The A. fumigatus strains used during the study are listed in the Supplementary Table 1. The culture used were mainly complete and/or minimal media. The complete medium consists of, YAG (2% w/v dextrose, 0.5% w/v yeast extract, 0.1% v/v element trace solution, 2% w/v agar), YUU (Medium YAG plus uridine and uracil 1.2 g/L), and YG or YG + UU media with the same composition, but without agar. The minimum medium (MM) consists of 1% glucose, 0.1 element trace solution % v/v, 20X salt solutions, pH 6.5 with or without the addition of 2% (w/v) agar (Kafer, 1977). The strains were grown at 37°C for 5 days.

All sequences are in accordance with the AspGD database (http://www.aspgd.org/). The deletion cassette was constructed by in vivo recombination in S. cerevisiae as described (Colot et al., 2006). About 1.0 kb of the flanking region of the 5′-UTR and 3′-UTR regions of the target ORF regions were selected for primer design. The 5′Fw and 3Rv primers contained a short sequence homologous to plasmid pRS426 and a checkmark pyrG or pPTRA. Both 5′and 3′-UTR fragments were amplified by PCR from A. fumigatus genomic DNA (gDNA). The pyrG gene placed inside the cassette as a prototrophic marker was amplified from the vector pCDA21. The deletion cassette was generated by transformation into S. cerevisiae SC94721, using the lithium acetate method (Schiestl and Gietz, 1989), plus the fragments, together with the plasmid pRS426 digested in two sites with the enzyme BamHI and EcoRI (New England Biolabs UK). The DNA of the transformants was extracted using the method described (Goldman et al., 2003). The deletion cassette was amplified by PCR from these plasmids using TaKaRa Ex Taq ™ DNA polymerase (Clontech Takara Bio) and used for the transformation of A. fumigatus. The same protocol was applied to the construction of the double mutants. However, the check mark used for the gene deletion was amplified from the plasmid pPTRI. Colonies of null mutants were selected by purifying colonies in MM with 1 μg/ml pyriamiamine hydrobromide (Sigma). Southern blot analysis demonstrated that the transformation cassette was homologous. Deletion strains of a single gene were complemented by co-transformation, ~1.0 kb of the 5′ and 3′-UTR plus ORF together with the plasmid pPTRI according to Herrera-Estrella et al. (1990) with modifications. Complements were selected by purifying colonies in MM with 1 ug/ml of piritiamine hydrobromide.

The homologous and/or heterologous recombination to gene replacement was confirmed by PCR and Southern blot. The absence of mRNA from the target genes was verified by qPCR. All primers used for the constructions are listed in Supplementary Table 2.

The characterization of sensitivity and/or resistance was evaluated by radial growth comparing mutant and wild strains. Five microliter of a 2 × 10⁷ suspension were dropped into MM with or without stressors, the plates were cultured for 5 days at 37°C. After that period, the radial diameter was measured and the values were used for statistical analysis. The different stressful conditions were ions a with the addition of calcofluor white (CFW), Congo Red (CR), menadione, sorbitol, and caspofungin.

To assess growth in different carbon sources, we use modified MM. Sodium acetate (1%) or ethanol (1%) was added to the MM instead of glucose (1%) (Kafer, 1977). The strains were grown for 5 days at 37°C.

Data were analyzed using (Prism, GraphPad) “Two-way ANOVA” followed by “Bonferroni post-tests.” Significance levels were ^*p < 0.1, ^**p < 0.01, and ^***p < 0.001. All the strains were compared with the single mutants and the Wild-type.

