form CoRAS1 and CoRAS2 decrease abnormal appressorium compared with the coira1 mutant in the presence of excess cAMP. Therefore, we generated these transformants. The Ras protein shows a high degree of amino acid conservation. On the basis of the human Ras gene mutation information, we constructed the dominant active forms CoRas1 (RAS1 G17V ) and CoRas2 (RAS2 Q65L ) alleles by replacing Gln-65 with Leu and Gly-17 with Val, respectively. Moreover, based on C. trifolii and Candida albicans Ras gene mutation information, we constructed the dominant negative forms CoRas1 (RAS1 S22N ) and CoRas2 (RAS2 G19A ) alleles by replacing Ser-22 with Asn and Gly-19 with Ala, respectively [6], [30]- [32]. Next, we generated the dominant active and negative forms of CoRAS1 DARS1 and DARS2 showed normal appressorium formation on the glass slides in distilled water ( Figure 4) and normal penetration hyphae similar to the wild-type on the cellulose membranes ( Figure S7), however, DARS2 caused the formation of smaller lesions compared with the wild-type and DARS1 on the cucumber cotyledons ( Figure S8). In the presence of 10 mM cAMP, DARS2 formed abnormal appressoria, similar to those observed for the coira1 mutants. In contrast, iDNRS2 suppressed the abnormal appressorium formation compared with that of the coira1 mutants ( Figure 4). These data indicated that CoIra1 could control the cAMP signaling pathway through CoRas2 during the process of appressorium formation. We also checked the intracellular cAMP accumulation in the coira1 mutant and the dominant active and negative CoRAS1 and CoRAS2 introduced transformants in vegetative hyphae whether the cAMP signaling Percentages of conidial germination, appressorium formation, and abnormal appressorium formation in the C. orbiculare WT and coira1 mutants on multiwell glass slides. Approximately 100 conidia of each strain were observed per well on multiwell slide glass. Three replicates were examined. Three independent experiments were conducted, and standard errors are shown. Black bar, conidial germination; gray bar, appressorium formation; white bar, abnormal appressorium formation. doi:10.1371/journal.pone.0109045.g001 pathway was regulated by those genes. The coira1 mutants accumulated higher levels of cAMP compared with the wild-type ( Figure 5). Moreover, intracellular cAMP levels in the DARS1 and DARS2 mutants were similar to those of the coira1 mutant. These data indicated that intracellular cAMP levels in the vegetative hyphae were controlled by CoIra1, CoRas1, and CoRas2.

CoRAS2 is involved in conidial germination and pathogenicity
CoIra1 regulates intracellular cAMP levels through Ras proteins. Therefore, to analyze the functional roles of CoRAS1 and CoRAS2, we aimed to generate coras1 and coras2 mutants by AtMT. Whereas we successfully developed coras2 mutants, the generation of coras1 mutants was challenging, indicating that CoRAS1 could be an essential gene in C. orbiculare. To investigate whether CoRAS2 is involved in infection-related morphogenesis, we observed conidial germination and appressorium formation on glass slides. Strikingly, conidial germination was not observed on glass slides for coras2 mutants, indicating that CoRAS2 is involved in conidial germination ( Figure S9). To investigate whether the pathogenicity of the coras2 mutant was attenuated during the infection of the cucumber cotyledons, conidial suspensions were inoculated onto the cucumber cotyledons. The coras2 mutant was defective in pathogenesis to the cucumber cotyledons, forming small speck lesion ( Figure S10). Microscopic observation revealed that most conidia in the coras2 mutant were defective in conidial germination as in vitro condition, indicating that the attenuated pathogenicity of this mutant was due to a defect in conidial germination and that small speck lesion was apparently caused by small numbers of appressorium forming conidia. CoRas2 is an upstream regulator of the MAPK signaling pathway From previous reports, the MAPK CoMekk1-Cmk1 signaling pathway has been shown to play pivotal roles in conidial germination and appressorium formation in several Colletotrichum speciese [2], [33]. In C. orbiculare, the Ste11 homolog CoMEKK1 encodes a Ras-association domain. Thus, to analyze whether CoRas2 is an upstream regulator of CoMekk1, we generated a comekk1 transformant DMK1/DARS2 expressing a dominant active form of the CoRAS2 allele. In addition, we generated the coras2 mutant DRS2/DAMK1 expressing a dominant active form of the CoMEKK1 allele. A normal conidial germination and appressorium formation was observed on the glass slide for the DRS2/DAMK1 mutant ( Figure 6). On the other hand, the defective conidial germination of the coras2 mutant remained in the DMK1/DARS2 mutant, showing a similar frequency of conidial germination as that in the coras2 mutant. These data suggested that CoRAS2 is an upstream regulator of the MAPK CoMekk1-Cmk1 signaling pathway.
