The functional relationship between fungal hydrophobins was studied by complementation analysis of an mpg1− gene disruption mutant in Magnaporthe grisea. MPG1 encodes a hydrophobin required for full pathogenicity of the fungus, efficient elaboration of its infection structures and conidial rodlet protein production. Seven heterologous hydrophobin genes were selected which play distinct roles in conidiogenesis, fruit body development, aerial hyphae formation and infection structure elaboration in diverse fungal species. Each hydrophobin was introduced into an mpg1− mutant by transformation. Only one hydrophobin gene, SC1 from Schizophyllum commune, was able partially to complement mpg1− mutant phenotypes when regulated by its own promoter. In contrast, six of the transformants expressing hydrophobin genes controlled by the MPG1 promoter (SC1 and SC4 from S.commune, rodA and dewA from Aspergillus nidulans, EAS from Neurospora crassa and ssgA from Metarhizium anisopliae) could partially complement each of the diverse functions of MPG1. Complementation was always associated with partial restoration of a rodlet protein layer, characteristic of the particular hydrophobin being expressed, and with hydrophobin surface assembly during infection structure formation. This provides the first genetic evidence that diverse hydrophobin‐encoding genes encode functionally related proteins and suggests that, although very diverse in amino acid sequence, the hydrophobins constitute a closely related group of morphogenetic proteins.
Fungal hydrophobins are small secreted, hydrophobic proteins which are fundamental to the developmental biology of fungi. More than 20 hydrophobin‐encoding genes have now been recognized and they may prove to be ubiquitous in filamentous fungi (Wessels, 1997). In most cases, hydrophobin genes have been identified as mRNAs abundantly transcribed during particular developmental processes such as sporulation, fruit body formation or fungal infection of plants and animals. Genetic studies have now implicated hydrophobins in all of these morphogenetic processes.
Hydrophobin genes have been mutated using targeted gene disruption and cause significant effects on development. In Aspergillus nidulans and Neurospora crassa, for example, null alleles at the rodA and EAS genes respectively cause spores to lose their surface hydrophobicity and clump together (Stringer et al., 1991; Bell‐Pederson et al., 1992; Lauter et al., 1992). This ‘easily wettable’ phenotype is due to the absence of a hydrophobin‐encoded protein layer on the surface of conidia (Templeton et al., 1995). In the mushroom Schizophyllum commune, four hydrophobin genes have been described—SC1, SC3, SC4 and SC6 (Wessels et al., 1991; Wessels, 1997)—and are differentially expressed during development of fruit bodies. SC3, a hydrophobin which is produced during hyphal growth, is required for production of upwardly projecting aerial hyphae as sc3− gene disruption mutants are unable to form hydrophobic aerial structures (van Wetter et al., 1996).
SC3 was the first hydrophobin to be purified, and this showed that hydrophobins can respond to interfaces between water and air, or between water and solid surfaces (Wösten et al., 1993, 1994b, 1995). Exposure to such an interface caused spontaneous aggregation of SC3 monomers resulting in an amphipathic protein polymer. The hydrophobic side of aggregated SC3 hydrophobin possessed a rodlet architecture identical to that found on aerial hyphae of S.commune (Wösten et al., 1993), suggesting that SC3 is secreted from fungal cells and forms the hydrophobic coating of hyphae after interfacial self‐assembly (Wösten et al., 1994a). Significantly, SC3 self‐assembly also occurred in response to hydrophobic surfaces, and sc3− mutants were deficient in their attachment to hydrophobic substrates (Wösten et al., 1994b, 1995).
In addition to development, hydrophobins play roles in fungal pathogenicity. Targeted gene replacement of MPG1 in the rice blast fungus Magnaporthe grisea produced mutants with reduced pathogenicity (Talbot et al., 1993). This was correlated with a reduced ability to form infection structures called appressoria. Magnaporthe grisea appressoria are dome‐shaped cells produced from the tips of fungal hyphae in response to the hydrophobic rice leaf surface (Hamer et al., 1988; Talbot, 1995). Their role is to generate turgor and mechanically breach the rice cuticle (Howard and Valent, 1996; de Jong et al., 1997). Purification of MPG1 indicated that hydrophobin self‐assembly occurs on the rice leaf surface and the resulting amphipathic hydrophobin layer then acts as a conformational cue for appressorium development (Talbot et al., 1996). However, MPG1 also encodes a spore wall rodlet protein (like EAS and rodA) and is required for efficient production of conidia. In M.grisea, therefore, it appears that a hydrophobin functions during both conidiogenesis and appressorium development.
The diversity of hydrophobins described so far suggests that these morphogenetic proteins serve a large number of specific functions in filamentous fungi (Talbot, 1997; Wessels, 1997). Hydrophobins are indeed very diverse proteins, as the 20 hydrophobins described to date share only 4.3% amino acid identity (Wessels, 1997). They do, however, share a common hydropathy profile, and all hydrophobins have eight cysteine residues characteristically spaced in their amino acid sequences.
