The strains used in this study are detailed in Table 1. A. fumigatus conidia were harvested in 0.2% (vol/vol) Tween 20, resuspended in double-distilled water (DDW) and counted with a hemocytometer. For continuous growth, A. fumigatus strains were grown on YAG medium, that consists of 0.5% (wt/vol) yeast extract, 1% (wt/vol) glucose, and 10 mM MgCl₂, supplemented with trace elements, vitamins, and 1.5% (wt/vol) agar when needed (Bainbridge, 1971). SM medium consisted of 1% (wt/vol) glucose, 1% (wt/vol) SM (Difco, Livonia, MI, United States), 0.1% (wt/vol) Casamino Acids (Difco), 7 mM KCl, 2 mM MgSO₄ and 50 mM Na₂HPO₄-NaH₂PO₄ buffer (pH 5.3), supplemented with vitamins, trace elements and 1.5% agar when needed. NaNO₃-depleted Aspergillus minimal medium (Weidner et al., 1998) was used for collagen medium with 0.1% (wt/vol) yeast extract, 0.5% (wt/vol) glucose and collagen as sole carbon and nitrogen source. Peptone medium contained 1% (wt/vol) glucose, 0.4% peptone (Difco), 7 mM KCl, 2 mM MgSO₄ and 50 mM Na₂HPO₄-NaH₂PO₄ buffer (pH 5.3), supplemented with vitamins and trace elements. Genetically modified organisms and pathogens used in this study were maintained in accordance with TAU Institutional Policies.

All strains were prepared in the Ku80 null background, strain AkuB^KU80 (da Silva Ferreira et al., 2006). Full details of the construction and verification of the strains shown in Table 1, including a list of the primers used (Supplementary Table S1), are provided in the Supplementary Data Section.

Freshly harvested A. fumigatus conidia (10⁷/ml) were suspended in DDW +0.5% (vol/vol) Tween 20 at 37°C. At different time-points, aliquots were diluted, plated on YAG plates and the number of colonies counted.

Freshly harvested A. fumigatus conidia (5 × 10⁷/ml) were suspended in 0.5 ml DDW +0.1% Tween 20 and mixed with 0.5 ml (packed volume) of acid washed glass beads, 150–212 μm (Sigma–Aldrich Corp., St. Louis, MO, United States). They were then vortexed on medium strength for up to 10 min. At each time point a sample was taken, diluted and plated on YAG plates. The plates were incubated at 37°C for 24–36 h, colonies were counted and survival rates were calculated as the percentage of viable spores.

Proteolytic activity on solid medium was assessed by spotting conidia on SM plates containing 0.1% Tween 20. The colonies were grown for 48 h at 37°C, then transferred to room temperature for another 48 h and subsequently photographed. Supernatants were collected from Aspergillus cultures grown in liquid SM for 48 h at 37°C. Azocasein (Sigma) was dissolved at a concentration of 5 mg/ml in assay buffer containing 50 mM Tris (pH 7.5), 0.2 M NaCl, 5 mM CaCl₂ and 0.05% Triton X-100 as previously described (Kogan et al., 2004). The azocasein solution (400 μl) was mixed with 100 μl portions of supernatants from Aspergillus cultures and incubated by shaking for 90 min at 37°C. The reactions were stopped by adding of 150 μl 12% (vol/vol) trichloroacetic acid, and the reaction mixtures were allowed to stand at room temperature for 30 min. Tubes were then centrifuged for 3 min at 8,000 g, and 100 μl of each supernatant was added to 100 μl of 1 M NaOH. The absorbance of released azo dye at 436 nm was determined with a spectrophotometer. Proteolytic activity on bovine serum albumin (BSA) was measured by growing A. fumigatus in 24-well plates with 1 ml/well liquid peptone medium containing 0.1% BSA, respectively, for 24–72 h at 37°C. Supernatants were boiled in sample buffer and run on an 8% SDS-PAGE gel followed by Coomassie staining to visualize the proteins.