The MAP kinase Cmk1 is activated through threonine/tyrosine phosphorylation catalyzed by MAPKK, which at the same time is activated through serine phosphorylation catalyzed by CoMekk1.
To determine whether the phosphorylation of Cmk1 in the coira1 and coras2 mutants was affected in the vegetative hyphae, we investigated the Thr-Gln-Tyr residue phosphorylation of Cmk1 using an anti-TpEY specific antibody. The phosphorylation levels of Cmk1 in the coira1 mutant and DARS2 were higher compared with those in the wild-type, whereas phosphorylation levels of Cmk1 in iDNRS2 were lower compared with those in the coira1 mutant, which was similar to those of the wild-type ( Figure 7). These data indicated that CoIra1 negatively regulates the phosphorylation of Cmk1 through CoRas2. Interestingly, the phosphorylation level of Cmk1 in DARS1 was higher compared with that in the wild-type, indicating that CoRas1 positively regulates the phosphorylation of MAPK Cmk1.

CoRas2 localization in the coira1 mutant was similar to the active CoRas2 localization pattern in vegetative hyphae
To analyze the cellular CoRas2 localization, a red fluorescent protein (RFP) gene was fused to the C-terminus of the CoRAS2 gene. In S. cerevisiae and C. albicans, the Ras protein conserves a CAAX motif (C, cysteine; A, aliphatic amino-acids; X; methionine or serine), which is important for post-translational modification, including farnesylation and palmitoylation, to ensure specific membrane localization and function [29], [34]. We assumed that the RFP gene fused C-terminus of CoRAS2 gene may block the localization and function of CoRas2. Thus, the RFP gene fused to the N-terminus of CoRAS2 gene was constructed, and the RFP-CoRAS2 gene controlled under the CoRAS2 native promoter was transformed into the coras2 mutant (DRS2/RFP-RS2) and the wild-type (WT/RFP-RS2). Both DRS2/RFP-RS2 and WT/ RFP-RS2 retained normal infection-related morphogenesis and pathogenicity. The signals of RFP-CoRas2 were detected within vesicle-like structures in conidia (Figure 8). During the process of premature appressoria formation, signals of the RFP-CoRas2 were localized mainly in subcellular compartments resembling vacuole-like structures in conidia, but these signals were not detected in matured appressoria. In vegetative hyphae, RFP-CoRas2 proteins showed no specific localization and their signals were uniformly distributed throughout the entire cell ( Figure 8). To analyze the cellular localization of the active form of CoRas2, the RFP gene was fused to the N-terminus of the dominant active form of the CoRAS2 gene. The RFP-DACoRAS2 gene controlled under the native CoRAS2 promoter was transformed into the wild-type (WT/RFP-DARS2). During the process of appressoria formation, the signal pattern of the RFP-DACoRas2 was similar to that of the RFP-CoRas2 ( Figure S11), however, the signals of the RFP-DACoRas2 in vegetative hyphae were detected predominantly in the plasma membrane, unlike those of native CoRas2. To analyze whether CoIra1 negatively regulates CoRas2, we generated an iRFP-RS2 transformant expressing RFP-CoRAS2 in the coira1 mutant and observed the cellular localization of CoRas2 in this mutant. During appressoria formation, the signals of RFP-CoRas2 in the coira1 mutant were detected as vesicle like structures as well as RFP-CoRas2 and RFP-DACoRas2. In vegetative hyphae, the signals of RFP-CoRas2 in the coira1 mutant were detected at the plasma membrane as well as RFP-DACoRas2 localization ( Figure 9). Engagement of CoIra1 in the regulation of CoRas2 during appressorium formation was not directly suggested by these data from the viewpoint of the cellular localization of Ras protein, whereas it was suggested that CoIra1 functions as a negative regulator for CoRas2 in vegetative hyphae. Furthermore, to analyze whether CoIra1 regulates the CoRas2 in the plasma membrane in vegetative hyphae, we generated the Vc-RS2/IRA-Vn transformant, expressing the Cterminal domain (159-238) VENUS that was fused with CoRAS2 and the N-terminal domain (1-158) of VENUS fused with CoIRA1 in the wild-type for bimolecular fluorescence complementation (BiFC) assays [35]. BiFC fluorescence was detected at the plasma membrane in vegetative hyphae. This result indicated that CoIra1 regulates CoRas2 at the plasma membrane in vegetative hyphae ( Figure 10).