Two possibilities exist regarding the functional relatedness of hydrophobins. The different functions of hydrophobins may, for example, simply reflect the particular developmental stage at which they are produced. In this case, any hydrophobin gene would be able to complement the mutant phenotype of another, as long as it is expressed at the appropriate developmental stage. Conversely, it may be that the considerable divergence in hydrophobin sequence reflects real functional diversity. If this is the case, then complementation of hydrophobin mutant phenotypes would perhaps be restricted to only hydrophobins fulfilling the same function in different fungi. We decided to address these questions by complementation of the mpg1− mutant phenotypes in the rice blast fungus M.grisea.
In this report, we describe the introduction of seven hydrophobin‐encoding genes into M.grisea mpg1− mutants. We show that six of the seven hydrophobins we tested could substitute, at least partially, for MPG1 during fungal pathogenicity when expressed as MPG1 promoter fusions. Complementation of MPG1 mutant phenotypes consistently was associated with formation of a rodlet protein layer, indicating that hydrophobin self‐assembly is fundamental to the function of MPG1. This provides the first evidence that hydrophobins are functionally related and suggests that although very diverse in amino acid sequence, they form a closely related group of morphogenetic proteins.
In order to test functional relationships between the fungal hydrophobins, we selected diverse heterologous hydrophobin‐encoding genes and introduced these individually into an M.grisea mpg1− null mutant. Complementation experiments were designed such that each hydrophobin‐encoding gene was expressed in M.grisea, either under the control of its own endogenous promoter or as a gene fusion to the MPG1 promoter. In this way, the relative contribution of the regulated expression of MPG1, or the characteristics of its hydrophobin gene product to biological function, could be assessed.
Identification of a functional MPG1 promoter
A functional MPG1 promoter initially was isolated for expression of alternative hydrophobin genes and to establish the precise pattern of expression of MPG1 in relation to its known functions (Talbot et al., 1993, 1996). A 1.28 kb promoter fragment from the 5′ end of the MPG1 gene was isolated and fused to a synthetic allele of the GFP reporter gene, encoding green fluorescent protein (Prasher et al., 1992; Stearns, 1995; Chiu et al., 1996), providing a non‐invasive reporter of MPG1 expression. The sGFP gene was subcloned in‐frame with the translation initiation codon at the 3′ end of the MPG1 promoter and introduced into an M.grisea transformation vector conferring hygromycin B resistance. In a control experiment, sGFP was fused to the constitutively expressed A.nidulans gpd promoter and transformed into a wild‐type M.grisea strain, Guy‐11 (see Materials and methods). Transformants were selected and analysed by Southern blot to ensure that they contained single insertions. A total of eight M.grisea transformants containing single copy integrations of the plasmids were selected for further analysis, four containing the MPG1(p)::sGFP::Hph construct, and four the gpd(p)::sGFP::hph construct.
Conidia from MPG1(p)::sGFP::Hph transformants were incubated on rice leaves, or sterilized onion epidermis, for 24 h and infection structures were allowed to form. GFP expression could be observed clearly in appressoria of all MPG1(p)::sGFP::Hph transformants, but fluorescence did not migrate into infection pegs or hyphae which entered the plant epidermis (Figure 1A and B). This strongly indicates that MPG1 expression is restricted to early infection‐related development, prior to plant penetration. Rice leaf infections were incubated for a further 72 h in conditions of high humidity, allowing disease lesions to form and produce aerially borne conidia (Figure 1C). Conidia from MPG1(p)::sGFP::Hph showed extensive GFP fluorescence, indicating that high level MPG1 expression also occurs during conidiogenesis (Figure 1D). Very little variation was observed between the four MPG1(p)::sGFP::Hph transformants examined. In control experiments, gpd(p)::sGFP::Hph showed GFP fluorescence at all stages of development (not shown).
Expression of diverse hydrophobin‐encoding genes in an mpg1::Hph null mutant of M.grisea
Having isolated a functional MPG1 promoter, seven alternative hydrophobin‐encoding genes were selected for complementation tests. These were SC1, SC3 and SC4 from S.commune, the rodA and dewA genes from A.nidulans, EAS from N.crassa and the ssgA from Metarhizium anisopliae. Heterologous hydrophobin genes were selected to encompass the diversity of functions so far attributed to hydrophobins: SC1 and SC4 represent dikaryon‐specific fruit body hydrophobins; rodA, dewA and EAS are all spore wall rodlet proteins; and ssgA has been implicated in appressorium development by the insect pathogen M.anisopliae, suggesting a function similar to MPG1. The origins and functions of all hydrophobins used in this study are given in Table I.