Aspergillus fumigatus wild-type and mutant strains were grown for 72 h at 37°C on YAG agar plates. Fixation and processing of samples for SEM was performed as described earlier (Fichtman et al., 2014; Hover et al., 2016). Briefly, small areas of conidiating mycelium were carefully excised from the zone of interaction and vapor fixed with 8% (vol/vol) paraformaldehyde and 4% (vol/vol) glutaraldehyde dissolved in water for 1 h in a closed chamber. Secondary vapor fixation was then carried out in an aqueous solution of 2% (wt/vol) osmium tetroxide for 1 h. The samples were submerged for 10 min in DDW and then dehydrated (10 min, twice for each step) under a series of ethanol concentrations (7.5, 15, 30, 50, 70, 90, 95, and 100%). Next, samples underwent critical point drying (CPD) using a K850 CPD dryer (Quorum Technologies, United Kingdom). Coating was done with 3 nm iridium using a Q150T coater (Quorum Technologies, United Kingdom). Samples were imaged with a Merlin scanning electron microscope (Zeiss, Germany).

Aspergillus fumigatus strains at a concentration of 10⁶ condia/ml were grown in liquid collagen medium for 72 h at 37°C. Total protein was TCA-precipitated from the CFs, digested with trypsin, iTRAQ-labeled and analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) as detailed in the Supplementary Data.

For the cyclophosphamide/cortisone acetate neutropenic model (Ejzykowicz et al., 2009), 6-week-old female ICR mice were injected intraperitoneally with cyclophosphamide (150 mg/kg in PBS) at 3 days prior to infection, on the day of infection and at 3 days post-infection. Cortisone acetate (150 mg/kg PBS with 0.1% Tween 20) was injected subcutaneously at 3 days prior to conidial infection.

For intranasal infection, mice were anesthetized by intraperitoneal injection of a solution of 250 μl xylazine (VMD, Arendonk, Belgium) and ketamine (Imalgene, Fort Dodge, IA, United States) at a concentration of 1.0 and 10 mg/ml, respectively (dissolved in PBS). Following anesthesia, the mice were inoculated intranasally with 2.5 × 10⁵ (intranasal model) or intravenously (IV) through the tail vein (IV model) with freshly harvested conidia of AkuB^KU80, ΔXprG, or ΔXprG/ΔPrtT in PBS +0.1% (vol/vol) Tween 20. The inoculum was verified by quantitative culture. The animals were monitored for survival for up to 18 days. For infection of immunocompetent mice, 6-week-old female ICR mice were inoculated intranasally with 1 × 10⁷ conidia and sacrificed 24 h or 48 h later. Excised lungs were ground and plated on YAG agar plates for CFU enumeration. Histological analysis was performed with Gomori methenamine silver stain (GMS, stains fungal elements black) or haematoxylin and eosin stain (H&E, stains host-cell nuclei purple, cytosol pink). Statistical analysis of mouse survival was performed with GraphPad Prism 4 software (GraphPad Software, San Diego, CA, United States). Animal studies were authorized by the Tel Aviv University Animal Welfare Committee according to the Israel Ministry of Health guidelines and carried out in accordance with Tel Aviv University institutional policies.

A single A. fumigatus xprG ortholog (AFUA_8G04050) was identified by amino-acid similarity to A. nidulans XprG (70% identity). It is predicted to be 1752 nucleotides long and to contain one intron. A. fumigatus XprG encodes a protein 583 amino acids in length, containing a predicted NDT80/PhoG like DNA-binding domain (amino-acid residues 156–327) and a MAP65/ASE1 domain (amino acid residues 336–505) shared by microtubule-binding proteins. The similarity to NDT80 family genes vib-1 from N. crassa and S. cerevisiae ndt80 is limited to the PhoG like DNA-binding domain. To analyze the function of XprG and its interaction with PrtT in A. fumigatus, we prepared ΔPrtT, ΔXprG, ΔXprG/ΔPrtT null mutants and corresponding reconstituted strains PrtT-KI, XprG-KI, and PrtT-KI/ΔXprG by transformation with a circular plasmid containing the PrtT gene and the phleomycin resistance cassete for selection (see Supplementary Data for full details about strain construction and verification).