CoIra1 colocalizes with CoRas2 in pregerminated conidia and appressoria that initiate the differentiation of infection hyphae
To elucidate the cellular localization of CoIra1, the VENUS fluorescence gene was fused to the C-terminal of the CoIRA1 gene [36]. VENUS fused with CoIRA1, expressing under a native CoIRA1 promoter was transformed into the wild-type (IRA1-VENUS). CoIra1-VENUS was detected in a vesicle-like structure of conidia ( Figure 11). In the germinated conidia, CoIra1-VENUS was detected at the tip of the germ tubes and distributed uniformly in the conidial cells. In developing appressoria, CoIra1-VENUS did not show any specific localization, and uniform signals were distributed throughout the appressoria except the presumable lipid bodies, similar to those on cucumber leaves. In developing infection hyphae, CoIra1-VENUS showed a uniform distribution either in the initial or late infection hyphae in cucumber cotyledons ( Figure S12). Then, to elucidate the intracellular colocalization of CoIra1 and CoRas2, we generated the RFP-RS2/IRA1-VENUS and RFP-DARS2/IRA1-VENUS transformants by introducing the RFP fused with CoRAS2 and VENUS fused with CoIRA1 genes into the wild-type, and the RFP fused with a dominant active form CoRAS2 and VENUS fused with CoIRA1 genes into the wild-type, respectively. RFP-CoRas2 and CoIra1-VENUS colocalized at vesicle-like structures in conidia and germinated conidia (Figure 11), however, the RFP-CoRas2 and CoIra1-VENUS did not show specific colocalization in developing appressoria, similar to RFP-DARS2/IRA1-VE-NUS ( Figure S11). Surprisingly, after 48 h of conidial incubation on glass slides, RFP-CoRas2 and CoIra1-VENUS were colocalized in a vesicle-like structures in appressoria ( Figure 12). Conclusively, CoIra1 colocalizes with CoRas2 in pregerminated conidia and in appressoria that initiate differentiation of infection hyphae. CoIra1 regulates the assembly of actin at the appressorium pore In M. oryzae, it has been reported that proper assembly of the Factin network in the appressorium pore is required for host infection and MAPK Mst12, the C. orbiculare Cst1 homolog, is known to be involved in the proper assembly of the F-actin network in the appressorium pore [37], [38]. Moreover, it has been reported that cAMP signaling is involved in the remodeling of the actin structure in S. cerevisiae [39]. Our data showed that CoIra1 regulated the cAMP and MAPK signaling pathways ( Figure 5 and 7). To elucidate whether CoIra1 colocalize with Factin, we generated the LA/IRA1-VENUS transformant by introducing Lifeact-RFP into it. In C. orbiculare, Lifeact, which binds to F-actin, forms vesicle-like structures in pregerminated conidia and then Lifeact-RFP mainly localizes in the appressorium pore during appressoria development. We examined the localization of CoIra1-VENUS and Lifeact-RFP in pregerminated conidia and appressorium pores. CoIra1-VENUS was colocalized with Lifeact-RFP in pregerminated conidia, however, CoIra1-VENUS was not clearly colocalized with Lifeact-RFP in the appressorium pore ( Figure 13). To elucidate whether CoIra1 is involved in the assembly of the F-actin in the appressorium pore, we generated the iLA transformant by introducing Lifeact-RFP into the coira1 mutant. In C. orbiculare, the wild-type showed a specific assembly of F-actin in the appressorium pore on the cucumber cotyledons, however, the frequency of this assembly in the coira1 mutant was lower than that in the wild-type ( Figure 14). These data indicated that CoIra1 is involved in the assembly of the F-actin in the appressorium pore.