Genomic clones of SC1, SC3, SC4, rodA, dewA and EAS were selected containing all necessary components for expression in the corresponding native fungus (see Table II). Each genomic fragment was subcloned to an M.grisea transformation vector containing a selective marker conferring resistance to either bleomycin or bialophos (Table II). The resulting plasmids were then transformed into the mpg1::Hph gene replacement mutant 53‐R‐39 (Talbot et al., 1996). Transformants were analysed by Southern blot to identify transformants carrying single‐copy integrations of each hydrophobin gene (data not shown), and two recombinant transformants expressing each heterologous hydrophobin were selected for phenotypic analysis (Table II).
At the same time, genomic clones of SC1, SC3, SC4, rodA, dewA and EAS, and a cDNA clone of ssgA were selected for construction of translational fusions with the MPG1 promoter fragment. Each hydrophobin gene was subcloned in‐frame to the MPG1 promoter fragment (see Materials and methods) and introduced into the M.grisea transformation vector pCB1265 conferring bialophos resistance (Sweigard et al., 1997). Bialophos‐resistant transformants were selected and Southern blotted to identify those with single‐copy integrations of each MPG1 promoter gene fusion (not shown). Two transformants carrying single‐copy integrations for each of the seven MPG1 promoter–hydrophobin fusions were then selected for phenotypic analysis. Details of the 26 recombinant strains generated in this study are given in Table II.
RNA gel blot analysis was carried out to ensure MPG1‐like patterns of expression in each gene fusion transformant. Starvation stress is known to induce MPG1 expression (Talbot et al., 1993; Beckerman and Ebbole, 1996: Lau and Hamer, 1996) and, therefore, transformants were exposed to starvation stress by transfer of fungal mycelium to medium lacking either a nitrogen or carbon source for 12 h of growth. At this time, RNA was extracted, fractionated by gel electrophoresis, blotted and probed with the corresponding hydrophobin gene for each transformant. An example, showing expression of the 846 bp rodA transcript of transformant ARODA.24 during nitrogen and carbon source starvation, is shown in Figure 2. All transformants containing MPG1 promoter fusions were tested in this way with identical results.
Expression of heterologous hydrophobin genes regulated by their own promoters was not detected by RNA gel blot analysis (using total RNA) in transformants either during infection‐related development or during starvation stress (data not shown).
Complementation of the conidiation‐deficient phenotype of an mpg1::Hph deletion mutant by heterologous hydrophobin‐encoding genes
Conidiation is severely reduced in mpg1::Hph deletion mutants, and conidiating cultures have an ‘easily wettable’ phenotype (Talbot et al., 1993). The easily wettable phenotype is due to loss of the MPG1‐encoded rodlet protein layer found on the surface of M.grisea conidia (Talbot et al., 1996). This also reduces the efficiency of conidiogenesis, allowing only a single conidium to differentiate from each conidiophore, in contrast to conidiophores of isogenic wild‐type strains of M.grisea which produce 4–5 conidia in a sympodial array (Ou, 1985; Talbot et al., 1996). Deletion of MPG1 also causes a reduction in pathogenicity, resulting in ∼80% reduction in the usual number of disease lesions on infected rice seedlings (Talbot et al., 1993, 1996). This is associated with reduction in the ability of mpg1::Hph mutants to elaborate appressoria (Talbot et al., 1993, 1996). Recombinant transformants expressing each hydrophobin gene were therefore examined to determine their ability to substitute for MPG1 in each of its attributed functions. As a control in these experiments, two transformants were also analysed routinely where MPG1 had been re‐introduced into an mpg1− mutant (Table II). We previously reported full complementation of all mpg1− mutant phenotypes in these transformants (Talbot et al., 1996), which are phenotypically indistinguishable.
Conidiogenesis was assessed by taking plate cultures of M.grisea grown for a uniform time, flooding them with water and gently removing conidia with a glass rod. When transformants expressing hydrophobins under control of their own promoters were analysed, the only transformants to show partial complementation of conidial numbers were those expressing the S.commune hydrophobin, SC1 (Figure 3A). Two transformants, MJK7.8 and MJK7.16 [SC1(p)::SC1::ble], produced mean conidial numbers of 9.5×105/ml compared with the wild‐type strain Guy‐11, which produced a mean of 8×106 conidia/ml, and the isogenic mpg1::Hph mutant 53‐R‐39 which produced a mean of 3×105 conidia/ml (Figure 3A). Both of the SC1(p)::SC1::ble transformants were also complemented for the easily wettable phenotype, being able to support a droplet of water on the culture surface (Figure 3C).
Transformants containing MPG1 promoter gene fusions produced considerably greater numbers of conidia compared with the isogenic mpg1− mutant 53‐R‐39 (Figure 3B). The highest conidial numbers were found in recombinants expressing MPG1(p)::EAS, MPG1(p)::dewA and MPG1(p)::SC1 gene fusions. Conidial numbers in these transformants were >3×106/ml and therefore not significantly different from wild‐type when compared by the Student's t‐test (P >0.05). In contrast, MPG1(p)::SC3::bar transformants produced conidial numbers which were significantly different from wild‐type (t = 35.1; P <0.001; df = 5), but not significantly different from the isogenic mpg1− mutant (t = 0.4; P >0.05; df = 5). Consistent with this, MPG1(p)::SC3::bar transformants were easily wettable while the remaining transformants were identical to wild‐type M.grisea with respect to surface hydrophobicity (Figure 3C).
