gfsA encodes a novel galactofuranosyltransferase involved in biosynthesis of galactofuranose antigen of O-glycan in Aspergillus nidulans and A. fumigatus

Abstract

The cell walls of filamentous fungi in the genus Aspergillus have galactofuranose-containing polysaccharides and glycoconjugates, including O-glycans, N-glycans, fungal-type galactomannan, and glycosylinositolphosphoceramide, which are important for cell wall integrity. Here, we attempted to identify galactofuranosyltransferases that couple galactofuranose monomers onto other wall components in Aspergillus nidulans. Using reverse-genetic and biochemical approaches, we identified that the AN8677 gene encoded a galactofuranosyltransferase, which we called GfsA, involved in galactofuranose (Galf) antigen biosynthesis. Disruption of gfsA reduced binding of β-Galf-specific antibody EB-A2 to O-glycosylated WscA protein and galactomannoproteins. The results of an in-vitro galactofuranose antigen synthase assay revealed that GfsA has β1,5- or β1,6- galactofuranosyltransferase activity for O-glycans in glycoproteins, uses UDP-D-galactofuranose as a sugar donor, and requires a divalent manganese cation for activity. GfsA was found to be localized at the Golgi apparatus based on cellular fractionation experiments. ΔgfsA cells exhibited an abnormal morphology characterized by poor hyphal extension, hyphal curvature, and limited formation of conidia. Several gfsA orthologs were identified in members of the Pezizomycotina subphylum of Ascomycota, including the human pathogen Aspergillus fumigatus. To our knowledge, this is the first characterization of a fungal β-galactofuranosyltransferase, which was shown to be involved in galactofuranose antigen biosynthesis of O-glycans in the Golgi.

Introduction

Polysaccharides and glycoconjugates are present in the plasma membrane, cell wall, and extracellular environment, and play important roles in many biological events. Filamentous fungal cell walls contain many polysaccharides, including β1,3-, β1,6-, and α1,3-linked glucans, chitin, galactosaminogalactan, and galactomannan, as well as O- and N-glycans linked to proteins (Bernard and Latgé, 2001; de Groot et al., 2009; Gastebois et al., 2009; Latgé, 2009; Latgé, 2010; Fontaine et al., 2011; Jin, 2012). The components and polymeric organization of the Aspergillus cell wall have been determined, and several glycosyltransferase genes involved in the synthesis of these components have been identified and characterized (Kelly et al., 1996; Motoyama et al., 1997; Horiuchi et al., 1999; Shaw and Momany, 2002; de Groot et al., 2009; Yoshimi et al., 2013).

Galactofuranose (Galf) is a component of several polysaccharides and glycoconjugates in a number of filamentous fungi, including various Aspergillus species, bacteria, trypanosomatids, and nematodes, but is absent in yeast, mammals, and plants (Latgé, 2009; Tefsen et al., 2012). Galf residues are immunogenic in mammals (Reiss and Lehmann, 1979; Yuen et al., 2001; Morelle et al., 2005) and are speculated to be involved in pathogenicity in humans. Therefore, the inhibition and detection of the biosynthesis of Galf might be exploited for chemotherapy and as a diagnostic tool for Aspergillus infection (Pedersen and Turco, 2003; Latgé, 2009; Tefsen et al., 2012).

Galf residues are frequently found in Aspergillus glycoproteins, including N-glycans and O-mannose glycans, which modify many cell wall proteins and extracellular enzymes (Wallis et al., 1999; Wallis et al., 2001; Goto, 2007; Tefsen et al., 2012). A single 1,2-linked Galf residue is present at the non-reducing terminus of N-glycans (Takayanagi et al., 1994; Wallis et al., 2001; Morelle et al., 2005; Schmalhorst et al., 2008), and terminally β1,5-linked Galf residues are present in O-mannose glycans (Leitao et al., 2003) (Fig. 1). O-mannose glycan modification of proteins is widely distributed in bacteria and eukaryotic cells, from yeast to mammals, and is required for the formation of filamentous fungal cell walls (Shaw and Momany, 2002; Oka et al., 2004; Oka et al., 2005; Zhou et al., 2007; Mouyna et al., 2010; Goto et al., 2009; Kriangkripipat and Momany, 2009; Lommel and Strahl, 2009). The initial reaction in the transfer of mannose to serine or threonine residues is catalyzed in the endoplasmic reticulum by protein-O-D-mannosyltransferase (Pmt), which requires dolichol phosphate-mannose as an immediate sugar donor (Strahl et al., 1999). Although O-glycans are elongated by glycosyltransferases in the Golgi apparatus, the responsible enzymes, such as galactofuranosyltransferases, remain to be identified in filamentous fungi.

Figure 1. Structures of galactofuranose-containing O-glycan, N-glycan, and fungal-type galactomannan in Aspergillus spp.

The O-glycan structure was described by Leitao et al. (Leitao et al., 2003). O-glycans exhibit structural diversity dependent on the species. N-glycan structures were described by Takayanagi et al. (Takayanagi et al., 1994) and Schmalhorst et al. (Schmalhorst et al., 2008) and the structure of fungal-type galactomannan was described by Latgé et al. (Latgé et al., 1994). GlcNAc, N-acetyglucosamine; Man, mannose; Galf, galactofuranose; Ser, serine; Thr, threonine; Asn, aspartic acid.

Galf is also found in fungal-type galactomannan of Aspergillus (Fig. 1). Fungal-type galactomannan of Aspergillus is composed of a linear mannan core with an α1,2-linked mannotetraose repeating unit attached via an α1,6-linkage and β1,5-Galf oligomers (galactofuran side chain) of up to five residues that are attached to the mannan backbone via β1,6- or β1,3-linkages (Latgé et al., 1994; Fig. 1). Galactomannan is anchored to the plasma membrane by glycosylphosphatidylinositol and is covalently bound to the non-reducing end of a short β1,3-glucan chain in the cell wall (Fontaine et al., 2000; Costachel et al., 2005). However, galactomannan has not been identified in the glycosylphosphatidylinositol anchors of Aspergillus spp. (Tefsen et al., 2012).

Although little is known about the enzymes involved in Galf initiation, extension, and termination, several enzymes involved in the synthesis of Galf-containing sugar chains have been studied in fungi. For example, UDP-glucose 4-epimerase (UgeA) is responsible for the conversion of UDP-glucose to UDP-galactopyranose (Galp) (El-Ganiny et al., 2010), which is then converted to UDP-Galf by UDP-Galp mutase (UgmA/GlfA) in the cytosol (Bakker et al., 2005; Damveld et al., 2008; El-Ganiny et al., 2008; Schmalhorst et al., 2008). The synthesized UDP-Galf is then transported into the Golgi lumen by the Golgi-localized UDP-Galf transporter (UgtA/GlfB) (Engel et al., 2009; Afroz et al., 2011), which is unique to Aspergillus spp. and is required for Galf deposition in the cell wall. UDP-Galf plays an important role in the Galf biosynthetic pathway, where it acts as a sugar donor for galactofuranosyltransferases (Engel et al., 2009; Afroz et al., 2011). An Aspergillus fumigatus glfA deletion mutant strain was shown to lack Galf residues and displayed attenuated virulence in a mouse model of invasive aspergillosis (Schmalhorst et al., 2008). In contrast, Lamarre reported that glfA deletion in a different A. fumigatus strain results in increased adhesion to host cells due to a lack of galactofuran (Lamarre et al., 2009), and no attenuation of virulence. Heesemann reported that the monoclonal antibody IgM L10-1, which specifically reacts with Galf-containing glycostructures, fails to inhibit hyphal growth of A. fumigatus and is not able to protect infected mice (Heeseman et al., 2011) Furthermore, Aspergillus nidulans Galf-deficient mutants exhibit compact colony growth, and abnormal conidiation and hyphal morphology (El-Ganiny et al., 2008). Taken together, these results suggest that Galf residues play important roles in cell wall integrity and pathogenicity in Aspergillus spp. To elucidate the individual role of Galf-containing polysaccharides and glycoconjugates in cell wall biosynthesis, it is necessary to identify and characterize the galactofuranosyltransferases that catalyze the addition of Galf residues from UDP-Galf.

Several bacterial genes reportedly encode galactofuranosyltransferases. For example, the product of the glfT gene is required for mycobacterial galactan polymerization and has both β1,5- and β1,6-Galf transferase activities (Kremer et al., 2001). In addition, WbbI in Escherichia coli performs β1,6-coupling of Galf to α-glucose (Wing et al., 2006) and WbbO in Klebsiella pneumoniae couples Galf to Galp by a β1,3-linkage (Guan et al., 2001). To date, LPG1 of Leishmania sp. is the only Galf transferase described in eukaryotes and the encoding gene was isolated by functional complementation of the Leishmania donovani R2D2 mutant (Ryan et al., 1993). Evidence suggests that LPG1 adds β1,3-Galf residue onto mannose (Huang and Turco, 1993). Although many laboratories have likely searched for galactofuranosyltransferases involved in the synthesis of Galf-containing polysaccharides and glycoconjugates, to our knowledge, the identification of such an enzyme has not been reported.

In the present report, we attempted to identify galactofuranosyltransferase genes involved in the synthesis of Galf antigen in A. nidulans and A. fumigatus using reverse-genetic and biochemical approaches. We speculated that galactofuranosyltransferase candidate genes might be among the set of A. nidulans genes encoding unidentified and uncharacterized putative glycosyltransferases that are not orthologs of genes identified in other organisms. By constructing and screening a gene-disruptant library of candidate genes involved in the biosynthesis of Galf antigen, we identified that the AN8677 gene encoded a galactofuranosyltransferase, which we called GfsA, involved in Galf antigen of O-glycan synthesis. Sucrose density gradient centrifugation analysis revealed that Galf is localized to Golgi apparatus, similar to the UDP-galactofuranose transporter UgtA/GlfB (Engel et al., 2009; Afroz et al., 2011). Based on these results, we updated the biosynthetic pathway for Galf antigen in Aspergillus to include the galactofuranosyltransferase GfsA.

