The Schizosaccharomyces pombe and Saccharomyces cerevisiae strains used are listed in Table S1 (14, 18). Rich media YES (yeast extract plus supplements) and YPD (yeast extract, Bacto™ peptone, dextrose) as well as minimal defined media EMM (Edinburgh minimal medium) and SD (synthetic defined) medium were used as described previously (18). Phosphate-free minimal medium based on EMM contained 14.6 mM sodium acetate instead of disodium hydrogen phosphate and potassium hydrogen phthalate. For shutdown experiments using the nmt81 promoter, cells were restreaked twice on EMM plates containing 5 mg/L thiamine. Standard procedures were used for yeast transformation as well as for genetic DNA manipulations. All cloning procedures were carried out in the Escherichia coli host DH5α (Thermo Fischer Scientific, Rockford, IL, USA).

S. pombe strains CMY1 and CMY2: a disruption cassette replacing the entire open reading frame of SPAC637.06 by the kanMX gene was amplified from genomic DNA of the heterozygous SPAC637.06 mutant strain BG_1883 (Bioneer Cooperation, Daejeon, South Korea) using primers Oligo 1305 and Oligo1306. A mass of 5 µg of the resulting DNA fragment was used to transform the wild-type (WT) yeast strain FY527 or the oma1Δ mutant strain SBY90. Transformants that integrated the disruption cassette were selected on YES medium containing G418. Genomic DNA was prepared from the resulting transformants and successful targeting of SPAC637.06 was confirmed by PCR using primers Oligo1231 and Oligo1232. All oligonucleotide sequences used in this study are available upon request.

The thiamine repressible nmt81-oma2⁺ strain TWY16 was transformed with S. pombe genomic libraries contained in pURSP1 and pURSP2, respectively (22). Before transformation, cells were grown in medium lacking thiamine. After transformation, cells were plated on selective medium lacking uracil and containing thiamine, then incubated for three days at 32 °C. For plasmid isolation, individual clones were grown overnight in 3 mL of cultures and harvested. Pellets were resuspended in 100 µL of buffer containing 50 mM Tris-HCl, pH=8.0, 50 mM EDTA, 8 % sucrose and 5 % Triton X-100. Cells were disrupted with glass beads (d=0.5 mm) using a RiboLyser (Hybaid, Altrincham, UK). After separation from the glass beads, samples were incubated for 3 min in a boiling water bath, chilled on ice and cell debris was pelleted for 10 min at 20 000×g. From the supernatant, plasmid DNA was recovered by ethanol precipitation and retransformed in E. coli. Isolated plasmids were retransformed in TWY16 and three independent clones were tested to reconfirm their suppressor ability on restrictive plates containing thiamine. To identify the genes contained in the suppressor plasmid, genomic sequences were determined by Sanger sequencing using M13fwd and M13rev universal primers.

Plasmids pTW51-1 and pTW60 have been described previously (18).

Plasmid pI-1/9: Plasmid pI-1/9 was directly derived from the multicopy suppressor screen with a genomic DNA fragment and only the phx1⁺ (SPAC32A11.03c) coding sequence.

Plasmid pCM6: The plasmid pII-1/2 containing the entire open reading frame of mpg1⁺ (SPCC1906.01) and a truncated coding sequence of SPCC1906.02c was isolated from the multicopy suppressor screen. This plasmid was cut with PstI to remove the SPCC1906.02c-containing fragment and religated, resulting in plasmid pCM6.

Plasmid pII-2/17: The plasmid pII-2/17 was directly derived from the multicopy suppressor screen and contains a genomic DNA fragment with only the gmh5⁺ (SPAC637.06) coding sequence.

Plasmid pCM10 and pCM11: For overexpression of gmh5⁺ and a gmh5⁺-hemagglutinin (HA) epitope-tag fusion in S. pombe and S. cerevisiae, primer pairs Oligo1246 combined with Oligo1247 or Oligo1248 were used to amplify a 1.0 kb DNA fragment containing the gmh5⁺ coding sequence. The resulting PCR products were cut with NotI and SalI or NotI and XhoI and subcloned into pREP3adh (18) cut with the same enzymes. The resulting plasmids pCM10 and pCM11 encode for Gmh5p and a Gmh5p-HA-tag fusion protein under the transcriptional control of the alcohol dehydrogenase 1 (adh1) promoter.

