The Glycosylphosphatidylinositol-Anchored DFG Family Is Essential for the Insertion of Galactomannan into the β-(1,3)-Glucan–Chitin Core of the Cell Wall of Aspergillus fumigatus

The fungal cell wall is a complex and dynamic entity essential for the development of fungi. It is composed mainly of polysaccharides that are synthetized by protein complexes. Enzymes involved in postsynthesis polysaccharide modifications, such as cleavage, elongation, branching, and cross-linking, are essential for fungal life. Here, we investigated in Aspergillus fumigatus the role of the members of the Dfg family, one of the 4 GPI-anchored protein families common to yeast and molds involved in cell wall remodeling. Molecular and biochemical approaches showed that DFG members are required for filamentous growth, conidiation, and cell wall organization and are essential for the life of this fungal pathogen.

polysaccharides encountered in the fungal kingdom, have been identified previously (3)(4)(5). However, the transglycosidases responsible for the branching and cross-linking of these linear polysaccharides in a 3D rigid skeleton are only beginning to be discovered (6). Early studies have shown the merits of concurrent analyses of putative transglycosylases in different fungal species such as Saccharomyces cerevisiae and Aspergillus fumigatus with different cell wall compositions but similar central polysaccharide 3D cores. Such analyses has led to the selection of four groups of glycosylphosphatidylinositol (GPI)-anchored proteins common to all fungi which could potentially have a central role in the transglycosylation of the cell wall structural polysaccharides. First, a family of ␤-(1,3)-glucanosyltransferase activity discovered in A. fumigatus and encoded by the GEL gene or GAS gene is responsible for the elongation of the ␤-(1,3)-glucans and is essential in the biosynthesis of the cell wall (53,54,55). This family contains two subgroups that depend on the presence of a carbohydrate binding domain, which is responsible for a dual form of enzyme activity, ␤-glucan elongation and branching (6). Second, the Sps2p/Ecm33p family of GPI-anchored proteins has been described previously in both S. cerevisiae and A. fumigatus (7,8). Members of the Sps2p family play an essential role in the formation of the ascospore cell wall in S. cerevisiae, whereas in A. fumigatus, Ecm33p is important for conidial morphogenesis and virulence. However, its enzymatic function remains unknown. Third, the Crhp family in S. cerevisiae is composed of three genes which are involved in the linkage of chitin to ␤-(1,6)-glucan in the cell wall (9,10). Although also present in A. fumigatus, it is obvious that the Crh proteins do not have the same function in S. cerevisiae and A. fumigatus since there is no ␤-(1,6)-glucan in the cell wall of A. fumigatus. Finally, the two genes in the DFG (for "defective in filamentous growth") family encode two GPI-anchored proteins with redundant activities in both S. cerevisiae and Candida albicans (11,12). Although single knockouts of DFG5 and DCW1 are viable, a double knockout was synthetically lethal in both S. cerevisiae and C. albicans (11,12). Cells depleted in either Dfg5p or Dcw1p released GPI-anchored cell wall proteins (GPI-CWPs) into the medium and showed increased cell volume, suggesting an alteration of the cell wall organization. However, the exact biochemical function of the Dfg proteins in yeast as well as in filamentous fungi remains unknown.
To understand the function of Dfg proteins in A. fumigatus, single-deletion and multiple dfg mutants targeting deletions of the members of the entire DFG family were constructed and analyzed. DFG multiple deletion resulted in the total loss of the cell wall galactomannan (GM), which was associated with severe growth phenotypes.

