Strains are listed in Table 1. Czapek-Dox (CD) medium was used as the standard culture, as described previously (Fujioka et al., 2007; Miyazawa et al., 2018).

We newly constructed agsA^OE and agsB^OE strains for this study. To generate agsA^OE, pAPyT-agsA plasmids (Miyazawa et al., 2018) were digested with NotI and transformed into a disrupted agsB (ΔagsB) strain (Supplementary Figure 1A). Correct integration of agsA overexpression cassettes was confirmed by PCR (Supplementary Figure 1B). To generate agsB^OE, the disrupted agsA (ΔagsA) strain was first generated using the Cre/loxP marker recycling system (Zhang et al., 2017a). The pAPG-cre/DagsA plasmid (Miyazawa et al., 2018) was digested with EcoRI and transformed into the ABPU1 (argB⁺) strain. Candidate strains were selected on CD medium without uridine and uracil, and then cultured on CD medium with uridine and uracil and 1% xylose to induce Cre expression (Supplementary Figure 1C). Strains that required uridine and uracil were isolated, and then replacement of the agsA gene was confirmed by PCR (Supplementary Figure 1D). The pAPyT-agsB plasmid was digested with NotI and transformed into the ΔagsA strain (Supplementary Figure 1E). Correct integration of agsB overexpression cassettes was confirmed by PCR (Supplementary Figure 1B).

The amyD^OE strain was constructed by replacing the native promoter with the constitutive tef1 promoter. The sequences of the primers are listed in Supplementary Table 1. To generate amyD^OE, the plasmid pAPT-amyD was constructed (Supplementary Figure 2A). The 5′-non-coding region (amplicon 1) and the coding region (amplicon 2) of amyD were amplified from A. nidulans ABPU1 genomic DNA. The pyrG marker (amplicon 3) was amplified from the pAPG-cre/DagsA plasmid. The tef1 promoter (amplicon 4) was amplified from the pAPyT-agsB plasmid. The four amplicons and a SacI-digested pUC19 vector were fused using an In-Fusion HD Cloning Kit (Clontech Laboratories, Inc., Mountain View, CA, USA). The resulting plasmid was digested with SacI, and transformed into the ABPU1 (argB⁺), agsA^OE, and agsB^OE strains (Supplementary Figure 2B). Correct integration of the cassette was confirmed by PCR (Supplementary Figure 2C).

In the first round of PCR, gene fragments containing the 5′-non-coding region (amplicon 1) and the coding region (amplicon 2) of amyD were amplified from ABPU1 genomic DNA, and the pyrG gene (amplicon 3) was amplified from A. oryzae genomic DNA (Supplementary Figure 2D). The three resulting fragments were gel-purified and fused into a disruption cassette in the second round of PCR. The resulting PCR product was gel-purified and transformed into the ABPU1 (argB⁺), agsA^OE, and agsB^OE strains (Supplementary Figure 2E). Replacement of the amyD gene was confirmed by PCR (Supplementary Figure 2F).