Results

Selection of candidate genes involved in the synthesis of galactofuranose antigens

A BLAST search (Altschul et al., 1997) of the Aspergillus genome did not identify any orthologs of Leishmania or bacterial galactofuranosyltransferase genes. Thus, we used a reverse-genetic approach that did not rely on primary sequence similarity and instead was based on the fact that galactofuranose residues are found in some filamentous fungi, protozoa, and nematodes (Beverley et al., 2005), but not in yeasts, mammals, or plants (Tefsen et al., 2012). A search of the CAZy database (carbohydrate-active enzymes database; http://www.cazy.org/) revealed that approximately 90 genes in the A. nidulans genome encode putative glycosyltransferases (Campbell et al., 1997). We selected candidate genes that might be involved in the synthesis of Galf antigen using the following approach. First, we excluded glycosyltransferase family 1 (GT1) genes, which are thought to be involved in the synthesis of glycosides. Second, we excluded orthologs of genes identified in other organisms, including Saccharomyces cerevisiae. Finally, we selected genes common to filamentous fungi that contain Galf antigen in their cell walls. Based on these criteria, 11 candidate genes encoding putative galactofuranosyltransferases in A. nidulans were selected (Table S1).

Screening of genes involved in the synthesis of galactofuranose antigen

To analyze the function of the 11 putative glycosyltransferase-encoding genes, each gene was disrupted in A. nidulans AKU89 by gene replacement with argB (Table S1 and Table S2). Chromosomal targeting of the 11 genes was confirmed by PCR using the primers ANxxxx-1 and argB-R, and argB-F and ANxxxx-4 (Yu et al., 2004; Table S4, Supplemental Fig. 1A and 1B), and at least two independent isogenic disruptant strains were isolated for each candidate gene. The colony phenotypes of all disruptant strains were observed after 3 days of growth at 30°C on minimal medium. Notably, strain ΔAN8677 formed small white colonies, a phenotype similar to that of strain ΔugmA, which lacks Galf (Fig. 2A; El-Ganiny et al., 2008; Alam et al., 2012).

Figure 2. Screening of genes involved in the synthesis of galactofuranose antigen.

(A) Colony phenotypic analysis of disruptants of the indicated candidate genes. The image was taken after conidia were incubated on minimal medium for 2 days. (B) Immunoblotting analysis of galactomannoproteins from each disruptant. The presence of Galf was detected with antibody EB-A2. Lanes 1–13, 5 μg of galactomannoproteins extracted from the wild-type (wt) strain (lane 1), strain ΔugmA (lane 2), or individual disruptants (lanes 3–15) was loaded in each lane. (C) D-galactopyranose content of galactomannoproteins of wt (bar 1), strain ΔAN8677 (ΔgfsA) (bar 2), and strain ΔugmA (bar 3). The D-galactopyranose content was determined as described in the Experimental procedures. Error bars indicate standard error.

We next tested the affinity of EB-A2, a monoclonal antibody that binds Galf residues with high specificity, towards galactomannoproteins extracted from the disruptant strains (Fig. 2B). Clinical screening with EB-A2 is considered to be a highly sensitive and specific assay for early detection of invasive aspergillosis (Klont et al., 2004), and the specificity of EB-A2 for Galf antigen is well characterized (Stynen et al., 1992; Yuen et al., 2001; Leitao et al., 2003; Morelle et al., 2005). The main epitope of EB-A2 is reportedly a tetrasaccharide of β1,5-linked Galf (galactofuran side chain) (Stynen et al., 1992); however, EB-A2 also reacts with a single terminal non-reducing Galf residue in N-glycans and terminal β1,5-linked Galf residues in O-glycans (Yuen et al., 2001; Leitao et al., 2003; Morelle et al., 2005). Based on these characterizations, EB-A2 is considered to represent a selective anti-β-Galf antibody.

Previous work has shown that EB-A2 bound to cell wall proteins of Aspergillus spp. is detectable as a smear that migrates between 40 to >200 kDa, whereas binding is absent in Galf-defective mutants (Stynen et al., 1992; Schmalhorst et al., 2008; Engel et al., 2009; Afroz et al., 2011). Here, immunoblot analysis was used to examine the binding of EB-A2 to galactomannoproteins extracted from the 11 A. nidulans disruptants (Fig. 2B). The analysis revealed a marked reduction in the EB-A2 signal for strain ΔAN8677 compared to wild type (wt), suggesting that AN8677 is involved in the biosynthesis of Galf antigen (Fig. 2B). Detection of EB-A2-positive signals in the wt and other disruptant strains demonstrated that Galf antigen was present (Fig. 2B).

We next measured the levels of D-galactopyranose in galactomannoproteins extracted from the wt and ΔAN8677 strains. It is generally known that furanoses, included in polymers, are detected as pyranoses after hydrolysis. The D-galactopyranose content of strain ΔAN8677 was only 70% of that observed for wt (Fig. 2C), providing further evidence that the levels of Galf antigen are reduced in strain ΔAN8677. Taken together, these results suggested that AN8677 is involved in the biosynthesis of Galf antigen and was therefore named gfsA (galactofuranose antigen synthase).

Southern blotting was used to examine the structure of the gfsA locus in strain ΔAN8677 (ΔgfsA) (Fig. 3). The analysis revealed that site-specific recombination with the argB cassette occurred at the chromosomal gfsA locus (Fig. 3A and 3B). Next, we introduced the entire gfsA gene from A. nidulans into strain ΔgfsA, yielding strain ΔgfsA+pPTR-II-gfsA. As expected, introduction of A. nidulans gfsA into ΔgfsA rescued the abnormal colony phenotype and conidia formation at 30°C (Fig. 3C). In addition, Galf antigen in galactomannoproteins from strain ΔgfsA+pPTR-II-gfsA was detected by immunoblotting with EB-A2 at signal intensities that were comparable to wt (Fig. 3D). These results are consistent with the speculation that gfsA is responsible for the defect in strain ΔgfsA.

Figure 3. Disruption of gfsA and functional rescue of a gfsA disruptant with A. nidulans gfsA.

(A) Schematic representation of gfsA disruption. (B) Confirmation of gfsA gene disruption by Southern blot analysis. Southern blotting was performed on EcoRV (EV)-digested genomic DNA from wt and the gfsA disruptant. (C) Colony phenotypic analysis of strains wt+pPTR-II, ΔgfsA+pPTR-II, and ΔgfsA+pPTR-II-gfsA. Conidia were incubated on minimal medium at 30°C for 3 days. (D) Functional rescue of the gfsA disruptant with plasmid pPTR-II-gfsA. The results of immunoblot analysis for galactomannoproteins from strains wt+pPTR-II (lane 1), ΔgfsA+pPTR-II (lane 2), and ΔgfsA+pPTR-II-gfsA (lane 3) with antibody EB-A2 are shown.

To investigate the origin of the signal associated with the binding of EB-A2, we performed immunoblot analysis with proteinase K-treated galactomannoproteins extracted from wt (Supplemental Fig. 2). We found that EB-A2 signal intensity decreased after proteinase K treatment, indicating that the signal originated from Galf antigen attached to galactomannoproteins (Supplemental Fig. 2).

To determine if Galf residues are attached to N-glycans and O-glycans in galactomannoproteins extracted from the wt strain, we first released N-glycans and O-glycans from galactomannoproteins by PNGase F treatment and β-elimination, respectively (Fig. 4A). Treatment of galactomannoproteins with PNGase F did not affect EB-A2 signal intensity, suggesting that the signal is not related to the Galf residues attached to N-glycans (Fig. 4A, lanes 1 and 2). By contrast, β-elimination treatment of galactomannoproteins resulted in a decrease in signal intensity (Fig. 4A, lanes 1, 3, and 4). Next, we tested the reactivity of EB-A2 to proteins extracted from A. nidulans strains disrupted for pmtA and pmtC, which encode enzymes that initiate the synthesis of O-mannose-type glycans (Shaw and Momany, 2002; Oka et al., 2004; Goto et al., 2009; Kriangkripipat and Momany, 2009). These disruptants have reduced levels of protein O-mannosylation (Oka et al., 2004; Goto et al., 2009; Kriangkripipat and Momany, 2009). We found that the signal intensity of both ΔpmtA and ΔpmtC strains was lower as compared to wt, indicating that the signal originated from Galf residues in O-glycans (Fig. 4B). In addition, we confirmed that Galf residues in O-mannosylated proteins were reduced in the ΔgfsA strain as compared with wt (Fig. 4B).

Figure 4. Estimation of the origin of the EB-A2 signal.

(A) Detection of Galf antigen in galactomannoproteins with EB-A2 after treatment with PNGase F and β-elimination. Five μg of galactomannoproteins extracted from wt (lanes 1–4) were separated by SDS-PAGE prior to immunoblotting with EB-A2. Lanes 2 and 4, PNGase F-treated proteins; lanes 3 and 4; β-elimination treated proteins. (B) Detection of Galf antigen with EB-A2 in galactomannoproteins from O-glycan-deficient mutants. Five μg of galactomannoproteins extracted from wt (lanes 1), ΔpmtA (lane 2), ΔpmtC (lane 3), ΔgfsA (lane 4), or ΔugmA (lane 5) were separated by SDS-PAGE prior to immunoblotting with EB-A2. (C) In-vivo galactofuranose synthase activity of GfsA toward O-glycosylated WscA. A. nidulans strain AKU89-AnwscA expressing wscA-ha was used as the parental strain (Goto et al., 2009). The gfsA and ugmA genes were disrupted in strain AKU89-AnwscA to construct the strains ΔgfsA-AnwscA and ΔugmA-AnwscA. WscA-HA was purified by rabbit polyclonal anti-HA-Tag-agarose and treated with PNGase F to remove N-glycans. A total of 2.5 μg WscA-HA from AKU89-AnwscA (lanes 1 and 4), ΔgfsA-AnwscA (lanes 2 and 5) and ΔugmA-AnwscA (lanes 3 and 6) were separated by 5–20% SDS-PAGE and detected by immunoblotting with mouse monoclonal anti-HA antibody (left panel) and EB-A2 (right panel). (D) Comparison of the intensity ratios of EB-A2 to anti-HA signals for WscA-HA from wt (bar 1), strain ΔgfsA (bar 2) and strain ΔugmA (bar 3). Densitometric quantification of immunoblotting bands was performed using Quantity One Ver. 4.4.1 (Bio-Rad). The intensity ratios of the EB-A2 to anti-HA signals for WscA-HA in Fig. 4C were calculated, and the ratio for wt was normalized to 1.0.