RESULTS
The DFG family in A. fumigatus. A BLAST query of the S. cerevisiae Dfg5p/Dcw1p protein sequences against the A. fumigatus genome database (https://fungi.ensembl .org/Aspergillus_fumigatusa1163/Info/Index) identified seven paralogs. The percentages of identity and similarity ranged between 15% and 42% and between 30% and 60%, respectively (see Table S1 in the supplemental material). In silico analysis revealed that all of the proteins encoded by these genes contained a secretory signal peptide at the N-terminal region (identified by the use of the SignalP website [http://www.cbs .dtu.dk/services/SignalP/]), a hydrophobic region at the C-terminal region with an -site characteristic of GPI-anchored proteins (except Dfg2p) (identified by the use of the BigPI website [http://mendel.imp.ac.at/gpi/fungi_server.html] and the PredGPI website [http://gpcr.biocomp.unibo.it/predgpi/pred.htm]), and the glycosyl hydrolase domain GH76 characteristic of Dfg5p/Dcw1p proteins from yeast (see Fig. S1 in the supplemental material). The expression levels of the DFG genes seen during different stages of development of A. fumigatus (swollen conidia, germinating conidia, and mycelium grown in Sabouraud liquid medium; sporulating mycelium and conidia cultivated on Sabouraud agar medium) were analyzed by quantitative real-time PCR (qRT-PCR) (Fig. 1). With the exception of DFG6, all of the DFG genes were expressed in the different fungal stages but their relative expression levels were highly dependent on the growth stage considered (Fig. 1). DFG3 was the gene that was the most highly expressed in germinating conidia. Three DFG genes (DFG1, DFG2, and DFG5) were highly expressed in conidia and sporulating mycelium, while DFG4 was highly expressed only in conidia and DFG7 only in sporulating mycelium.
Mycelial growth and hyphal morphology of ⌬dfg mutants. At 37°C, no growth differences were observed for the single-deletion strains (Δdfg1, Δdfg2, Δdfg4, Δdfg5, and Δdfg7) and multiple-deletion strains (Δdfg5/2, Δdfg5/2/1, Δdfg5/2/1/4) in comparison to the parental strain during growth on malt agar medium ( Fig. 2A). Only the single-knockout Δdfg3 mutant and the multiple-knockout mutants containing a DFG3 deletion (Δdfg5/2/1/3, Δdfg5/2/1/3/4, and Δdfg5/2/1/3/4/7) displayed a significant reduction in vegetative mycelial growth compared to the parental strain and the DFG3 revertant strain (Fig. 2). In the sextuple-deletion mutant, the absence of DFG6 expression was verified by qRT-PCR at different stages of development of A. fumigatus (not shown). This result indicated that the sextuple-deletion mutant can be considered to be a mutant lacking the entire DFG gene family. Even though all mutants were evaluated, only the phenotypic analyses of the ⌬dfg3 single-deletion mutant and the ⌬dfg5/2/1/ 3/4/7 sextuple-deletion mutant are reported here. The differences in the amounts of mycelium produced by the Δdfg mutants and the parental strain in Sabouraud liquid medium were much lower than those seen in agar media (Fig. 3B). However, the mycelia of these mutants grown in both solid and liquid media were highly branched (Fig. 2B). In liquid cultures, this hyperbranching phenotype resulted in the production of a myriad of very small and tight fungal balls which looked different from the homogeneous mycelial mass produced by the parental strain (Fig. 3A). Conidiation, conidial morphology of ⌬dfg mutants, and susceptibility to drugs. The capacity of the different mutants to conidiate was assayed on malt agar-containing tubes following incubation for 10 days at room temperature. At 37°C, the slant was entirely covered by the fungus and no conidiation difference was observed for the single-deletion strains (Δdfg1, Δdfg2, Δdfg4, Δdfg5, and Δdfg7) and multiple-deletion strains (Δdfg5/2, Δdfg5/2/1, and Δdfg5/2/1/4) compared to the parental strain. Interestingly, the single-deletion ⌬dfg3 mutant produced as many conidia as the parental strain whereas the sextuple-deletion Δdfg5/2/1/3/4/7 mutant displayed a drastic reduction in conidiation (Table 1). This conidiation defect could be compensated for by the addition of 6% KCl (Table 1). Similarly, 52% of the resting conidia from the sextuple-deletion Δdfg5/2/1/3/4/7 mutant were intracellularly stained with fluorescein isothiocyanate (FITC), whereas the resting conidia from the parental and ⌬dfg3 strains were weakly stained by FITC at the cell surface (Fig. S3). Moreover, conidia of the sextuple-deletion Δdfg5/2/1/3/4/7 mutant germinated faster (ϳ18% germ tubes were formed after 4 h of incubation at 37°C) than those of the Δdfg3 and parental strains, which showed nearly no germinated conidia after the same 4 h of incubation time (Fig. 4A). While conidia of the parental and mutant strains were similar in size in the resting stage, the swollen conidia of the sextuple-deletion Δdfg5/2/1/3/4/7 mutant were approximately 1.5 times larger than those of the parental strain (Fig. 4B). In line with what was seen for FITC staining, calcofluor white (CFW) staining indicated that the cell walls of sextupledeletion Δdfg5/2/1/3/4/7 mutant were more permeable than those of the parental strain, since intracellular labeling of around 50% of the swollen conidia was seen following 5 min of exposure to the stain (Fig. 4B). The single-deletion Δdfg3 mutant and the sextuple-deletion Δdfg5/2/1/3/4/7 mutant were more susceptible to voriconazole and itraconazole and to the cell wall-disturbing calcofluor white dye than the parental strain on RPMI medium after 72 h of incubation at 37°C (Fig. 5). All these conidial phenotypes suggested that the cell wall integrity was altered in the Δdfg5/2/1/3/4/7 mutant and that the alterations were associated with defects in cell wall permeability.