The strains were inoculated (1 × 10⁷ conidia) in 250-ml erlenmeyer's flasks with 50 ml of YG medium and grown in a rotatory shaker (200 rpm) at 37°C for 16 h before being exposed to different stressing agents, such as Congo Red, sorbitol. The total protein extracts were obtained through mycelia ground with The macerated mycelia were resuspended (500 mg) in 1 ml of B250 lysis buffer (250 mM NaCl, 100 mM Tris-HCl pH 7.5, 10% glycerol, 1 Mm EDTA and 0.1% NP-40) supplemented with 1.5 ml/L of 1M DTT, 100 mL of complete protease inhibitor Cocktail without EDTA (Roche), 3 ml/L of 0.5M Benzamidine, 10 ml/L of 100X phosphatase inhibitors (1M NaF, 5M sodium orthovanadate, 8M glycerol phosphate) and 10 mL/L of 100 mM PMSF. The samples were centrifuged at 13,000 g for 10 min at 4°C. The supernatants were removed and quantified by Bradford assay (BioRad). Western blots were performed as previously described (de Assis et al., 2018).

To perform the co-immunoprecipitation (co-IP) assays, AtfA-D:GFP strains had SakA marked with 3xHA at the C-terminus (SakA:3xHA). The strains were grown for 16 h in liquid YG and were further exposed to two conditions, these being, sorbitol 1.2 M (10 min) and CR 300 ug/ml (10 min). The co-IP experiments were carried out with GFP-Trap (Chromoteck) according to Manfiolli et al. (2017). To perform co-IP tests, the mycelia were frozen with liquid nitrogen and macerated with the aid of a crucible and pistil. The samples were normalized and the same amount of protein was added to 20 μL of GFP-trap, the resin was washed three times with resuspension buffer before incubation. The protein extracts and the resin were then incubated with shaking at 4°C for 4 h. After incubation, the resin was washed three times in B250 buffer. To release proteins from the resin, the samples were incubated with sample buffer and boiled at 98°C for 5 min. The proteins were transferred from a 10% SDS-PAGE gel to a nitrocellulose membrane for a Western blot assay using a Trans-Blot turbo transfer system (Bio-Rad). GFP-labeled AtfA-D were detected using a rabbit anti-GFP antibody (Cell Signaling Technology) at a dilution of 1:1,000. For the detection of HA-labeled proteins, a mouse anti-HA monoclonal antibody (Sigma) was used in the 1:2,000 dilution. Both primary antibodies were identified with the conjugated anti-mouse IgG antibody (Cell Signaling Technology) used at 1:1,000 dilution as a secondary antibody.

The strains were inoculated (1 × 10⁷ conidia) in 250 ml erlenmeyer's flasks with 50 ml of YG medium and grown in a rotatory shaker (200 rpm) at 37°C for 24 h. After that period, the mycelia were then transferred to minimum medium plus 1% glucose as a carbon source for another 24 h. The amount of free glucose remaining in the medium was measured using the GOD-PAP Glucose Liquid Stable Monoreagent kit (Laboratories LaborLab Ltd. Guarulhos, São Paulo, Brazil), according to the manufacturer's instructions. Glucose uptake was calculated by determining the difference in glucose present in the initial medium and after 24 h of incubation.

The strains were inoculated (1 × 10⁷ conidia) in 250 ml erlenmeyer's flasks with 50 ml of YG medium and grown in a rotatory shaker (200 rpm) at 37°C for 24 h. After that period, the mycelia were transferred to minimum medium plus 1% glucose as a carbon source for 24 h. Then the mycelia were filtered and taken to the lyophilizate for 48 h. The dry weight of each strain was compared to the dry weight of the wild type mycelium to normalization.

Intracellular glycogen and trehalose levels were measured in mycelia grown 24 h in MM supplemented with glucose as a carbon source as previously described (Manfiolli et al., 2019).

Galleria mellonella larvae were obtained by reproducing adult larvae (Fuchs et al., 2010). The larvae used for the infection were in the last larval stage of development (sixth week). All selected larvae weighed ~300 mg, and were restricted to food for 24 h before the experiment. The fresh conidia of each A. fumigatus strain were counted using a hemocytometer. The initial concentration of conidial suspensions for infections were 2 × 10⁸ conidia/ml, total of 5 μl (1 × 10⁶ conidia/larvae) of each suspension was inoculated by larva. The control group was composed of larvae whose 5 μl of PBS were injected to observe death from physical trauma. The inoculum was carried out with the Hamilton syringe (7000.5 KH) through the last left ear. After infection, the larvae were kept with food restrictions, at 37°C in Petri dishes in the dark, and scored daily. The larvae were considered dead due to lack of movement in response to touch. The viability of the inoculum administered was determined by serial dilution of the conidia in YAG medium and incubating the plates at 37°C for 72 h.

