Bipolaris maydis HITO7711 (mating type MAT1-2) was used as the wild-type strain throughout this study. Albino strain alb3 (mating type MAT1-1) was used in a crossing assay (Tanaka et al., 1991). The strain lacking the CHK1 gene (Δchk1) was used as a negative control in a hydrophobicity assay (Izumitsu et al., 2009; Degani et al., 2013). All gene disruptants of B. maydis were derived from HITO7711. MASHIKI2-2 (MAT1-1) was used as an alternative wild-type strain in a following crossing. For the crossing assay, the quadruple disruptant of hydrophobin genes (MAT1-1) was obtained by crossing the quadruple disruptant (MAT1-2) and MASHIKI2-2.

Aspergillus nidulans ABPU1 strain (ΔligD), which is highly efficient for gene disruption via homologous recombination, was used as a parental strain to obtain a dewA gene-disruptant (Hagiwara et al., 2007; Yoshimi et al., 2013). The BPU1 strain carrying the argB gene was used as a control (CNT) strain for phenotypic comparisons (Hagiwara et al., 2007; Miyazawa et al., 2021).

All strains were maintained on complete medium agar (CMA; Tanaka et al., 1991) or vegetable juice agar (VJA) medium at 25°C. VJA contained vegetable juice (Kagome, Tokyo, Japan) instead of V8 juice but otherwise had the same composition as V8 agar medium (Ribeiro, 1978); no difference in the growth of B. maydis was detected between V8 agar and VJA (data not shown). In the crossing assay, Sachs’s agar medium was used. In a stress tolerance assay, CMA was used. Czapek–Dox (CD) medium was used for culturing A. nidulans (Fujioka et al., 2007).

Hydrophobins are small secreted proteins with a signal peptide and a conserved characteristic C-terminal sequence. To identify the hydrophobins of B. maydis, we downloaded the translated sequence from gene catalog of Cochliobolus heterostrophus C5 v2.0 from the website of the DOE Joint Genome Institute (JGI, https://mycocosm.jgi.doe.gov/CocheC5_3/CocheC5_3.home.html, Ohm et al., 2012; Condon et al., 2013). In this dataset, we first searched for proteins that were cysteine rich (≥4) and small (≤250 amino acids). Next, we checked whether those proteins possess signal peptides using SignalP 5.0 (https://services.healthtech.dtu.dk/services/SignalP-5.0/). From the refined data, we visually selected proteins with a cysteine pattern specific to hydrophobins (-C-CC-C-C-CC-C-). Additionally, we performed the same search using the updated v3.0 dataset (https://mycocosm.jgi.doe.gov/CocheC5_4m/CocheC5_4m.home.html, Haridas et al., 2023). Hydropathy plots were investigated using ProtScale (https://web.expasy.org/protscale/) based on the Kyte and Doolittle (1982) method. The amino acid sequences of B. maydis and P. oryzae hydrophobins were aligned with MUSCLE using MEGA X (Edgar, 2004; Kumar et al., 2018). In the alignment and the analysis of the hydropathy plots, the amino acid sequences of P. oryzae Mpg1 (class I) and Mhp1 (class II) were used as references. For in silico characterization, InterProScan (https://www.ebi.ac.uk/interpro/search/sequence/) was used to analyze each B. maydis hydrophobin.

The wild-type strain was incubated under three different conditions: liquid culture, on solid medium, or on a host plant (maize leaves). In liquid culture, the conidia (final concentration, 10⁵ conidia/mL) were added to 100 mL CM liquid medium and incubated with shaking. On solid medium, 100 conidia were inoculated onto cellophane membrane placed on the top of Petri plates filled with CMA. On the host plant, conidial suspension (20 μL of 5.0 × 10⁵ conidia/mL) was inoculated at five points on a single detached 1-month-old leaf. Subsequently, five lesion areas were cut off from leaves and collected. All incubations were carried out at 25°C for 3 days in the dark and were repeated 3 times. Total RNA was extracted from mycelia or leaf lesion areas using Trizol, according to Yoshida and Tanaka (2019). cDNA was synthesized from the total RNA with SuperScript IV reverse transcriptase (Thermo Fisher Scientific, Waltham, MA, USA) using the (dT)21VN primer.

