Analyses of the three 1-Cys Peroxiredoxins from Aspergillus fumigatus reveal that cytosolic Prx1 is central to H2O2 metabolism and virulence

Standing among the front defense strategies against pathogens, host phagocytic cells release various oxidants. Therefore, pathogens have to cope with stressful conditions at the site of infection. Peroxiredoxins (Prx) are highly reactive and abundant peroxidases that can support virulence and persistence of pathogens in distinct hosts. Here, we revealed that the opportunistic human pathogen A. fumigatus presents three 1-Cys Prx (Prx6 subfamily), which is unprecedented. We showed that PrxB and PrxC were in mitochondria, while Prx1 was in cytosol. As observed for other Prxs, recombinant Prx1 and PrxC decomposed H2O2 at elevated velocities (rate constants in the 107 M−1s−1 range). Deletion mutants for each Prx displayed higher sensitivity to oxidative challenge in comparison with the wild-type strain. Additionally, cytosolic Prx1 was important for A. fumigatus survival upon electron transport dysfunction. Expression of Prxs was dependent on the SakAHOG1 MAP kinase and the Yap1YAP1 transcription factor, a global regulator of the oxidative stress response in fungi. Finally, cytosolic Prx1 played a major role in pathogenicity, since it is required for full virulence, using a neutropenic mouse infection model. Our data indicate that the three 1-Cys Prxs act together to maintain the redox balance of A. fumigatus.


Results
The A. fumigatus genome has three 1-Cys Prx genes. Initially, the repertoire of antioxidant enzymes in the A. fumigatus A1163 genome was analyzed using Prx enzymes from Saccharomyces cerevisiae and Homo sapiens as query sequences. This analysis revealed that A. fumigatus possesses six putative Prx-encoding genes. Three of them encoding atypical 2-Cys Prxs (having an intramolecular disulfide as a catalytic intermediate): AFUB_049980, AFUB_063010 and the AFUB_096050 14 . In contrast, no typical 2-Cys Prx (Prx1/AhpC subgroup 29 ) was found. We also identified three genes encoding 1-Cys isoforms: AFUB_065670, AFUB_062560 and AFUB_080670. Notably, this is the first description of a genome containing three genes encoding 1-Cys Prxs that belong to the Prx6 subgroup 29 . We named AFUB_062560 and AFUB_080670 as prxB and prxC, respectively to be consistent with A. fumigatus nomenclature. The gene AFUB_065670 was previously named as prx1 30 .
We focused our studies on the three 1-Cys Prx enzymes, since they are less studied than the 2-Cys Prx counterparts. Moreover, the number of these isoforms found in the A. fumigatus genome is unprecedented, frequently only one 1-Cys Prx isoform or none is found in the majority of organisms. Our results revealed that except for A. fumigatus, all Aspergillus species available in the Aspergillus Genome Database (http://www.aspergillusgenome. org) possess at least two isoforms of 1-Cys Prx with amino acid identity raging from 70-90% with the A. fumigatus 1-Cys Prx counterparts. In each species, one of the two isoforms presented a putative N-terminal mitochondrial target sequence (probability > 90%). Noteworthy, only in A. fumigatus strains (A1163 and Af293), three 1-Cys Prx isoforms were detected. These data reinforces that the number of isoforms of 1-Cys Prx is unusual in Aspergilli, especially in A. fumigatus.
The three A. fumigatus 1-Cys Prx genes presented moderate similarity with the human orthologue (Prx6), ranging from 37% (prx1) to 51% identity (prxB and prxC). In addition, they contain a C P embedded in the highly conserved PVC P TTE motif, which is archetypal to the Prx6 subclass ( Fig. 1) 25,29,31 .