Introduction of heterologous hydrophobin‐encoding genes restores pathogenicity of an mpg1::Hph deletion mutant
The ability of transformants to cause rice blast disease was determined by infecting rice seedlings of a susceptible rice cultivar CO‐39. Conidial suspensions were prepared and adjusted to 1×104 conidia/ml. Fourteen‐day‐old rice seedlings were then sprayed with the suspension and the disease allowed to progress for 96 h. At this time, rice blast infections with the wild‐type M.grisea strain Guy‐11 produced large numbers of oval‐shaped lesions on rice leaves. In mpg1− mutants, the frequency of lesions was between 10 and 20% of the number generated by an isogenic wild‐type strain of M.grisea (Talbot et al., 1993, 1996). To assess pathogenicity of transformants expressing alternative hydrophobins, the lesion number was recorded on 5 cm leaf sections as shown in Figure 4A and B. Pathogenicity was partially restored for SC1(p)::SC1::ble transformants, but none of the other transformants carrying hydrophobin genes expressed under endogenous promoters showed significant increases in lesion number compared with the mpg1::Hph mutant (P >0.1; df = 40) (Figure 4A and C). In contrast, six of the transformants expressing hydrophobin genes controlled by the MPG1 promoter showed partial restoration of pathogenicity. The highest number of disease lesions was produced by MPG1(p)::EAS and MPG1(p)::SC1 transformants which were not significantly different from lesion numbers generated by Guy‐11 (Figure 4B and D). MPG1(p)::SC3 transformants were not restored for pathogenicity and showed similar lesion numbers to the mpg1::Hph null mutant (Figure 4B). We conclude that MPG1 promoter‐driven expression of dewA, rodA, SC1, SC4, EAS and ssgA can restore the activity of MPG1 in pathogenicity of M.grisea.
Restoration of appressorium development by expression of heterologous hydrophobins in an mpg1::Hph mutant
Reduced pathogenicity of mpg1− mutants is associated with a decrease in the ability of the fungus to elaborate appressoria (Talbot et al., 1993). Transformants were therefore tested for their ability to develop appressoria on hydrophobic surfaces. Consistent with restoration of other Mpg1 mutant phenotypes, SC1(p)::SC1::ble transformants showed partial restoration of this ability (Figure 5A). None of the remaining transformants, however, produced significantly greater numbers of appressoria compared with the isogenic mpg1::Hph mutant. Transformants expressing hydrophobin genes under control of the MPG1 promoter showed restoration of appressorium formation on hydrophobic surfaces, with the exception of MPG1(p)::SC3 transformants. These produced numbers of appressoria which were not significantly different from the mpg1− mutant, 53‐R‐39 (t = 2.96; P >0.04; df = 5). Appressoria were produced in the greatest numbers by MPG1(p)::EAS, MPG1(p)::SC1 and MPG1(p)::dewA transformants (Figure 5B). Appressorium production was therefore restored to near wild‐type levels by expression of EAS, SC1 and dewA under control of the MPG1 promoter, and partially restored by expression of SC4, ssgA and rodA.
A control experiment was carried out to determine whether the variation in appressorium formation was associated with introduction of the heterologous hydrophobins. Because transformation in M.grisea is integrative, it has the potential to cause mutations and, therefore, a control experiment was necessary to test whether appressorium deficiency was due simply to the insertion of plasmid DNA. Appressorium development in M.grisea can be induced by exogenous application of cAMP (Lee and Dean, 1993) and, previously, we and others showed that MPG1 acts upstream of the cAMP signalling pathway as appressorium deficiency in mpg1− mutants can be restored by exogenous cAMP (Beckerman and Ebbole, 1996; Talbot et al., 1996). A 10 mM aliquot of cAMP was therefore added to germinating conidia of all hydrophobin transformants. Under these conditions, all transformants generated in this study were able to form appressoria (data not shown).
Complementation of mpg1− by heterologous hydrophobins is correlated with production of a rodlet protein
Rodlet proteins are well known ultrastructural characteristics of aerial structures of fungi and have been shown to be the products of hydrophobins by purification and peptide sequencing (Wösten et al., 1993; Templeton et al., 1995), immunolocalization (Wösten et al., 1994a) or by their absence in hydrophobin null mutants (Stringer et al., 1991; Bell‐Pederson et al., 1992; Talbot et al., 1996). MPG1 encodes a rodlet protein composed of 5–7 nm interwoven rodlets which coats the conidial cell wall (Talbot et al., 1996), as shown in Figure 6. This rodlet layer is completely absent from mpg1::Hph mutants (Figure 6D) and distinct from the rodA‐encoded rodlets found on A.nidulans conidia or EAS‐encoded rodlets of N.crassa (Figure 6A and B). In both cases, the size of rodlets is greater than those of M.grisea and the arrangement is characteristic of each species.