Plasmid pCM12: The 1.5 kb coding region of the S. cerevisiae SUC2 gene was amplified by PCR from S. cerevisiae genomic DNA using Oligo1309 and Oligo1310. The primer pair was used to introduce a 5’ NotI restriction site and a 6× His epitope-tag and XhoI restriction site. The resulting PCR product was digested with NotI and XhoI and subcloned into pREP3adh cut with the same enzymes. The resulting plasmid encodes a Suc2-6xHis-3xHA fusion protein under the transcription control of the adh1 promoter.

Plasmid pGma12HA: The primer pair Oligo1297 and Oligo1299 was used to amplify a 1.1 kb PCR fragment containing the coding region of gma12⁺ excluding the stop-codon. S. pombe genomic DNA was used as a template. The resulting PCR product was subcloned into pREP3adh using NotI and XhoI restriction sites. The resulting plasmid encodes for a Gma12p-3xHA fusion protein driven by the adh1 promoter.

Assays were performed in 96-well microtiter plates in a final volume of 200 µL of YES medium or YPD medium for S. pombe or S. cerevisiae, respectively. Serial dilutions of caffeine (0 to 17 mM; Sigma-Aldrich, Merck, St. Louis, MO, USA), hygromycin (0 to 16.25 µg/mL; Sigma-Aldrich, Merck) and the PMT-inhibitor R3A-5a (0 to 32 µM; Merck KGaA, Darmstadt, Germany (23)) were prepared in EMM or SD medium and inoculated from a logarithmically growing culture to a final absorbance A_{600 nm}=0.02. The 96-well microtiter plates were incubated for 24 h at 30 °C and cell growth was determined by measuring the absorbance in a FLUOstar Omega (BMG Labtech, Ortenberg, Germany) microplate reader at 600 nm. All conditions were tested in triplicates and experiments were repeated twice.

Crude yeast membranes used in Western blot analysis and mannosyltransferase assays were prepared as described previously (24). In brief, a total of cells corresponding to a combined A_{600 nm} of 200 were harvested from a logarithmically growing culture and washed once with ice-cold buffer containing 50 mM Tris-HCl, pH=7.4, and 5 mM MgCl₂. Cells were resuspended in a buffer containing 50 mM Tris-HCl, pH=7.4, 5 mM MgCl₂, and 1 mM phenylmethylsulfonylfluoride (PMSF), 1 mM benzamidine, 50 µg/mL tosyl-l-phenylalanine chloromethyl ketone (TPCK), 0.25 mM tosyl-l-lysine chloromethyl ketone (TLCK), 10 µg/mL antipain, 1 µg/mL leupeptin and 1 µg/mL pepstatin as protease inhibitors (PI). An equal volume of glass beads (d=0.5 mm) was added, and cells lysed in a RiboLyser (Hybaid; 4×25 s at level 4.5 with 1-minute intervals on ice). The tube was punctured at the bottom, and the lysate was allowed to flow into a fresh tube. Cell debris was removed by centrifugation (model 5430R; Eppendorf, Hamburg, Germany; 5 min at 1500×g and 4 °C). Microsomal membranes were collected by centrifugation (Eppendorf 5430R; 60 min at 30 000×g and 4 °C). The membrane fraction was resuspended in a buffer containing 20 mM Tris-HCl, pH=8.0, 10 mM EDTA, and PI equivalent to A_{600 nm} (cell)=1.0 per 1 µL. For cell wall preparations, cell lysates were pelleted for 20 s at 20 000×g and 4 °C, pellets were washed twice in ice-cold buffer containing 50 mM Tris-HCl, pH=7.4, and 5 mM MgCl₂ plus PI and resuspended in the same buffer to 1.0 A_{600 nm} cell equivalents.