The aforementioned null and reconstituted strains were point inoculated on YAG agar plates and examined for differences in morphology after growth for 72 h at 37°C. Results show that the ΔXprG, ΔXprG/ΔPrtT null strains produced lighter colored conidia as compared to the AkuB^KU80, ΔPrtT, and XprG-KI strains (Figure 1A). Examination of the conidiophore structure by light microscopy and SEM revealed that the ΔXprG and ΔXprG/ΔPrtT strains produced significantly smaller, more compact conidiophores (P < 0.02) with shorter conidial chains (Figures 1B,E). The conidial hydrophobin outer rodlet layer remained intact and unchanged (Figure 1C) and the surface hydrophobicity of the mutant colonies, based on their ability to exclude water, was unchanged (not shown). Reflecting the shorter conidial chains, the number of conidia produced by the ΔXprG and ΔXprG/ΔPrtT strains was significantly reduced three–fourfold per plate (P < 0.05) compared to AkuB^KU80 and ΔPrtT (Figure 1D). Radial growth rates of the ΔPrtT, ΔXprG, ΔXprG/ΔPrtT strains were similar to that of AkuB^KU80 at both 37 and 48°C (Data not shown). The lighter conidial color of the PrtT-KI complemented strain compared to the deleted strain (Figure 1A) is unexpected and may be due to multiple integration of the complementing plasmid.

We evaluated the ability of the mutant conidia to withstand osmotic or detergent disruption by measuring their survival in water or in water containing Tween 20 detergent. Conidia from the ΔXprG and ΔXprG/ΔPrtT strains exhibited similar increased susceptibility to the detergent, losing >90% viability after 4 h as compared to loss of only <20% viability in the AkuB^KU80 and ΔPrtT strains (Figure 2A). There were no differences in conidial stability during storage in DDW, suggesting the mutants are not osmotically sensitive (data not shown). The ability of the mutant conidia to withstand physical disruption was assessed by subjecting them to agitation in the presence of glass beads. Conidia from the ΔXprG and ΔXprG/ΔPrtT strains exhibited increased susceptibility to glass-bead agitation, losing all viability after 4 min agitation, compared to 10 min for the AkuB^KU80 and ΔPrtT strains (Figure 2B).

Taken together these results show that deletion of xprG in the AkuB^KU80 and ΔPrtT strains resulted in the production of smaller conidiophores containing less and more fragile spores.

To investigate the effect of xprG deletion on secreted protease activity, we point inoculated the strains on SM agar plates and visually determined the formation of a proteolytic halo due to casein degradation around the fungal colony. As we have previously shown (Sharon et al., 2009), deletion of prtT resulted in the formation of a very weak halo compared to the control strain AkuB^KU80, whereas deletion of xprG (ΔXprG) led to a partially reduced halo. Deletion of both xprG and prtT (ΔXprG/ΔPrtT) completely abolished the formation of a proteolytic halo (Figure 3A) and reduced azocasein degradation by over 90% (Figure 3B). Halo formation was restored in the reconstituted strain XprG-KI and slightly increased in the PrtT-KI and PrtT-KI/ΔXprG strains. This increase, however, was not seen in the azocasein or BSA degradation assays described below. We used a more sensitive SDS-PAGE-based BSA degradation assay to better differentiate between the proteolytic activities of the ΔXprG,ΔPrtT and ΔXprG/ΔPrtT mutants (Figure 3C). Strains were grown for 24–72 h on peptone medium containing 0.1% BSA. The supernatant was separated on an SDS-PAGE gel followed by Coomassie staining to visualize the degradation of the BSA protein. The control AkuB^KU80 and reconstituted strains substantially degraded the BSA substrate protein after 48 h of incubation, whereas the ΔPrtT, ΔXprG, and ΔXprG/ΔPrtT strains did not (Figure 3C). After 72 h of incubation, all strains except ΔXprG/ΔPrtT had completely degraded the BSA. Taken together, our findings indicate that deletion of both prtT and xprG results in an additive reduction in the ability to degrade casein or albumin. Previous analyses using protease inhibitors and deletion mutants have demonstrated that secreted A. fumigatus serine proteases, and in particular Alp1, are primarily responsible for the degradation of these substrates (Jaton-Ogay et al., 1994; Tomee et al., 1997; Kogan et al., 2004).

To identify the repertoire of proteins secreted by the ΔPrtT, ΔXprG, and ΔXprG/ΔPrtT mutants compared to the control AkuB^KU80 strain, conidia were cultured in liquid MM containing collagen for 72 h at 37°C with shaking. Supernatants were TCA-precipitated, treated after Yeung and Stanley (2009), followed by a Wessel-Flügge precipitation, labeled with iTRAQ 4-plex isobaric labeling for quantitative proteomics approach and analyzed by LC-MS/MS. A total of 938 proteins were identified in the fungal supernatants, 274 of which were predicted to be secreted by in silico analysis (secretion signal found by signalP for 236 proteins and targetP for 274 proteins) (Supplementary Table S2).