When heterologous hydrophobins were expressed in M.grisea under control of their own promoters, a rodlet layer was only observed on conidia of SC1(p)::SC1::ble transformants (Figure 6E). In the remaining 10 transformants examined, the surfaces of conidia were indistinguishable from those of 53‐R‐39 (Figure 6D).
Expression of hydrophobin genes under control of the MPG1 promoter led to partial restoration of a conidial rodlet layer for six of the seven hydrophobins tested (Figure 6G–L). In general, the conidial rodlet layers of transformants were indistinct compared with those of wild‐type M.grisea, but it was possible to observe considerable variation in the size and arrangement of rodlets (Table III). Rodlets encoded by dewA, SC1, SC4 and ssgA have not been described previously, and so characteristics of the layers observed on M.grisea conidia cannot be compared directly with rodlet layers in the host fungus. EAS‐encoded rodlets and rodA‐encoded rodlets, however, did appear to adopt broadly similar conformations in M.grisea transformants when compared with N.crassa and A.nidulans, respectively, although the rodlet layers were less distinct. We conclude that heterologous expression of hydrophobins in an M.grisea mpg1− mutant can result in partial restoration of a rodlet layer.
Complemented mpg1::Hph mutants exhibit hydrophobin‐mediated self‐assembly during appressorium formation
The MPG1 hydrophobin previously has been proposed to act as a developmental sensor of hydrophobic surfaces (Beckerman and Ebbole, 1996; Talbot et al., 1996). Evidence for this role was based on assays of appressoria and germ tube adhesion to Teflon coverslips following washes in 60% ethanol and boiling in 2% SDS (Talbot et al., 1996). Because assembled hydrophobins are insoluble in ethanol and hot SDS, while most proteins and carbohydrates are not, these washes remove fungal structures unless attached to the surface in a hydrophobin‐dependent manner. In such an assay, wild‐type M.grisea appressoria and germ tubes remain attached to hydrophobic surface while those of mpg1− mutants do not (Talbot et al., 1996), consistent with MPG1‐mediated attachment.
In this study, hydrophobin adhesion assays were carried out for each hydrophobin transformant to determine whether remediation of appressorium deficiency and pathogenicity was due to surface self‐assembly of the heterologous hydrophobin. Germ tubes and appressoria of complemented MPG1(p)::EAS, MPG1(p)::dewA and MPG1(p)::SC1 transformants showed wild‐type levels of attachment to the hydrophobic surface (Figure 7). However, germ tubes and appressoria of the non‐complemented MPG1(p)::SC3 were unable to attach strongly to the surface after treatment with ethanol and hot SDS. Taken together, these observations confirm that efficient formation of appressoria in M.grisea requires surface self‐assembly of MPG1, and show that this function may be partially restored by correct expression of heterologous hydrophobins such as EAS, dewA and SC1.
In this study, we have demonstrated that hydrophobins constitute a functional group of proteins and that diverse hydrophobins can partially complement mpg1− mutant phenotypes if expressed at the appropriate developmental stage.
Our aim in these experiments was to establish whether regulated expression of a hydrophobin gene, or the particular characteristics of its encoded hydrophobin, were of greatest significance in determining its biological function. As only one hydrophobin gene, SC1 from the basidiomycete S.commune, was able to complement mpg1− when expressed under the control of its endogenous promoter, it would appear that regulated expression plays a role in determining hydrophobin function. This is consistent with the lack of detectable levels of mRNA expression in these transformants. The significance of gene regulation in determining hydrophobin function could be tested directly by expressing hydrophobin genes, including MPG1, under a constitutive M.grisea promoter. In this way, the effects of high level hydrophobin expression, in the absence of developmental changes in the fungus, could be investigated.
The failure of EAS, dewA, rodA and ssgA to complement mpg1−, when controlled by their native promoters, was somewhat surprising given that ascomycete promoters work well among related species (for a review, see Fincham, 1989) and that these hydrophobins are normally expressed during conidiation (Stringer et al., 1991; Bell‐Pederson et al., 1992; Lauter et al., 1992; Stringer and Timberlake, 1995) or appressorium development (St. Leger et al., 1992).
The general lack of complementation by natively regulated hydrophobin genes highlighted the importance of specific regulation for the function of MPG1 and the necessity for MPG1 promoter fusions in any effective comparison of heterologous hydrophobins.
