SUN Proteins Belong to a Novel Family of β-(1,3)-Glucan-modifying Enzymes Involved in Fungal Morphogenesis*

Background: SUN proteins are involved in yeast morphogenesis, but their function is unknown. Results: SUN protein plays a role in the Aspergillus fumigatus morphogenesis. Biochemical properties of recombinant SUN proteins were elucidated. Conclusion: Both Candida albicans and Aspergillus fumigatus SUN proteins show a β-(1,3)-glucanase activity. Significance: The mode of action of SUN proteins on β-(1,3)-glucan is unique, new, and original. In yeasts, the family of SUN proteins has been involved in cell wall biogenesis. Here, we report the characterization of SUN proteins in a filamentous fungus, Aspergillus fumigatus. The function of the two A. fumigatus SUN genes was investigated by combining reverse genetics and biochemistry. During conidial swelling and mycelial growth, the expression of AfSUN1 was strongly induced, whereas the expression of AfSUN2 was not detectable. Deletion of AfSUN1 negatively affected hyphal growth and conidiation. A closer examination of the morphological defects revealed swollen hyphae, leaky tips, intrahyphal growth, and double cell wall, suggesting that, like in yeast, AfSun1p is associated with cell wall biogenesis. In contrast to AfSUN1, deletion of AfSUN2 either in the parental strain or in the AfSUN1 single mutant strain did not affect colony and hyphal morphology. Biochemical characterization of the recombinant AfSun1p and Candida albicans Sun41p showed that both proteins had a unique hydrolysis pattern: acting on β-(1,3)-oligomers from dimer up to insoluble β-(1,3)-glucan. Referring to the CAZy database, it is clear that fungal SUN proteins represent a new family of glucan hydrolases (GH132) and play an important morphogenetic role in fungal cell wall biogenesis and septation.

In yeasts, the family of SUN proteins has been involved in cell wall biogenesis. Here, we report the characterization of SUN proteins in a filamentous fungus, Aspergillus fumigatus. The function of the two A. fumigatus SUN genes was investigated by combining reverse genetics and biochemistry. During conidial swelling and mycelial growth, the expression of AfSUN1 was strongly induced, whereas the expression of AfSUN2 was not detectable. Deletion of AfSUN1 negatively affected hyphal growth and conidiation. A closer examination of the morphological defects revealed swollen hyphae, leaky tips, intrahyphal growth, and double cell wall, suggesting that, like in yeast, AfSun1p is associated with cell wall biogenesis. In contrast to AfSUN1, deletion of AfSUN2 either in the parental strain or in the AfSUN1 single mutant strain did not affect colony and hyphal morphology. Biochemical characterization of the recombinant AfSun1p and Candida albicans Sun41p showed that both proteins had a unique hydrolysis pattern: acting on ␤-(1,3)-oligomers from dimer up to insoluble ␤-(1,3)-glucan.

Referring to the CAZy database, it is clear that fungal SUN proteins represent a new family of glucan hydrolases (GH132) and play an important morphogenetic role in fungal cell wall biogenesis and septation.
In Saccharomyces cerevisiae, four paralogous SUN genes, namely SIM1, UTH1, NCA3, and SUN4 (1)(2)(3), are classified into two groups. Members of group I encode proteins with a conserved C-terminal region of ϳ250 amino acids corresponding to the SUN domain or Pfam-PF03856 (4) comprising four Cys residues in a Cys-X 5 -Cys-X 3 -Cys-X 24 -Cys motif (5). The N-terminal region is less conserved, ranging from 80 to 168 amino acids with a signal peptide and a low complexity region rich in serine and threonine residues. On the contrary, in the group II members, the SUN domain (especially the Cys-rich motif) and the low complexity region are degenerated.
Mounting evidences strongly suggest that the yeast SUN proteins play a role in cell wall biogenesis, septum integrity and cell separation (3,6). In S. cerevisiae, Sun4p is localized in the cell wall and is required, together with Uth1p, for septation (3). In Schizosaccharomyces pombe, deletion of the single class I SUN gene, psu1, is associated with cell wall defects during cell separation, resulting in swollen cells that eventually undergo lysis (7). In Candida albicans, inactivation of SUN41 leads to a defect in the separation of daughter cells from mother cells, whereas simultaneous inactivation of SUN41 and SUN42 is lethal in the absence of osmotic protection. Cell wall defects seen in this double mutant are mainly localized in the region surrounding the septa in mother yeast cells and subapical hyphal compartments (8 -10). Proteomic analyses have shown that Sun41p and Sun42p are found in the C. albicans cell wall and secretome (9,11,12). However, no biochemical function has been associated with any of these SUN proteins, and no direct role in cell wall biosynthesis has been demonstrated. SUN proteins have not been studied in the filamentous ascomycetes. Here, we report the characterization of the SUN genes of Aspergillus fumigatus. The A. fumigatus genome harbors two SUN genes: the class I AfSUN1 and the class II AfSUN2. In the present study, we showed that only AfSUN1 plays a role during morphogenesis. Using the recombinant Sun1p of A. fumigatus and the orthologous Sun41p of C. albicans, it is shown for the first time that SUN proteins bind and hydrolyze ␤-(1,3)-glucan in a very specific manner.