The sequences of the primers are listed in Supplementary Table 1. A GPI-anchor modification site, the ω-site, was predicted with the GPI Prediction Server v. 3.0 (https://mendel.imp.ac.at/gpi/gpi_server.html), and the best score for the ω-site was Asn535 of AmyD. To remove the GPI anchor of AmyD, 54 nucleotides corresponding to the 18 amino acid residues from Asn535 in AmyD were deleted from the authentic amyD gene (Supplementary Figure 3A). To create complementary genes that have full-length open reading frames of either amyD or the gene without the GPI anchor–coding region, the plasmids pAHT-amyD, pAHdPT-amyD, pAHT-amyD(ΔGPI), and pAHdPT-amyD(ΔGPI) were first constructed (Supplementary Figure 3A). To construct pAHT-amyD, primers IF-Ptef1-hph-Fw and IF-amyD-up-hph-Rv were amplified by PCR using pAPT-amyD as a template (amplicon 1). The hygromycin-resistance gene hph (amplicon 2) was amplified with primers 397–5 and 397–3 from pSK397 (Krappmann et al., 2006). The two amplicons were fused using a NEBuilder HiFi DNA Assembly kit (New England Biolabs, Ipswich, MA, USA) according to the manufacturer's instructions. Then, to delete the GPI anchor–encoding region of amyD, PCR amplification was performed with primers ANamyD-dGPI-Fw and ANamyD-dGPI-Rv from the resulting pAHT-amyD plasmid with PrimeSTAR Max DNA Polymerase (Takara Bio Inc., Kusatsu, Japan). The amplified fragment was transformed into DH5α competent cells, and the pAHT-amyD(ΔGPI) plasmid was obtained (Supplementary Figure 3A). To construct pAHdPT-amyD, the first half (amplicon 1) and the second half (amplicon 2) of pyrG were amplified from A. oryzae genomic DNA. The fragment containing hph, tef1 promoter, and amyD (amplicon 3) was amplified from pAHT-amyD. The three amplicons were fused using a NEBuilder kit. For pAHdPT-amyD(ΔGPI) construction, the fragment containing hph, tef1 promoter, and amyD lacking its GPI anchor–coding region (amplicon 3′) was amplified from pAHT-amyD(ΔGPI). The three amplicons and the SacI-digested pUC19 vector were fused using an In-Fusion HD Cloning Kit (Supplementary Figure 3B). The pAHdPT-amyD and pAHdPT-amyD(ΔGPI) plasmids were digested with SacI and transformed into the ΔamyD and agsB^OE ΔamyD strains (Supplementary Figure 3C). Correct integration of the cassettes was confirmed by PCR (Supplementary Figure 3D).

Cell walls were fractionated as previously described with some modification (Miyazawa et al., 2018). Mycelia cultured for 24 h in CD medium were collected by filtering through Miracloth (Merck Millipore, Darmstadt, Germany), washed with water, and freeze-dried. The mycelia were then pulverized in a MM400 bench-top mixer mill (Retch, Haan, Germany). The powder (1 g) was suspended in 25 mL of chloroform–methanol (3:1 vol/vol) and stirred at room temperature for 12 h to remove the total polar lipid content of the mycelial cells. The mixture was centrifuged (10,000 × g, 10 min). The residue was suspended in chloroform–methanol, and the delipidizing procedure was repeated. Then the de-polar lipid residue was suspended in 40 mL of 0.1 M Na phosphate buffer (pH 7.0), and cell-wall components were fractionated by hot-water and alkali treatments, as described previously (Miyazawa et al., 2018). Hot-water–soluble, alkali-soluble, and alkali-insoluble fractions were obtained from this fractionation, and the alkali-soluble fraction was further separated into a fraction soluble in water at neutral pH (AS1) and an insoluble fraction (AS2). The monosaccharide composition of AS2 fractions was quantified according to Miyazawa et al. (2018).

To obtain mycelia cultured for 16 h, conidia (final conc. 5.0 × 10⁵/mL) were inoculated into 200 mL CD medium and rotated at 160 rpm at 37°C. The mycelia were collected and fractionated as described above.

The AS2 fraction of each strain (50 mg) was suspended in 1 mL of 1 M NaOH/D₂O and dissolved by vortexing. One drop of DMSO-d₆ (deuterated dimethyl sulfoxide) was then added to each fraction and the solutions were centrifuged (3,000 × g, 5 min) to remove insoluble debris. ¹³C NMR spectra of the supernatants were obtained using a JNM-ECX400P spectrometer (JEOL, Tokyo, Japan) at 400 MHz at 35°C (72,000 scans). Chemical shifts were recorded relative to the resonance of DMSO-d₆.

The MM of alkali-soluble glucan was determined by gel permeation chromatography (GPC) according to the methods of Puanglek et al. (2016), with some modification. A GPC-101 system (Showa Denko Co. Ltd., Tokyo, Japan) with an ERC-3125S degasser (Showa Denko) and an RI-71S refractive index detector (Showa Denko) was used for the measurement. It was fitted with a GPC KD-G 4A guard column (Showa Denko) and a GPC KD-805 column (8.0 × 300 mm; Showa Denko). The eluent was 20 mM LiCl in N, N-dimethylacetamide (DMAc), and the flow rate was 0.6 mL/min at 40°C. The detector was normalized with polystyrene standards (SM-105; Showa Denko). With SmartChrom software (Jasco, Tokyo, Japan), the GPC profile was divided into virtual time slices (n_i) with the height of each virtual slice from the base line (H_i) corresponding to a certain MM (M_i) obtained by calibrating the column. From these values, the number-average MM (M_n) and weight-average MM (M_w) were calculated as:

Polydispersity was calculated as M_w/M_n.