Impact of DFG deletions on cell wall galactomannan (GM). In A. fumigatus, GM is either cross-linked to cell wall ␤-(1,3)-glucans or membrane bound through a GPI anchor (LGM) (14)(15)(16). The chemical compositions of the alkali-insoluble (AI) and alkali-soluble (AS) fractions of the A. fumigatus cell wall of parental and mutant strains are shown in Table 2. A decrease in the level of cell wall galactomannan (GM) was observed in both the AI and AS fractions in the ⌬dfg3 mutant, and its complete absence was noted in the AI fraction of the Δdfg5/2/1/3/4/7 mutant. Quantification of GM carried out with the Δdfg mutants was confirmed in situ, as shown by the absence of labeling of the GM with an antigalactofuran monoclonal antibody at the cell surface of the Δdfg3 mutant (Fig. 6). The reduction in the level of cell wall GM was primarily compensated for by an increase in the glucose content within the AI fraction and by an increase in the galactosamine content within the AS fraction, corresponding to larger amounts of ␤-(1,3)-glucan and galactosaminogalactan, respectively ( Table 2). The single-deletion Δdfg3 mutant showed a significant reduction in GM content (40% of the level seen with the AI fraction) compared with the parental strain, whereas no cell wall cross-linked GM was detected in the sextuple-deletion mutant (Fig. 7A). In addition, quantification of the GM in the single-deletion ⌬dfg3 mutant and the multiple-deletion ⌬dfg5/2/1/3, ⌬dfg5/2/1/3/4, and ⌬dfg5/2/1/3/4/7 mutants showed an additive effect consisting of a concomitant decrease in the level of cell wall GM content corresponding to an increase of the number of DFG deletions.
In contrast to the cell wall GM, the GM content was slightly increased in membrane preparations (Fig. 7B). The amount of GM in the ⌬dfg3 mutant was close to 110% of the GM content in the membrane of the parental strain and reached 160% of the parental strain level in the sextuple-deletion mutant (Fig. 7B), In addition, no significant differences were observed in the molar ratios of the hexose content of the membrane-bound GM of the parental and the sextuple-deletion Δdfg5/2/1/3/4/7 strains. Structural analysis of membrane-bound GM showed the presence of terminal galactofuranose, 5-Osubstituted galactofuranose, 2-O-substituted mannose, 6-O-substituted mannose, 2,6di-O-substituted mannose, and 2,3-di-O-substituted mannose residues, typical of the GM sequence (14,16). The lipid anchor of LGM was isolated from the sextuple-deletion mutant and analyzed by electrospray-mass spectrometry (ES-MS). The major pseudomolecular ion [M-H]at m/z ϭ 1,086 characterized the GPI-related anchor of LGM with the presence of a glucosamine residue linked to an inositolphosphoceramide moiety, which was composed of a C 18 -phytosphingosine and a 2-hydroxy-C 24:0 fatty acid, as previously described (16) (Fig. S4). Taking into account all these structural data, the level of GM from the Δdfg5/2/1/3/4/7 mutant is identical to that from the parental strain, indicating that the intracellular biosynthetic pathway of GM was not altered by DFG deletions.