The procedure for generating disruptants and locations of primers to confirm gene disruptions are shown in Supplementary Figure S1A. All primers are listed in Supplementary Table S1. A disruption cassette was constructed for each gene encoding a putative hydrophobin. PCR to construct these cassettes was performed with Takara Ex Taq (Takara Bio, Kusatsu, Japan). All PCR confirmations of gene disruption were performed with SapphireAmp Fast PCR Master Mix (Takara Bio).

HYP1 disruptant (Δhyp1): The genomic DNA of B. maydis HITO7711 was obtained as described by Izumitsu et al. (2012). The upstream (primer set Hyp1-f1/Hyp-r1) and downstream regions (Hyp-f2/Hyp-r2) of the HYP1 gene ORF were amplified from genomic DNA. The neomycin phosphotransferase II gene (NPTII) cassette was amplified from the pZGenI plasmid (primer set NPTandHPH-f1/NPTII-r1; Sumita et al., 2017). The three amplified products were fused by fusion-PCR (Hyp1-f3/Hyp1-r3 primer set). The resulting cassette was purified by ethanol precipitation and used for HYP1 disruption. The transformation was based on the PEG-mediated protoplast transformation method described in Izumitsu et al. (2009). The replacement of HYP1 by the disruption cassette was confirmed by PCR with three primer sets (Hyp1-f1/NPTII-chk1, Hyp1-f4/Hyp1-r4 and NPTandHPH-chk2/Hyp1-r1). A similar approach was used to disrupt HYP2–HYP4 individually and to generate Δhyp2–Δhyp4 strains.

Double, triple, and quadruple disruptants: To generate a double disruptant, Δhyp2Δhyp3, the HYP2 gene was replaced in the parental Δhyp3 strain using a disruption cassette with a nourseothricin-resistance gene (NAT). The NAT gene was amplified from the pZNAT1 plasmid with the primer set NATandBAR-f1/NAT-r1 (Izumitsu et al., 2009). To generate the triple disruptant, Δhyp2Δhyp3Δhyp4, the HYP4 gene of the double disruptant Δhyp2Δhyp3 was replaced with the bialaphos-resistance gene (BAR) cassette. The BAR gene cassette was amplified from the pCB1546 plasmid with the primer set NATandBAR-f1/BAR-r1 (Sweigard et al., 1997). To generate the quadruple disruptant, Δhyp1Δhyp2Δhyp3Δhyp4, the HYP1 gene of the triple disruptant was replaced with a hygromycin B phosphotransferase gene (HPH) cassette. The HPH cassette was amplified from the plasmid pCB1004 (Carroll et al., 1994).

CMA was supplemented with 0.05% SDS, 0.015% H₂O₂, or 25 μg/mL Calcofluor White (CFW; Sigma-Aldrich, Saint Louis, MO, USA). Each strain was incubated for 7 days at 25°C, and their colony diameters were measured. To quantify the reduction in radial-growth of colonies, we calculated the percentage decrease relative to the untreated control.

Maize plants (Zea mays cv. Takanestar) were grown at 25°C for 1 month. Freshly detached leaves were placed in plastic cases lined with wet paper towels. Conidia of each strain were harvested from colonies grown on VJA for 7 days. Twenty μL of 5.0 × 10⁵ conidia/mL suspensions (in 0.01% Tween-20) were dropped onto the leaves and incubated in the dark at 25°C for 3 days. We measured and photographed the lesions on the leaves.