Strains and Culture Media
A. fumigatus mutants constructed in this study were derived from the strain CEA17⌬ku80 (13); all strains were maintained on 2% malt agar slants supplemented, when necessary, with 150 g/ml hygromycin B (Sigma) and/or 20 g/ml phleomycin (InvivoGen). Minimal medium was used for the transformation experiments (14). Cultures were grown in liquid Sabouraud medium (2% glucose containing 1% mycopeptone) for DNA extraction as well as phenotypic analyses or in YPD (1% yeast extract, 2% Bacto peptone, 1% glucose) for RNA extraction. C. albicans used in this study was the strain BWP17 (15), which was maintained at 30°C on YPD.

Construction of A. fumigatus Deletion and Complementation Cassettes by Fusion PCR
The deletion and complementation cassettes used in this work were constructed by fusion PCR as described earlier (18). Primer positions are illustrated in supplemental Fig. S1, and the primer sequences are shown in supplemental Table S1. The Escherichia coli hph gene, coding for hygromycin B phosphotransferase, obtained from the plasmid pAN7-1 (19) was used to replace SUN1. SUN2 was replaced by a lox disruption cassette-borne ble marker (encoding a phleomycin binding protein), obtained from the plasmid pSK341 (a kind gift from S. Krapmann, Georg-August-University Göttingen, Germany). In a first round of PCR, flanking regions 1 and 2 (amplicons 1 and 3, respectively) were amplified from the CEA17⌬ku80 genomic DNA prepared according to Girardin et al. (20), and selection markers (HPH and BLE; amplicon 2) were amplified using plasmids pAN7-1 and pSK341, respectively, and 60-bp chimeric oligonucleotides (primers SunB, SunC, SunD, and SunE). PCR was performed as follows: 30 cycles of amplification for 30 s at 95°C and 3 min at 68°C (Advantage 2 polymerase, Clontech). The resulting three PCR products were gel-purified and used as templates for a second PCR using the SunA and SunF primers. The PCR parameters were the same as described above except for a 6-min annealing and extension step.
A SUN1 complementation cassette was also constructed using the fusion PCR method (18) as described above. This cassette contained the 5Ј-flanking region of AfSUN1, the AfSUN1 gene, the actin terminator, the phleomycin resistance marker, and the 3Ј-flanking region of AfSUN1 (supplemental Fig. S1C).

A. fumigatus Transformation
The fusion PCR products (1-2 g) were used to transform either CEA17⌬ku80 conidia or ⌬sun1 conidia using the electroporation method described by Sanchez et al. (21) with modifications (22,23). Transformants were selected on minimal medium agarose (0.7%) ϩ 150 g/ml hygromycin B for AfSUN1 deletion or minimal medium agarose ϩ 20 g/ml phleomycin for AfSUN2 deletion and incubated at room temperature for 1 week. Genomic DNAs from hygromycin-or phleomycin-resistant transformants and the parental strain were prepared as described by Girardin et al. (20), digested with restriction enzymes (Roche Applied Science), and verified by Southern blot analysis (supplemental Figs. S2 and S3). For Southern blot analysis, 5 g of digested genomic DNA was loaded on a 0.7% agarose gel, blotted on a nylon membrane (Hybond N ϩ , GE Healthcare), and revealed using probes 32 P-labeled using the Rediprime kit (Amersham Biosciences).