Functional relatedness of the fungal hydrophobins
Complementation analysis of mpg1− with MPG1 promoter‐regulated hydrophobin genes suggests that hydrophobins are a closely related group of proteins in spite of their very considerable amino acid divergence. This indicates that conservation of eight cysteine residues and a similar distribution of hydrophobic and hydrophilic amino acids is sufficient for hydrophobin function, irrespective of low amino acid identity. Complementation tests did, however, define two functional groups of hydrophobins. Hydrophobin genes normally expressed in fruit bodies, conidia or infection structures, such as rodA, dewA, EAS, ssgA, SC1 and SC4 (Wessels, 1997), were all partially able to complement mpg1− mutant phenotypes, whereas SC3, which is expressed in fungal mycelium and required for formation of aerial hyphae (van Wetter et al., 1996) and surface attachment (Wösten et al., 1994b), was unable to complement.
We cannot, at this stage, preclude that lack of complementation of mpg1− by SC3 was due to poor levels of secretion of the protein, perhaps due to the SC3 signal peptide being poorly recognized in M.grisea. Given the efficient expression and secretion of all other hydrophobins in M.grisea transformants, however, this seem unlikely. Interestingly, comparison of putative amino acid sequences for the seven hydrophobin genes does not reveal the putative functional relationship suggested by complementation tests. SC3, for example, is most homologous to the other S.commune hydrophobins SC4 and SC1 (Wessels, 1997), while MPG1 shows greatest amino acid identity with ssgA (Talbot et al., 1993). It may be worth speculating, therefore, that the function of a hydrophobin is determined mainly by its ability to link with the underlying fungal cell wall following self‐assembly, perhaps mediated by carbohydrate interactions. This is consistent with lack of rodlet formation in MPG1(p)::SC3 transformants. SC3 is known to be glycosylated (de Vocht et al., 1998), and other hydrophobins have sequence motifs consistent with N‐ and O‐linked glycosylation (Stringer et al., 1991; Bell‐Pederson et al., 1992). Correct glycosylation may, therefore, be essential for linkage to fungal cell walls and it may be that lack of complementation by SC3 is due to incorrect glycosylation of the hydrophobin in M.grisea. This could also account for rodlet protein layers being only partially restored in heterologous hydrophobin transformants. A comparison of the glycosylation of purified hydrophobins is therefore likely to prove informative in comparison with the functional complementation data presented here.
Recently, two further hydrophobins expressed during mycelial growth of fungi have been described; the HCf‐1 hydrophobin from Cladosporium fulvum (Spanu, 1997) and the CO‐1 hydrophobin from Coprinus cinereus (Asgeirsdottir et al., 1997). If SC3 does form a separate functional group of hyphal hydrophobins as suggested by this study, then these hydrophobins should also fail to complement mpg1 and form a functional group with SC3. Experiments to test this idea currently are underway.
Evolution of fungal hydrophobins
The data presented show that very different heterologous hydrophobins can complement for a hydrophobin functioning in fungal pathogenesis. This suggests that hydrophobins have been co‐opted into serving diverse developmental functions in fungi by evolution of specific developmental expression patterns. In the case of MPG1, for example, a single hydrophobin mediates efficient production of conidia, their surface hydrophobicity and the formation of infection structures, based largely on its regulated expression. The considerable divergence in hydrophobin sequence may, therefore, simply reflect a relative lack of selective pressure on hydrophobins as long as they are able to carry out interfacial self‐assembly (Wösten et al., 1993) and linkage to the underlying fungal cell wall. This serves not only to emphasize developmental regulation of hydrophobin expression in conditioning function, but also the likely widespread conservation of the interfacial self‐assembly process in the action of fungal hydrophobins.
Materials and methods
Strains of M.grisea used in this study are stored in the laboratory of N.J.Talbot. The fungus was grown routinely on complete medium (Talbot et al., 1993) using standard procedures (Crawford et al., 1986). Long‐term storage of M.grisea was carried out by growing the fungus through sterile filter paper discs, desiccating these for 48 h and storing them at −20°C. The fertile rice pathogenic strain Guy‐11 (Notteghem and Silue, 1992) was used as the wild‐type strain in this study. The mpg1− mutant progeny 53‐R‐39 was selected from a cross between an original mpg1::Hph transformant, TM400‐2, and a strain of opposite mating type strain, TH3 (Talbot et al., 1993, 1996).
Construction of MPG1(p)::sGFP gene fusion
A 1.28 kb PstI–NcoI restriction fragment from the 5′ end of MPG1 was isolated from pNT800, a genomic subclone of MPG1 previously shown to complement an mpg1− mutant (Talbot et al., 1996). The NcoI site was introduced at the translation initiation codon by PCR using the primer 5′‐ACCTAATTCACCATGGTCTCCCTC‐3′ and vector‐specific T7 primer (Stratagene). The fragment was digested with PstI and NcoI and cloned into pBS SK+ (Stratagene) to generate pNT190. The MPG1 promoter fragment was sequenced by the dideoxy chain termination method using Sequenase (Amersham) and compared with the full DNA sequence of pNT800 to ensure fidelity with the genomic sequence.