We previously reported that A. nidulans WscA, a putative stress sensor protein, is N- and O-glycosylated (Goto et al., 2009; Fugatami et al., 2011). We therefore constructed wt, ΔgfsA, and ΔugmA strains expressing a WscA-HA fusion protein, which was then purified from each strain subjected to immunoblotting analysis with anti-HA antibody and EB-A2 after PNGase F treatment (Fig. 4C). The EB-A2 signal intensity of WscA-HA from ΔgfsA was less than that from wt (Fig. 4C, lanes 4 and 5). Densitometric quantification of the immunoblot bands was performed using Quantity One Ver. 4.4.1 (Bio-Rad). The signal intensity ratios for EB-A2 and anti-HA were then calculated and the ratio for wt was normalized to 1.0. The ratio of intensities of EB-A2 to anti-HA from ΔgfsA was less than half of that from wt (Fig. 4D). These data indicate that the gfsA gene is involved in the synthesis of the Galf antigen in O-glycans.

Features of GfsA

The gfsA gene encodes a 532-amino acid protein with a putative molecular mass of 60.9 kDa (Supplemental Fig. 3). gfsA cDNA was amplified by PCR using a 24-h developmental cDNA library (obtained from the FGSC) as a template. Comparison of the cDNA and genome sequences revealed that no introns were present in the gfsA gene. Analysis of secondary structure using SOSUI (Hirokawa et al., 1998) revealed that GfsA has a putative signal sequence (amino acids 1-17) and transmembrane domain (amino acids 18–40) at the N-terminus, suggesting that GfsA is a type II membrane protein. GfsA also has a metal-binding DXD motif (amino acids 256–258), which is conserved among GfsAs from other organisms, and two potential N-glycosylation sites (amino acids 93–95 and 414–416) (Supplemental Fig. 3). Two putative paralogs, AN5663 and AN2015, were identified that have 27.5% and 26.0% sequence identity, respectively, with GfsA. However, disruption of these genes in A. nidulans did not result in any alteration of the colony phenotype under our experimental conditions (Fig. 2A).

BLAST analysis indicated that GfsA proteins are widely distributed in the subphylum Pezizomycotina (Table S3). However, no orthologs of GfsA were found in the genomes of Caenorhabditis elegans, Trypanosoma cruzi, or bacteria, including E. coli, Klebsiella pneumoniae, and Mycobacterium tuberculosis. GfsA has 87% sequence identity with A. flavus NRRL 3357 (AFL2G_12474), 84% with A. niger CBS 513.88 (An12g08720), 84% with A. fumigatus A1163 (AFUB_096220), 84% with Af293 (Afu6g02120), 83% with A. oryzae RIB40 (AO090120000096), 76% with Penicillium chrysogenum Wisconsin 54-1255 (Pc22g24320), 62% with Ajellomyces capsulatus NAm1 (HCAG_03193), 57% with Coccidioides immitis RS (CIMG_05833), 56% with Paracoccidioides brasiliensis Pb03 (PABG_06492), 55% with Trichophyton rubrum CBS118892 (TERG_00129), 32% with Neurospora crassa OR74A (NCU02213), and 32% with Fusarium graminearum PH-1 (FGSC_05770).

Structural and functional analyses of GfsA

To enable a better understanding of GfsA, we constructed a strain expressing 3xFLAG-tagged GfsA by chromosomal tagging (Fig. 5A). The phenotype of the tagged strain was identical to wt, indicating that 3xFLAG-tagged GfsA has similar functional to that of wt GfsA. To confirm expression of 3xFLAG-tagged GfsA, a solubilized protein was prepared from the tagged strain and analyzed with anti-FLAG antibody (Fig. 5B). The strain expressing 3xFLAG-tagged GfsA produced a protein of approximately 67 kDa (Fig. 5B, lane 1). After endoglycosidase H (Endo Hf) treatment, 3xFLAG-tagged GfsA showed faster mobility during SDS-PAGE, with an apparent molecular weight of approximately 61 kDa (Fig. 5B, lane 2). These results show that GfsA is N-glycosylated.

Figure 5. Construction of GfsA-3xFLAG-expressing strain using PCR-based chromosomal tagging and purification of GfsA-3xFLAG protein.

(A) Schematic representation of GfsA-3xFLAG strain construction. (B) Detection of GfsA protein in strain GfsA-3xFLAG. Microsomal fractions were mock treated (lane 1) or treated with endoglycosidase H (lane 2), separated by 5–20% SDS-PAGE, and analyzed by immunoblotting with anti-FLAG antibody. (C) Purification of GfsA-3xFLAG protein. GfsA-3xFLAG was purified from strain GfsA-3xFLAG (ΔugmA) (lanes 1 and 3) and the mock sample was similarly prepared from strain ΔugmA (lanes 2 and 4). A total of 0.28 mg (silver staining) or 2.8 μg (immunoblotting) proteins were separated by 5%-20% SDS-PAGE, and were then assayed by silver staining (left panel) or immunoblot analysis with anti-FLAG antibody (right panel). GfsA-3xFLAG was detected as a 67-kDa protein. Asterisk indicates degraded GfsA-3xFLAG. (D) In-vitro analysis of Galf antigen synthase activity of purified GfsA-3xFLAG by immunoblotting with EB-A2. Enzymatically synthesized UDP-Galf was used as a sugar donor, and galactomannoproteins from strain ΔgfsA (GMPΔgfsA) were used as acceptor substrates. Enzyme activities were assayed as described in the Experimental procedures. Reaction products using purified GfsA-3xFLAG (14 μg) (lane 1) and the control sample (lane 2) are shown. (E) In-vitro analysis of Galf antigen synthase activity for the purified GfsA protein fraction by immunoblotting with EB-A2. Enzymatically synthesized UDP-Galf was used as a sugar donor and purified de-N-glycosylated WscA-HA protein from ΔgfsA was used as an acceptor substrate. Enzyme activities were assayed as described in the Experimental procedures. Reaction products using purified GfsA-3xFLAG protein (14 μg) (lanes 1, 3, and 4), lacking the purified GfsA-3xFLAG protein (lane 2), purified WscA-HA protein from ΔgfsA (0.25 μg) after PNGase F treatment (lanes 1, 2, and 4), lacking the purified WscA-HA protein (lane 3), 0.72 mM UDP-Galf (lanes 1, 2 and 3), and lacking UDP-Galf (lane 4) are shown.

To determine the enzymatic function of GfsA, we assayed Galf antigen synthase activity using an in-vitro assay that was developed for measuring galactofuranosyltransferase activity using EB-A2 antibody. To exclude the influence of intracellular Galf antigen, the assay was conducted in the ΔugmA background (ΔugmA and GfsA-3xFLAG (ΔugmA) strains). GfsA proteins were first immunoprecipitated from strains GfsA-3xFLAG (ΔugmA) and ΔugmA (negative control) using anti-FLAG beads and then subjected to SDS-PAGE and silver staining (Fig. 5C, left panel). Purified GfsA was detected as two distinct protein bands of approximately 67 kDa (Fig. 5C, lane 1). The upper 67-kDa band is intact GfsA-3xFLAG and the lower band is a degradation product or an insufficiently N-glycosylated product that retained reactivity to FLAG antibody (Fig. 5C, lanes 1 and 3). Although a few weakly stained bands of lower molecular weight can be observed in the gel, these bands were also present in the control sample. These results indicate that GfsA-3xFLAG was highly purified.

Galactofuranosyltransferase activity of the purified protein was assayed using 0.72 mM UDP-Galf as a sugar donor, 0.5 mM Mn²⁺ as a co-factor, and 2.5 μg of galactomannoproteins from strain ΔgfsA (GMPΔgfsA). A strong signal of EB-A2 binding was detected in immunoblots when purified GfsA-3xFLAG was used (Fig. 5D, lane 1), whereas no detectable galactofuranosyltransferase activity was observed in the control fraction (Fig. 5D, lane 2). These results indicated that the purified GfsA fraction was not contaminated with any other galactofuranosyltransferase and that GfsA has galactofuranosyltransferase activity. To confirm these results, we also assayed the galactofuranosyltransferase activity of solubilized proteins extracted from ΔugmA and ΔugmAΔgfsA (Supplemental Fig. 4). A signal intensity of EB-A2 staining on immunoblots was only detected for solubilized proteins extracted from ΔugmA (Supplemental Fig. 4, lane 1). In contrast, the solubilized protein fraction obtained from ΔugmAΔgfsA did not have any galactofuranosyltransferase activity (Supplemental Fig. 4, lane 3). This relatively simple experiment clearly demonstrates that proteins other than GfsA have no, or only very minimal, galactofuranosyltransferase activity.