All the Prx enzymes characterized to date present a catalytic triad that also comprises a Thr (in some cases replaced by a Ser) and an Arg residue 18 . As expected, these three residues are conserved in Prx1 and PrxC, but surprisingly, the catalytic triad Thr is substituted by a His residue in the primary sequence of PrxB. Unfortunately, we were unable to obtain the soluble form of recombinant PrxB to investigate the ability of this enzyme to reduce hydrogen peroxide (see below). PrxC presents a N-terminal extension (Fig. 1), suggesting mitochondrial localization, as is seen for the 1-Cys Prx from S. cerevisiae (Prx1) 32 . Indeed, PrxC displayed a high probability (0.99) of being targeted to mitochondria with the predicted cleavage site located at the residue 59. A. fumigatus Prx1 and PrxC are able to efficiently decompose H 2 O 2 . Next, we attempted to express the three 1-Cys Prx enzymes in Escherichia coli strains, but after several trials no soluble PrxB was obtained, precluding further characterization. In contrast, Prx1 and PrxC were successfully expressed and purified ( Supplementary Fig. S1). Then, their ability to decompose H 2 O 2 was analyzed by a competitive assay against horseradish peroxidase (HRP). The rate constants for the reactions of Prx1 and PrxC with H 2 O 2 were both in the 10 7 M −1 s −1 range (Fig. 2), therefore indicating that these two enzymes are highly efficient peroxidases. Identical residues among sequences are depicted in blue. The yellow box denotes the conserved motif PVC P TTE in the Prx6 subclass. The colored arrowheads indicate the conserved amino acids of the catalytic triad: Thr (red), C P (orange) and Arg (blue). The colored bars denote the mitochondria signal peptide (red), residues insertions (green) or deletions (orange). The secondary structure assignments based on the structure of H. sapiens Prx6 (PDB code: 1PRX) are shown below the amino acid sequences. The species abbreviation initials are given in italics and are as follows: Af_prx1 = A. fumigatus prx1 (AFUB_065670); Af_prxB = A. fumigatus prxB (AFUB_062560); Af_ prxC = A. fumigatus prxC (AFUB_080670); Sc_Prx1 = S. cerevisiae Prx1 (NP_009489. Their abilities to reduce H 2 O 2 were similar to other Prxs previously described 21,33,34 and comparable to catalases and glutathione peroxidases 35,36 . Original absorbance spectra for the HRP competitive assay are available in Supplementary Figure S2.
Unfortunately, we were unable to identify reductants for Prx1 and PrxC, which preclude as from performing bi-substrate, steady-state kinetics. Preliminary studies indicated that thioredoxin and GSH are poor reducing substrates for these peroxidases.

1-Cys
Prxs differentially contribute to oxidative challenge tolerance. Subsequently, prx1, prxB and prxC were deleted (Supplementary Fig. S3A-K) and phenotypes were investigated. Each individual gene deletion strain was also complemented with the corresponding wild-type allele to discard the occurrence of possible secondary mutations. The mRNA levels for each Prx were assessed by RT-qPCR. As expected, they were similar in the complemented and wild-type strains, while absent in the corresponding null mutant (data not shown).
We also attempted to isolate double and triple mutants for prx1, prxB and prxC genes, but we only succeeded in obtaining the ΔprxB ΔprxC double mutant ( Supplementary Fig. S3M). Remarkably, any double mutant containing the deletion of the prx1 gene seemed to be non-viable when we platted the transformed protoplasts in the regeneration medium (YG supplemented with 0.6 M KCl as osmotic stabilizer). Therefore, we isolated a conditional mutant for prx1 under the control of the nitrate reductase promoter (niiA) 37 to be used in the construction of double mutants. However, under repressive conditions (Minimal Medium supplemented with 50 mM ammonium tartrate), the niiA::prx1 mutant did not exhibit phenotypes similar to the null mutant Δprx1 in the presence of oxidants (data not shown). Possibly, transcriptional repression achieved by the niiA promoter was not strong enough to generate a loss-of-function phenotype. Further experimentation is required to understand the essentiality of the prx1 gene in backgrounds already deficient for the other 1-Cys Prxs.