Cultures of S. cerevisiae BY4741 strain transformed with pREP3adh (empty vector), pGma12 (gma12⁺) or pCM10 (gmh5⁺) were grown overnight in SD medium lacking leucine to final A_{600 nm} of 1.0. A total of cells corresponding to a combined A_{600 nm} of 200 was harvested by centrifugation (Eppendorf 5430R) at 3000×g for 10 min at 4 °C and washed once with TMS buffer (20 mM Tris-HCl, pH=7.5, 5 mM MgCl₂ and 250 mM sucrose). Cell pellets were resuspended in 200 µL TMS buffer plus PI, and an equal volume of glass beads (d=0.5 mm) was added. Cells were lysed using a RiboLyser (Hybaid) in four 30-second intervals (level 4.5) interrupted by 1 min cooling on ice. The bottoms of the reaction tubes were punctured and lysates were collected. Cell debris was removed by centrifugation for 10 min at 3000×g and 4 °C and microsomal membranes were harvested from the supernatant at 30 000×g for 60 min at 4 °C. The membrane fraction was resuspended in TMS to a final concentration of 20 µg/µL, Triton X-100 was added to a final amount of 2 % (m/V) and microsomes were solubilized for 30 min at 4 °C with mild agitation. Solubilized samples were cleared by centrifugation at 30 000×g for 60 min at 4 °C, and the protein concentration was determined by DC protein assay (BioRad, Hercules, CA, USA). Then, galactosyltransferase assays of 50 µL of total volume were performed containing 0.1 M Hepes-NaOH, pH=7.0, 1 mM MnCl₂, 25 nmol UDP-[3H]galactose (specific activity 1 mCi/mmol (1 Ci=3.7·10¹⁰ Bq); BioTrend, Köln, Germany), 5 µmol of a given sugar acceptor and 50 µg of solubilized protein extracts. Acceptors used were: α-methyl-d-mannoside (Sigma-Aldrich, Merck), α-methyl-d-galactoside (Sigma-Aldrich, Merck) and α-1,6-mannobiose (Sigma-Aldrich, Merck). The samples were incubated for 60 min at 30 °C. Each reaction was terminated by the addition of 200 µL of ice-cold water and loaded on an 800 µL Dowex-1 (Cl-form; Sigma-Aldrich, Merck) column packed in a Pasteur pipette. Each column was washed twice with 1 mL of water, eluates were combined, and radioactivity in the eluates was determined in a liquid scintillation counter (Beckman, Brea, CA, USA). Each reaction was performed in triplicate and the experiment was repeated twice.

Cultures of S. cerevisiae BY4741 strain transformed with pREP3adh (empty vector) and pCM11 (Gmh5p-HA) were grown overnight in SD medium lacking leucine to final A_{600 nm}=1.0. Microsomal membrane fractions were obtained as described above. Microsomes were resuspended in 300 µL buffer containing 20 mM Tris-HCl, pH=7.5, 5 mM MnCl₂, 250 mM sucrose, 1 % digitonine, and PI, and solubilized for 1 h at 4 °C with mild agitation. Cell debris was removed by centrifugation (Eppendorf 5430R) at 20 000×g for 30 min at 4 °C and the supernatant was added to 50 µL of anti-HA monoclonal antibody covalently coupled to Sepharose (Roche, Mannheim, Germany). Proteins were immunoprecipitated overnight at 4 °C and washed three times with a buffer containing 50 mM HEPES, pH=7.2, 0.4 % digitonin and 10 mM MnCl₂, and resuspended in 50 µL of the same buffer. Mannosyltransferase assays of 50 µL volume contained 50 mM HEPES, pH=7.2, 0.4 % digitonin, 10 mM MnCl₂, 25 nmol GDP-[3H]mannose (specific activity 1 mCi/mmol (1 Ci=3.7·10¹⁰ Bq); BioTrend), 10 mM α-methyl-d-mannoside, and 5 µL of anti-HA sepharose. Reactions were incubated for 1 h at 30 °C with agitation. Reactions were terminated by adding 200 µL of ice-cold water and loaded on an 800 µL Dowex-1 (Cl-form, Sigma-Aldrich, Merck) column packed in a Pasteur pipette. Each column was washed twice with 1 mL of water, eluates were combined, and radioactivity in the eluates was determined in a liquid scintillation counter (Beckman). Each reaction was performed in triplicate and the experiment was repeated twice.

For heterologous expression of the S. cerevisiae protein chitinase, S. pombe strains FY527, omh1Δ and CMY1 (gmh5Δ) were transformed with the plasmid pFML1-CTS1 (14), and chitinase was isolated as described elsewhere (14). In brief, 5 mL cultures were grown to saturation in EMM lacking leucine, harvested by centrifugation (Eppendorf 5430R) at 3000×g for 10 min at room temperature and resuspended in 10 mL EMM containing 0.1 % glucose and 2 % raffinose instead of 2 % glucose. Cultures were allowed to express chitinase for 48 h. The medium was then cleared by centrifugation (Eppendorf 5430R; 3000×g for 10 min at room temperature), 20 mg of crab shell chitin was added and the samples were incubated overnight at 4 °C with agitation. The chitin was pelleted by centrifugation (15 000×g for 15 min at 4 °C), washed three times with ice-cold buffer containing 50 mM Tris-HCl, pH=7.5, 150 mM NaCl, and chitinase was eluted in 50 µL 2× SDS sample buffer at 95 °C for 5 min. A volume of 10 µL of each sample was separated on 6 % SDS-PAGE and analyzed by Western blot.