Smith degradation of the alkali-soluble glucan was performed as described (Miyazawa et al., 2018). In brief, the AS2 fraction (20 mg) was suspended in 0.1 M acetate buffer (pH 3.9), oxidized with sodium periodate, reduced with sodium borohydride, hydrolyzed with trifluoroacetic acid, and freeze-dried. These procedures resulted in selective hydrolyzing of the 1,4-linked glucose residues, which contain vicinal hydroxyl groups, but not the 1,3-glucose residues in alkali-soluble glucan. The Smith-degraded sample was dissolved in DMAc containing LiCl for GPC analysis.

Mycelial cells cultured for 16 h in CD liquid medium were dropped on a glass slide and dried at 55°C for 15 min. The cells were fixed and labeled with α-1,3-glucan-binding domain-fused green fluorescent protein (AGBD-GFP) (Suyotha et al., 2013) for α-1,3-glucan, fluorophore-labeled antibody for β-1,3-glucan, and fluorophore-labeled lectin for chitin. The cells were then imaged by confocal scanning microscopy as described (Miyazawa et al., 2018). Enzymatic digestion of β-1,3-glucan in the hyphal cells was performed as described (Miyazawa et al., 2018).

Culture broth in CD liquid medium incubated for 24 h was filtered through Miracloth. Proteins in the supernatant were precipitated with trichloroacetic acid, separated by SDS-PAGE, and then transferred onto a polyvinylidene difluoride membrane. The membrane was blocked with the polyvinylidene difluoride blocking reagent for Can Get Signal (Toyobo). Antibodies of rabbit IgG against AmyD were developed with synthesized peptides (NH₂-C+SGERAGELDVPMSK-COOH) (Eurofins Genomics, Tokyo, Japan) and used as the primary antibody, diluted with Can Get Signal (Toyobo). Antibody binding was visualized using a horseradish peroxidase–conjugated goat anti-rabbit IgG secondary antibody (Pierce Biotechnology, Rockford, IL, USA) and an ImmunoStar LD chemiluminescent substrate (Fujifilm Wako Pure Chemical Corp., Osaka, Japan).

Supernatant obtained from CD liquid medium cultured for 24 h was concentrated 167 times and buffer-exchanged to 10 mM Tris·HCl (pH 8.0) in Nanosep Centrifugal Devices 10K (Pall, Port Washington, NY, USA). A 20-μL mixture containing 1 mM para-nitrophenyl (pNP)-α-maltopentaoside and 1 mM acarbose (Fujifilm Wako Pure Chemical Corp.), and 5 μL of the concentrated culture supernatant in 50 mM acetic acid/sodium acetate buffer (pH 5.5) was incubated for 20 min at 40°C. Samples (3 μL) were withdrawn from the reaction mixture and immediately inactivated by adding 40 μL of methanol. pNP-α-Maltopentaoside was prepared by following the method of Usui and Murata (1988). Then, 157 μL of water was added to each sample solution, which was analyzed by HPLC with a Jasco Intelligent System liquid chromatograph (Jasco). The bound material was eluted with 20% methanol at a flow rate of 1.0 mL/min at 40°C. The elution profiles were detected at 300 nm with a Unison UK C-18 column (4.6 × 250 mm, Imtakt, Kyoto, Japan).

We constructed amyD^OE and ΔamyD strains by introducing the amyD cassettes for overexpression and disruption into the wild-type, agsA^OE, and agsB^OE strains (Supplementary Figure 2). The expression level of amyD in each strain was quantified in hyphal cells. Whereas, each disrupted strain (ΔamyD, agsA^OE ΔamyD, and agsB^OE ΔamyD) showed scarce amyD expression, each overexpressing strain (amyD^OE, agsA^OE amyD^OE, and agsB^OE amyD^OE) showed significantly higher amyD expression than their parental strain (Figure 1).