We next attempted to determine if WscA-HA protein from strain ΔgfsA can serve as an acceptor substrate for GfsA-3xFLAG protein. The galactofuranosyltransferase activity of purified GfsA protein was assayed in the presence of 0.25 μg WscA-HA protein obtained from strain ΔgfsA (Fig. 5E). Prior to its use an acceptor substrate, WscA-HA was treated with PNGase F to remove N-glycans. Analysis of the reaction products of the assay by immunoblotting with EB-A2 antibody showed that purified GfsA-3xFLAG protein had galactofuranosyltransferase activity in the presence of both WscA-HA and UDP-Galf (Fig. 5E, lane 1). In contrast, reaction mixtures without GfsA-3xFLAG, WscA-HA, or UDP-Galf did not have detectable enzymatic activity, as indicated by the absence of an EB-A2 staining signal (Fig. 5E, lanes 2, 3, and 4).

Biochemical analyses of GfsA from A. nidulans were also performed using the established in-vitro assay. EB-A2 signal intensity clearly increased in a time-dependent manner (Fig. 6A, lanes 1–5). GfsA was inactive toward galactomannoproteins from ΔugmA (GMPΔugmA) (Fig. 6B, lane 4), and was also relatively inactive in the presence of UDP-Galp, GDP-mannose, and UDP-glucose compared to UDP-Galf, which was associated with a strong EB-A2 signal (Fig. 6C, lanes 3-5). GfsA was also inactive in the presence of 10 mM EDTA, which can chelate Mn²⁺ (Fig. 6D, lane 2). Taken together, these results show that UDP-Galf and Mn²⁺ are required for GfsA activity.

Figure 6. Biochemical analysis of GfsA protein.

(A) GfsA activity increased in a time-dependent manner in an in-vitro assay. Enzymatically synthesized UDP-Galf was used as a sugar donor. Galactomannoproteins from ΔgfsA (GMPΔgfsA) was used as acceptor substrates. Enzyme activities were assayed as described in the Experimental procedures. Reaction products were stained with EB-A2 antibody after incubation with purified GfsA-3xFLAG for 0 h (lane 1), 1 h (lane 2), 3 h (lane 3), 6 h (lane 4), or 24 h (lane 5). (B) Substrate requirement for GfsA activity. Galactomannoproteins from GMPΔgfsA or ΔugmA (GMPΔugmA) were used as acceptor substrates. Reaction products using GMPΔgfsA and UDP-Galf (lane 1), lacking UDP-Galf (lane 2), lacking GMPΔgfsA (lane 3), and using GMPΔugmA in place of GMPΔgfsA (lane 4) are shown. (C) Sugar nucleotide requirement for GfsA activity. Reaction products using 0.72 mM UDP-Galf (lane 1), lacking UDP-Galf (lane 2), 0.72 mM UDP-Galp (lane 3), 0.72 mM GDP-Man (lane 4), and 0.72 mM UDP-Glc (lane 5) are shown. (D) Reaction products in the absence (lane 1) or presence (lane 2) of 10 mM EDTA to chelate Mn²⁺ are shown.

Subcellular localization of GfsA protein

To determine the subcellular localization of GfsA in A. nidulans, we performed sucrose density gradient centrifugation with total subcellular fractions extracted from the A. nidulans strain expressing GfsA-3xFLAG (Fig. 7). The presence of GfsA and distribution of marker proteins were analyzed by immunoblotting with antibodies against UgtA/GlfB (a marker for the Golgi apparatus) and BipA (ER lumen). GfsA was predominantly detected in fractions 9 to 11. This pattern was similar to that observed for UgtA/GlfB, suggesting that GfsA colocalizes with UgtA/GlfB (Fig. 7). In contrast, BipA was detected at similar levels in fractions 1 to 9 (Fig. 7). Thus, the distribution pattern observed for GfsA indicates that GfsA is principally localized to the Golgi apparatus.

Figure 7. Subcellular localization of GfsA.

Sucrose density gradient centrifugation analysis. After centrifugation, each resulting gradient was separated into 16 fractions as described in the Experimental procedures. Anti-UgtA/GlfB was used to detect markers for the Golgi apparatus, anti-BipA was used for endoplasmic reticulum markers, and anti-FLAG was used for GfsA.

Abnormal cell morphology of strain ΔgfsA

The disruption of gfsA inhibited hyphal extension and conidial formation. We observed that strain ΔgfsA forms smaller colonies than wt after cultivation on minimal medium at 30°C for 3 days (Fig. 8A). The colony growth rates of the wt and ΔgfsA strains were 0.40 ± 0.070 and 0.27 ± 0.060 mm/h, respectively. The presence of an osmotic stabilizer (0.6 M KCl) slightly restored colony size in ΔgfsA (Fig. 8A). Moreover, under high-temperature conditions (42°C), the growth defects of strain ΔgfsA were restored (Fig. 8A).

Figure 8. Phenotypic analyses of strain ΔgfsA.

(A) Colony phenotypic analysis of strain ΔgfsA. Strains were grown on minimal medium in the presence (+) or absence (−) of 0.6 M KCl at 30 or 42°C for 3 days. (B) Conidia formation by strain ΔgfsA. Efficiency of conidiation was analyzed as described in the Experimental procedures. Values in parentheses are percentages of growth relative to 7.65 × 10⁵ conidia/mm² (30°C) and 3.51 × 10⁶ conidia/mm² (42°C) for the wt strain. (C) Sensitivity of ΔgfsA strain to various drugs. Strains were grown in the presence of SDS (0.002%), Congo red (30 μg ml⁻¹), and calcofluor white (30 μg ml⁻¹) on CM medium for 3 days at 30°C.

We also examined the conidiation efficiency of the ΔgfsA and wt strains. Formation of normal green conidia was markedly repressed in ΔgfsA grown on minimal medium at 30°C for 3 days (Fig. 8A). The efficiency of conidiation in ΔgfsA was reduced to approximately 11% of that of the wt strain, but recovered to as much as 86% of wt on minimal medium plates in the presence of 0.6 M KCl at 42°C (Fig. 8B). Next, we tested the sensitivity of strain ΔgfsA to SDS, Congo Red, and calcofluor white treatments (Fig. 8C). We found that the growth of strain ΔgfsA was more sensitive than the wt strain to treatment with 0.002% (w/v) SDS (Fig. 8C), but was more resistant to 30 μg ml⁻¹ calcofluor white (Fig. 8C). Hyphal growth of the ΔgfsA strain was not significantly affected by treatment with 30 μg ml⁻¹ Congo red (Fig. 8C).

Conidiophores grown on minimal medium plates at 30°C for 3 days were next examined by scanning electron microscopy. Conidiophores of ΔgfsA were sparse and scattered compared with those of wt (Fig. 9A (a) and (b)). In addition, the number of conidia per conidiophore was smaller in ΔgfsA than in wt (Fig. 9A (c) and (d)). The structure of hyphae in liquid minimal medium was also investigated by fluorescence microscopy. The hyphae of wt A. nidulans grew linearly (Fig. 9B (a)), whereas strain ΔgfsA formed curved hyphae with abnormal branching. The distance between septa in hyphae of strain ΔgfsA was shorter than that in hyphae of wt (Fig. 9B (b)). Together, these observations suggest that gfsA disruption causes abnormal hyphal formation, but does not influence septum formation.

Figure 9. Microscopic analyses of hyphae and conidiophores of strain ΔgfsA strain.

(A) Density of conidiophores of the wt (a) and ΔgfsA (b) strains. Conidiophores and mycelia were cultured on minimal medium for 3 days prior to scanning electron microscope imaging. Conidiophore morphology of wt (c) and ΔgfsA strains (d). Conidiophores were cultured on minimal medium for 3 days prior to imaging. (B) Mycelial morphology of wt (a) and ΔgfsA (b) observed under a confocal laser-scanning microscope. The mycelia were grown for 24 h and harvested. Staining of chitin by calcofluor white was used as a marker for mycelia.

Transcriptional and protein expression analyses of gfsA

To elucidate the growth stage at which gfsA is expressed, we analyzed its transcription using real-time reverse transcription-PCR after 18, 24, 36, and 48 h of culture (Fig. 10A). The transcriptional levels of gfsA were relatively similar between 18 through 24 h, but markedly decreased between 24 and 48 h. We also assayed GfsA protein levels at 18, 24, 36 and 48 h (Fig. 10B). GfsA was expressed at high levels at 18 and 24 h, but the levels decreased after 24 h, whereas γ-actin was expressed almost constitutively expressed (Fig. 10B). This result was comparable with the result of gfsA transcription analysis. In addition, to determine if gfsA is transcribed in conidia, we compared transcription of gfsA in conidia with that in mycelia. We first tested the quality of the conidia RNA sample by assaying levels of the conidia-specific spoC1-C1C transcript. The results indicated that the sample RNA was successfully extracted from conidia (Fig. 10C, right panel). Moreover, the level of gfsA RNA in the conidia was higher than that in the mycelia after cultivation in minimal liquid medium for 18 h (Fig. 10C, left panel). These results are consistent with the speculation that GfsA has a significant role not only during hyphal development, but also during conidiation and in the conidia.

Figure 10. Transcriptional and protein expression analyses of gfsA.

(A) Real-time reverse transcription (RT)-PCR analysis of gfsA expression in the wt strain. Conidia (2 × 10⁷) were inoculated into 100 ml liquid minimal medium and grown for 18, 24, 36, or 48 h at 30°C. gfsA expression was quantified by real-time RT-PCR analysis. The y-axis shows the level of mRNA relative to the histone 2B gene. (B) GfsA protein expression analysis in the GfsA-3xFLAG strain. Conidia (2 × 10⁷) were inoculated into 100 ml liquid minimal medium and grown for 18, 24, 36 or 48 h at 30°C. The expression levels of GfsA (upper panel) and γ-actin (lower panel) were determined by immunoblot analysis. γ-actin was used as a protein loading control. (C) Transcription analysis of gfsA in conidia. RNA was extracted from mycelia collected after 18 h of cultivation and conidia were harvested on a minimal medium plate after 3 days. Expression of gfsA, spoC1-C1C, and argB were quantified by real-time RT-PCR analysis. The argB gene was used previously as a control for transcriptional analysis in conidia (Stephenes et al., 1999; Momany et al., 2001), and spoC1-C1C is an established marker for conidia (Orr and Timberlake, 1982; Stephenes et al., 1999). The y-axis shows the level of mRNA relative to the argB gene.