In a previous publication by our research group (Silva et al., 2017), we showed that the genes encoding the TFs AtfA-D have their mRNA accumulation dependent on MAP kinases SakA and MpkC and possibly function downstream the HOG pathway regulating genes involved in the adaptation of A. fumigatus to osmotic stress. These four TFs belong to a small bZIP family whose proteins are between 185 and 555 aa (Figure 1A). Multiple alignment of the bZIP domain of AtfA has about 58–65% identity with AtfB-D. The bZip domains were identified by SMART (http://smart.embl-heidelberg.de) and Figure 1A shows the linear protein structures with the correspondent bZip domain position. Protein modeling predictions based on structural templates allowed to observe the AtfA-D structural configuration (Figure 1B) and the binding site deduced region (Figure 1C). To further characterize these TFs, we constructed double deletions mutants for these four null TF genes, therefore generating six double mutants (ΔatfA ΔatfB, ΔatfA ΔatfC, ΔatfA ΔatfD, ΔatfB ΔatfC, ΔatfB ΔatfD, and ΔatfC ΔatfD) and evaluated the possible genetic interactions among the mutants in the presence of different stressing conditions.

Growth of the double deletion mutants in minimum medium (MM) for 5 days at 37°C provided initial evidence of the differences and synergism between single TF mutants (Figures 1D,E). The mutant ΔatfA displayed reduced growth when compared to the wild-type and the other single deletion mutants (Figures 1D,E). Deleting either atfB or atfD in the ΔatfA background restored wild-type phenotype suggesting these two mutations can suppress the ΔatfA growth defect (Figures 1D,E); however, deleting atfC in the same background mimicked the ΔatfA phenotype (Figures 1D,E). Reduced growth was also observed for the ΔatfB ΔatfC mutant (Figures 1D,E). Interestingly, deleting atfD in both ΔatfA, and ΔatfB backgrounds resulted in sectorized growth, what was not observed for ΔatfC ΔatfD (Figure 1D). Table 1 shows the different genetic interactions between the mutants.

In the presence of CFW, the double ΔatfA ΔatfB is as sensitive as ΔatfA and ΔatfB to CFW (90 μg/ml) while ΔatfA ΔatfC recapitulates ΔatfA growth (Figure 2A). The ΔatfB ΔatfC mutant is more sensitive to CFW (90 μg/ml) than the corresponding single mutants, suggesting additive interaction between these two mutants (Figure 2A, Table 1). Deleting atfD in the ΔatfA background restored wild-type phenotype suggesting this mutation can suppress the ΔatfA growth defect (Figure 2A). The ΔatfB ΔatfD and ΔatfC ΔatfD mutants are as sensitive as the single mutants, suggesting the single mutants are not interacting (Figure 2A, Table 1). In the presence of CR (50 μg/ml), all the double mutants behave like the single mutants, except for ΔatfA ΔatfC and ΔatfB ΔatfC that are more sensitive than the corresponding single mutants, suggesting additive interaction (Figure 2B, Table 1). The double mutants ΔatfB ΔatfC, ΔatfB ΔatfD and ΔatfC ΔatfD are more resistant to higher osmotic concentrations (1.2 M sorbitol) mimicking the single mutant phenotypes of ΔatfC and ΔatfD (Figure 2C, Table 1). In contrast, deletion of atfA in ΔatfC and ΔatfD background increase the sensitivity to 1.2 M sorbitol, suggesting a suppression phenotype (Figure 2C, Table 1).