Interestingly, the amount of GM in the culture filtrate of the Δdfg3 and Δdfg5/2/1/ 3/4/7 mutants increased substantially in comparison to the level seen with the parental strain (Fig. 7C). An additive effect with a concomitant increase in the free GM content in the culture filtrate was seen with the increase of the number of DFG deletions.
In conclusion, the intracellular biosynthetic pathway of GM was not altered by DFG deletions qualitatively or quantitatively. In contrast, the DFG deletion resulted in a defect in insertion of this polysaccharide into the cell wall and in associated extracellular release of the GM into the external medium. Even though only the deletion of the DFG 3 gene resulted in a morphogenetic phenotype among all the single-deletion Δdfg mutants, the comparison of the GM phenotypes of the Δdfg3 mutant and the Δdfg5/ 2/1/3/4/7 multiple-deletion mutant indicated additivity of the functions of these DFG proteins in the insertion of GM into the cell wall.
Complementation of the ⌬dfg5/dcw1 yeast mutant by AfDFG3. A doubleknockout ⌬dcw1/⌬dfg5 mutant was shown previously to be synthetically lethal in yeast (11). Functional complementation by A. fumigatus DFG3 (AfDFG3) in S. cerevisiae was investigated using a thermosensitive dcw1 ts /⌬dfg5 mutant. This mutant was unable to grow on solid minimum medium at 37°C and produced a 15% proportion of enlarged yeast cells in liquid medium at 37°C. The introduction of the AfDFG3 gene into the mutant strain restored normal growth at 37°C at a level that was similar to that seen with the control DCW1/⌬dfg5 mutant strain (Fig. 8), showing that the AfDFG3 genes and S. cerevisiae DCW1 (ScDCW1) genes share similar biological activities.  Cell Wall Galactomannan and DFG in A. fumigatus

DISCUSSION
The phenotypes of ⌬dfg mutants are different among yeast and mold. In this report, we describe the results of characterization of the DFG family in A. fumigatus. The deletion of DFG genes led to a major reduction in vegetative growth with hyperbranched hyphae. In addition, it was shown that the genes of the DFG family have additive biological activities, as shown also in S. cerevisiae. Nine DFG paralogs have been annotated in the genome of Neurospora crassa, and among the single-deletion mu- tants, only two have been studied, both of which showed altered morphology (17). Of the two mutants, the ⌬dfg5 mutant (NCU03770) showed a reduction in colony size and a hyperbranched mycelial pattern, in similarity to the ⌬dfg3 mutant in A. fumigatus. The ⌬dcw1 mutant (NCU08127) did not show a significant growth phenotype in Neurospora, but the growth phenotype of the ⌬dfg5/⌬dcw1 double-deletion mutant of N. crassa showed more-severe effects than were seen with the strain with a single ⌬dfg5 deletion, with increased susceptibility to cell wall-perturbing agents, such as caspofungin (17). In yeast, single deletion of the paralogous DFG5 gene or DCW1 gene also leads to cell wall alterations and growth defects. The Sc⌬dfg5 mutant has been reported previously to be defective in filamentous growth, cell polarity, and elongation (18). Mutant Sc⌬dcw1 (for "defective in cell wall") was hypersensitive to zymolyase, a cell wall-digesting enzyme, suggesting a role in cell wall organization (19). In C. albicans, a mutant lacking DFG5 was defective in hyphal formation at alkaline pH whereas a mutant lacking DCW1 had no obvious phenotype. In both C. albicans and S. cerevisiae, deletions of DFG5 and DCW1 were synthetically lethal (12,19). In A. fumigatus, the entire deletion of all members of the DFG family (since DFG6 remained unexpressed in the sextuple-deletion mutant) was not lethal. Our data showed, as often reported for orthologous genes of yeast and filamentous fungi, that the phenotypes of their respective mutants were different. Among the single-deletion mutants, only DFG3 deletion led to a growth defect; the major phenotypic difference was seen in the sextuple-deletion mutant, where additive effects on mycelial morphology, germination rate, drug sensitivity, cell wall composition, and permeability were observed with increasing numbers of DFG gene deletions.