Strains FY127 (WT) and CMY1 (gmh5Δ) were transformed with plasmid pCM12 to heterologously express S. cerevisiae invertase Suc2p. Cell walls from these transformants were isolated as described above. To remove N-glycosidically linked glycans, cell walls of 5.0 A_{600 nm} equivalents were treated with endoglycosidase H (Endo H; New England Biolabs GmbH, Frankfurt am Main, Germany) according to the manufacturer’s instructions. Reactions omitting the glycosidase served as a control. From each reaction, proteins were extracted with 1× SDS sample buffer, and 1.0 A_{600 nm} cell equivalents were separated on SDS gels and analyzed by Western blot.

Protein extracts were separated by SDS-PAGE and transferred to nitrocellulose membrane. Monoclonal mouse anti‐HA (16B12; Abcam, Cambridge, UK), polyclonal rabbit anti-Cts1 (gift from W. Tanner) and anti‐invertase (gift from L. Lehle) antibodies were used at 1:8000 (anti-HA) and 1:5000 (others) dilutions, respectively. A horseradish peroxidase conjugated secondary anti-mouse or anti-rabbit antibodies (Sigma-Aldrich, Merck) were used at 1:5000 dilution. Protein–antibody complexes were visualized by enhanced chemiluminescence using the Amersham ECL system (GE Healthcare, Chicago, IL, USA).

Cells were grown in 5 mL YES medium to an A_{600 nm}=1.5, and N(cell)=1.5·10⁸ were pelleted, resuspended in 5 mL of MM-P medium, and incubated for 3 h at 30 °C to induce production of acid phosphatase. The cells were then collected by centrifugation (Eppendorf 5430R; 3000×g for 10 min at 4 °C), washed once with 62.5 mM Tris-HCl, pH=6.8, and suspended in 200 μL ice-cold lysis buffer (62.5 mM Tris-HCl, pH=6.8, 1 mM EDTA, 2 mM PMSF, 0.1 mM dithiothreitol and 10 % glycerol). Cell lysates were prepared by vigorously mixing the cell suspension five times with glass beads (d=0.5 mm) for 30 s at 4 °C using a RiboLyser (Hybaid). The lysates were centrifuged at 15 000×g for 10 min, and the supernatants were recovered and centrifuged again at 15 000×g for 20 min. The supernatants were recovered again and protein was determined using a BCA™ protein assay kit (Thermo Fisher Scientific, Rockford, IL, USA). A mass of 30 μg of total protein from each sample were mixed with 1/3 volume of 0.01 % bromophenol blue, 15 % glycerol and 62.5 mM Tris-HCl, pH=6.8, and were immediately subjected to electrophoresis on a 6 % polyacrylamide native gel. The electrophoresis buffer and the method for staining for acid phosphatase activity were used as described elsewhere (25).

Cells were grown overnight to an A_{600 nm} of 1.0. A volume of 1 mL of the cell suspension was harvested by centrifugation (Eppendorf 5430R) at 3000×g for 3 min. Cell pellets were washed once with distilled water and resuspended in 100 µL of staining solution containing either 0.5 mg/mL aniline blue (Sigma-Aldrich, Merck; dissolved in distilled water) or 5 µg/mL filipin complex (Sigma-Aldrich, Merck; dissolved in DMSO). Aniline blue served both as a β-1,3-glucan stain and as a vital dye for viability assays. Samples (15 µL) were mounted and analyzed on a Leica DM IBR fluorescence microscope (Leica, Wetzlar, Germany) equipped with Leica DFC 350 FX monochrome CMOS camera.

Phylogenetic analysis of galactosyl- and mannosyltransferases was carried out using Clustal Omega (26). Phylogenetic tree data were visualized using Phylodendron (27). Input protein sequences used were: S. cerevisiae: ScHoc1 (UniPro P47124), ScMnn10 (UniProt P50108), ScMnn11 (UniProt P46985), ScMnn9 (UniProt P39107), ScAnp1 (UniProt P32629); S. pombe: SpGmh1 (PomBase SPAC5H10.1), SpGmh2 (PomBase SPAC5H10.13c), SpGmh3 (PomBase SPAC22E12.06c), SpGmh4 (Pombase SPBC8D2.17), SpGmh5 (PomBase SPAC637.06), SpGmh6 (PomBase SPBC1289.13c), SpGma12 (PomBase SPCC736.04c); Candida albicans: CaMnn10 (UniProt Q59UR2); Kluyveromyces lactis: KlMnn10 (UniProt Q70KX5).