A. fumigatus Growth
Growth kinetics were performed by inoculating 10 5 conidia in 50 ml of Sabouraud liquid medium in a shaking incubator (37°C, 150 rpm). Mycelia were collected by filtration under vacuum after 24, 36, and 48 h and dried at 80°C overnight, and dry weights were recorded. For mycelial growth on the plates, colony diameters were measured. Germination tests were performed by placing 5 l of a conidial (10 6 ) suspension on Sabouraud solid medium at 37°C and counting the number of germinated conidia every hour. For microscopic observation, the parental as well as the mutant strains were grown in 24-well plates in 1 ml of Sabouraud medium with 10 4 conidia per well and incubated for 24 h at 37°C without shaking.

Sensitivity to Antifungal Drugs
Sensitivity to Calcofluor white and Congo red was tested as described previously (17). The effect of echinocandins (caspofungin and micafungin) was checked by E-test as per the manufacturer's instructions (bioMérieux). Conidial suspensions (2 ϫ 10 7 ) were spread on 1% yeast extract agar plates, and the E-test strip was placed. After a 48-h incubation at 37°C, minimum inhibitory concentrations were determined as the lowest drug concentration at which the border of the elliptical inhibition zone intercepted the scale on the antifungal strip.

Electron Microscopy Analysis
Conidia (2 ϫ 10 7 ) were incubated in 40 ml of Sabouraud liquid medium for 36 and 44 h at 37°C. The mycelium was collected and fixed at 4°C for 2 days in 4% glutaraldehyde in 0.1 M cacodylate, pH 7.4. After several washings with 0.1 M cacodylate buffer, samples were post-fixed for 45 min in the same buffer containing 2% osmium tetroxide (Merck, Darmstadt, Germany). During dehydration in a graded ethanol series, the 50% (v/v) ethanol incubation was prolonged for a period of 1 h, and 1% of 3-glycidoxypropyl trimethoxysilane (SPI-CHEM, West Chester, PA) was added during this incubation. Then, samples were embedded in Epon resin. Contrasted ultrathin sections (60 nm) were observed under a JEM 1010 transmission electron microscope (under Jeol, Tokyo, Japan).

Expression of AfSUN1 and CaSUN41 in Pichia pastoris and Purification of Recombinant AfSun1p and CaSun41p
P. pastoris GS115 strain (Invitrogen) and the expression vector pHILS1 (Invitrogen) were used to express recombinant AfSun1p and CaSun41p (r-AfSun1p and r-CaSun41p, respectively). 5 The open reading frame of AfSUN1 fused to a histidine tag (His 6 tag) was obtained after PCR amplification on A. fumigatus CEA17⌬ku80 cDNA with primers Sunprot1 and Sunprot2-HIS containing XhoI and BamHI restriction sites, respectively (supplemental Table S1). The open reading frame of CaSUN41 fused to a His 6 tag was also obtained by PCR amplification with primers Sun41-fw-ATG and Sun41-rev-HIS, both containing an XhoI restriction site (supplemental Table S1). The resulting PCR products were digested with XhoI and BamHI (in the case of AfSUN1) or XhoI (in the case of CaSUN41) and cloned in pHILS1 digested by the same enzymes, yielding pHILS1-AfSUN1 and pHILS1-CaSUN41, respectively. pHILS1-AfSUN1 and pHILS1-CaSUN41 were linearized by BglII or StuI, respectively, purified with phenol-chloroform extraction, and precipitated with ethanol before transformation of P. pastoris by the lithium chloride method (Invitrogen). Transformants were plated on a histidine-deficient medium and screened on minimal methanol medium (Invitrogen) for the insertion of the construct in the AOX1 locus of P. pastoris. Both recombinant proteins were obtained after culturing the transformed P. pastoris in buffered complex methanol medium (Invitrogen) at 30°C for 72 h followed by the addition of methanol (1%) to the culture medium every 24 h. Purification of secreted r-AfSun1p and r-CaSun41p was performed using ProBond nickel beads following the manufacturer's instructions (Invitrogen), and the eluates containing recombinant proteins were dialyzed against sodium acetate buffer (pH 6.5).