Function of the Dfg proteins. On the basis of sequence homology, Dfg proteins have been assigned to the GH76 glycoside hydrolase family in the CAZy database (20). Dfg proteins were predicted to act as endo-␣-(1,6)-mannanases based on the homologies of the sequence of a single member of this family with that of the Aman6 protein from Bacillus circulans (21,22). However, to date, there has been no biochemical evidence found to indicate that any fungal Dfg protein acts as an endo-␣-(1,6)mannanase or possesses any hydrolytic activity with respect to cell wall polysaccharides. In our hands, efforts to produce recombinant AfDfg3p were not successful. Since ScDcw1p showed 32% identity with AfDfg3p (see Table S1 in the supplemental material) and the S. cerevisiae dcw1 ts /⌬dfg5 mutant was functionally complemented by AfDFG3 (Fig. 8), we investigated the enzymatic activity of the recombinant ScDcw1 protein. The glycosylhydrolase activity of the recombinant ScDcw1 protein was assayed with several ␣-mannosides (see Text S1 in the supplemental material). The transglycosylase activity has been also investigated using galactomannan as a donor and soluble ␤-(1,3)-glucans or the cell wall AI fraction from S. cerevisiae as an acceptor (Text S1). No hydrolase or transfer activity has been detected with any of these substrates, even with the sensitive fluorometry assay used with the bacterial GH76 member (23) (Text S1). The enzyme activity of fungal GH76 family members therefore remains uncharacterized but certainly does not represent that of an ␣-mannanase.
The increased secretion of GPI-anchored proteins in the ⌬dfg5 mutant or the ⌬dcw1 mutant in S. cerevisiae suggested that these proteins were involved in the cross-linking of GPI-CWPs to the cell wall glucans (17,24). However, in addition to the absence of any biochemical demonstration of such transglycosylase activity, ascribing a cross-linking function remains controversial. In C. albicans, shutoff strains for DFG5 and DCW1 were previously shown to exhibit the release of both GPI-anchored and non-GPI cell wall mannoproteins (25). Moreover, in previous studies, secretion of very few GPI proteins, including Acw1p (ϭECM33), Gel1p in N. crassa, and Cwp1p in S. cerevisiae (11,17), was demonstrated, whereas localization of the major GPI-anchored protein, Gas1p, remained unchanged in both Dfg5 and Dcw1 single-deletion mutants in S. cerevisiae (26). In N. crassa, the same proteins were released in the culture supernatant of mutants with deletions of very different genes, such as those encoding a putative hyaluronic synthase (CPS1) or mannosyltransferase (OCH1) (27,28). In A. fumigatus, DFG deletions led also to a 3-fold increase in the amount of secreted proteins (see Fig. S5 in the supplemental material). However, the six overproduced proteins were mainly non-GPIanchored proteins such as endopolygalacturonase, endochitinases, and pectate lyase (Fig. S5). Such a modification of the secreted protein pattern has been indeed reported in other A. fumigatus mutants after genes coding for proteins with a known function in cell wall biosynthesis were deleted (29,30). Taking the data together, the results suggest that Dfg proteins are not involved in the cell wall localization of GPI-anchored proteins, at least in filamentous fungi. The modification of the cell wall structure due to the loss of cell wall galactomannan in the ⌬dfg3 single-deletion and multiple-deletion mutants leads to facilitation of the release of secreted proteins in transit through the cell wall. These cell wall modifications lead to a change in permeability such as was demonstrated when the Δdfg3 single-deletion mutant and the Δdfg5/2/1/3/4/7 multiple-deletion mutant were incubated with CFW or FITC ( Fig. 5; see also Fig. S3). Intracellular labeling with FITC and CFW has been repeatedly performed with cell wall mutants of A. fumigatus to demonstrate the effect of a gene deletion on cell wall permeability (5,8,(29)(30)(31)(32). However, we have no idea of the nature of the material stained intracellularly by CFW or FITC.