The sGFP allele (Chiu et al., 1996) was obtained as a gene fusion under control of the A.nidulans gpd promoter and carrying the A.nidulans trpC terminator (Punt et al., 1987), kindly supplied by Dr Marc‐Henri Lebrun (Université de Paris‐Sud, France). A 3.75 kb EcoRI–HindIII fragment containing the gpd(p)::sGFP gene fusion was subcloned into pCB1004, an M.grisea transformation vector conferring hygromycin B resistance (Sweigard et al., 1997), to create pMJK‐80. This construct subsequently was used in control transformation experiments to provide constitutive sGFP expression.
The MPG1(p)::sGFP expression vector, pMJK96, was generated by excision of gpd(p) from pMJK‐80 as a 2.5 kb BstXI–NcoI fragment and replacement with MPG1(p), subcloned as a 1.3 kb BstXI–NcoI fragment from pNT190.
Construction of hydrophobin expression vectors
Transformation vectors containing hydrophobin genes under control of their own promoter were constructed by subcloning the corresponding genomic clones into transformation vector pAN8.1 which confers bleomycin resistance (Punt et al, 1987), or into the vector pCB1265 which confers bialophos resistance (Sweigard et al., 1997).
Expression vectors containing hydrophobin genes as translational fusions to the MPG1 promoter were generated by amplification of each hydrophobin gene coding sequence from genomic clones. Restriction sites were introduced during PCR amplification. Introduction at the start codon of an NcoI site (or sites compatible with NcoI) allowed each gene to be subcloned in‐frame to the MPG1 promoter at their translation initiation codon. At the 3′ end of each hydrophobin gene, a restriction site was introduced which was present in the transformation vector, but not internal to the MPG1 promoter or the hydrophobin gene. In all cases, an ApaI (5′‐GGGCCC‐3′) site was thus generated. PCR amplification from plasmid templates was performed in a Perkin Elmer Gene Amp PCR 2400 with 25 cycles of amplification using the proofreading enzyme Pfu polymerase (Stratagene). This ensured fidelity of the amplified sequence. Each cycle consisted of 30 s denaturation at 94°C, 30 s annealing at 50°C and 1 min extension at 72°C. In all cases, a 3 min denaturation step at 94°C preceded the first amplification cycle and a 15 min extension at 72°C was performed after completion.
Oligonucleotides were designed according to the published sequence of each hydrophobin. For rodA (Stringer et al., 1991) amplification, the primers used were RodA‐5, 5′‐CCACCTCTTCATCATGAAGTTCTCCAT‐3′, and RodA‐3, 5′‐CTTGGGCCCGAAGTAAAAGATAATAAACA‐3′, to introduce sites for BspHI and ApaI, respectively (underlined). For amplification of SC1 (Schuren and Wessels, 1990), primers used were SC1‐5, 5′‐AGCAACAACCATGGCCTTCTCGCTCGCC‐3′, and SC1‐3, 5′‐CTTGGGCCCGTGCGTGTATACCGACCTT‐3′, introducing sites for NcoI and ApaI, respectively. For SC4 (Schuren and Wessels. 1990) amplification, primers used were SC4‐5, 5′‐CTTCATGAGATTCTCGCTCGCTCTGCTTGCT‐3′, and SC4‐3, 5′‐CTTGGGCCCGTCGACCACGTGCGCGCG‐3′, to introduce sites for BspHI and ApaI. For amplification of ssgA (St. Leger et al., 1992), primers used were SSGA‐5, 5′‐TCCTTCAACATGTTCAAGGCTCTCATCG‐3′, and SSGA‐3, 5′‐CTTGGGCCCAAGAGACAGAGTAAATTTGT‐3′, to introduce sites for AflIII and ApaI, respectively. For EAS (Bell‐Pederson et al., 1992; Lauter et al., 1992) amplification, primers used were EAS‐5, 5′‐CCCAACCGCCATCATGAAGTTCACCAGC‐3′, and EAS‐3, 5′‐CTTGGGCCCAGGGTTAAGAGAGTGTATTA‐3′, to introduce sites for BspHI and ApaI, respectively. For amplification of dewA (Stringer and Timberlake, 1995), primers used were dewA‐5, 5′‐AACTCATCAACATGTCCCGCTTCATCGTC‐3′, and dewA‐3, 5′‐TGTGGGCCCGAACAACACCAATTATTATT‐3′, introducing sites for AflIII and ApaI, respectively.
In all cases, amplification products were excised and digested with appropriate restriction enzymes and subcloned into pNT190 to generate an MPG1 promoter translational fusion at the NcoI site. Each gene fusion was then subcloned to pCB1265 (Sweigard et al., 1997) for M.grisea transformantion.
The subcloning of SC3 (Wösten et al., 1994) into pCB1265 was carried out by ligation of a 874 bp NcoI restriction fragment containing the SC3 open reading frame (Schuren and Wessels, 1990) directly into pNT190. Recombinant plasmids were restriction mapped to ensure the fragment was in the correct orientation. The MPG1(p)::SC3 gene fusion was then excised as a 2.15 kb PstI–ApaI fragment and subcloned into pCB1265. For details of all recombinant DNAs used in this study, see Table II.