Deglycosylation of r-AfSun1p
N-Deglycosylation was performed using a recombinant N-glycosidase F (Roche Applied Science) according to the manufacturer's instructions. Total deglycosylation was carried out using the trifluoromethanesulfonic acid reagent (24). 10 g of purified protein was freeze-dried and kept under vacuum in presence of P 2 O 5 . The sample was treated with 50 l of a trifluoromethanesulfonic acid/anisole solution (2:1, v/v) in an ice bath under argon atmosphere for 3 h. The reaction was stopped by the addition of 60% pyridine in an ethanol/dry ice bath until pH 6.0 was achieved. Deglycosylated protein samples were dialyzed against water and analyzed by SDS-PAGE.

Cell Wall Analysis
A. fumigatus was grown in liquid Sabouraud medium at 37°C for 20 or 36 h, mycelia were harvested, washed with water, and disrupted using glass beads in a FastPrep (MP Biomedicals), and the disrupted suspension was centrifuged (3000 ϫ g, 10 min). The cell wall fraction (pellet) was washed with water and boiled in 50 mM Tris-HCl buffer (pH 7.5) containing 50 mM EDTA, 2% SDS and 40 mM ␤-mercaptoethanol (15 min, ϫ2). The sediment obtained after centrifugation (3000 ϫ g, 10 min) was washed with water and incubated in 1 M NaOH containing 0.5 M NaBH 4 at 65°C (1 h, ϫ2). The insoluble pellet obtained upon centrifugation (3000 ϫ g, 10 min) was washed with water to neutrality and freeze-dried (alkali-insoluble (AI) fraction). In the supernatant, excess of NaBH 4 was removed using acetic acid followed by dialysis against water and freeze-drying (alkalisoluble (AS) fraction). The monosaccharide composition in the AI and AS fractions was determined by gas-liquid chromatography (GLC; Perichrom PR2100 (Saulx-les-Chartreux, France) equipped with flame ionization detector and fused silica capillary column (30 m ϫ 0.32 mm inner diameter) filled with BP1)) as described earlier (25).

Analysis of the r-AfSun1p and r-CaSun41p Activities
The activity of both recombinant proteins was checked on the fungal cell wall components obtained as described above. C. albicans cell wall fractions were obtained as described in Wang et al. (26). In brief, the reaction mixture contained alkaliinsoluble (AI) or alkali-soluble (AS) fractions extracted from the fungal cell wall (200 g), r-AfSun1p (1 g) or r-CaSun41p (1 g) in a total volume of 100 l (acetate buffer, 50 mM, pH 6.0). Following incubation at 37°C overnight, reducing sugars released from the solubilized cell wall materials were analyzed using the p-aminobenzoic acid (PABA) method. Briefly, 50 l of the solubilized material was boiled for 10 min with 950 l of PABA reagent (50 mM sodium sulfite, 250 mM NaOH, 25 mM sodium citrate, 10 mM calcium chloride, and 10 mg ϫ ml Ϫ1 PABA), the optical density was measured at 414 nm, and the amount of reducing sugar was calculated using glucose as the standard. Further, enzyme kinetic studies were performed using AI fraction and the PABA method.

Substrate Binding Assays
Pulldown Assay-10 g of r-AfSun1p was incubated with 100 g of cell wall ␤-(1,3)-glucan in a total volume of 100 l of acetate buffer (50 mM, pH 5.5; 37°C for 1 h). Upon centrifugation, protein concentration in the supernatant was determined by Bradford assay. The sediment obtained was washed twice with wash buffer (10 mM Tris (pH 7.5), 500 mM NaCl, 0.02% Tween 20) and then boiled with SDS sample buffer. Protein in the sample buffer as well as in the supernatant was analyzed by SDS-PAGE (15% gel).
Surface Plasmon Resonance Assay-SPR assays were performed on a Biacore X100 biosensor instrument (GE Healthcare). Recombinant AfSun1p diluted in 50 mM acetate buffer (pH 4.5) containing 100 M EDTA to a concentration of 25 g/ml was coupled to the surface of a CM5 chip (Biacore, Inc., Piscataway, NJ) by standard amine chemistry (N-hydroxysuccinimide-1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide, Biacore, Inc.) to a level of ϳ2000 response units. The remaining active sites were quenched using 1 M ethanolamine. A reference flow cell consisting of an activated and quenched surface without r-AfSun1p was created to normalize readings. The running solvent was acetate buffer (50 mM, pH 5.5) containing 100 M EDTA at a flow rate of 10 l/min. Laminarioligosaccharides of DPs 2, 6, 12, and 23 were diluted to a final concentration of 200 M in the running buffer. Both injection and dissociation of the laminarioligosaccharides were analyzed for 570 s at 37°C. Regeneration of the surface was performed by injecting 50 mM glycine-NaOH, pH 9.5, for 60 s at a flow rate of 10 l/min; ␤-(1,6)-oligosaccharides of DP10 were used as the control.