Functional analyses of a gfsA ortholog in A. fumigatus

It has been suggested that Galf-containing molecules are related to virulence and infection of A. fumigatus, which is an opportunistic human pathogen (Schmalhorst et al., 2008; Lamarre et al., 2009). We identified a putative ortholog of gfsA (AFUB_096220) in A. fumigatus strain A1163. To analyze the function of this gene, which we termed AfgfsA, we generated an AfgfsA disruptant using A. fumigatus strains A1160 and A1151 as parental strains. Colonies formed by strain ΔAfgfsA in the A1160 background were smaller than those formed by strain A1160 at 47°C (Fig. 11A) and conidia formation was reduced. These phenotypes were restored in the presence of 0.6 M KCl at 47°C. To confirm that the mutation could be complemented, we introduced the plasmid pPTR-II-gfsA into strain ΔAfgfsA. The abnormal colony phenotype was restored in the ΔAfgfsA strain expressing A. nidulans gfsA (Fig. 11B). In addition, Galf antigens in galactomannoproteins were detected by immunoblotting analysis with EB-A2. In the ΔAfgfsA strain expressing gfsA from A. nidulans, the signal intensity of EB-A2 immunodetection was also recovered to wt levels, indicating that A. nidulans gfsA can synthesize the Galf antigen in both A. nidulans and A. fumigatus (Fig. 11C). This result also indicates that AfgfsA encodes a galactofuranosyltransferase.

Figure 11. Functional analyses of the A. fumigatus AfgfsA gene.

(A) Colony phenotypic analysis of wt and ΔAfgfsA strains. Strains were grown on minimal medium in the presence (+) or absence (−) of 0.6 M KCl at 30, 37, or 47°C for 3 days. (B) Functional rescue of a ΔAfgfsA strain with plasmid pPTR-II-gfsA. Strains were grown on minimal medium at 47°C for 3 days. (C) Immunoblot analysis of galactomannoproteins from strains A1160+pPTR-II, ΔAfgfsA+pPTR-II, and ΔAfgfsA+pPTR-II-gfsA. Galactofuranose antigen was detected with EB-A2 antibody. Galactomannoproteins (5 μg) were extracted from strains A1160+pPTR-II (lane 1), ΔAfgfsA+pPTR-II (lane 2), and ΔAfgfsA+pPTR-II-gfsA (lane 3). (D) Sensitivity to micafungin and voriconazole assayed by the Etest. Conidia (6 × 10⁵/plate) were evenly spread on MEA plates and test strips were placed at the center of the plate before incubation at 37°C for 24 h. Arrows indicate MIC values.

Finally, we tested the sensitivity of A1151, ΔglfA, and ΔAfgfsA to micafungin and voriconazole on MEA medium. Plates with conidia covering the surface were test strips were incubated with test strips of the two antibiotics for 24 h at 37°C. Under these conditions, A. fumigatus strain A1151 formed a clear zone of growth inhibition around voriconazole (minimum inhibitory concentration [MIC], 0.125 μg ml⁻¹) and micafungin (MIC, 0.008 μg ml⁻¹). However, strain ΔAfgfsA showed a larger zone of growth inhibition around voriconazole (MIC, 0.064 μg ml⁻¹l) and micafungin (MIC, 0.004 μg ml⁻¹), similar to that formed by strain ΔglfA strain (Fig. 11D), indicating that ΔAfgfsA has increased sensitivity to antifungal agents.

Discussion

Strain ΔugmA displays an abnormal phenotype that appears to result from the loss of all Galf-containing polysaccharides and glycoconjugates, including O-glycans, N-glycans, galactofuran side chains, and glycosylinositolphosphoceramides. To help clarify the roles of individual Galf-containing polysaccharides and glycoconjugates, it is necessary to identify and characterize genes that encode galactofuranosyltransferases involved in the synthesis of these molecules. Towards this goal, here, we attempted to identify a galactofuranosyltransferase gene involved in Galf antigen synthesis in the filamentous fungus A. nidulans. Using reverse-genetic and biochemical approaches, we provide evidence that gfsA encodes a galactofuranosyltransferase with the ability to synthesize the Galf antigen of O-glycans. To our knowledge, this is the first report describing a galactofuranosyltransferase-encoding gene that functions in the fungal Galf synthetic pathway. Engel and colleagues proposed a novel schematic model of galactofuranosylation (Engel et al., 2009). Based on our present results, we have updated the model of galactofuranosylation in fungal cells to include GfsA protein (Fig. 12). In the updated model, UDP-Galf is first synthesized from UDP-glucose via UDP-Galp in the cytosol by UgeA and UgmA/GlfA (Damveld et al., 2008; Schmalhorst et al., 2008; El-Ganiny et al., 2008; Lamarre et al., 2009; El-Ganiny et al., 2010) and is then transported into the Golgi apparatus via the antiporter protein UgtA/GlfB (Engel et al., 2009; Afroz et al., 2011). GfsA then synthesizes Galf antigen of O-glycans on galactomannoproteins in the Golgi apparatus during hyphal tip growth. Finally, the synthesized galactofuranosylated proteins are transported by vesicles and localized to the cell surface (Fig. 12).

Figure 12. Galactofuranose (Galf) antigen in the O-glycan biosynthesis pathway.

Updated model of biosynthesis of the Galf antigen of O-glycan. UMP, uridine 5′-monophosphate; UDP, uridine 5′-diphosphate; Galf, galactofuranose; Galp, galactopyranose; Glc, glucose.

The reactivity of monoclonal antibody EB-A2 with Galf antigen present on galactomannoproteins was reduced in strain ΔgfsA, suggesting that gfsA is involved in the synthesis of Galf antigen (Fig. 2B). EB-A2 reactivity towards purified WscA protein, an O-glycosylated protein, was also reduced following disruption of gfsA (Fig. 4D). In addition, the results of in-vitro assays provide evidence that purified GfsA protein has Galf antigen synthase activity toward WscA protein and GMPΔgfsA, using UDP-Galf as a sugar donor (Fig. 5D and E). However, GfsA does not show activity toward GMPΔugmA (Fig. 6B, lane 4). Based on the structure of Galf antigen in O-glycans (Fig. 1), at least two galactofuranosyltransferases appear to be involved in the synthesis of Galf antigen; one forms a β1,6-linkage between Galf and the α1,6-mannobiose backbone, and the other forms β1,5-Galf linkages between Galf residues at the non-reducing termini. As the latter reaction requires a Galf residue as an acceptor substrate for catalysis, galactomannnoproteins extracted from ΔugmA strain cannot serve as a substrate for this enzyme. Therefore, it is highly likely that GfsA is a novel β1,5-galactofuranosyltransferase involved in O-glycan biosynthesis. However, details studies of the type of linkage for the Galf residue and the substrate structure of GfsA are warranted.

A. nidulans strains ΔugmA and ΔugtA exhibit defects in colony growth, hyphal morphogenesis, and conidiation (El-Ganiny et al., 2008; Afroz et al., 2011). Strain ΔgfsA exhibits defects similar to those of ΔugmA (Fig. 1A), including abnormal cell morphology, poor hyphal extension, hyphal curvature, and limited conidia formation (Figs. 8A, 8B, 9A and 9B). Similar phenotypes were also observed in strains ΔugmA and ΔpmtA, indicating that Galf residues in O-mannose type glycans are required for normal cell morphology (Oka et al., 2004; El-Ganiny et al., 2008). If GfsA is also a galactosyltransferase involved in the synthesis of galactofuran side chains, our findings also indicate that these side chains are also required for normal cell morphology.

Genes involved in the Galf biosynthetic pathway are thought to be functionally linked to cell wall formation and maintenance (El-Ganiny et al., 2008; El-Ganiny et al., 2010; Afroz et al., 2011). The growth defect of strain ΔgfsA was not observed under conditions of high temperature and hyperosmolarity, a finding that is also reported for mutants with defects in cell wall integrity (Horiuchi et al., 1999; Oka et al., 2004; Goto et al., 2009; Futagami et al., 2011). The molecular mechanisms that control the ability of Aspergillus spp. to maintain cell wall integrity under stress have not been fully elucidated. Nevertheless, it is reasonable to speculate that in strain ΔgfsA, the triggering of stress responses, such as those normally induced by heat and hyperosmotic stress, might help to compensate for the loss of Galf antigen in cell wall O-glycans. Strain ΔgfsA shows sensitivity to Congo red (30 μg ml⁻¹), an inhibitor of glucan synthesis, but is resistant to calcofluor white (30 μg ml⁻¹), an inhibitor of chitin synthesis. These results imply that the balance of cell wall components is altered in strain ΔgfsA, perhaps due to the reduction in Galf antigen and/or an increase in chitin levels. El-Ganiny reported that the abnormal hyphal phenotype that is observed in the A. nidulans ΔugmA strain under normal growth conditions can be partially suppressed by growth on medium containing 10 μg ml⁻¹, but not 30 μg ml⁻¹, calcofluor white (El-Ganiny et al., 2008). This finding is different from that observed for strain ΔgfsA, suggesting that the function of the Galf antigen of O-glycans differs from those of other antigens, including the galactofuran side chain, N-glycan, and glycosylinositolphosphoceramide, at least as they relate to resistance or sensitivity to calcofluor white.