Discussion
CoIra1 is involved in the crosstalk between cAMP and MAPK signaling pathways through CoRas2 in C. orbiculare In S. cerevisiae, the activation of the cAMP-dependent pathway causes cells to undergo unipolar growth, which is coupled with an elongated growth that is controlled by the filamentous MAPK pathway [40]. In C. orbiculare, cAMP-PKA signal transduction is involved in conidial germination, and the MAPK cascade is involved in appressorium development [1]. In S. cerevisiae, an increase in the intracellular levels of RAS-GTP against RAS-GDP is observed in ira1 and ira2 mutants, activating various target effectors [41]. Our data indicated that intracellular cAMP levels and the phosphorylation of MAPK Cmk1 in the coira1 mutant, DARS1, and DARS2 was higher than that in the wildtype. Interestingly, the phosphorylation of Cmk1 in the coras2 mutant was higher compared with the wild-type, although CoRas2 is an upstream regulator of CoMekk1-Cmk1. In M. oryzae, the intracellular cAMP levels of the dominant active MEK Mst7 S212D T216E strain are lower than those of the wild-type [42]. Recently, the functional analysis of the adenylate cyclase-associated protein encoding CAP1, the ortholog of yeast Srv2, Cap1 may play a role in the feedback inhibition of MoRas2 signaling when Pmk1 MAP kinase is activated [12]. In yeast and phytopathogenic fungi, cAMP signaling is intimately associated with a MAPK homologous with PMK1 for regulating various developmental and plantinfection processes [43]- [45]. We assumed that the constitutively active forms CoRas1 and CoRas2 induce excessive activation of the cAMP-PKA and MAPK CoMekk1-Cmk1 signaling pathways, resulting in the interruption of the cAMP-PKA and MAPK signaling pathways ( Figure 15). Interestingly, the level of cAMP and the phosphorylation of Cmk1 in DARS1 was higher than that in the DARS2; however, the pathogenicity of DARS1 was not attenuated during the infection of cucumber leaves ( Figure S11). Therefore, CoRas1 may be involved in the cAMP-PKA and MAPK CoMekk1-Cmk1 signaling pathways in vegetative hyphae. To understand Ras-mediated signaling pathways clearly, further functional analysis of the relationship between CoRas1 and CoRas2 during infection is required.

CoIra1 interacts with CoRas2 and negatively regulates CoRas2
Ras serves as a molecular switch, coupling activated membrane receptors to the downstream signaling molecule, by alternating between the GTP-bound (active) and the GDP bound (inactive) conformations. In S. cerevisiae, Ira1/2 negatively regulates Ras1/ 2; therefore, the ira1/2 mutants accumulate high amounts of Ras-GTP compared with the wild-type [7]. In the present study, the localization pattern of CoRas2 in the coira1 mutant was similar to that in the active form CoRas2 in vegetative hyphae. Moreover, BiFC assays supported that CoIra1 regulates CoRas2 at the plasma membrane of vegetative hyphae. It is likely that CoIra1 negatively regulates CoRas2 in vegetative hyphae. In Aspergillus fumigatus, RasA localizes in plasma membranes in hyphae where it associates with and stimulates targeted effectors, resulting in the   regulation of polarized morphogenesis [46]. We considered that CoIra1 negatively regulates CoRas2, which is required for the proper morphogenesis of vegetative hyphae. However, this morphogenesis in the coira1 mutant and the dominant active form CoRas2 expressed strain was similar to that in the wild-type. In C. orbiculare, the conidia in the MAPK cmk1 disruption mutant fails to germinate, however, by adding 0.1-% yeast extract, conidial germination efficiency can be restored [33]. Thus, there is a strong possibility that a nutrient-specific signaling pathway regulates conidial germination in C. orbiculare. Therefore, we speculated the presence of a bypass pathway that regulates vegetative hyphae morphogenesis independent of the CoIra1-CoRas2 signaling pathway in the presence of nutrients.