Identification of the SUN Genes in A. fumigatus-A BLAST
search against the A. fumigatus genome identified two SUN domain-containing proteins encoded by AfSUN1 (AFUA_ 7G05450) and AfSUN2 (AFUA_1G13940). As for the other euascomycetes (8), A. fumigatus contained only one class I SUN protein Sun1p. AfSUN1 displayed a 1245-bp open reading frame (ORF) coding for a protein of 415 amino acids with a predicted molecular mass of 43.5 kDa. The AfSun1p SUN domain, containing the canonical Cys-X 5 -Cys-X 3 -Cys-X 24 -Cys motif, spanned residues 111-414 (Fig. 1). The sequence contained a signal peptide at the N terminus (amino acids residues In contrast, AfSun2p showed the characteristics of group II SUN domain proteins with a degenerated motif located at the N-terminal end of the protein (residues 17-301). The 1413-bp AfSUN2 ORF encoded a protein with 471 amino acid residues with a calculated molecular mass of 48.6 kDa. AfSun2p also displayed a signal peptide at the N terminus (corresponding to the amino acid residues 1-21). In addition, AfSun2p showed all the characteristics of glycosylphosphatidylinositol-anchored proteins: a hydrophobic N-terminal region (amino acid residues 1-17), a C-terminal region (452-471 amino acids) rich in serine and threonine, and a cleavage site for a carboxyl-peptidase and a -site on the 446-amino acid residues (30). However, unlike AfSun1p, AfSun2p did not show any N-glycosylation site.
Expression of the AfSUN Genes-AfSUN1 expression was detected starting 4 h after the initiation of spore germination and thereafter during mycelial growth. Quantitative PCR analysis showed a linear increase in the AfSUN1 expression during growth until 36 h beyond which there was a slow decrease (Fig.  2). In contrast, AfSUN2 expression was not detectable in the conditions tested.
Genetic Inactivation of AfSUN Genes-To analyze the functions of AfSUN1 and AfSUN2, deletion mutants were constructed by replacing each gene with an antifungal resistance cassette in the parental A. fumigatus strain, CEA17⌬ku80. PCR and Southern blot analyses showed correct and unique integration of the drug resistance cassettes at the target locus for AfSUN1 and AfSUN2 (supplemental Fig. S2). Deletion of AfSUN2 in the ⌬Afsun1 mutant strain was also performed. PCR and Southern blot analysis showed correct and unique integration of the phleomycin resistance cassette at the AfSUN2 locus in the ⌬Afsun1 mutant background (supplemental Fig. S2). AfSUN1 gene complementation was performed in the ⌬Afsun1 mutant strain by reintroduction of the wild-type AfSUN1 gene into the genome. RT-PCR confirmed expression of the AfSUN1 gene in the complemented ⌬sun1 mutant strain (supplemental Fig. S3).
AfSUN1 Is Required for Normal Growth and Correct Hyphal Morphogenesis-A modest but significant reduction in the dry mass was observed after 36 h of growth of the ⌬Afsun1 and ⌬Afsun1/Afsun2 mutants in the Sabouraud liquid medium at 37°C (Fig. 3A). Mycelial growth of these mutants was also affected on the agar medium. After 36 h of incubation on Sabouraud solid medium, diameters of the ⌬Afsun1 and ⌬Afsun1/ Afsun2 mutant colonies were 1.5 times smaller than that of the parental strain at 37°C and 2 times smaller at 50°C (Fig. 3B). In contrast, no difference could be observed for the ⌬Afsun2 mutant when compared with the parental strain (Fig. 3B). The growth defects on agar plates were independent of the pH and the presence of osmotic stabilizers (sorbitol or NaCl; data not shown). In agreement with the growth defects, an alteration of the mycelial morphology was seen in the mycelium of the ⌬Afsun1 mutant. After 16 