Superimpositions were performed with program Coot (28) and structure predictions were done with AlphaFold (29). Structural models and their superimpositions were visualized using PyMOL (30).

Statistical analyses were performed using GraphPad Prism v. 8.0 (31). For glycosyltransferase assays data were evaluated using one-way ANOVA followed by Tukey´s post-hoc test to create confidence intervals for all pairwise differences between the different conditions. For inhibitor assays, non-linear expression analysis was performed and IC₅₀ values were determined. Differences in IC₅₀ values across different conditions were determined by one-way ANOVA followed by a Tukey´s post-hoc test.

We previously demonstrated that in the model yeast S. pombe decreased O-mannosylation severely affects cell wall morphology and stability, especially during septum formation (18). To characterize these defects in more detail, logarithmically growing WT, oma1Δ and oma4Δ mutant cells were stained with aniline blue. The dye preferentially binds to linear β-1,3-glucan (32), mainly localized in the primary septum of dividing cells, where it can be detected from initiation to completion of septum formation (33). In contrast to WT cells with defined staining of the septum, around 30 % of oma1Δ (out of 847 cells) and 40 % oma4Δ (out of 261 cells) cells, respectively, showed significant accumulation of β-1,3-glucan at one or both cell poles, or at the septum (Fig. 1a). Staining of sterol-rich areas of the plasma membrane confirmed local thickening of the cell wall in these mutants (Fig. 1a). Furthermore, vital staining with aniline blue revealed that around 7 % (30 out of 455 cells) of oma1Δ and 20 % (31 out of 153 cells) of oma4Δ cells die during vegetative growth (Fig. S1).

To define the lethal phenotype caused by defective O-mannosylation more precisely, we took advantage of oma2⁺, the essential member of the PMT family in S. pombe (18). We analyzed a conditional lethal mutant expressing the oma2⁺ gene under the control of the nmt81 promoter, which is repressed by thiamine (34, 35). As shown in Fig. 1b, after 24-hour incubation in thiamine-containing medium, more than 80 % of the nmt81-oma2⁺ cells were dead (210 out of 253 cells) with most cells lysing at the septum during or after cell separation (170 out of 253 cells). In some cells, large β-1,3-glucan depositions could be observed in the cell wall, similar to oma1Δ and oma4Δ cells.

In S. pombe, β-1,3-glucan serves as a primary structural component of the cell wall (21). When the synthesis of other cell wall constituents, such as α-glucan or glycoproteins, is compromised, the cell activates compensatory mechanisms to maintain integrity (21). Central to this response is the upregulation of β-1,3-glucan synthesis, which is regulated by signalling pathways involving Rho1 GTPase and protein kinase C homologues, which respond to cell wall stress (36). This adaptive increase in β-glucan content reinforces the cell wall, ensuring cellular morphology and viability under stress conditions. The accumulation of β-glucan at the septum and cell poles and frequent cell lysis at the septum in O-mannosylation mutants (Fig. 1a and Fig. 1b) highlights the crucial role of O-mannosylation in cell wall organization. Consistent with this interpretation, transcriptome analyses of mutant oma421, expressing the oma1⁺, oma2⁺ and oma4⁺ genes under the control of the nmt81 promoter, indicated an induction of genes related to Rho1/PKC signalling pathways when grown in the presence of thiamine. A detailed description of these data will be presented elsewhere.