Dfg proteins and the cell wall GM in A. fumigatus. GM is a polymer composed of a linear ␣-(1,2)/␣-(1,6)-mannan chain with short side chains of ␤-(1,5)-galactofuran (Galf) (33). This polysaccharide can be covalently bound to ␤-(1,3)-glucans in the cell wall, bound to the plasma membrane by a GPI anchor, or present in the extracellular environment. Despite of its importance in fungal morphogenesis and in the host immune response (31), GM biosynthesis is poorly understood. Previous studies have shown that GM biosynthesis takes place in the Golgi apparatus, into which sugar-donor (UDP-galactofuranose and GDP-mannose) are transported prior to polysaccharide polymerization (34)(35)(36). The polymerization of galactofuran was due to the action of a specific galactofuranosyltransferase, GfsA (37,38). Two Ktr/Mnn2 mannosyltransferases have been recently identified to be essential to the mannan polymerization (31,39). Interestingly, the absence of Ktr mannosyltransferases led to the absence of cell wall GM and to filamentous growth with an hyperbranched mycelium and a conidiation defect (31). The absence of cell wall cross-linked GM in Δdfg mutants could suggest a putative role of Dfg family members in mannan polymerization. However, our biochemical characterization showed that the dfg mutants still produced membrane-bound GM with the same chemical structure as that of the wild type (WT), showing that Dfg proteins are not involved in GM biosynthesis. Our study results show that these DFG proteins play a key role in the cross-linking of GM to ␤-(1,3)-glucans through an as-yet-undiscovered form of remodelling activity. The study also confirmed that all of the polysaccharide components of the cell wall skeleton [␤-(1,3)-glucans, chitin, and galactomannan] are essential for fungal life whereas the alkali amorphous content has a nonstructural function and is mainly involved in the communications with the external milieu (32).

MATERIALS AND METHODS
Culture conditions. Parental and mutant strains of A. fumigatus were grown at 37°C in Aspergillus minimal medium (AMM; 1% glucose and 5 mM ammonium tartrate), Sabouraud medium (2% glucose, 1% Mycopeptone [Difco]), RPMI medium (Sigma), or 2% malt medium (Cristomalt). Media were either used in liquid form or supplemented with 2% agar. When necessary, 6% KCl was added to the media to enhance conidiation. Conidia were collected from agar medium plates after 10 days of growth at 37°C, using water containing 0.05% Tween 20.
Construction and complementation of the ⌬dfg mutants. The single-deletion and multipledeletion mutants were constructed in the CEA17_ΔakuB KU80 background (40) using the ␤-rec/six sitespecific recombination system (13). The self-excising ␤-rec/six blaster cassette containing the hygromycin resistance marker was released from pSK529 plasmid using FspI restriction enzyme. The dfg replacement cassette containing the marker module flanked by 5= and 3= homologous regions of the target gene was generated by using primers listed in Table S3 in the supplemental material and cloned into the pUC19 vector using GeneArt seamless cloning and assembly (Life Technologies, Carlsbad, CA, USA). The corresponding replacement cassettes were released from the resulting vector via the use of either EcoRV or FspI, respectively. The CEA17ΔakuB KU80 parental strain was transformed with the dfg replacement cassettes by electroporation to generate the dfg single-deletion mutants. Transformants were analyzed by PCR and Southern blotting using the digoxigenin (DIG) probe protocol (Roche Diagnostics) (see Fig. S2 in the supplemental material). For the construction of multiple-deletion strains, single-deletion mutants were cultivated in the presence of 2% xylose-containing minimal medium, which allows the excision of the selection marker by triggering recombination between the six recognition regions in the ␤-rec/six cassette. Because deletion of DFG4 results in resistance to hygromycin, the sextuple-deletion mutant was constructed with the phleomycin resistance marker (41). The deletion cassette was made by fusion PCR with primers listed in Table S3 by using the double-joint PCR method.
Complementation of the Δdfg3 mutant was carried out by reintroduction of the parental copy of the gene flanked by the hygromycin resistance cassette and a 3= flanking region ( Fig. S2; see also Table S3). The complementation cassette was transformed into the cassette-excised Δdfg3 mutant. The presence of the parental copy of the gene at the DFG3 locus was confirmed by Southern blot analysis (Fig. S2). All mutants are detailed in Table S2.