Protoplast preparation and transformations were performed as described previously (Talbot et al., 1993). Constructs were transformed into the Mpg1 mutant 53‐R‐39 and transformants selected for bleomycin resistance at 35 μg/ml bleomycin in the presence of 100 μg/ml caffeine, or for bialophos resistance at 35 μg/ml. Putative transformants were confirmed by DNA gel blot hybridization, and those carrying single plasmid integrations were selected.
Fourteen‐day‐old rice seedlings were infected with suspensions of M.grisea conidia prepared in 0.1% gelatin at a concentration of 1×104 conidia/ml. Two‐week‐old seedlings of the susceptible rice cultivar CO‐39 were sprayed using an artists‘ airbrush (Badger Co., Franklin Park, IL). Plants were incubated in plastic bags for 18 h to maintain high humidity and then transferred to controlled environment chambers at 24°C, 84% relative humidity, with 900 μE/m2 tungsten illumination and a 14 h day length. Plants were incubated for 96–120 h for full disease symptoms to become apparent. The first disease symptoms were observed 96 h after seedling inoculation. Lesion densities were scored routinely from 40 randomly chosen 5 cm leaf tips, and means and standard deviations determined. Results from pathogenicity assays were compared using the Student's t‐test and two non‐parametric tests; the Mann–Whitney two sample test and the Kruskal–Wallis test (Sokal and Rohlf, 1981).
Assays for infection‐related morphogenesis
Appressorial development by M.grisea was observed on Teflon membranes (DuPont) as described by Hamer et al. (1988). A 200 μl drop of a conidial suspension at a concentration of 1×105/ml was placed on the surface of a Teflon coverslip and left in a humid environment at 24°C. The frequency of appressorium formation was determined by counting the number of appressoria that had developed from 300 conidia after 14 h (Talbot et al., 1993). Remediation experiments with cAMP were carried out by incubation of conidia in 10 mM cAMP (Sigma) solutions as described by Lee and Dean (1993). Results from appressorium assays were compared using the Student's t‐test.
Nucleic acid isolations and analysis
Genomic DNA was extracted from fungal mycelium using a CTAB (hexadecyltrimethylammonium bromide) procedure described by Talbot et al. (1993). Gel electrophoresis, restriction enzyme digestion and DNA gel blot hybridizations were all carried out using standard procedures (Sambrook et al., 1989). DNA hybridization probes were labelled by the random primer method (Feinberg and Vogelstein, 1983) using the Stratagene Prime‐It kit (Stratagene), and high stringency washes were carried out as described previously (Talbot et al., 1993). RNA was isolated from hyphal cultures of M.grisea by the method of Timberlake (1980). RNA gel electrophoresis and RNA gel hybridization were carried out using standard procedures (Sambrook et al., 1989).
The surface of M.grisea conidia were viewed as replicas made after freeze etching. For freeze etching, conidia were fixed in 3% glutaraldehyde in 100 mM potassium phosphate buffer (pH 7.0) and washed three times in buffer in a modification of the procedure of Stringer et al. (1991). Samples were then cryoprotected by sequential infiltration with 10% (for 1 h) and 20% glycerol (overnight). Conidia were then frozen in Freon 22 (ICI, Runcorn, UK) and nitrogen slush. Freeze‐fracturing was carried out in a Baltzer's BA 301 (Baltzer Pfeiffer GmbH, Leichenstein) and conidia were then shadowed with carbon and platinum at 45°C. A backing layer of pure carbon was added at 90°C and the replicas floated onto distilled water. Replicas were cleaned overnight in 50% chromic acid and washed several times in distilled water before being picked up onto copper grids and viewed with a Jeol 100C transmission electron microscope (Jeol, Tokyo, Japan).
Fungal attachment assays
Experiments to investigate hydrophobin‐mediated attachment to hydrophobic surfaces were carried out by sequential treatment of germinated conidia–germ tube–appressoria combinations prepared on Teflon membranes (DuPont). Samples were treated with 60% ethanol for 20 min at room temperature with gentle shaking and then boiling in 2% SDS for 20 min in a water bath (Talbot et al., 1996). Teflon membranes were rinsed gently in sterile distilled water and examined by microscopy. Several independent experiments were carried out to determine the effect of each treatment singly. The percentage of adhering conidia–germ tube–appressoria combinations from each strain was then recorded.
We gratefully acknowledge Howard Stebbings (Exeter), John Hamer (Purdue) and Joseph Wessels (Groningen) for critical reading of the manuscript. This work was supported by grants to N.J.T. from the Biotechnology and Biological Sciences Research Council (P01550), the Royal Society and the University of Exeter Research Fund. N.J.T. is a Nuffield Foundation Science Research Fellow.
- Copyright © 1998 European Molecular Biology Organization