Next, phenotypes were evaluated in cells grown in minimal medium (MM). In non-stressing conditions, the deletion mutants displayed radial growth comparable to the wild-type and complemented strains (Fig. 3). The growth kinetics, germination rates and conidiation levels were also analyzed but no significant differences were observed for the three 1-Cys Prx null mutants in comparison to the wild-type strain (data not shown). However, the three Prx null mutants displayed increased sensitivity to paraquat and menadione ( Fig. 4 and Supplementary  Fig. S4), two compounds that continuously generated superoxide and H 2 O 2 by the redox cycling. Surprisingly, the tolerance of the double mutant ΔprxB ΔprxC to paraquat was comparable to the wild-type strain, suggesting compensatory effects such as increased expression of other antioxidant enzyme. As a general trend, the sensitivities to oxidative challenge were more evident for the Δprx1 mutant.
In contrast, the three Prx mutants were as resistant as wild-type cells to exogenous H 2 O 2 treatment ( Supplementary Fig. S5). Possibly, the three 1-Cys Prx enzymes might be more adapted to the removal of H 2 O 2 continuously formed by the autoxidation of redox compounds, such as paraquat, menadione (Fig. 4) and components of the respiratory chain.
Since glucose represses the expression of genes involved in mitochondrial biogenesis, respiratory metabolism and antioxidant defense in S. cerevisiae 38 , we hypothesized that sensitivity to H 2 O 2 would be altered in cells growing in non-fermentative carbon sources, such as glycerol and lactate. However, again, no increased sensitivity was observed, even in the double mutant ΔprxB ΔprxC ( Supplementary Fig. S5A-B).
Additionally, prx1, prxB and prxC were dispensable for voriconazole and caspofungin susceptibility (data not shown) while a slight sensitivity to amphotericin B and SDS was observed for the Δprx1 mutant ( Supplementary  Fig. S6).
In summary, our phenotypic analysis indicated that all three 1-Cys Prx are relevant in conditions where H 2 O 2 is continuously formed by the redox cycling of paraquat and menadione and Δprx1 is more sensitive to these stressful conditions than the other two deletion mutants.  Fig. S3N-S). The sensitivities of these strains to oxidative insult were comparable to the wild-type strain indicating that they are fully functional ( Supplementary Fig. S7). The Prx1::GFP strain produced a strong and diffuse fluorescent signal without a noticeably sub-cellular accumulation, consistent with a cytosolic localization (Fig. 5). In contrast, the fluorescence of PrxB-GFP co-localized with the mitochondrial network, but the signal intensity was weaker than that of Prx1-GFP and PrxC-GFP. The PrxC-GFP fusion protein produced the most intense fluorescent signal, which also co-localized with Mitotracker, being evenly distributed along mitochondria (Fig. 5A). This was expected, since as mentioned above, the product of the prxC gene presents a predicted presequence signal for mitochondrial localization (Fig. 1). As an alternative method, we performed cell fractionation of the mycelium and samples were then analyzed by Western blot using anti-GFP to detect each of the 1-Cys Prx. According to the microscopic studies, Prx1 was detected in the cytosolic fractions, while PrxB and PrxC were present in the mitochondrial fractions (Fig. 5B). Therefore, the three 1-Cys Prxs are abundant in the cytosol and mitochondrial compartments even in the absence of exogenous oxidative challenge. Expression patterns of 1-Cys Prxs. To gain insights on the physiological roles of A. fumigatus 1-Cys Prxs, the expression patterns of their respective genes were characterized. Paraquat was chosen as the oxidant molecule since the single deletion mutants displayed increased sensitivity to this drug (Fig. 4). As a general trend, prx1, prxB and prxC genes were all induced by paraquat in a time dependent manner. However, there are considerable differences among them, such as the low induction of prxC (approximately 1.9-fold) in comparison to prx1 (approximately 5.0-fold) (Fig. 6). Expressions of 1-Cys Prx genes in the mutant strains were also investigated and compensatory effects were observed only in few conditions. For instance, we observed an increase in prxB mRNA abundance 30 minutes after paraquat exposure in the Δprx1 strain. In contrast, the expression of prxC is comparable in the Δprx1 and wild-type strains.