Complementation of Sc⌬dcw1 ts /⌬dfg5 mutant by AfDFG3. S. cerevisiae BY230 strain was a derivative of S288c in which both DCW1 and DFG5 chromosomal loci were deleted. Since they are synthetically lethal, the strain therefore also carried p413TEF.DCW1, a URA3 plasmid (42) containing the DCW1 gene. The DCW1 deletion in a Δdfg5 strain was obtained from the haploid BY4741 deletion collection. A DCW1 deletion plasmid was generated by cloning 5= and 3= flanking regions of DCW1 into YIPLAC211 (43). The flanking regions were generated by PCR with primers 1975PCR with primers , 1976PCR with primers , 1977PCR with primers , and 1978 (Table S3). Following digestion with Mlu1, this plasmid was integrated at the DCW1 locus of the ⌬dfg5 strain. This integrant was transformed with a CEN-ARS HIS3 plasmid, p413TEF.DCW1, carrying the DCW1 open reading frame (ORF) cloned as a BamH1-Xho1 fragment (generated with primers 2309 and 2310 and sequence verified). The genomic locus was then deleted by selection on plates with 5-fluoro-orotic acid followed by screening for the ORF deletion by PCR using flanking primers 2306 and 2307 to generate strain JT347 (DCW1/⌬dfg5). In strain JT346 (dcw1 ts /dfg5⌬), plasmid shuffling was used to replace the p413TEF.DCW1 plasmid with p413TEF.dcw1 ts , a temperature-sensitive allele unable to support growth at 37°C (J. Trow and B. P. Cormack, unpublished data). Yeast strain JT346 (dcw1 ts /⌬dfg5) was used for complementation performed with AfDFG3. Yeasts were grown at 30°C and 220 rpm in either a standard Cell Wall Galactomannan and DFG in A. fumigatus YEPD medium (10 g/liter yeast extract, 20 g/liter Bacto peptone, 20 g/liter glucose) or in YNB medium (1.7 g/liter yeast nitrogen base without amino acids and ammonium, 5 g/liter ammonium sulfate, 20 g/liter glucose) supplemented with the auxotrophic requirements.
The AfDFG3 cDNA was synthetized by Life Technology SAS (St Aubin, France). A NotI restriction site and an XhoI restriction site were incorporated at the 5= and 3= ends of the cDNA, respectively, and cloned into pMA plasmid containing an ampicillin resistance marker. The AfDFG3 cDNA was subcloned in plasmid pREP3-ADH (44) containing a LEU2 gene to obtain pREP3-AfDFG3 after digestion by NotI/XhoI. Then, the dcw1 ts /⌬dfg5 mutant was transformed with 3 g of plasmid pREP23-AfDFG3 following the lithium acetate method (45). As a control, the dcw1 ts /⌬dfg5 mutant was transformed with the pREP3-ADH plasmid alone. The transformants were selected on YNB plates without leucine, and the levels of expression of the AfDFG3 genes in the S. cerevisiae mutant were checked by RT-PCR using primer pair AfDFG3comp1 and AfDFG3comp2 (Table S3). Phenotypes of the JT347 (DCW1/⌬dfg5), JT346 (dcw1 ts / ⌬dfg5), and complemented (dcw1 ts /⌬dfg5::AfDFG3) strains were analyzed on YNB medium at 30°C and 37°C.
Quantitative real-time PCR analysis. Fungal material was disrupted by the use of 0.5-mm-diameter glass beads in 500 l of saturated phenol (Interchim, Montluçon, France) (pH 4.5), and RNA was isolated as described earlier (5) or by using a Qiagen RNeasy minikit. Quantitative PCR assays were performed as previously described (5). The expression ratios were normalized to EF1␣ expression levels and were calculated according to the 2 ΔCT (threshold cycle) method (46). The absence of genomic DNA contamination was verified with negative controls without reverse transcriptase. Three independent biological replicates were performed. The specificity of each primer (Table S3) was checked by agarose gel electrophoresis of RT-PCR products.
Phenotype of the ⌬dfg strains. Mycelial growth of the different strains was measured on 2% malt solid media after 48 h of incubation at 37°C. Agar pieces containing the entire colony of each of the strains were boiled 5 min in water to eliminate medium and extensively washed with hot water, and mycelial dry weight was recorded after overnight incubation at 80°C. The morphology of the mutant strains was determined during a kinetic study of the fungus grown in Sabouraud medium after staining of the fungus with calcofluor white (final concentration of 5 g/ml). Conidia and mycelium were observed with a fluorescence microscope (Evos; Life Technologies) (excitation delta [ ex], 357/44 nm; emission delta [em], 447/66 nm). To quantify growth, 24-h-old mycelial cultures that had been maintained at 37°C were filtered, washed, and dried overnight at 80°C and the dry weights were recorded. The conidiation rates were estimated following inoculation of conidial suspensions (100 l, 10 4 /ml) into three tubes containing 2% malt agar (10 ml/tube) or malt agar containing 6% KCl. After 10 days at 25°C, conidia were recovered with aqueous 0.05% Tween 20 solution and counted using a hemocytometer.