There was no significant difference in radial growth among the strains grown on agar plates for 5 days (Supplementary Figure 4). In liquid culture, the wild-type and ΔamyD strains formed tightly aggregated hyphal pellets; however, the hyphae of the amyD^OE strain were almost fully dispersed (Figure 2). He et al. reported that the phenotype of their amyD^OE strain resembles that of the ΔagsB strain in A. nidulans (He et al., 2014), which is consistent with our results (Figure 2). In agreement with our previous results (Miyazawa et al., 2018), the agsA^OE and agsB^OE strains formed, respectively, loosely and tightly aggregated pellets (Figure 2). Disruption of amyD did not affect the phenotypes of the agsA^OE and agsB^OE strains (Figure 2). Also, overexpression of amyD scarcely affected the phenotypes of the agsA^OE and agsB^OE strains (Figure 2).

Cell-wall components of each strain were fractionated by a hot water–alkali treatment method, each fraction was weighed, and the monosaccharide composition of the AS2 fraction was quantified. The amount of glucose in the AS2 fraction was significantly lower in the amyD^OE strain than in the wild-type strain (Figure 3; P < 0.05). That in the ΔamyD strain was similar to that in the wild-type strain (Figure 3). Those in the agsA^OE amyD^OE and agsA^OE ΔamyD strains, which were constructed from the parental strain agsA^OE, were almost the same (Figure 3). It was significantly lower in the agsB^OE amyD^OE strain than in the agsB^OE and agsB^OE ΔamyD strains (Figure 3; P < 0.05). These results indicate that AmyD acts to decrease the amount of alkali-soluble glucan in the wild-type and agsB^OE strains, but not in the agsA^OE strain, even when amyD is overexpressed.

By ¹³C NMR analysis, the primary component in the AS2 fraction of the wild-type, amyD^OE, and ΔamyD strains was found to be α-1,3-glucan, suggesting that amyD did not affect the primary components of alkali-soluble glucan (Supplementary Figure 5). To reveal whether the MM of alkali-soluble glucan was affected by disruption or overexpression of amyD, we determined the MM of alkali-soluble glucan in each strain by GPC analysis. Polystyrene (MM, 13,900–3,850,000) was used as a standard molecule to calibrate the column for size exclusion analysis. Although M_w is used to assess the physical properties of a polymer, the calculation of M_w favors molecules with a larger MM. Since M_n is the average of the MM values of the individual macromolecules, here we use M_n as the MM of alkali-soluble glucan. The M_n of the alkali-soluble glucan was 1,260,000 ± 270,000 in the agsA^OE strain and 312,000 ± 5,000 in agsB^OE strain (Figures 4A,B; Table 2), consistent with our previous results (Miyazawa et al., 2018). Although the M_n of alkali-soluble glucan in the agsA^OE amyD^OE strain (1,110,000 ± 110,000) was similar to that in the parental (agsA^OE) strain, that of agsA^OE ΔamyD was significantly greater (2,250,000 ± 130,000) than that of agsA^OE (Figure 4A; Table 2; P < 0.05). In addition, the M_n of agsB^OE amyD^OE (140,000 ± 8,000) was significantly less than that of the parental (agsB^OE) strain (Figure 4B; Table 2; P < 0.05). The M_n of alkali-soluble glucan in agsB^OE ΔamyD (358,000 ± 19,000) was similar to that in agsB^OE (Figure 4B; Table 2). Lastly, the M_n of alkali-soluble glucan in the wild-type (2,280,000 ± 320,000) and ΔamyD (2,390,000 ± 400,000) was larger than that in agsB^OE (312,000 ± 5,000; Figure 4C; Table 2). The amyD^OE strain had a primary peak at around 17 min (Mn1, 32,900 ± 300) and a secondary peak at 11 min (Mn2, 2,210,000 ± 700,000). These results suggest that AmyD degraded the alkali-soluble glucan eluted around 11 min to produce alkali-soluble glucan with a smaller MM (Figure 4C; Table 2).