CoIRA1 regulates proper infection-related morphogenesis
Our data indicated that the pathogenicity of the coira1 mutant was attenuated during the infection of the cucumber cotyledons.  Plants have evolved with a variety of defense mechanisms against attacking phytopathogenic fungi. These include the deposition of cell wall reinforcement components (papillae), hypersensitive cell death, and the synthesis of an antimicrobial secondary metabolite. However, callose deposition was not observed following attempted penetration from the appressoria of the coira1 mutant. Further-more, the pathogenesis of the coira1 mutant was not restored during the infection of heat-shocked cucumber cotyledons. Therefore, there is a small possibility that the observed reduction of pathogenicity in the coira1 mutant was caused by the plant defense response. In C. orbiculare, the generation of the appressorium turgor pressure is required for mechanical penetration of the plant surface [47], however, appressorial turgor of the coira1 mutant was not affected, as shown by the cytorrhysis assay [27]. The pathogenicity of the coira1 mutant was attenuated compared with that of the wild-type during infection of wounded leaves, thereby resulting in the reduction in the development of hyphae causing infection in the host plants. In C. orbiculare, the catalytic subunit of PKA Cpk1 and adenylate cyclase Cac1 is involved in infectious growth in plants [3]. We postulate that excessive cAMP levels in the coira1 mutant would affect the process of the infection hyphae development in the host plant. The vegetative hyphae of the coira1 mutant and the wild-type showed normal morphology; however, the coira1 mutant formed bulbshaped hyphae on the cellulose membranes. In response to nitrogen starvation, diploid S. cerevisiae cells undergo pseudohyphal growth, which is enhanced by the expression of the dominant active allele RAS2 [48]. Further investigation has revealed that the pseudohyphal growth is controlled by both Kss1 MAPK and cAMP-PKA signaling [49]. Similarly, C. albicans strains carrying the activating RAS1 V13 allele formed more abundant hyphae in a shorter time than that in the wild-type strain. Thus, activated Ras is a key factor for regulating hyphal morphogenesis in fungi. In this study, the molecular mechanism of bulb-shape penetration hyphae caused by the coira1 mutation remains unclear. Therefore, further functional analysis of CoIra1 will be helpful for understanding the regulation of penetration hyphae morphogenesis for the requirement of proper infection of the host plant.

CoIra1 and CoRas2 is involved in the appressoriamediated differentiation of infection hyphae development
CoRas2 and CoIra1 did not show specific colocalization in developing appressoria. Consistently, CoIra1 and the active form CoRas2 did not colocalize in developing appressoria. Therefore, during appressorium formation, CoRas2 may be regulated by other factors different from CoIra1. In C. orbiculare, 48 to 72 h post-inoculation is a crucial phase for appressorium-mediated penetration. Our data indicated that CoIra1 and CoRas2 colocalized in a vesicle-like structure in the appressoria 48 h post-inoculation. Moreover, in M. oryzae, the correct organization of F-actin in the appressorium pore is important for penetration peg development [37]. Our data also indicated that CoIra1 is involved in the assembly of the F-actin in the appressorium pore. Therefore, we assume that CoIra1 could be involved in the F-actin organization in appressoria, which is required for penetration peg emergence. However, it is likely that CoIra1 does not directly regulate F-actin, because it does not colocalize with the F-actin in the appressoria pore. In M. oryzae, NADPH oxidases regulate the septin-mediated assembly of F-actin in the appressorium pore [50]. Moreover, in C. neoformans, the GTP-bound form of Ras1 interacts with Rho-GEF Cdc24 that mediates the activation of Cdc42 and Rac proteins, and Cdc42 is involved in cytokinesis and bud morphogenesis through the organization of septin proteins [51]. In Epichloë festucae, NoxA is activated