To further analyze compensatory mechanisms, the expression of other antioxidant genes was evaluated, such as the catalases and superoxide dismutases 39,40 . Furthermore, we analyzed the expression of the transcription factor yap1 required for oxidative stress tolerance 30 and the putative mitochondrial cytochrome-c involved in the response to oxidative stress (yeast CCP1 homolog) 30 . As expected, all the genes tested showed significant up-regulation after paraquat exposure ( Supplementary Fig. S9). Since these antioxidant genes were still induced in the 1-Cys Prx deletion strains, no evidence for compensatory effects was obtained.
The expression of 1-Cys Prxs was also evaluated at the protein level by Western blot. The detected bands migrated as predicted by their aminoacid sequences fused with GFP (Fig. 7). The expression of Prx proteins increased in response to both H 2 O 2 and paraquat, but following distinct patterns. Expression levels of Prx1 and PrxB progressively increased. Interestingly, PrxC abundance in non-stressing conditions was higher than PrxB, but PrxB was highly inducible. Indeed, the expression of PrxC follows a constitutive pattern, although in the case of paraquat, induction was observed for treatments at 2.5 mM concentration.
Prx1 is important for survival upon electron transport chain dysfunction. The role of the mitochondria and ROS in human fungal pathogenesis and drug resistance is an important area of investigation [41][42][43] . Therefore, we examined the effects of pharmacological inhibitors of the electron transport chain on the viability of the 1-Cys Prx deficient strains. In the presence of varying concentrations of rotenone, antimycin A and the uncoupler FCCP [carbonyl cyanide 4-trifluoromethoxy) phenylhydrazone], all three mutant strains presented sensitivity similar to wild-type cells ( Supplementary Fig. S11). In contrast, the Δprx1 mutant displayed impaired tolerance to malonate, oligomycin and SHAM (salicylhydroxamic acid), which inhibits complex II, FoF1-ATP synthase and the alternative oxidase (AOX), respectively ( Fig. 8A and Supplementary Fig. S12), indicating an important role of Prx1 in cell adaptation under respiratory deficiency. Again, this phenotype was not observed in the double mutant ΔprxB ΔprxC, further suggesting a prominent function of prx1.
AOX plays a major role in limiting mitochondrial ROS formation and consequently oxidative stress in plants 44,45 and in A. fumigatus 41,46 . Therefore, we investigated if deletion of 1-Cys Prx genes could increase the expression of AOX from A. fumigatus (AoxA). Indeed, aoxA mRNA levels were twice to three times higher in the ΔprxB and ΔprxC mutants compared with the wild-type strain, in non-stressing condition (Fig. 8B). In contrast, only in the Δprx1 background, the aoxA gene was not induced by paraquat, suggesting a possible genetic interaction between Prx1 and AoxA. The lower levels of AoxA protein in the mitochondrial fractions found only for the Δprx1 mutant (when compared to the wild-type strain) further support this finding (Fig. 8C). It is possible that another factor is involved in the Prx1-AoxA interaction, as this Cys-based peroxidase is cytosolic and AoxA is mitochondrial.  in iprodione and fludioxonil resistance [47][48][49][50][51] . Here, we found that the expression of the three 1-Cys Prxs genes were up-regulated by fludioxonil in the wild-type strain; prxB achieving very high levels of induction by this drug (Fig. 9A). The inactivation of sakA, but not mpkC, significantly attenuated the transcriptional responses for all three Prxs to fludioxonil and paraquat (p ≤ 0.05; Fig. 9A,B). Noteworthy, the expression of the three 1-Cys Prxs, especially the prx1 gene, were even higher in the ΔmpkC mutant background in the presence of fludioxonil. Moreover, Δprx1 was the most sensitive mutant to fludioxonil and iprodione, although ΔprxB and ΔprxC were also less tolerant than the wild-type strain to these fungicides (Fig. 9C,D). Interestingly, there was an epistatic interaction between prxB and prxC for the tolerance to fludioxonil (Fig. 9C,D). The levels of SakA phosphorylation increased in the time course experiment of fludioxonil treatment in the wild-type strain ( Supplementary Fig. S14). Further increase in SakA phosphorylation upon fludioxonil treatment was observed only in the Δprx1 background after 15 minutes of exposure, which again suggests a leading role of Prx1 in fludioxonil and ROS tolerance. Collectively, these results indicate that SakA/MpkC are important for 1-Cys Prx regulation in A. fumigatus.