Disruption of ugmA leads to a complete loss of galactofuranose residues in the cell, because UDP-Galf, which is the sugar donor for all galactofuranosyltransferases, is not synthesized. In contrast, galactofuranose residues were detected in strain ΔgfsA (Fig. 2C), providing further evidence that GfsA protein is a galactofuranosyltransferase that can transfer galactofuranose from UDP-Galf to O-glycans. Consistent with this speculation, the defects observed in strain ΔgfsA were less severe than those observed for ΔugmA (Fig. 2A), demonstrating that other galactofuranose residues, in addition to O-glycans, are also important for normal cell growth.

We also found that galactomannnoproteins extracted from ΔgfsA are weakly recognized by EB-A2, even though no signal was detected with proteins from ΔugmA (Fig. 3D, lane 2; Fig. 4C, lane 5). A signal was seen for WscA purified from the ΔgfsA mutant because a larger number of WscA molecules were present on the gel compared to individual galactomannoproteins (Fig. 4C, lane 5). This observation indicates that galactofuranosyltransferase activity remained in the ΔgfsA cells. However, galactofuranosyltransferase activity was not detected in the solubilized protein fraction extracted from ΔugmAΔgfsA (Supplemental Fig. 4, lane 3). Although these results are at first contradictory, they suggest that GfsA plays a major role in the transfer of galactofuranose residue to O-glycans in cells and that the conditions of our in-vitro assay were not suitable to detect the remaining galactofuranose activity in strain ΔugmA.

The gfsA gene was identified among candidate genes selected from the CAZy database and belongs to the GT31 family, which includes galactosyltransferases such as glycoprotein-N-acetylgalactosamine-β1,3-galactosyltransferase (beta3GalT1) (Bardoni et al., 1999), hydroxyproline-O-galactosyltransferase (Basu et al., 2013), and glycosaminoglycan-β1,3-galactosyltransferase (Zhou et al., 1999). In the A. nidulans genome, there are seven genes encoding GT31 family proteins (AN2015, AN4824, AN5663, AN7535, AN8627, AN11144, and AN11697) in addition to gfsA. Of these genes, AN2015 and AN5663 appear to be the most closely related to gfsA. Thus, it is likely that these genes are also involved in the addition of Galf residues to O-glycans. These genes will be examined in future experiments in an effort to identify additional galactofuranosyltransferases involved in the synthesis of Galf-containing polysaccharides and glycoconjugates.

Fungal-type galactofuran side chains of galactomannan are β1,5-linked Galf oligomers composed of up to five residues. Galactofuran side chains are linked to α1,2-mannan by β1,6- or β1,3-linkages (Latgé et al., 1994). In yeast, the mannosyltransferase Mnn1 is responsible for the synthesis of both O-glycans and the outer chain of N-glycans (Lussier et al., 1994). Because the structure of galactofuran side chains is similar to the structure of O-glycans (Fig. 1), we cannot exclude the possibility that GfsA, and/or its putative paralogs, have the ability to biosynthesize fungal-type galactofuran side chains of galactomannan.

Genes related to gfsA are widely distributed in the subphylum Pezizomycotina of Ascomycota, but are not found in the subphyla Saccharomycotina or Taphrinomycotina (Table S3). In contrast, basidiomycete fungi, including Cryptococcus neoformans and Ustilago maydis, have both ugmA/glfA and ugtA/glfB orthologs, but no ortholog of gfsA (Table S3) (Beverley et al., 2005). Consistent with this finding, no binding of EB-A2 to exopolysaccharides from C. neoformans, Candida albicans, Saccharomyces cerevisiae or Rhizopus stolonifer was observed (Stynen et al., 1992; Kappe and Schulze-Berge, 1993). To our knowledge, Galf antigen structures are not found in C. neoformans, despite the fact that a non-reducing terminal Galf residue is observed in galactoxylomannans from C. neoformans (James and Cherniak, 1992). These observations are consistent with C. neoformans having both ugmA and ugtA/glfB orthologs, but no gfsA ortholog. In the phylum Zygomycota, which includes fungi such as Rhizopus oryzae RA 99-880, no genes involved in Galf synthesis have been detected, including gfsA orthologs (Table S3). It seems likely that fungal species containing the gfsA gene would also have both ugmA/glfA and ugtA/glfB, as the products of these genes act upstream of GfaA in the Galf synthesis pathway.

The Pezizomycotina subphylum contains fungi pathogenic to humans, animals, and plants. It has been suggested that Galf-containing polysaccharides and glycoconjugates from these fungi are related to virulence and infection (Pedersen and Turco, 2003; Latgé, 2009; Tefsen et al., 2012). Moreover, in the present study, we found that strain ΔAfgfsA of the opportunistic human pathogen A. fumigatus exhibits a small-colony phenotype and has increased sensitivity to antifungal treatments. These characteristics suggest that GfsA activity might be an appropriate new target for antifungal treatment (Fig. 11A and D). We anticipate that our findings will stimulate functional investigation into the medical and biotechnological implications of Galf-containing polysaccharides and glycoconjugates.

Experimental procedures

Microorganisms and growth conditions

The A. nidulans and A. fumigatus strains used in this study are listed in Table S2. A. nidulans strain AKU89 was previously described (Goto et al., 2009). A. fumigatus A1160 and A1151 (da Silva Ferreira et al., 2006) were obtained from the Fungal Genetics Stock Center (FGSC). Strains were grown on minimal medium (1% (w/v) glucose, 0.6% (w/v) NaNO₃, 0.052% (w/v) KCl, 0.052% (w/v) MgSO₄·7H₂O, 0.152% (w/v) KH₂PO₄, biotin (trace), and Hunter’s trace elements, pH 6.5) (Barratt et al., 1965), complete medium (CM) with appropriate supplements for nutritional markers as described previously (Kaminskyj, 2001), YG medium (0.5% yeast extract and 1% glucose) or MEA medium (2% malt extract, 2% glucose, 0.1% peptone, and 2% agar, pH 6.0). Growth experiments to allow hyphal development in submerged culture were begun by inoculation of 100 ml minimal medium with 2 × 10⁷ conidia in 500-ml culture flasks. The flasks were shaken at 126 rpm at 30°C for A. nidulans and 37°C for A. fumigatus. Standard transformation procedures for Aspergillus strains were used (Yelton et al., 1984). Plasmids were amplified in Escherichia coli JM109 or XL1-Blue.

Construction of deletion cassettes and gene disruption strains

A. nidulans AKU89, A. fumigatus A1160, and A. fumigatus A1151 were used as wild-type (wt) control strains (Table S2). The pyrG gene was amplified by PCR using A. nidulans FGSC26 genomic DNA as template and primers AnpyrG-F and AnpyrG-R (Table S4). The amplified fragments were inserted into the EcoRV sites of pGEM-5Zf(+) using the pGEM-T Easy Vector System (Promega Co., Madison, WI) to yield pSH1. Genes were disrupted in A. nidulans by argB or ptrA insertion. AfgfsA was disrupted in strain A1160 by pyrG insertion, and glfA and AfgfsA were disrupted in strain A1151 by ptrA insertion. DNA fragments for gene disruption were constructed using a “double-joint” PCR method, as described previously (Yu et al., 2004). All PCR was performed using Phusion High-Fidelity DNA Polymerase (Daiichi Pure Chemicals Co., Ltd, Tokyo, Japan). Primers used in this study are listed in Table S4. Eight primers (from ANxxxx-1 to ANxxxx-8 for A. nidulans genes or from AFUB_xxxxxx-1 to AFUB_xxxxxx-8 for A. fumigatus genes; xxxx or xxxxxx indicates the systematic gene name) were used to construct deletion cassettes. The 5′- and 3′-flanking regions (approximately 0.9-1.2 kb each) of each gene were PCR amplified from genomic DNA with primer pairs ANxxxx-1/ANxxxx-2 or AFUB_xxxxxx-1/AFUB_xxxxxx-2 and ANxxxx-3/ANxxxx-4 or AFUB_xxxxxx-3/ AFUB_xxxxxx-4, respectively (Table S4). The argB, pyrG, and ptrA genes used as selective markers were amplified using plasmids pDC1 (Aramayo et al., 1989, obtained from the FGSC), pSH1, and pPTR-I (Takara Bio, Inc., Otsu, Japan), respectively, as template and the primer pairs ANxxxx-5/ANxxxx-6 or AFUB_xxxxxx-5/AFUB_xxxxxx-6. The three amplified fragments were purified and mixed and a second PCR was performed without specific primers, as the overhanging chimeric extensions act as primers, to assemble each fragment. A third PCR was performed with the nested primer pairs ANxxxx-7/ANxxxx-8 or AFUB_xxxxxx-7/AFUB_xxxxxx-8 and the products of the second PCR as template to generate the final deletion construct. The amplified final deletion constructs were purified by QIAquick Gel Extraction Kit (Qiagen, Hilden, Germany) and used directly for transformation. Transformants were grown on minimal medium plates containing 0.6 M KCl as an osmotic stabilizer under appropriate selection conditions and single colonies were isolated twice before further analysis.

Disruption of target genes was confirmed by PCR with the primer pairs ANxxxx-1/argB-R and argB-F/ANxxxx-4 or ANxxxx-1/ptrA-R and ptrA-F/ANxxxx-4 for A. nidulans, AFUB_xxxxxx-1/pyrG-R and pyrG-F/AFUB_xxxxxx-4 for A. fumigatus A1160, and AFUB_xxxxxx-1/ptrA-R and ptrA-F/AFUB_xxxxxx-4 for A. fumigatus A1151. Disruption of gfsA in transformants was confirmed by Southern blotting. The 5′-UTR of gfsA was amplified by PCR with the primer pair gfsA-probe-F/gfsA-probe-R. The amplified fragment was labeled with digoxigenin and then used as a probe.

To construct an argB⁺ derivative of the A. nidulans parent strain AKU89, the entire argB gene was amplified from genomic DNA of A. nidulans FGSC26 by PCR with the primer pair argB-RESCUE-F and argB-RESCUE-R (Table S4). The 2.5-kb amplified fragment was purified using a QIAquick Gel Extraction Kit (Qiagen) and used directly for transformation to generate strain AKUARG89.