The susceptibility of strains to antifungal and cell wall-disturbing compounds was estimated by spotting 10-fold serial dilutions of conidia (starting from 2 ϫ 10 6 spores) onto RPMI plates supplemented with the following drugs: itraconazole [ Carbohydrate analysis of the cell wall, membrane fractions, and culture supernatant. Following 24 h of growth in Sabouraud liquid medium at 37°C and 150 rpm, mycelia and culture supernatants were separated by filtration. Macromolecules from the culture medium were precipitated by the use of three volumes of ethanol at 4°C overnight and collected by centrifugation (5 min, 4,000 ϫ g). Cell walls and membranes were obtained after mycelium disruption and centrifugation (5,16). Lipogalactomannan (LGM) was purified from each membrane preparation as previously described (16). The amount of galactomannan present in the culture filtrate was estimated using a Platelia kit as described by the manufacturer (Bio-Rad, Marnes la Coquette, France) with purified GM as the standard. Prior to the GM assay, proteins from the culture supernatant were eliminated by solid-phase extraction (SPE) on Sep-Pak classic C 18 cartridges (Waters) as previously described (48). Polysaccharides from the cell wall were separated based on their alkali solubility (5). Neutral hexoses and osamines were quantified by colorimetric and chromatographic assays (5,49,50). Proteins were quantified by use of the bicinchoninic acid (BCA) assay (Thermo Scientific), analyzed by SDS-PAGE, and identified by tandem mass spectrometry (MS/MS) analysis.
Galactomannan analysis. The analysis of glycosidic linkages in LGM was performed by methylation (51). The lipid anchor of LGM was released by mild acid hydrolysis (50 mM HCl, 100°C for 15 h), purified on a silica gel column, and analyzed by electrospray-mass spectrometry (16). Galactomannan (GM) amounts were estimated using a Platelia kit according to the instructions of the manufacturer (Bio-Rad, Marnes la Coquette, France) with purified GM as the standard. GM from the membrane preparation was solubilized with 2% Triton X-100. After trypsin digestion (1 mg trypsin for 10 mg of protein at 37°C for 24 h), GM was purified by SPE on Sep Pak classic C 18 cartridges and eluted with 5% propanol-1 (48). Mass spectrometry analysis of the lipid anchor was carried out on a Synapt G2Si instrument (Waters Corp., Milford, MA, USA). The source temperature was set to 80°C. The capillary and cone voltages were set to 1,500 and 40 V. Time of flight (TOF) data were collected between m/z 50 and 2,000, at a low level of collision energy (10 eV), in negative mode. Argon was used as the collision gas. Scans were collected for 1 s. An external calibration was done with clusters of NaI, and the mass range of calibration was m/z 50 to 2,000. Mass Lynx 4.1 was used for both acquisition and data processing. Samples were dissolved in chloroform/methanol (1/4 [vol/vol]) and introduced in nanoelectrospray mode via the use of a coated, medium-sized nano-electrospray ionization (ESI) capillary (Proxeon).
Fluorescence microscopy. Mycelia of the Δdfg and parental strains were fixed using p-formaldehyde (at a concentration of 2.5% in phosphate-buffered saline [PBS]) for one night at 4°C, washed three times with 0.1 M NH 4 Cl-PBS and once with PBS, and then incubated with antigalactomannan antibody (52). Galactomannan was labeled with a rat anti-galactofuran monoclonal antibody (EBA2; a kind gift of M. Tabouret, Bio-Rad, Steenvorde) and a secondary FITC-conjugated goat anti-rat (Sigma) antibody. A monoclonal antibody with the same isotype was used as a negative control.
Statistical analysis. At least three biological replicates were performed per experiment; the statistical significance of the results was evaluated by one-way variance analysis using JMP1 software (SAS Institute, Cary, NC, USA).