The C. neoformans strains used in this study are listed in Supplementary Table S1. The C. neoformans var. grubii H99 strain was obtained from the Fungal Genetics Stock Center (Kansas City, USA).

The strains were cultured on YPD medium (2% w/v glucose, 2% w/v peptone, and 1% w/v yeast extract). To produce GXMGal, the strains were cultured in 10% Sabouraud liquid medium [0.4% w/v glucose, 0.1% w/v peptone, 0.1% w/v tryptone, and 50 mM 3-morpholinopropanesulfonic acid (pH 7.3)].

CAP59 (CNAG_00721) was disrupted in C. neoformans H99 by inserting NEO using the CRISPR/Cas9 system (Huang et al., 2022). A gene replacement cassette encompassing a 50-bp homology arm at the 5′ end and a 50-bp homology arm at the 3′ end of CAP59 was amplified by recombinant PCR using pNEO_6xHA (So et al., 2017) as a template and the cap59-del-F–cap59-del-R primer pair (Supplementary Table S2). The Cas9 expression cassette was amplified by PCR using pBHM2403 as a template and the M13-F–M13-R primer pair. The sgRNA expression cassette was amplified in two PCR steps. An sgRNA scaffold containing a 20-bp target sequence and U6 promoter and an sgRNA scaffold containing a 20-bp target sequence and U6 terminator was amplified by PCR using pBHM2329 as a template and the M13-F–cap59-gRNA-R1 and cap59-gRNA-F2–M13-R primer pairs. The PCR fragments were combined by fusion PCR using the U6-F–U6-R primer pair. All PCR fragments were introduced into C. neoformans by electroporation using Gene Pulser II (Bio-Rad, Hercules, CA), yielding cap59Δ strain. Transformants were selected using YPD agar plates supplemented with 200 μg/mL G418. The introduction of NEO into each locus was confirmed by PCR using the cap59-comf-F–cap59-comf-R primer pair (Supplementary Figure S1).

GGT1 (CNAG_01385), GGT2 (CNAG_01050), and GGT3 (CNAG_06918) were disrupted in C. neoformans H99 and cap59Δ strains by inserting HYG. Gene replacement cassettes encompassing a 50-bp homology arm at the 5′ end and a 50-bp homology arm at the 3′ end of GGT1, GGT2, and GGT3 were amplified by recombinant PCR using pHYG_GFP (So et al., 2017) as a template and the xxxx-del-F–xxxx-del-R (where “xxxx” indicates GGT1, GGT2, or GGT3) primer pair (Supplementary Table S2). An sgRNA scaffold containing a 20-bp target sequence and U6 promoter and an sgRNA scaffold containing a 20-bp target sequence and U6 terminator were amplified by PCR using pBHM2329 as a template and the M13-F–xxxx-gRNA-R1 and xxxx-gRNA-F2–M13-R primer pairs. All PCR fragments were introduced into C. neoformans H99 and cap59Δ by electroporation using Gene Pulser II (Bio-Rad, Hercules, CA), yielding ggt1Δ, ggt2Δ, ggt3Δ, cap59Δggt1Δ, cap59Δggt2Δ, and cap59Δggt3Δ strains. Moreover, transformants were selected on YPD agar plates supplemented with 200 μg/mL hygromycin B. The introduction of HYG into each locus was confirmed by PCR using the xxxx-comf-F–xxxx-comf-R primer pair (Supplementary Figure S1).

GGT3 was disrupted in C. neoformans cap59Δggt2Δ strain by inserting NAT. A gene replacement cassette encompassing a 50-bp homology arm at the 5′ end and a 50-bp homology arm at the 3′ end of GGT3 was amplified by recombinant PCR using pNAT_mCherry (So et al., 2017) as a template and the ggt3-del-F–ggt3-del-R primer pair (Supplementary Table S2). Transformants were selected on YPD agar plates supplemented with 100 μg/mL nourseothricin sulfate. The introduction of NAT into each locus was confirmed by PCR using the ggt3-comf-F–ggt3-comf-R primer pair (Supplementary Figure S2).