Bipolaris maydis forms a characteristic sexual organ called a pseudothecium; usually, it is black in the wild-type strain and light tawny in the albino strain. The ability of any strain to form pseudothecia can be evaluated by using the albino strain in a crossing assay. The crossing assay was performed according to Tanaka et al. (1991). Dried maize leaves were placed on Sachs’s agar medium, then each disruptant (MAT1-2) and the albino strain (MAT1-1) were inoculated diagonally across the leaves and incubated at 25°C in the dark for 30 days. The pseudothecia were photographed. Both black and light tawny pseudothecia were harvested and dissected in a drop of 10% glycerol on a glass slide to observe the asci and ascospores within the pseudothecia, respectively, under a microscope (Leica DML, Leica Microsystems, Wetzlar, Germany).

The gene disruption strategies and locations of primers to confirm the insertion are shown in Supplementary Figure S2. PCR for plasmid construction was performed with PrimeSTAR GXL DNA polymerase (Takara Bio). All PCR confirmations were performed with Takara Ex Premier DNA Polymerase (Takara Bio). NEBuilder HiFi DNA Assembly Master Mix (New England Biolabs, Ipswich, MA, USA) was used for fusions. Fungal transformations were performed using the PEG-mediated protoplast transformation method as described in Yoshimi et al. (2013).

To construct a disruption cassette of the A. nidulans dewA gene (for class I hydrophobin), the upstream and downstream regions of this gene were amplified from genomic DNA with the primer sets dewA-f1/dewA-r1 and dewA-f2/dewA-r2. The argB gene of A. oryzae (used in this work as selection marker) was amplified from the pAORB plasmid with a primer set AoargB-f1/AoargB-r1 (Furukawa et al., 2002). The three amplicons were fused into a BamHI-linearized pUC19 plasmid. The dewA gene disruption cassette was amplified from the resulting plasmid with the primer set pUC-chk1/pUC-chk2, purified by ethanol precipitation and integrated into the ABPU1 (ligD) strain. The candidate disruptants were inoculated and incubated on selective CD medium without arginine. The replacement of the dewA gene with the argB gene was confirmed with the dewA-f3/AoargB-chk1, dewA-f4/dewA-r4, and AoargB-chk2/dewA-r3 primer sets.

For complementation assays with the dewA and B. maydis hydrophobin genes, we constructed a plasmid to reintroduce the dewA gene and four plasmids to introduce B. maydis hydrophobin genes into the ΔdewA strain. To generate the pUC-pyrG plasmid, the entire A. nidulans pyrG gene with the putative promoter was amplified with the primer set pyrG-f1/pyrG-r1, and the PCR fragment was fused into a BamHI-linearized pUC19 vector. To generate the dewA complementation plasmid, the entire dewA gene with the promoter and ORF was amplified with the primer set dewA-pyrG-f1/dewA-pyrG-r1 and fused into a SmaI-linearized pUC-pyrG plasmid.

To generate the HYP1 complementation plasmid, the HYP1 gene with the 3′ untranslated region from the start codon was amplified from B. maydis genomic DNA with the primer set HYP1-start-f1/HYP1-pyrG-r1. The putative promoter region (ca. 1000 bp upstream of dewA ORF) was amplified with the primer set dewA-pyrG-f1/dewA-HYP1-r1. These two PCR fragments were fused into the SmaI-linearized pUC19-pyrG plasmid. Complementation plasmids for the remaining three B. maydis hydrophobins were constructed in a similar fashion.

To generate dewA^comp and B. maydis hydrophobin-complemented strains (HYP1–4^comp), the corresponding complementation plasmids were introduced into the ΔdewA strain. Resulting candidate strains were inoculated and incubated on selective CD medium without arginine, uridine, and uracil. Insertion of each gene was confirmed with the corresponding primer set f4/r4.

The CNT, ΔdewA, dewA^comp, and HYP1–4^comp strains were incubated on CD medium for 1 week. Conidia (final concentration, 10⁵ conidia/mL) from each colony were added to 20 mL of CD liquid medium and incubated with shaking for 24 hours. The extraction of total RNA from the mycelia and the synthesis of cDNA were performed as described above. The histone H2B gene was used as internal control (Yoshimi et al., 2013).