h of growth, the hyphae of the ⌬Afsun1 mutant showed leaky hyphal tips (ϳ15-20%), intrahyphal growth, and short intercalary cells with closely arranged septa and swollen appearance (Fig. 3C, row i); after 48 h (Fig. 3C, row ii), there was an increase in the severity of the morphological alterations with an increased number of leaky tips (ϳ35-40%), whereas the parental strain hyphae were normal. In the ⌬Afsun1/⌬Afsun2 double mutant, no additional defect in hyphal integrity and morphogenesis was observed when compared with the ⌬Afsun1 mutant strain (data not shown).
Morphological defects of the ⌬Afsun1 mutant were further confirmed by electron microscopy analysis. Intrahyphal growth was shown by the doubling of the cell wall in the ⌬Afsun1 mutant (Fig. 3C, row iii). Interestingly, a high number of Woronin bodies were seen at the septal region of the ⌬Afsun1 mutant, suggesting that a defect in the closure of the septal pores could be at the origin of intracellular hyphae.
Conidiation was also affected with a 2-fold reduction for the ⌬Afsun1 and the ⌬Afsun1/Afsun2 mutants when compared with the parental strain (data not shown). However, conidia produced were normal, and there were also no differences in the germination of the single and double ⌬Afsun strains and the parental strain spores (data not shown). Reintroduction of the AfSUN1 in the ⌬Afsun1 mutant restored parental phenotypes (data not shown).
Biochemical Function of the SUN Proteins AfSun1p and CaSun41p-The phenotypes (swollen and vacuolated hyphae with double cell wall, and leaky tips; Fig. 3) observed during vegetative growth suggested a cell wall defect in the ⌬Afsun1 and ⌬Afsun1/⌬Afsun2 mutant strains. However, when cell wall analysis was performed at different times of growth (20 and 36 h), no differences were seen in the cell wall composition (alkali-insoluble/alkali-soluble ratio or monosaccharide composition) of the ⌬Afsun1 mutant or the parental or AfSUN1complemented strains (supplemental Fig. S4; data not shown). These results suggested that Sun1p does not have a direct effect on overall cell wall polysaccharide biosynthesis.
To investigate further the function of AfSun1p, a recombinant protein r-AfSun1p carrying a His 6 tag was expressed in the yeast P. pastoris and affinity-purified (see "Experimental Procedures"). The recombinant r-AfSun1p was highly glycosylated and had an apparent molecular mass of 68.5 kDa. Upon deglycosylation, r-AfSun1p migrated on a SDS-PAGE at 44 kDa (data not shown) in agreement with the predicted molecular mass of AfSun1p.
When incubated with the AS and AI fractions obtained from the parental A. fumigatus cell wall, r-AfSun1p showed a hydrolytic activity only toward the AI fraction (7-9% reducing sugars released; determined by the PABA method), which consists mainly of ␤-(1,3)-glucan and chitin. However, r-AfSun1p did not show any activity toward pure chitin. Further characteristics of the ␤-(1,3)-glucan hydrolysis were studied using the AI fraction as the substrate. r-AfSun1p was active in the pH and temperature ranges of 5.0 -7.0 and 25-45°C, with pH and temperature optima of 5.5 and 37°C, respectively. The specific activity was 3.2 nmol of reducing equivalent/min Ϫ1 g Ϫ1 protein. When incubated with curdlan (a linear ␤-(1,3)-glucan) and schizophyllan (a branched polysaccharide having ␤-(1,6)-side chain on every third glucose residue of the ␤-(1,3)-backbone) (31), r-AfSun1p showed hydrolytic activity only toward curdlan (12-16% reducing sugars released, as determined by the PABA method), indicating that AfSun1p acted only on linear ␤-(1,3) glucans.