Although the alkali-soluble glucan in the wild-type strains was synthesized mainly by AgsB, its MM was larger than that in the agsB^OE strain (Table 2). Additionally, when agsB was overexpressed, the amount of α-1,3-glucan was three times that in the wild-type (Figure 3). We supposed that some unknown glycosyl modification enzymes may contribute to the increase in MM of α-1,3-glucan in the wild-type, and that because the agsB^OE strain produces more α-1,3-glucan, there is little modification by the unknown enzymes. Therefore, we determined the MM of alkali-soluble glucan extracted from 16-h cultured mycelia, which should be less affected by the modification enzyme than the 24-h cultured mycelia (He et al., 2017). Unexpectedly, the M_n of the alkali-soluble glucan in the mycelia cultured for 16 h was 1,980,000 ± 320,000, which was similar to that in the mycelia cultured for 24 h (1,930,000 ± 280,000; Supplementary Figure 6; Supplementary Table 2). We then evaluated the MM of alkali-soluble glucan in A4, which is the Glasgow wild-type of A. nidulans (Table 1), and found it had M_n = 2,224,000 ± 390,000 (Supplementary Table 3), which is similar to that in the wild-type strain.

To validate whether the degree of polymerization of α-1,3-glucan subunits in the alkali-soluble glucan was altered when the MM was changed by amyD disruption or overexpression, we applied Smith degradation to the alkali-soluble glucan from each strain to selectively cleave 1,4-linked glucan, and then determined the MM by GPC. One subunit of α-1,3-glucan in the alkali-soluble glucan is composed of ≈200 glucose residues (Choma et al., 2013; Miyazawa et al., 2018). The Smith-degraded alkali-soluble glucan in each strain had almost the same MM, equivalent to 300–400 glucose residues (Supplementary Figure 7; Supplementary Table 4), which suggests that AmyD activity does not decrease the degree of polymerization of the glucose residues in each α-1,3-glucan subunit.

We previously revealed that spatial localization of α-1,3-glucan in the cell wall changes according to its MM (Miyazawa et al., 2018); α-1,3-glucans in agsB^OE cells are localized in the outer layer in the cell wall, whereas most of those in the agsA^OE cells are masked by a β-1,3-glucan layer. In this study, disruption or overexpression of amyD altered the MM of alkali-soluble glucan (Figure 4; Table 2); therefore, we analyzed whether this alteration affected the spatial localization of α-1,3-glucan in the cell wall. In agreement with previous results (Miyazawa et al., 2018), the α-1,3-glucans with AGBD-GFP labeling showed clearly in the wild-type and agsB^OE cells, but only weakly in agsA^OE cells (Figure 5). The ΔamyD and amyD^OE cells were also labeled with AGBD-GFP (Figure 5); fluorescent intensity in amyD^OE was relatively low, which might be caused by a decrease in the amount of alkali-soluble glucan in the cell wall of amyD^OE cells. The labeling with AGBD-GFP in agsA^OE amyD^OE and agsA^OE ΔamyD cells was weak, as was that in the cells of the parental agsA^OE strain (Figure 5). The agsB^OE ΔamyD cells were clearly labeled with AGBD-GFP, as in the parental agsB^OE (Figure 5). The AGBD-GFP labeling was slightly weaker in agsB^OE amyD^OE than in agsB^OE, which might be attributable to a decrease in the amount of α-1,3-glucan. After treatment with β-1,3-glucanase, α-1,3-glucans of the hyphal cells in agsA^OE, agsA^OE amyD^OE, and agsA^OE ΔamyD cells were clearly labeled with AGBD-GFP (Supplementary Figure 8), suggesting that these strains have α-1,3-glucan in the inner layer of the cell wall in their hyphal cells. Taken together, these findings indicate that disruption or overexpression of amyD gene scarcely affected the spatial localization of α-1,3-glucan in the cell wall.