To find out which proteins and pathways are affected when O-mannosylation is decreased, we screened for multicopy suppressors of the conditionally lethal phenotype of mutant nmt81-oma2⁺. The S. pombe genomic DNA libraries pURSP1 and pURSP2 (22) were used to transform nmt81-oma2⁺ cells. Approximately 106 400 individual transformants were further selected on thiamine-containing medium. After a second round of selection, library plasmids were recovered and retransformed into strain nmt81-oma2⁺. Genes of interest were subcloned into the vector pUR19 (22), and individually validated as suppressors of nmt81-oma2⁺. Among the identified suppressors were the homeobox protein SPAC32A11.03c/Phx1p, which is essential for long-term survival and meiotic sporulation (37), the GDP-mannose-1-phosphate guanylyltransferase SPCC1906.01/Mpg1p required for proper septum structure and cell cycle progression (38), and SPAC637.06/Gmh5p, a so far uncharacterized predicted α-1,2-galactosyltransferase (15, 39) (Fig. 2a). Phx1p (SPAC32A11.03c) is a transcription factor essential for long-term survival and stress adaptation in S. pombe (37). Its multicopy suppression of O-mannosyltransferase mutants may result from enhanced stress tolerance and metabolic flexibility, indirectly supporting cell viability under glycosylation stress. Overexpression of SPCC1906.01/Mpg1p, which encodes a GDP-mannose-1-phosphate guanyltransferase essential for GDP-mannose biosynthesis also suppressed the lethal phenotype of mutant oma2Δ (Fig. 2b). This effect is most likely due to increased availability of this donor substrate for various glycosylation steps.

In contrast to mpg1⁺, gmh5⁺ overexpression does not suppress the lethality of oma2Δ (Fig. 2b). The different outcomes of the complementation tests (compare Fig. 2a and Fig. 2b) can be explained by residual expression of oma2⁺ in the thiamine-repressible strain due to promoter leakiness, while in the oma2Δ mutant, Oma2p activity is completely lost after 5’-FOA treatment. Consequently, suppressor gmh5⁺ can only complement the thiamine-repressed strain, but not the full deletion of oma2⁺. Furthermore, hypersensitivity of the viable oma1Δ and oma4Δ mutants to the drug caffeine (18) is significantly reduced when gmh5⁺ is overexpressed, although not to full extent. Apparent IC₅₀ (half maximal inhibitory concentration) values change from 4.30 to 6.81 mM in oma1Δ, and from 5.00 to 7.17 mM in oma4Δ mutants expressing gmh5⁺ on pREP3adh, but are still below the IC₅₀ value of 8.55 mM in WT cells (Fig. 2c). Based on these findings, we consider gmh5⁺ to be a moderate suppressor of O-mannosylation deficiency. In the following study we are focusing on Gmh5p.

SPAC637.06/Gmh5p is a member of the glycosyltransferase family 34 (GT34), including the S. pombe glycosyltransferases Gma12p and Gmh1p to Gmh6p (40). These enzymes show structural features typical for type-II Golgi glycosyltransferases with a single transmembrane domain flanked by a short N-terminal cytoplasmic and a C-terminal catalytic domain and a typical DxD motif for divalent metal ion binding (41). Gmh1p, Gmh2p, Gmh3p, Gmh6p and Gma12p are confirmed α-1,2-galactosyltransferases acting in the extension of N-linked outer chain and O-mannosyl glycans (15, 16). Gmh5p was initially predicted to act as α-1,2-galactosyltransferase. However, in vivo galactosylation of N- and O-linked glycans was unaffected in a S. pombe gmh5⁺ deletion mutant (15); thus, the enzymatic activity of Gmh5p remains elusive.

The GT34 family includes the mannosyltransferases Mnn10p and Mnn11p from S. cerevisiae (42, 43). Phylogenetic analysis of protein sequences of various galactosyl- and mannosyltransferases of this family from different yeast species revealed the highest homology of Gmh5p to Mnn10p α-1,6-mannosyltransferases (Fig. 3a), with a sequence identity of 38 % to Mnn10p from S. cerevisiae, compared to only 29 % identity to the α-1,2-galactosyltransferase Gma12p from S. pombe.

Like other GT34 family members, Gmh5p is predicted to be a retaining glycosyltransferase with a GT-A fold, including a Rossmann-like domain (40, 44). Gmh5p features a DxD-like motif for divalent metal ion binding, although the second aspartate is replaced by threonine (DxT), a variation also observed in other glycosyltransferases (44). Currently the only GT34 enzyme with an experimentally determined 3D structure is the xyloglucan-xylosyltransferase 1 (XXT1; PDB ID: 6bsu) from Arabidopsis thaliana (45), but no experimental structures are available for fungal GT34 enzymes. However, AlphaFold (29) predictions provide highly reliable models of the core domains for Gmh5p, Mnn10p, and the confirmed α-1,2-galactosyltransferase Gma12p (46). Structural superimposition of Gmh5p and Mnn10p yields a root mean square deviation (RMSD) of 1.2 Å over 265 Cα atoms, indicating a very close structural similarity (Fig. 3b). In contrast, the RMSD between Gmh5p and Gma12p is 1.5 Å over 219 Cα atoms (Fig. 3c), reflecting greater divergence. A detailed comparison of the active sites within the GT34 family shows that in contrast to XXT1, Gmh5p and Mnn10p are nearly identical in their active site architecture, especially in the nucleotide-binding site (Fig. 3d). All enzymes share the GT-A fold and the metal-binding motif, suggesting similar mechanisms for donor substrate binding. However, there is a key difference in nucleotide specificity of the enzymes: while XXT1 from Arabidopsis uses UDP-xylose (47), Mnn10p uses GDP-mannose (9) as a donor substrate. Overall, the nucleotide-binding site of Gmh5p is highly similar to Mnn10p, suggesting that Gmh5p uses GDP-mannose rather than UDP-galactose as donor substrate and is thus more likely to act as a mannosyl- than a galactosyltransferase.