Construction of strains ΔgfsA-AnwscA and ΔugmA-AnwscA

A. nidulans AKU89-AnwscA (Goto et al., 2009) was used as a parental strain for the construction of strains ΔgfsA-AnwscA and ΔugmA-AnwscA (Table S2). The gfsA and ugmA genes were disrupted in AKU89-AnwscA by ptrA insertion. Eight primers (from AN8677-1 (gfsA::ptrA) to AN8677-8 (gfsA::ptrA) or AN3112-1 (ugmA::ptrA) to AN3112-8 (ugmA::ptrA)) were used to construct a deletion cassette. DNA fragments for gene disruption were constructed using a double-joint PCR method (described above). Disruption of gfsA and ugmA was confirmed by PCR with the primer pairs AN8677-1 (gfsA::ptrA)/ptrA-R and ptrA-F/AN8677-4 (gfsA::ptrA) or AN3112-1 (ugmA::ptrA)/ptrA-R and ptrA-F/AN3112-4 (ugmA::ptrA).

Purification of WscA-HA protein

HA-tagged WscA was purified from strains AKU89-AnwscA, ΔgfsA-AnwscA, and ΔugmA-AnwscA that were grown in 100 ml minimal medium at 30°C for 24 h and harvested by filtration. The cells (1 g wet cells) were ground in liquid nitrogen with a mortar and pestle, and the lysed cells were resuspended in buffer B (50 mM HEPES-NaOH (pH 6.8), 100 mM NaCl, 30 mM KCl, and 1% Triton X-100). After cell debris was removed by centrifugation at 3,000 × g for 10 min, 20 μl of rabbit polyclonal anti-HA-Tag-agarose (Medical & Biological Laboratories Co., Ltd., Nagoya, Japan) was added to the supernatant and the mixture was gently shaken for 3 h. The anti-HA-Tag-agarose was collected by centrifugation at 1,400 × g for 10 min, and was then washed five times with 1.5 ml buffer B.

Amplification, cloning, and sequencing of gfsA cDNA

The gfsA cDNA was amplified by PCR using a 24-h developmental cDNA library from A. nidulans that was constructed by R. Aramayo (obtained from the FGSC) as a template and the primers AN8677-cDNA-F/AN8677-cDNA-R (Table S4). The amplified cDNA fragments were inserted into the EcoRV sites in pGEM-5Zf(+) using the pGEM-T Easy Vector Systems (Promega Co.) to yield pTOGFSA. The sequence of gfsA was confirmed using an Applied Biosystems 3130xl Genetic Analyzer (Life Technologies Co., Carlsbad, CA).

Extraction of galactomannoproteins

Galactomannoproteins were extracted according to the method described by Chiba and Jigami (Chiba and Jigami, 2008). Aspergillus cells were grown in minimal medium at 30°C or 37°C for 24 h. The cells were harvested by filtration, washed twice with distilled water, resuspended in 100 mM citrate buffer (pH 7.0), and then autoclaved at 121°C for 120 min. After cell debris was removed by filtration, the extracted galactomannoproteins were precipitated by ethanol, dialyzed with dH₂O, and then lyophilized. The protein concentration of the purified galactomannoproteins was measured based on absorbance at 280 nm (A₂₈₀).

Determination of D-galactopyranose content of galactomannoproteins

The extracted galactomannoproteins (2.4 μg) were lyophilized, incubated in 4 M TFA at 100°C for 4 h, and then dried at room temperature. The hydrolysates were labeled with fluorescent p-aminobenzoic acid ethyl ester (ABEE) using an ABEE Labeling Kit (J-Oil Mills, Inc., Tokyo, Japan) according to the manufacturer’s protocol. Protein levels were measured using a Qubit protein assay kit (Life Technologies Co.). ABEE labeled D-galactopyranose was analyzed using a high-performance liquid chromatography system equipped with a Honenpak C18 column (4.6 mm × 75 mm) (J-Oil Mills, Inc.). D-galactose (Sigma Aldrich, St. Louis, MO, USA) was used as a standard for quantification.

Release of O- and N-glycans from extracted galactomannoproteins

To release N-glycan, galactomannoproteins (72 μg) were treated with PNGase F (New England Biolabs, Ipswich, MA) according to the manufacturer’s protocol. To release O-glycan from cell wall proteins, β-elimination was performed using the GlycoProfile Beta-Elimination Kit (Sigma-Aldrich) according to the manufacturer’s protocol. The β-elimination reaction was performed at 25°C for 24 h.

Construction of an expression vector for gfsA

pPTR-II (Takara Bio, Inc.), which is an autonomously replicating plasmid containing the AMA1 sequence (Kubodera et al., 2002), was used an expression vector for gfsA. The gfsA gene was amplified by PCR using A. nidulans genomic DNA as template and the primers gfsA-In-fusion-F and gfsA-In-fusion-R. The amplified fragments were inserted into the SmaI sites of pPTR-II using an In-Fusion™ Advantage PCR Cloning Kit (Clontech Laboratories, Inc., Mountain View, CA) to yield pPTR-II-gfsA. A. nidulans strain ΔgfsA was transformed with pPTR-II-gfsA or pPTR-II. Transformants were selected on minimal medium supplemented with 0.1 mg ml⁻¹ pyrithiamine and 0.6 M KCl. For each construct, three different transformants were randomly selected for subsequent experiments.

Construction of GfsA-3xFLAG-expressing strain

All PCR was performed using Phusion High-Fidelity DNA Polymerase (Daiichi Pure Chemicals Co., Ltd). A 3xFLAG-tagged DNA fragment was synthesized by PCR without template using the primer pair 3xFLAG-F/3xFLAG-R. The amplified fragments were inserted into the SphI sites of pDC1 to yield pDC1-3xFLAG. The open reading frame and upstream promoter of gfsA were amplified from genomic DNA by PCR with the primer pair AN8677-3xFLAG-1/AN8677-3xFLAG-2. Similarly, the terminator regions of the gfsA gene were amplified with the primer pair AN8677-3xFLAG-3/AN8677-3xFLAG-4. The argB gene, including the 3xFLAG tag, was amplified with the primer pair AN8677-3xFLAG-5/AN8677-3xFLAG-6 using plasmid pDC1-3xFLAG as template. The three DNA fragments were mixed and a second PCR was performed with the nested primers AN8677-3xFLAG-7/AN8677-3xFLAG-8 to generate the final DNA construct for chromosomal tagging. The final amplified deletion constructs were purified using a QIAquick Gel Extraction Kit (Qiagen) and used directly for transformation.

Construction of strain GfsA-3xFLAG (ΔugmA)

AThe ugmA gene was disrupted in A. nidulans GfsA-3xFLAG (Table S2) by ptrA insertion. Eight primers (from AN3112-1 (ugmA::ptrA) to AN3112-8 (ugmA::ptrA)) were used to construct a deletion cassette. DNA fragments for gene disruption were constructed using the double-joint PCR method (described above). Disruption of ugmA was confirmed by PCR with the primer pairs AN3112-1 (ugmA::ptrA)/ptrA-R and ptrA-F/AN3112-4 (ugmA::ptrA).

Preparation of solubilized protein

Cells from GfsA-3xFLAG (ΔugmA), ΔugmA or ΔugmAΔgfsA were collected by centrifugation and were then ground in liquid nitrogen with a mortar and pestle. The lysed cells were resuspended in buffer A (50 mM HEPES-NaOH (pH 6.8), 100 mM NaCl, 30 mM KCl, 1 mM MnCl₂, and 5% glycerol (w/v)). All procedures were performed on ice or at 4°C. Cell debris was removed by centrifugation at 10,000 × g for 10 min, and the obtained supernatants were further centrifuged at 100,000 × g for 30 min. The resultant pellet was resuspended in buffer A with 0.2% CHAPSO (Dojindo Laboratories, Kumamoto, Japan). To obtain solubilized membrane proteins, the sample was centrifuged at 100,000 × g for 30 min after stirring for 1 h.

Purification of GfsA-3xFLAG protein

GfsA protein purified using anti-FLAG beads from strain GfsA-3xFLAG (ΔugmA) was used as a source of enzyme. A protein sample purified from strain ΔugmA using the same procedure was used as a negative control. The strains GfsA-3xFLAG (ΔugmA) and ΔugmA were grown in 2 l minimal medium at 30°C for 24 h and then harvested by filtration. The solubilized proteins were prepared from 25 g of wet cells as described above. Mouse-IgG-agarose (100 μl; Sigma-Aldrich) was added to the supernatant and the mixture was gently shaken for 1 h. Mouse-IgG-agarose was removed by centrifugation at 1,400 × g for 10 min. A total of 200 μl anti-FLAG M2 affinity gel (Sigma-Aldrich) was then added to the supernatant and the resultant mixture was gently shaken for 1 h. Anti-FLAG M2 affinity gel was collected by centrifugation at 1,400 × g for 10 min and then washed five times with 50 ml buffer A containing 0.2% CHAPSO. GfsA protein was then eluted with 200 μl buffer A with 0.2% CHAPSO containing 0.5 μg μl⁻¹ 3xFLAG peptide (Sigma-Aldrich). The eluted proteins obtained from strains GfsA-3xFLAG (ΔugmA) and ΔugmA were designated GfsA-3xFLAG and control, respectively. The protein concentration was determined using the bicinchoninic acid protein assay reagent (Thermo Fisher Scientific, Waltham, MA) with bovine serum albumin as a standard.