Colonies of the wild-type strain and all disruptants of B. maydis were grown on CMA at 25°C for 7 days. The colonies were punched 1 cm from their edge with a cork borer (diameter: 1 cm), and 10 µL of a solution containing 0.2% SDS and 50 mM ethylenediaminetetraacetic acid (EDTA) was dropped onto the mycelial disks. We took photos of these droplets every 10 min for 30 min to record their changes.

CNT, ΔdewA, dewA^comp, and HYP1–4^comp strains of A. nidulans were incubated on CD medium at 37°C for 4 days. The colonies were punched at the edge using a cork borer, aligning the borer’s perimeter with the colony’s perimeter. The hydrophobicity of each strain was assessed as above. Photographs were taken every minute for 5 min.

By searching the gene catalog (protein data) of C. heterostrophus C5 from the JGI website, we identified four candidate proteins with a signal peptide and a cysteine pattern characteristic of hydrophobins. The total length was 117 amino acids for Hyp1, 100 amino acids for Hyp2, 157 amino acids for Hyp3, and 182 amino acids for Hyp4. Each of the hydrophobin genes had two exons. The hydropathy profile of Hyp1 was similar to that of Mpg1 of P. oryzae while that of Hyp2 was similar to that of Mhp1 (Figure 1; Supplementary Figure S3). Those of Hyp3 and Hyp4 were also similar to that of Mhp1, except some regions (Figure 1; Supplementary Figure S3B). Hyp3 had a unique long sequence (71 amino acids) after the 8th cysteine, which contained a 9th cysteine (Supplementary Figure S3B). Hyp4 had 90 amino acids, approximately half of them glycines, between the signal peptide and the first cysteine (Supplementary Figure S3B). InterProScan searches showed that Hyp1 belongs to the hydrophobin family (InterPro IRR001338; class I), whereas Hyp2–Hyp4 belong to the cerato-ulmin hydrophobin family (InterPro IRR010636; class II). Cerato-ulmin is a representative class II hydrophobin produced by Ophiostoma ulmi (Stringer and Timberlake, 1993; Sbrana et al., 2010).

In mycelia of B. maydis incubated in liquid culture, on solid medium, or on maize leaves, all hydrophobin genes were expressed, but at different levels: the transcript levels were higher for HYP1 and HYP3 than for HYP2 and HYP4 (Figure 2), suggesting that this fungus uses mainly Hyp1 and Hyp3. The transcript levels of HYP1 in maize leaves and HYP3 gene in liquid culture were similar to that of β tubulin-encoding gene as an internal control. The transcript level of HYP4 was highest in maize leaves, but that of HYP2 did not differ significantly among the three conditions. The transcript levels of all hydrophobin genes, except for HYP2, differed significantly among the culture conditions, suggesting that each hydrophobin may be used differently under different conditions.

The single disruption of each hydrophobin gene in B. maydis was confirmed by PCR (Supplementary Figures S1C–F). A class II hydrophobin-deficient strain (hyp2Δhyp3Δhyp4) and a strain deficient in all four hydrophobins (Δhyp1Δhyp2Δhyp3Δhyp4) were also generated (Supplementary Figure S1A, B, G–I). All strains were inoculated on VJA and incubated at 25°C for 7 days. Radial growth was similar in all the disruptants and the wild-type strain (Figure 3; Table 1). Conidia of all strains were harvested from these colonies and analyzed. The conidia of the wild-type strain were spindle-shaped and composed of cells divided by multiple septa. No considerable differences in conidial morphology or the number of conidia were observed between the disruptants and the wild-type strain (Figure 3, Table 1). These results indicate that the loss of hydrophobins in this fungus does not affect normal radial growth and conidiation.

All strains were cultured on CMA in the absence or presence of stress-inducing compounds (CFW, NaOH, or SDS). The decrease in radial growth of all disruptants on each medium was not significantly different from those of the wild-type strain (Figure 4; Supplementary Figure S4). These results showed that hydrophobins in B. maydis do not contribute to these types of stress tolerance.