For complementation, analysis of GGT2 using a gene replacement cassette, encompassing a homology arm at the 5′ end of GGT2, wild-type GGT2 containing 3′-UTR, hygromycin B resistance gene (hph), and a homology arm at the 3′ end of GGT2, was constructed by recombinant PCR using H99 genomic DNA; pNAT_mCherry as a template; and the ggt2-comp-1–ggt2-comp-2, ggt2-comp-3–ggt2-comp-4, and ggt2-comp-5–ggt2-comp-6 primer pairs. The resultant DNA fragment was amplified with the ggt2-comp-1–ggt2-comp-6 primer pair. An sgRNA scaffold containing a 20-bp target sequence and U6 promoter and an sgRNA scaffold containing a 20-bp target sequence and U6 terminator were amplified by PCR using pBHM2329 as a template and the M13-F–HYG-gRNA-R1 and HYG-gRNA-F2–M13-R primer pairs. All PCR fragments were introduced into C. neoformans ggt2Δ and cap59Δggt2Δ by electroporation. Transformants were selected on YPD agar plates supplemented with 100 μg/mL nourseothricin sulfate. The introduction of NAT into each locus was confirmed by PCR using the ggt3-comp-comf-F–ggt3-comp-comf-R primer pair (Supplementary Figure S1).

C. neoformans strains were cultured in 3 mL of YPD liquid medium at 30°C for 24 h. The cells were then collected by centrifugation and washed three times with sterile phosphate-buffered saline (PBS), suspended in 2 mL of 10% Sabouraud liquid medium, and incubated at 30°C for 24 h to induce capsule production. The culture medium was diluted with PBS mixed at a ratio of 1:1 with India ink (Syogeikuretake Shikon BB1-18; Kuretake Co., Ltd. Nara, Japan) and incubated for 15 min. Images of stained cells were acquired using a microscope equipped with a digital camera. The diameters of the cells and capsules were measured immediately (50 cells), and the average diameter was calculated.

Purification of GXMGal was performed as described (Rocha et al., 2015). Briefly, cap59Δ strains were cultivated in 1 L of 10% Sabouraud medium at 30°C with shaking (160 rpm) for 5 days. The culture supernatant was collected by centrifugation, mixed with an equal volume of phenol:chloroform, and centrifuged. The collected supernatant was dialyzed overnight at 4°C using a Visking Tube (Nihon Medical Science, Inc. Japan). The polysaccharides were powdered by lyophilization. GXMGal powder was dissolved in 3% cetyltrimethylammonium bromide solution in 1% borate at pH 9.5. The GM fraction was collected as precipitate, washed with 75% ethanol, dialyzed with water, and lyophilized.

Glycosidic linkages were analyzed as previously described (Klutts and Doering, 2008; Katafuchi et al., 2017). Briefly, GXMGal was separately dissolved in dimethyl sulfoxide, followed by NaOH addition. After stirring for 3 h, methyl iodide was added, and the suspension was stirred for 24 h. The methylated products were extracted in chloroform and washed using dH₂O. Then, the methylated samples were hydrolyzed using 2 M trifluoroacetic acid, reduced, and acetylated. The partially methylated alditol acetates were analyzed by GC–MS using a capillary column (30 m × 0.25 mm; DB-5, Agilent, CA) with helium as the carrier gas at a gradient temperature program of 210°C–260°C at 5°C/min. The GC–MS analyses were performed using a JMS-K9 mass spectrometer (JEOL, Tokyo, Japan). NMR experiments were performed as previously described (Klutts and Doering, 2008; Katafuchi et al., 2017). The NMR spectra were recorded using a JNM-LA600 spectrometer (JEOL) at 45°C. Proton and carbon chemical shifts were referenced relative to internal acetone at δ 2.225 and 31.07 ppm, respectively.