We also investigated the response to two agents that cause oxidative stress, H₂O₂ and menadione (Figures 2D,E). Growth in different H₂O₂ concentrations showed that ΔatfA ΔatfC and ΔatfA ΔatfD recapitulate the ΔatfA phenotype, suggesting there are no genetic interactions between these mutants (Figure 2E, Table 1). In contrast deleting atfB in the ΔatfA background partially suppresses the H₂O₂ sensitivity while ΔatfB in the ΔatfC or ΔatfD background increases H₂O₂ sensitivity (Figure 2E, Table 1). In the presence of 0.05 mM menadione, the double mutants ΔatfA ΔatfB, ΔatfA ΔatfC, ΔatfB ΔatfC, and ΔatfA ΔatfD are as sensitive as the corresponding single mutants (Figure 2D, Table 1), In contrast the deletion of atfD in the ΔatfA or ΔatfD background increases the H₂O₂ sensitivity (Figure 2E, Table 1).

The results observed here for the single mutants are similar to those reported by Silva et al. (2017). However, the data obtained with the analysis of the double mutants indicate complex genetic interactions among these TFs that could impact the response to different kinds of stressful conditions.

Because the cell localization of TFs AtfA-D in A. fumigatus has not been previously described we investigate the nuclear localization of these TFs in functional AtfA-D:GPF strains (Figure 3A, Supplementary Figure 1). AtfA:GFP is ~97% located in the nucleus of A. fumigatus, independently of exposure or not to stressing conditions (Figures 3A–D). In contrast, about 20–30% of AtfB, -C, and -D:GFP locates into the nucleus (Figures 3A–D). However, under stressing conditions, these TFs tend to translocate to the nucleus. AtfB:GFP translocates to the nucleus about 60–80% upon the addition of sorbitol, H₂O₂, or caspofungin (Figures 3A–D). AtfC:GFP translocates about 65–80% upon addition of sorbitol and H₂O₂ but only 10% in the presence of caspofungin (Figures 3A–D). Finally, AtfD:GFP translocates 30–60 in the presence of sorbitol and H₂O₂, respectively, but 60% in the presence of caspofungin (Figures 3A–D).

These data suggest AtfA-D are possibly important for the transcriptional regulation of genes involved in the response to osmotic, oxidative, and cell wall stresses.

To verify if AtfA-D could influence carbon source utilization we checked the glucose consumption rate in liquid medium for single and double deletion mutants. Single mutants show slower glucose consumption when compared to the wild-type (Figures 4A–D). While the wild-type and ΔatfD exhaust all glucose in 24 h, the ΔatfA and ΔatfB strains consumed only 80% of glucose and ΔatfC consumed 90%. The ΔatfA ΔatfB, ΔatfA ΔatfC, ΔatfB ΔatfC mimicked the phenotypes of either ΔatfA, ΔatfB, or ΔatfC (Figures 4E,F,H, Table 1) while deletion of either atfA, atfB, or atfC in the ΔatfD recapitulates the ΔatfD phenotype (Figures 4G,I,J, Table 1). We measure the dry weight of mycelia to evaluate if a reduced glucose consumption would affect the growth of strains (Supplementary Figure 2). The single mutant ΔatfA shows a 50% reduction in dry weight when compared to the wild-type (Supplementary Figure 2A), while the double mutants ΔatfA ΔatfC, ΔatfA ΔatfD, ΔatfB ΔatfC, and ΔatfB ΔatfD show 40–50% dry weight reduction when compared to the wild-type strain (Supplementary Figure 2B).

Decreased glucose consumption for the single and double mutants led us to conclude that the deletion of atfA-D genes impacts carbon source utilization. In the presence of acetate 1% as single carbon source, he ΔatfA ΔatfB, ΔatfA ΔatfC, and ΔatfA ΔatfD mimick the ΔatfA 50% growth reduction on acetate (Figures 5A,B, Table 1) while ΔatfC, ΔatfD, and ΔatfC ΔatfD have about 40% increased growth on acetate compared to the wild-type strain (Figures 5A,B, Table 1). In the presence of ethanol 1% as single carbon source, the ΔatfC, ΔatfD, ΔatfB ΔatfC, ΔatfC ΔatfD mutants have about 40% growth reduction when compared to the wild-type strain (Figures 5C,D, Table 1). There is an additive interaction in the ΔatfA ΔatfB double mutant (Figures 5C,D, Table 1).