Since baker's yeast does not possess intrinsic galactosyltransfease activity, we further explored the enzymatic activity of Gmh5p in S. cerevisiae. For this purpose, Gmh5p and Gma12p, C-terminally tagged with the hemagglutinin (HA) epitope, were recombinantly expressed under the control of the adh1 promoter in S. cerevisiae WT strain BY4741 (Fig. S2a). Microsomal membranes were used as an enzyme source to measure galactosyl transfer from UDP-[3H]galactose to the putative acceptor substrates α-methyl-d-galactoside, α-methyl-d-mannoside and α-1,6-mannobiose. As shown in Fig. 4a, galactosyltransfer by Gma12p-HA to α-methyl-d-mannoside and to α-1,6-mannobiose could be detected, confirming previous results. However, Gmh5p-HA showed no activity towards the tested acceptor substrates. Thus, the results further support that Gmh5p does not act as α-1,2-galactosyltransferase.

To test for mannosyltransferase activity, Gmh5p-HA was solubilized from microsomal membranes using 1 % digitonine and immunopurified using anti-HA monoclonal antibodies attached to sepharose beads. Immunopurified Gmh5p-HA catalyzed the transfer of α-[3H]mannose from GDP-[3H]mannose to α-methyl-d-mannoside, which serves as a standard model substrate in such assays (Fig. 4b and Fig. S2b). These data further support mannosyltransferase activity Gmh5p.

These experimental results validate our structural predictions and clarify the previous ambiguity regarding the enzymatic specificity of Gmh5p. The activity of Gmh5p corresponds to its closer sequence and structural similarity to Mnn10p from S. cerevisiae, an α-1,6-mannosyltransferase. This provides conclusive evidence that Gmh5p functions as a mannosyl- and not as a galactosyltransferase.

Having defined Gmh5p as a mannosyltransferase, we examined whether Gmh5p is a functional homologue of S. cerevisiae Mnn10p. The Mnn10p protein is a major catalytic subunit of the Golgi-localized M-Pol II complex (Anp1p, Hoc1p Mnn9p to Mnn11p), responsible for the elongation of the α-1,6-mannan backbone of N-linked carbohydrates in baker's yeast (42). In S. cerevisiae, mnn10 mutants have a severely negative impact on N-linked glycosylation (48) and show strong sensitivity towards the aminoglycoside hygromycin B (43). This drug sensitivity was exploited to determine the functional conservation between S. cerevisiae Mnn10p and S. pombe Gmh5p. To that end, Gmh5p was expressed in S. cerevisiae mnn10Δ mutant cells. As shown in Fig. 5, hygromycin tolerance in mnn10Δ strains expressing gmh5⁺ nearly reached WT levels (IC₅₀: mnn10Δ, 1.34 µg/mL; mnn10Δ + gmh5⁺, 2.515 µg/mL (two individual transformants are shown in Fig. 5), WT, 3.00 µg/mL), qualifying Gmh5p as an Mnn10p orthologue.

This functional conservation between Gmh5p and Mnn10p is particularly interesting given their taxonomic distance. S. pombe and S. cerevisiae diverged approximately 400-600 million years ago and have distinct cell wall compositions (49, 50). The ability of Gmh5p to complement the mnn10Δ phenotype supports the conclusion that core aspects of N-glycan biosynthesis pathways remain conserved despite other evolutionary changes in cell wall structure and composition (12, 51). Interestingly, in S. cerevisiae pmtΔ mutants, the Mnn10p orthologue Mnn11p (21 % sequence identity) was shown to be destabilized in pmtΔ mutants, most likely due to reduced O-mannosylation (52). A similar mechanism may apply in S. pombe, where impaired O-mannosylation in omaΔ mutants could affect the stability of Gmh5p and thereby explain its role as a multicopy suppressor of omaΔ mutants. Exploring this hypothesis in S. pombe will be an interesting avenue for future research.