First, we searched the C. neoformans H99 genome for candidate genes encoding enzymes that transfer the β-galactosyl residue to the hydroxyl group at position 3 of the α-galactosyl residue in GXMGal. S. pombe Pvg3 is a glycosyltransferase that exhibits similar enzymatic activity. Therefore, we selected candidate genes by PSI-BLAST search using Pvg3 as a query. Two homologous proteins were selected (CNAG_01050 and CNAG_01385). CNAG_01385 is Ggt1––an α-mannoside β-(1 → 6)-galactosyltransferase involved in GIPC biosynthesis in C. neoformans. Then, a BLASTp search was performed using CNAG_01050 as a query to select candidate genes. CNAG_06918 was selected in addition to CNAG_01385. Therefore, we named CNAG_01050 and CNAG_06918 as Ggt2 and Ggt3, respectively, the two α-galactoside β-(1 → 3)-galactosyltransferase candidates. Ggt2 and Ggt3 are members of the GT31 family. Ggt2 has an amino acid sequence homology of 23% (in the 434–645 amino acid region of Ggt1) and 33% (in the 122–320 amino acid region of Ggt3) with Ggt1 and Ggt3, respectively (Figure 1). Ggt1 does not have a transmembrane domain, but analysis using DeepLoc 2.0 (Thumuluri et al., 2022) predicted a transmembrane region at 338–360 aa. Ggt2 and Ggt3 were predicted to be Golgi-localized type II membrane proteins with one transmembrane region at N-terminal 35–57 and 21–43 aa, respectively, according to DeepLoc 2.0 (Figure 1).

To investigate the effect of loss of presumptive GT31 family glycosyltransferases on GXMGal biosynthesis, we constructed GGT1, GGT2, and GGT3 triple disruptants and GGT2 and GGT3 double disruptants using cap59Δ as parental strain. All strains were cultured in 10% Sabouraud medium for 5 days, and GXMGal was purified from the culture supernatant. Purified GXMGal was analyzed by 1H-NMR (Figure 2). Chemical shifts at ~5.22, 5.16, and 4.99 ppm in cap59Δ, cap59Δggt1Δ, and cap59Δggt3Δ strains indicated -(1 → 3)-Manα-, terminal Manα, and -(1 → 3)-Manα-(2 ← 1)-galactomannan side chain of GXMGal, respectively (Klutts and Doering, 2008). In cap59Δggt2Δ and cap59Δggt2Δggt3Δ, chemical shifts appeared at 4.98 ppm, indicating the presence of the -(1 → 6)-Galα- side chain. These results indicate that loss of GGT2 results in a reduction or loss of the galactomannan side chain of GXMGal.

To clarify the physiological role of GGT in C. neoformans cells, ggt1, ggt2, and ggt3 single-disruptant strains were constructed using H99 as the parental strain. As uge1Δ and ugt1Δ cannot supply UDP-Gal to the Golgi, they are deficient in GXMGal and GIPC biosynthesis and exhibit a Ts phenotype at 37°C (Moyrand et al., 2008; Li et al., 2017). ggt1Δ lacks GIPC and exhibits a Ts phenotype at 37°C. Therefore, we observed growth of GGT disruptants at 37°C (Figure 3). At 30°C, ggt1Δ showed slightly delayed growth compared with the wild-type; moreover, ggt2Δ and ggt3Δ showed similar growth compared with the H99 strain. By contrast, at 37°C, ggt1Δ and ggt2Δ showed dramatically delayed growth, indicating a Ts phenotype. The growth of ggt3Δ and H99 was similar. These results indicate that the glycan structure synthesized by Ggt2 is important for high-temperature stress tolerance in C. neoformans.