To further characterize the mode of action and the minimum size of the ␤-(1,3)-glucan required for r-AfSun1p activity, p-NPG as well as soluble laminarioligosaccharides of varying DPs were used as the substrates. There was no hydrolytic activity by r-AfSun1p when p-NPG was used as substrate, confirming that AfSun1p is not a ␤-glucosidase. The products formed using laminarioligosaccharides as the substrates were analyzed on an anion-exchange chromatography column (Dionex). The smallest substrate for r-AfSun1p was found to be laminaribiose (DP2). Surprisingly, along with the release of glucose, there was also the formation of laminaritriose in minor quantity (as confirmed based on the retention time of a standard laminaritriose on the Dionex), suggesting an associated, but minor, transferase activity of r-AfSun1p in addition to its exo-␤-(1,3)-glucanase activity (Fig. 4A). When laminarioligosaccharides of higher DP (DPs 6 and 12) were used as the substrates, r-AfSun1p also showed exo-␤-(1,3)-glucanase and minor transferase activities (Fig. 4, B and C). With the course of time, there was an increase in the release of glucose by r-AfSun1p; however, the amount of substrate degraded never exceeded 10% of the initial concentrations. It was verified that this limited hydrolytic activity was not due to the presence of the product of hydrolysis because the addition of external glucose had no effect on the observed activity of r-AfSun1p (data not shown).
P. pastoris was used to produce recombinant AfSun1p; hence, the possibility of this expression system contaminating the observed AfSun1p glucanase activity was checked. Firstly, histidine-tagged r-AfSun1p was affinity-purified, which showed a single band on the SDS-PAGE when revealed by silver staining, and secondly, another histidine-tagged recombinant protein, r-AfAspF1 (a ribonuclease; AFUA_5G02330), when affinity-purified following the same protocol, did not hydrolyze laminarioligosaccharide. Moreover, the endogenous exo-␤-(1,3)-glucanase produced by P. pastoris is reported to hydrolyze chromogenic substrate p-NPG (32), whereas r-AfSun1p did not act on p-NPG. All these confirmed that the observed exo-␤-(1,3)-glucanase and minor transferase activities were exclusively associated with AfSun1p.
Substrate binding assays also showed a weak but specific binding toward ␤-(1,3)-glucan. Upon pulldown assay, ϳ20% of r-AfSun1p was bound to the cell wall ␤-(1,3)-glucan (Fig. 5). In accordance with the Dionex data, the surface plasmon resonance assay displayed a low binding of r-AfSun1p to laminaribiose, and this binding affinity increased with an increase in the size of the ␤-(1,3)-oligomers (DPs 6 and 12). However, a decrease in the binding affinity with the ␤-(1,3)-oligomer DP23 when compared with that of DP12 could be due to decrease in the solubility of DP23 when compared with DP12. r-AfSun1p did not show binding affinity toward ␤-(1,6)-linked oligosaccharide of DP10, confirming the specific binding of r-AfSun1p with the ␤-(1,3)-linked substrates (Fig. 5B).
Similar to r-AfSun1p, recombinant CaSun41p showed activity in the pH range 5.0 -6.0 with an optimum at 5.5 and between 35 and 40°C (here the alkali-insoluble fraction obtained from C. albicans cell walls was used as the substrate). The specific activity was found to be 4.4 nmol of reducing sugar equivalent/ min Ϫ1 g Ϫ1 proteins. No hydrolysis of p-NPG could be observed, and the activities toward laminarioligosaccharides were similar to those observed with r-AfSun1p (Fig. 4D). These results suggested that SUN proteins from both A. fumigatus and C. albicans displayed similar enzymatic activity.