Cell lysis cannot be completely excluded during cultivation and sample preparation. It is therefore possible that a proportion of the secretome consisted of proteins without a secretion signal, which had been released by cell lysis. However, the comparison of the abundance (based on normalized PSM values) of typical intracellular, non-secreted proteins like actin (Afu6g04740) or tubulin (Afu1g02550) with the abundance of a secreted protein like lap2 (Afu3g00650) revealed a depletion by a factor between 12 and 106. In addition, fungi are able to release typical intracellular proteins via extracellular vesicles to the extracellular space (Joffe et al., 2016). It can be therefore assumed that the contamination of the secretome by cell lysis was relatively low.

Deletion of prtT resulted in strongly reduced secretion of 22 proteases, including Lap1, Lap2, Cp1, Cp3 Alp1, Alp2, and Mep (Figure 4 and Supplementary Table S3). Deletion of xprG resulted in reductions in the secretion of 19 proteases, including most strongly Lap1, Sxa2, DppV, DppIV, Mep, and SedC. Notably, several proteases that were strongly reduced in the ΔPrtT mutant were only slightly reduced (Pep2, DapB, Cps1, Cp1, Cp3) or not reduced (CtsD, Ape3, Lap2, Alp1, Alp2) in the ΔXprG strain (Figure 4 and Supplementary Table S3). This provides a good explanation for the higher secreted protease activity of the ΔXprG vs. the ΔPrtT mutant, especially the lack of reduction in the major neutral serine protease Alp1 (Behnsen et al., 2010). Deletion of both prtT and xprG resulted in the strongest reductions in the secretion of 24 proteases comprising the entire joint dataset of the ΔPrtT and ΔXprG mutants, including 14 serine proteases, 7 metalloproteases and 3 aspartic proteases (Figure 4 and Supplementary Table S3). Interestingly, in all three mutant strains, the secretion of numerous glucanases, chitinases, and allergens was also reduced, indicating that these transcription factors not only regulate the expression of secreted proteases but also of cell wall enzymes and antigenic proteins (Supplementary Table S3). The significance of these results will be explained in the discussion. No proteins showed increased secretion in the mutants vs. the control AkuB^KU80 strain.

Previous studies have shown that deletion of prtT in A. fumigatus does not affect virulence in lung-infected neutropenic mice (Bergmann et al., 2009; Sharon et al., 2009). We hypothesized that the almost complete lack of secreted protease activity in the ΔXprG/ΔPrtT strain would result in a noticeable reduction in virulence. To test this, we compared the virulence of the control AkuB^KU80, ΔXprG and ΔXprG/ΔPrtT strains in two mouse models of invasive aspergillosis. Both models reproduce profound neutropenia (as found, for example following chemotherapy in leukemic patients), by immunosuppressing the mice with a combination of cortisone acetate and cyclophosphamide. In the first model, the mice were infected intranasally with freshly harvested conidia. In the second model reproducing disseminated infection, mice were infected IV. The number of live mice in each group was recorded daily throughout the experiment. The ΔPrtT strain was not included as we have previously shown it to be fully virulent in these models (Sharon et al., 2009). Figure 5 shows survival curves obtained during the course of the experiment for the lung (Figure 5A) and disseminated (Figure 5B) models. No significant differences in virulence were found for the ΔXprG and ΔXprG/ΔPrtT mutants compared to the control AkuB^KU80 strain (P-value > 0.3). Next, we analyzed the response of immunocompetent mice to intranasal pulmonary infection (10⁷ conidia/mouse) with ΔXprG/ΔPrtT and control AkuB^KU80 strains. Outputs included lung CFU (Figure 5C) and histology (Supplementary Figure S6), 24 and 48 h following infection. There was no significant difference in lung fungal load (P-value > 0.2) between the ΔXprG/ΔPrtT and control AkuB^KU80 strains (Figure 5C). Histology revealed abundant conidia in the bronchi of both mutant and control strains 24 h after infection (CMS stain, right panel, arrows) accompanied by a considerable influx of red blood cells and lymphocytes (H&E stain, left panel) (Supplementary Figure S6). As expected in immunocompetent mice, the number of conidia strongly decreased 48 h after infection, as did the number of red blood cells and lymphocytes (Supplementary Figure S6). Taken together, these results suggest that immunocompetent mice mount a similar and effective immune response to both ΔXprG/ΔPrtT and control AkuB^KU80 infection.