To investigate whether B. maydis hydrophobins are involved in pathogenicity, we performed a pathogenicity assay on maize leaves. Lesions formed by each single disruptant were similar to those formed by the wild-type strain (Figure 5A; Table 1). Similarly, no significant difference was observed in the size of lesions among the triple disruptant, quadruple disruptant, and the wild-type strain (Figure 5B; Table 1). These results show that the lack of hydrophobins does not affect the pathogenicity of B. maydis.

In the cross between the wild-type and albino strains, black pseudothecia and light tawny pseudothecia were formed along their contact line (Figure 6). When all disruptants were crossed with the albino strain, pseudothecia of black and light tawny colors were formed; therefore, the loss of hydrophobins did not affect the formation of female organs (Figure 6). To investigate whether hydrophobins are involved in ascus and ascospore development, pseudothecia of both colors were harvested and inspected. In all crosses with the albino strain, asci containing ascospores were formed in pseudothecia of both colors (Figure 6). When the quadruple disruptant was crossed with itself, asci containing ascospores were formed normally in pseudothecia (Supplementary Figure S5). These results suggested that hydrophobins in B. maydis are dispensable for sexual reproduction.

To investigate whether hydrophobins play a role in B. maydis colony hydrophobicity, solution containing 0.2% SDS and 50 mM EDTA was dropped on mycelial disks (Figure 7). The Δchk1 strain, disruptant of the CHK1 gene encoding MAPK, was used as the control because its colonies are significantly less hydrophobic than those of the wild type (Degani et al., 2013). The droplet on the mycelial disk of Δchk1 strain changed its shape after 10 min and disappeared completely after 20 min. The droplet on the wild-type strain remained unchanged after 30 min. In all disruptants, including the quadruple one, droplets also remained unchanged after 30 min (Figure 7). These results showed that B. maydis hydrophobins are not required for colony hydrophobicity.

From the above results, we hypothesized that the hydrophobins of B. maydis lost their ability to provide hydrophobicity. To test this hypothesis, we investigated whether B. maydis hydrophobin could restore the hydrophobicity of a hydrophobin-deficient mutant of A. nidulans. Aspergillus nidulans with a disruption of the dewA gene encoding a class I hydrophobin shows significant mycelial wettability (Grünbacher et al., 2014). First, we generated a dewA gene disruptant strain (Supplementary Figure S2A, B). The droplet on the CNT strain remained unchanged after 5 min, but the droplet on the ΔdewA strain disappeared completely after 2 min (Figure 8A). This result is consistent with previous studies (Stringer and Timberlake, 1995; Grünbacher et al., 2014). Next, we generated the dewA^comp strain by reintroducing the wild-type dewA gene into the ΔdewA strain and confirmed the reintroduction by PCR (Supplementary Figure S2C, D). The hydrophobicity of the dewA^comp strain was similar to that of the CNT strain (Figure 8A).

We also constructed complemented strains, in which the B. maydis hydrophobin genes under the control of the dewA promoter were integrated into the ΔdewA strain, and obtained one strain of HYP1^comp and HYP2^comp each, three HYP3^comp strains and two HYP4^comp strains (Supplementary Figure S2C, D). While the colony growth of the HYP1^comp and HYP4^comp strains was similar to that of the CNT strain, HYP2^comp and HYP3^comp strains showed slower colony growth and an abnormal colony morphology compared to the CNT strain (Figure 8B; Supplementary Figure S6A). The expression analysis revealed that B. maydis hydrophobin expression in each complemented strain was equivalent to or even exceeded the expression level of dewA gene in the CNT strain (Supplementary Figure S6B). Similarly to the dewA^comp strain, the HYP1^comp strain, complemented with class I hydrophobin HYP1 gene, regained its colony hydrophobicity (Figure 8A). Interestingly, although the degree of hydrophobicity varied, the class II hydrophobin-complemented strains (HYP2^comp, HYP3^comp, and HYP4^comp) also showed increased hydrophobicity compared to the ΔdewA strain (Figure 8A; Supplementary Figure S6C). This result showed that the hydrophobins of B. maydis have not lost their ability to confer colony hydrophobicity.