by a small GTPase RacA [52]. Therefore, in C. orbiculare, Ras-Cdc42-septin proteins and the Ras-Rac-Nox proteins signaling pathway may regulate the assembly of F-actin in the appressorium pore. Our future work will identify interaction factors between CoIra1 and Figure 15. The hypothetical model for CoIra1 and CoRas1/2 mediated signaling transduction pathway in C. orbiculare. The cAMP-PKA signaling pathway is involved in conidial germination, appressorium penetration and invasive hyphae formation. The MAPK CoMekk1-Cmk1 signaling pathway is involved in conidial germination and appressorium formation. The CoRas2 localization pattern in the coira1 mutant was similar to that in DARS2. Moreover, BiFC assays supported that CoIra1 interacted with CoRas2 in the plasma membrane. Therefore, CoIra1 negatively regulates CoRas2. The coira1 mutant and DARS2 significantly induced abnormal appressorium formation and the frequency of abnormal appressorium formation in iDNRS2 was lower compared with that in the coira1 mutant. Moreover, intracellular cAMP levels in the coira1 mutant and DARS2 was high compared with those in the wild type. Therefore, CoIra1 regulates cAMP-PKA signaling pathway through CoRas2. The conidia of the coras2 mutant failed to germinate, whereas DRS2/DAMK1 restored the phenotype of the coras2 mutant. Therefore, CoRas2 is an upstream regulator of the MAPK CoMekk1-Cmk1 signaling pathway. Interestingly, the phosphorylation of MAPK Cmk1 in the coras2 mutant was higher compared with that in the wildtype, although CoRas2 is an upstream regulator of CoMekk1-Cmk1. Therefore, that CoIra1 may be a key factor for regulating the crosstalk between the cAMP-PKA and CoMekk1-Cmk1 MAPK signaling pathway through CoRas2. Intracellular cAMP levels in DARS1 were higher compared with those in the wild-type in vegetative hyphae. However, DARS1 showed lower sensitivity to exogenous cAMP in appressorium development compared with DARS2. Moreover, the intensity of pathogenesis in DARS1 was similar to that in the wild-type. Therefore, CoRas1 could be involved in the cAMP-PKA signaling pathway in vegetative hyphae but not during infection related morphogenesis. doi:10.1371/journal.pone.0109045.g015 CoRas2, elucidating those that are involved in the assembly of Factin in the appressorial pore.  Figure S4 Appressorium cytorrhysis assay for the coira1 mutant. Appressoria formed on the multiwell slides were exposed to glycerol solutions ranging from 0 M to 4 M and the percentage of collapsed spherical appressorium was counted. Approximately 200 conidia were observed per well. Three replicates were examined. Three independent experiments were conducted, and standard errors are shown. Solid line, the wildtype; dotted line, the coira1 mutant. (TIF) Figure S5 Pathogenicity assay of the coira1 mutant in wounded leaves. Conidial suspensions of each strain were inoculated on wounded sites on the cotyledon of the cucumber prepared by scratching the leaves with a sterile toothpick. The leaves were incubated at 24uC for seven days. Strains: WT, wildtype 104-T; DL1, the coira1 mutant; CL1, the CoIRA1complemented transformant of DL1; DPS1, the pks1 mutant. (TIF) Figure S6 Host defense response was not induced by the penetration of the coira1 mutant. (A) Quantification of papilla formation at sites of attempted penetration by appressorium in C. orbiculare. At three days, leaf epidermal strips inoculated with each strain was stained with Aniline blue to reveal the papilla and observed with epi-fluorescence microscopy. Strains: WT, wild-type; DL1, the coira1 mutant; CL1, the CoIRA1-complemented transformant of DL1; DSD1, the ssd1 mutant. At least 200 appressoria were counted for each strain and standard deviations were calculated from three replicated experiments. (B) Pathogenicity assay of the coira1 mutant on heat-shock cotyledons after the heat treatment at 50uC for 30 s, cucumber cotyledons were inoculated with conidial suspensions. Strains: WT, wild-type; DL1, coira1 mutant; CL1, the CoIRA1complemented transformant of DL1; DSD1, the ssd1 mutant. Controls were not exposed to heat shock. (TIF)