In order to investigate changes in protein glycosylation in more detail, we created an S. pombe gmh5⁺ deletion strain (detailed in Materials and Methods). The known N-glycosylated S. pombe protein, acidic phosphatase (53) was induced in WT and gmh5Δ mutant strains. Cell lysates were analyzed by native polyacrylamide gel-electrophoresis followed by activity staining. Acid phosphatase extracted from gmh5Δ cells has increased electrophoretic mobility compared to WT cells (Fig. 6a), suggesting decreased N-glycosylation, which agrees with Gmh5p acting as an S. cerevisiae Mnn10p orthologue in the elongation of N-linked core glycans. In addition, the highly N-glycosylated and O-mannosylated S. cerevisiae enzymes invertase (54) and chitinase (24), respectively, were recombinantly expressed in S. pombe mutant gmh5Δ. Enriched protein fractions were isolated and proteins analyzed as described in Materials and Methods. Both enzymes were detected with protein-specific polyclonal antibodies. Like in S. cerevisiae, in S. pombe WT cells, invertase migrates as a broad band ranging from 100 to 250 kDa, which shifts to the mass of the unmodified protein (~60 kDa) after N-glycans were removed with Endo H (Fig. 6b). In mutant gmh5Δ, high molecular mass protein species >150 kDa are strongly reduced, which further corroborates the role of Gmh5p in the extension of outer chain of N-glycans. In contrast, the molecular mass of exclusively O-mannosylated chitinase is not changed in mutant gmh5Δ compared to WT (Fig. 6c). The observation that O-mannosylation of chitinase remains unaffected in gmh5Δ mutants while N-glycosylation of acid phosphatase and invertase is significantly impacted highlights the specificity of Gmh5p in the biosynthesis of N-linked glycans.

In S. pombe, N-linked outer chains can be elongated by the action of an M-Pol II-like complex composed of Mnn9p and Anp1p, which sequentially elongates the mannose chains (12). This complex is mirroring the function of M-Pol I (Anp1p and Van1p) and M-Pol II (including Anp1p, Mnn10p, Mnn11p and Hoc1p) in S. cerevisiae (8, 9, 42) but may differ in its assembly. Our data now strongly support that Gmh5p is an orthologue of S. cerevisiae Mnn10p and contributes to M-Pol II-like complex activity in S. pombe by further extending the mannose backbone. This is evident in Fig. 6a and Fig. 6b, where Western blot shows that although N-glycosylation of acid phosphatase and invertase is still present, the glycan chains are substantially shorter, as indicated by the reduced molecular mass of the glycoproteins. Whether these effects observed in mutant gmh5Δ can be complemented by heterologous expression of S. cerevisiae Mnn10p remains to be determined.

In S. cerevisiae, impairment of the M-Pol II complex, as in anp1Δ mutants, leads to synthetic growth defects under conditions of compromised O-mannosylation (55). In line with its function as an orthologue of S. cerevisiae Mnn10p, targeted disruption of gmh5⁺ in oma1Δ cells resulted in increased phenotypic severity of aberrant cell and cell wall morphologies (Fig. 7a). Furthermore, gmh5Δ mutants were highly susceptible to the protein O-mannosyltransferase inhibitor R3A-5a (23) (IC₅₀: WT, 3.08 µM; gmh5Δ, 1.45 µM (two individual mutant clones are shown in Fig. 7b).

The exacerbated phenotypic severity observed when gmh5⁺ is disrupted in oma1Δ cells shows an important interplay between O- and N-linked glycosylation pathways in maintaining cell wall integrity. Similarly, the hypersensitivity of gmh5Δ mutants to the inhibitor R3A-5a suggests a compensatory relationship between these pathways. Our previous work in S. cerevisiae already showed a compensatory response between the two types of glycosylation, specifically in the formation of high mannose outer chains. Transcription of several mannosyltransferase genes involved in the biosynthesis of N-linked high mannose carbohydrate chains was induced upon treatment with O-mannosylation inhibitors and in pmtΔ mutants (56). Similarly, inhibition of N-glycosylation by tunicamycin leads to increased transcription of PMTs and KTR1 involved in O-glycosylation (57), further suggesting a regulatory relationship between these pathways in yeast cell wall biogenesis.