In-vitro assay of galactofuranose antigen synthase activity

UDP-Galf was synthesized and then purified by HPLC, as previously described (Oppenheimer et al., 2010). UDP-Galf was used as a sugar donor and purified WscA-HA, galactomannoproteins from ΔgfsA (GMPΔgfsA), or galactomannoproteins from ΔugmA (GMPΔugmA) were used as acceptor substrates. Galf antigen synthase was assayed in a reaction mixture containing 25 mM HEPES-NaOH (pH 6.8), 50 mM NaCl, 15 mM KCl, 0.5 mM MnCl₂, 2.5% glycerol (w/v), 0.72 mM UDP-Galf acceptor substrate, and 14 μg purified GfsA-3xFLAG protein in a total volume of 10 μl. For the assay, 2.5 μg GMPΔgfsA, GMPΔugmA or 0.25 μg purified WscA-HA was used as an acceptor substrate. The mixture was incubated at 37°C and the reaction was stopped by incubation at 99°C for 5 min. The reaction products were analyzed by immunoblotting (described below).

Antibodies

The following antibodies were used for immunoblotting. Anti-BipA antibodies were described previously (Goto et al., 2004). Anti-FLAG M2 and anti-UgtA/GlfB antibodies were purchased from Sigma-Aldrich. Anti-UgtA/GlfB antibody was raised by subcutaneous immunization of a rabbit with a synthetic peptide (CEPTLPTVNPAVDKPEPPK) conjugated to KLH, and the serum was tested using an enzyme-linked immunosorbent assay. Anti-FLAG M2 mouse monoclonal antibody (1:5000; Sigma-Aldrich), anti-HA F-7 mouse monoclonal antibody (1:5000; Santa Cruz Biotechnology, Inc., Santa Cruz, CA), anti-γ-actin monoclonal antibody clone C4 (1:200; MP Biomedicals, LLC., Santa Ana, CA), UgtA/GlfB (1:1000), and BipA (1:200) were used as primary antibodies. Anti-mouse IgG conjugate HRP (Santa Cruz Biotechnology, Inc.) was used to detect anti-FLAG M2; anti-γ-actin and anti-UgtA/GlfB was used to detect UgtA/GlfB; and anti-rabbit IgG conjugate HRP (Santa Cruz Biotechnology, Inc.) was to detect anti-BipA. EB-A2 antibody, which is one of the components of Platelia Aspergillus EIA (Bio-Rad Laboratories, Hercules, CA), was used at a dilution of 1:10 to detect β-Galf antigen. All secondary antibodies were used at a dilution of 1:5000.

Immunoblotting analysis

Galactomannoproteins were separated by SDS-PAGE and then transferred to a polyvinylidene fluoride (PVDF) membrane using an AE-6677 electroblotter (Atto Co., Tokyo, Japan) at 100 mA for 1.5 h. After incubation of the membrane for 1 h in a blocking buffer containing 10 mM phosphate (pH 7.4), 4% skim milk (Wako Pure Chemical Industries, Ltd., Osaka, Japan), 0.1% Tween 20, and 0.9% (w/v) NaCl, the membrane was transferred to a 5-ml solution of primary antibody. The membrane was incubated for 12 h at 4°C, washed three times with a buffer containing 10 mM phosphate (pH 7.4), 0.1% Tween 20, and 0.9% (w/v) NaCl for a total of 30 min, and then incubated for 1 h with secondary antibody. An ECL Prime Western Blotting Detection System (GE Healthcare, Little Chalfont, UK) was used to visualize the immunoreactive proteins. Chemical fluorescent signals on the membrane were recorded using a MicroChemi imaging system (Berthold Technologies, Bad Wildbad, Germany). Imaging of the ECL-treated membranes was stopped before saturation of the signals.

Colony growth rate determination

Colony growth rates were measured as described previously (Kellner and Adams, 2002). Briefly, conidia from each strain were point-inoculated into the center of agar medium plates. Colony diameters were measured after 24, 48, 72, 96, and 120 h of incubation at 30°C. The growth rates were determined for each colony in millimeters per hour during each of the incubation intervals, i.e., 24–48, 48–72, 72–96, and 96–120 h, and were then averaged across the entire time interval. Measurements of growth rates for all individual strains were performed 20 times.

Analysis of the efficiency of conidiation

The efficiency of conidiation was analyzed as described previously (Oka et al., 2004). Briefly, approximately 10⁵ conidia were spread onto a minimal medium plate (90-mm diameter). After 5 days of incubation at 30°C, the conidia were suspended in 5 ml of 0.01% (w/v) Tween 20 and counted using a hemocytometer.

Microscopy

Hyphae cultured in liquid minimal medium were observed using a Fluoview FV10i confocal laser-scanning microscope (Olympus Co., Tokyo, Japan). The mycelia were harvested and incubated with 10 ng ml⁻¹ Fluorescent Brightener 28 (calcofluor white; Sigma-Aldrich) for 10 min. Conidiophores were observed using a VE-8800 scanning electron microscope (Keyence, Osaka, Japan).

Transcription analysis of gfsA

For RNA extraction from mycelia, conidia (2 × 10⁷) of A. nidulans wt strain (AKUARG89) were inoculated into 100 ml liquid minimal medium and were then grown for 18, 24, 36, and 48 h at 30°C. For RNA extraction, approximately 2 × 10⁵ conidia were spread onto an 84-mm agar plate of minimal medium. After a 3-d incubation at 30°C, conidia were collected and then ground for 90 sec at 2,000 rpm with a Multi-Beads Shocker (Yasui Kikai Co., Osaka, Japan). RNA was extracted using RNAiso PLUS (Takara Bio, Inc.) according to the manufacturer’s protocol and was then quantified using a Quant-iT RNA Assay Kit (Life Technologies Co.). cDNA was synthesized from the RNA using a PrimeScript Perfect Real-Time Reagent kit (Takara Bio, Inc.) according to the manufacturer’s protocol. Real-time reverse transcription (RT)-PCR analysis was performed using a LightCycler Quick 330 system (Roche Diagnostics) with SYBR Premix DimerEraser (Perfect Real Time; Takara Bio, Inc.). The following primers were used: gfsA-RT-F and gfsA-RT-R for gfsA, histone-RT-F and histone-RT-R for the histone H2B, argB-RT-F and argB-RT-R for argB, and spoC1-C1C-RT-F and spoC1-C1C-RT-R for spoC1-C1C (Table S4). The histone H2B gene was used to standardize the mRNA levels of the target genes (Fujioka et al., 2007).

Organelle separation

Organelles were separated on sucrose gradients as described previously (Engel et al., 2009). Briefly, A. nidulans cells were grown in minimal medium at 30°C for 24 h. All manipulations were performed on ice or at 4°C. Cells were harvested and ground in liquid nitrogen with a mortar and pestle. The lysed cells were suspended at 0.5 ml g⁻¹ in 50 mM MOPS-NaOH, pH 7.0. Cell debris was removed by centrifugation at 3,000 × g. A discontinuous gradient was formed by sequentially adding 2.4 mL each of 55% sucrose, 50% sucrose, 45% sucrose, 40% sucrose, and 35% sucrose. A total of 1.7 mL of supernatant was then loaded onto the upper 35% sucrose solution, and the tube was centrifuged at 100,000 × g for 18 h, and a total of 16 fractions were collected. Specific membrane fractions were identified by immunoblotting using antibodies against organelle-specific markers.

Sensitivity testing

To assay the sensitivity of A. fumigatus to antifungal treatment, Etests for micafungin and voriconazole was performed using the appropriate test strips (Sysmex bioMérieux Co., Ltd., Lyon, France) on MEA plates. For the assay, conidia were harvested, washed twice with distilled water, and then counted using a hemocytometer. Conidia (6 × 10⁵) were then evenly spread and test strips were placed in the center of the plate before incubation at 37°C for 24 h.

Software and database searches

The structure of GfsA was analyzed using the SOSUI tool (http://bp.nuap.nagoya-u.ac.jp/sosui/) (Hirokawa et al., 1998). For similarity searches, we used the Multi-fungi BLAST program (http://www.broadinstitute.org/annotation/genome/FGI_Blast/Blast.html) and the AspGD Multi-Genome Search using NCBI BLAST+ (http://www.aspgd.org/cgi-bin/compute/blast_clade.pl) (Altschul et al., 1997).

Supplementary Material

Supp Fig S1-S4 & Table S1-S4

Supplemental Figure 1. Disruption of candidate galactofuranosyltransferase genes. (A) Schematic representation of the DNA structure of the candidate genes in the disruptants. (B) Electorophoresis analyses of products amplified by PCR with the primer pairs ANxxxx-1/argB-R (5′-region) and argB-F/ANxxxx-4 (3′-region).

Supplemental Figure 2. Detection of Galf antigen in proteinase-K treated galactomannoproteins by EB-A2. The extracted galactomannoproteins (5 μg) were incubated at 37°C for 24 h with proteinase K (100 μg/mL) in 10 mM Tris-HCl (pH 7.8), 10 mM EDTA, and 0.5% SDS prior to immunoblotting analysis with EB-A2 antibody.

Supplemental Figure 3. Protein structure of GfsA. (A) Schematic representation of GfsA protein (amino acids 1–532). A putative transmembrane (TM) domain (black bar) and conserved DXD motif (gray bar) were identified. Horizontal bar, putative signal sequence. Asterisks, potential N-glycosylation attachment sites.

Supplemental Figure 4. In-vitro analysis of Galf antigen synthase activity of the solubilized proteins extracted from ΔugmA and ΔugmAΔgfsA. Enzymatically synthesized UDP-Galf was used as a sugar donor and galactomannoproteins from strain ΔgfsA (GMPΔgfsA) were used as acceptor substrates. Enzyme activities were assayed as described in the Experimental procedures. Reaction products using solubilized protein extracted from ΔugmA (0.1 mg) (lanes 1 and 2) and ΔugmAΔgfsA (0.1 mg) (lanes 3 and 4), and 0.72 mM UDP-Galf (lanes 1 and 3), lacking UDP-Galf (lanes 2 and 4) were analyzed by immunoblotting with EB-A2 antibody.