Taken together, these data suggest a multi-factorial interaction between TFs AtfA-D that contributes to modulate metabolism allowing the adaptation of the fungus to stressful conditions, therefore impacting A. fumigatus ability to use different carbon sources.

We have previously shown that MAPKs SakA and MpkC physically interact with protein kinase A and these protein kinases are necessary for normal accumulation/degradation of trehalose and glycogen (de Assis et al., 2018). The ΔatfD, ΔatfA ΔatfD, ΔatfB ΔatfD, and ΔatfC ΔatfD have about 30–50% reduction in trehalose accumulation when compared to the wild-type strain (Figures 6A,B, Table 1). There is about a 2-fold increased glycogen accumulation in ΔatfD and ΔatfA ΔatfC when compared to the wild-type (Figures 6C,D, Table 1). Interestingly, all the double mutants, except ΔatfA ΔatfC, have about 50–90% glycogen reduction when compared to the wild-type (Figures 6C,D, Table 1).

These data suggest that AtfA-D are important to modulate the metabolism of storage sugars, like trehalose and glycogen.

The single mutants ΔatfA and ΔatfB are either attenuated or avirulent in both Galleria mellonella larval model and in a neutropenic murine model (Silva et al., 2017). To evaluate the virulence of double deletion mutants we used the G. mellonella model of infection (Silva et al., 2017). The double mutant ΔatfA ΔatfB is avirulent and are epistatic to ΔatfA and ΔatfB (Figure 7A, Table 1) while atfA or atfB deleted in the ΔatfC or ΔatfD mimicked the avirulent phenotypes of ΔatfA and ΔatfB (Figures 7B–E, Table 1). Interestingly ΔatfC ΔatfD showed to be avirulent while ΔatfC suppressed the avirulent phenotype of ΔatfA (Figure 7F, Table 1).

These results suggest that AtfA-D interactions affect the A. fumigatus virulence in G. mellonella.

Transcriptional analysis of A. fumigatus identified atfA-D mRNA accumulation as dependent on SakA and MpkC (Silva et al., 2017). A. nidulans AtfA physically interacts with SakA upon oxidative stress (Lara-Rojas et al., 2011). A. fumigatus SakA:3xHA physically interacts not only with AtfA:GFP, but also with AtfB-D:GFP in the absence and presence of osmotic (10 min exposure to 1.2 M sorbitol) and cell wall (10 min exposure to congo red 300 μg/ml) stresses (Figures 8A,B, Supplementary Figure 3).

These results suggest that SakA is directly or indirectly interacting with AtfA-D raising the hypothesis that SakA is directly or indirectly involved in the activation of TFs AtfA-D in A. fumigatus.

We constructed six different combinations of functional strains expressing AtfA-D:GFP and AtfA-D:3xHA (AtfA:GFP AtfB:3xHA, AtfA:GFP AtfC:3xHA, AtfD:GFP AtfA:3xHA, AtfB:GFP AtfC:3xHA, AtfD:GFP AtfB:3xHA, and AtfD:GFP AtfC:3xHA; Figure 9, Supplementary Figure 4). Subsequently, we evaluated through co-IP experiments the hypothesis that AtfA-D could physically interact forming heterodimers during different stressing conditions. AtfA:GFP interacted with AtfB, AtfC, and AtfD, both at before (T0) and after the addition of sorbitol or CR (Figures 9A–C). The TFs AtfC and AtfD also showed physical interaction in all three conditions (Figure 9E). In addition, AtfB interacted with AtfC in the three conditions T0, sorbitol, and CR (Figure 9F). The additional bands observed for AtfD:GFP and AtfA:3xHA (Figure 9C) and AtfD:GFP (Figure 9D) could be GFP degradation products and/or AtfA/AtfD intermediary phosphorylated forms.

Together, these data strongly suggest that the interactions of TFs are of great importance for efficient gene transcription, allowing the fungus to quickly and robustly respond to different growth conditions.