AmyD is thought to contain a GPI anchor at the C-terminal region. Fungal GPI anchor–type proteins are transferred from the plasma membrane to the cell wall by the activity of the GH76 family (Vogt et al., 2020). We speculated that localization in the cell wall would be essential for AmyD to reach the substrate, alkali-soluble glucan, so we constructed overexpression strains of amyD with and without the GPI-anchor site. Because we noticed that overexpression of amyD alters the phenotype or the alkali-soluble glucan, we used ΔamyD and agsB^OE ΔamyD strains as hosts for the amyD^OE strains. The hyphae of ΔamyD formed pellets in shake-flask culture (Figure 6). Those of ΔamyD-amyD^OE were dispersed, as in amyD^OE (Figure 6). Those of ΔamyD-amyD^OE(ΔGPI) formed pellets, although the form was slightly different from that in the parental strain (Figure 6). Those of agsB^OE ΔamyD, agsB^OE ΔamyD-amyD^OE, and agsB^OE ΔamyD-amyD^OE(ΔGPI) formed similar pellets (Figure 6). Although the ΔamyD-amyD^OE hyphae had less AS2-Glc (1.13% ± 0.21%) than ΔamyD (5.68% ± 0.25%), the amount was restored in ΔamyD-amyD^OE(ΔGPI) hyphae (5.17% ± 0.46%; Figure 7A). These results suggest that the GPI anchor of AmyD has an important negative effect on α-1,3-glucan biosynthesis. The hyphae of agsB^OE ΔamyD-amyD^OE had marginally less AS2-Glc (16.2% ± 0.6%) than those of agsB^OE ΔamyD (17.6% ± 0.3%) and agsB^OE ΔamyD-amyD^OE(ΔGPI) (16.7% ± 0.5%; Figure 7B). We then evaluated the MM of alkali-soluble glucan in the cells of agsB^OE ΔamyD, agsB^OE ΔamyD-amyD^OE, and agsB^OE ΔamyD-amyD^OE(ΔGPI). The M_n of the alkali-soluble glucan in agsB^OE ΔamyD-amyD^OE cells (174,000 ± 8,000) was smaller than that in agsB^OE ΔamyD (270,000 ± 8,000; Figure 7C; Table 3; P < 0.05). The M_n of alkali-soluble glucan in agsB^OE ΔamyD-amyD^OE(ΔGPI) cells (349,000 ± 42,000) was similar to that in agsB^OE ΔamyD (Figure 7C; Table 3). These results suggest that the GPI anchor of AmyD is also important for regulating the MM of alkali-soluble glucan.

Western blotting showed that secretion of AmyD in the culture supernatant could be detected only in the ΔamyD-amyD^OE (ΔGPI) strain (Supplementary Figure 9A). Because AgtA in A. niger has relatively high transglycosylation activity on donor substrates with maltooligosaccharides (Van Der Kaaij et al., 2007), we evaluated the α-amylase activity in concentrated culture supernatants with pNP-α-maltopentaoside. Although various products possibly produced by coexisting α-glucosidase were detected in ΔamyD and both amyD-overexpressing strains, pNP-α-maltooctaoside (probably a transglycolylation product of AmyD) was detected only in the ΔamyD-amyD^OE (ΔGPI) strain (data not shown). A histidine residue in region I, which is a highly conserved region of α-amylases, is substituted to the asparagine residue in AmyD and proteins encoded by orthologs of amyD (Van Der Kaaij et al., 2007). Substitution of the histidine residues in α-amylase increases the inhibitory constant (Ki) of a representative α-glucosidase inhibitor, acarbose (Svensson, 1994). Therefore, we evaluated the α-amylase activity in concentrated culture supernatants in the presence of acarbose. As expected, the hydrolysis product of pNP-α-maltopentaoside was hardly detected in ΔamyD (Supplementary Figure 9B). pNP-α-maltoside and pNP-α-maltooctaoside were clearly detected in the supernatant of ΔamyD-amyD^OE (ΔGPI), but scarcely in that of ΔamyD-amyD^OE (Supplementary Figure 9B). These results suggest that although enzymatically active AmyD is secreted into the culture supernatant, it cannot decrease the MM of α-1,3-glucan, leading us to suppose that active AmyD needs to be localized on the plasma membrane or in the cell wall space to regulate the MM of α-1,3-glucan.