We examined the growth of ggt2Δ on media containing various drugs. ugt1 disruptant strains show sensitivity to NaCl, Congo red, H₂O₂, and sodium dodecyl sulfate (SDS) (Li et al., 2017). Therefore, we tested the sensitivity of ggt2Δ to these drugs (Figure 4). ggt2Δ was not significantly sensitive to any of the drugs. Conversely, the Ts phenotype of ggt2Δ was completely rescued by complementation with wild-type GGT2 and slightly rescued by 1 M sorbitol, which exerts high osmotic pressure. These results indicate that the galactomannan side chain of GXMGal biosynthesized by Ggt2 is important for high-temperature stress tolerance in C. neoformans.

To further analyze the ggt2Δ phenotype, the capsule structure was stained with India ink for microscopic observation (Supplementary Figure S3). Quantification of cell and capsule sizes revealed no significant differences between ggt2Δ and the wild-type strain. These findings indicated that Ggt2 absence did not affect GXM production.

To examine the role of Ggt2 in GXMGal biosynthesis, we complemented wild-type GGT2 in cap59Δggt2Δ and analyzed the structure of GXMGal. The structure of GXMGal produced by each strain was analyzed by ¹H-NMR (Figure 5). Sharp doublet peaks at 4.98 ppm of the α-galactan backbone seen in cap59Δggt2Δ disappeared due to GGT2 complementation, and the Man-derived chemical shift seen in wild-type GXMGal reappeared.

Methylation GC–MS analysis was used to investigate constituent sugars and the linkage mode of GXMGal produced by each strain (Table 1). In GXMGal produced by cap59Δggt2Δ, terminal Xyl (tXyl₁) residues were reduced to below the detection limit. Additionally, 4-substituted Gal (₄Gal₁) and 3,4-substituted Gal (_3,4Gal₁) (indicating β-Gal residues in the galactomannan side chain) and 3,6-substituted Gal (_3,6Gal₁) (indicating α-Gal residues attached to galactomannan side chains) were reduced to below detection limit. The ratio of 2,6-substituted Gal (indicating α-Gal with Galf side chain) and terminal Gal (tGal₁) and Galf (tGalf₁) increased. By contrast, the ratio of 3-substituted Gal (₃Gal₁) and 6-substituted Man (₆Gal₁), which are not present in GXMGal, increased drastically. Terminal Man (tMan₁), 3-substituted Man (₃Man₁), and 2,3-substituted Man (_2,3Man₁) decreased but did not disappear.

We analyzed the structure of GXMGal in detail by ¹³C-NMR (Figure 6). In GXMGal from cap59Δ, we detected a broad chemical shift at 100.6–105.1 ppm, possibly derived from position C-1 of Manα-(1 → 3)-, 2,3-O-substituted-Manα, Xylβ-(1 → 2)-, and Xylβ-(1 → 3)- in the galactomannan side chain (Klutts and Doering, 2008). These structures were suggested to be lost with the loss of GGT2. The chemical shifts at 78.5 and 77.1 ppm, which originate from position C-4 of 4-substituted Galβ and 3,4-substituted Galβ in wild-type GXMGal, disappeared in ggt2Δ. The chemical shift at 80.9 ppm originating from position C-3 of 3,6-substituted Galα disappeared in ggt2Δ. These results strongly support our hypothesis that Ggt2 catalyzes β-(1 → 3)-Gal transfer to α-galactoside––the initial reaction in the biosynthesis of the galactomannan side chain of GXMGal.

Sequences of Pvg3, Ggt1, Ggt2, and Ggt3 family proteins were used to construct an evolutionary phylogenetic tree (Figure 7). The data set for analysis was obtained from FungiDB using the amino acid sequences of S. pombe Pvg3 and C. neoformans Ggt1, Ggt2, and Ggt3 as search queries.¹ The protein sequences were clearly divided into the Pvg3, Ggt1, Ggt2, and Ggt3 clades (Figure 7). Ggt1 was widely distributed in Basidiomycota, including Ustilago, Coprinopsis, Schizophyllum, and Cryptococcus. Ggt2 was distributed in Pucciniomycetes and Tremellomycetes in Basidiomycota. Furthermore, Ggt3 was distributed only in Tremellomycetes in Basidiomycota.