DISCUSSION
In this study, for the first time, we report the role of SUN family genes in a filamentous ascomycete, A. fumigatus. Although two genes belonging to the SUN family were found in A. fumigatus, only AfSUN1 had a morphogenetic role, at least in the growth conditions tested. Deletion of AfSUN1 resulted in a variety of cell wall phenotypes, suggesting a role of AfSun1p in cell wall polarization and integrity. In particular, the appearance of intrahyphal hyphae suggested a role of AfSun1p during cell wall septation. This is further supported by the occurrence of many Woronin bodies in the septal region of the ⌬Afsun1 hyphae, an indication of difficulties in completion/closure of the septum (33). Similar phenotypes, especially intrahyphal hyphae, were observed in the csm-chitin synthase mutants of A. fumigatus (34,35). It was suggested that these chitin synthases could function as repair enzymes in septated compartments of the old hyphae. A role of SUN proteins in septation is also in agreement with the data obtained in other fungal species. Deletion of the S. pombe group I SUN gene, psu1, is lethal and associated with cell wall defects during cell separation (7). Similarly, simultaneous inactivation of the C. albicans SUN41 and SUN42 is synthetically lethal, leading to lysis of the mother cells after septation (8). A putative role in septum formation has also been proposed for ScSUN4 and ScUTH1 of S. cerevisiae (3). In Williopsis saturnus var. mrakii, the SUN family protein Wmsu1p has been reported to be involved in the cell wall structure (36,37). However, differences do exist among these fungal species in the phenotypes observed for the group I SUN protein mutants. AfSun1p is not essential for growth in contrast to SpPsu1p and the CaSun41p/CaSun42p pair. In A. fumigatus (present study) and in S. pombe (7), the defective phenotypes could not be rescued by the osmo-protectants (data not shown) as observed in C. albicans (8).
Moreover, functions of the group II and group I SUN proteins are obviously different. In A. fumigatus, we could not identify conditions upon which the group II AfSUN2 gene is expressed. Consistently, deletion of AfSUN2 did not result in any phenotypic difference relative to the parental strain. Several group II SUN proteins have been identified in yeasts: S. cerevisiae (Ymr244w), S. pombe (Adg3p), and C. albicans (Orf19.5896) (8). However, only Adg3p of S. pombe has been studied, and the corresponding mutant strain showed a slight delay in cell separation (38).
In databases, SUN proteins have been annotated as ␤-glucanase on the basis of sequence homologies with BglAp of Candida wickerhamii (39). However, these SUN proteins did not show similarity with already described glycoside hydrolases (GH) in the CAZyme database (40), and hence, they will be assigned to a new GH family (GH132). The CwBGLA and CwBGLB genes were identified during the search for proteins reacting with a polyclonal antibody raised against a purified extracellular ␤-glucosidase. However, the ␤-glucosidase activity has been demonstrated only for BglBp, which does not carry a SUN domain (39). Thus, until now, SUN proteins could not be considered as ␤-glucosidases as the biochemical evidence for such an activity was lacking. Here, we have characterized, for the first time, an enzymatic activity that is similar in SUN proteins from two different fungal species, A. fumigatus and C. albicans. AfSun1p and CaSun41p bind to and act on linear ␤-(1,3)-glucans in a specific manner. AfSun1p showed a hydrolytic activity only against ␤-(1,3)-glucan. Branching on the ␤-(1,3)-glucan negatively affected such hydrolytic activity. Moreover, earlier, Zverlov et al. (41) showed a laminaribiase activity associated with Thermotoga neapolitana BglBp; however, it also showed hydrolytic activity on the p-NPG, whereas r-AfSun1p acted on laminaribiose (Fig. 4A) and not on p-NPG, confirming a distinct role associated with this protein. The mode of action of the SUN proteins is unique. It is clear from the biochemical data obtained that our understanding of the enzymatic function of this protein is far from being complete. For example, in contrast to AfSun1p, true exo-glucanases of A. fumigatus degraded 100% of the substrate after 1 h of incubation (42). Similarly, when pulldown assay was performed with ␤-(1,3)-glucan binding receptors such as GNBP3 (43), 100% of the receptor is pulled down by the ␤-(1,3)-glucan, whereas with AfSun1p, it is only ϳ20%. Several reasons could be put forward to explain these discrepancies: (i) the recombinant proteins used are highly glycosylated, and this moiety could impact on the enzymatic function due to steric hindrance problems; and (ii) the activity has been tested only in vitro, and the lack of another cofactor may impede the identification of the true biochemical function of the protein. For example, in S. cerevisiae and S. pombe, the machinery constructing the septum is composed of several proteins acting together (44,45). We have verified that the addition of external chitohexaose or N-acetylglucosamine did not result in any additional product formation, confirming that r-AfSun1p has no glucan-chitin cross-linking activity. Moreover, except for a single transferred product, there was no major transferase activity associated with r-AfSun1p during the course of time, ruling out its ␤-(1,3)-glucan elongase activity (Fig. 4).
Although the biological function of these proteins in vivo is not fully understood, this study showed that these proteins bind and act on ␤-(1,3)-glucans and are probably involved in the septum closure. Consistent with this hypothesis, inactivation of genes involved in degradation of the primary septum (CaACE2 or CaCHT3) in C. albicans or physiological conditions preventing the activity of chitinases that act at the primary septum rescue the lethality associated with simultaneous inactivation of CaSUN41 and CaSUN42 (data not shown). This may suggest that in the cell wall, group I SUN proteins act to provide building blocks to other enzymes that are necessary for cell wall biogenesis and/or counteracting the activity of cell wall-degrading enzymes. This also highlights the need to develop approaches to study cell wall biosynthetic enzymes in situ because global chemical approaches are not able to pinpoint structural changes occurring at very specific locations such as the septum.