Yap1 is a transcription factor described as a primary determinant in the antioxidant response in yeast (reviewed in 52 ). In A. fumigatus, the close homolog yap1 is also involved in the response to oxidative challenge, but it is not required for virulence 30 . Therefore, we analyzed whether the expression of A. fumigatus 1-Cys Prx genes could also occur in a Yap1-dependent manner. The transcriptional induction of prxC and especially of prx1 by paraquat and fludioxonil were significantly reduced in the Δyap1 and ΔsakA strains (Fig. 9A,B). In contrast, the expression of prxB was only partially regulated by Yap1.
The Δprx1 strain has attenuated virulence in a neutropenic murine infection model. The involvement of the three 1-Cys Prxs in the pathogenicity of A. fumigatus was evaluated in a neutropenic murine model of invasive pulmonary aspergillosis. The Δprx1 infection resulted in a significant decrease in mortality rate, in a condition where wild-type infection resulted in 100% mortality 12 days post-infection (Fig. 10A). In contrast, no significant differences among the wild-type, ΔprxB and ΔprxC were observed in our experimental conditions (Fig. 10B). Virulence was fully restored in the Δprx1::prx1 + complemented strain. In addition, we asked whether these mutations would affect the host immune response. To investigate whether the three 1-Cys Prxs could influence the internalizing and fungicidal activity of macrophages, we compared phagocytosis index and killing levels of Bone Marrow Derived Macrophages (BMDMs) among the wild-type, mutants and complemented strains. Corroborating the survival experiments, both the phagocytosis index (Fig. 10C) and the killing levels (Fig. 10D) were significantly increased in Δprx1 strain when compared to wild-type strain and complemented strains. Again, no significant differences were verified in these parameters for the ΔprxB and ΔprxC mutants. We also measured the levels of TNF-α, IL-1β, and IL-12p40 from BMDMs after co-incubation with A. fumigatus conidia ( Supplementary Fig. S15), but no significant differences were verified. Overall, these experimental observations suggest that Prx1 is part of the virulence arsenal of A. fumigatus and highlight the importance of 1-Cys Prxs in host-pathogen interactions.

Discussion
Tolerance to oxidative challenge is one of the most recognized determinants of virulence in several pathogens. Although several antioxidant genes and transcriptional regulators have been characterized in A. fumigatus 30,39,40,53 , the exact contributions of these proteins in the cellular response to ROS remain elusive. This study, therefore, aimed to gain insights on how A. fumigatus antioxidant defenses are impacted by the peroxidase activity of 1-Cys Prxs.
We described here that Prx1, PrxB and PrxC hold the conserved PVC P TTE motif of the 1-Cys Prxs (Prx6 subfamily), as well as a fully conserved Arg residue (Fig. 1). These features are related with extremely high efficiency of Prx1 and PrxC to reduce H 2 O 2 (rate constants in the 10 7 M −1 .s −1 range, Fig. 2). In contrast, atypical 2-Cys Aspf3 from A. fumigatus presented rate constant several orders of magnitude lower (10 4 M −1 .s −1 range), but it should be considered that this parameter was obtained by steady state kinetics (kcat/Km), using heterologous proteins from E. coli to turn over the enzyme 14 . Nevertheless, Prx1 and PrxC seem to play a major role in H 2 O 2 metabolism as these enzymes are highly reactive (Fig. 2) and also abundant (Figs 5-7). Accordingly, we observed that the Prx null mutants showed decreased tolerance to paraquat and menadione, two compounds that undergo redox cycling, generating a flux of H 2 O 2 (Fig. 4). Unfortunately, we could not identify the nature of Prx1 and PrxC reducing agents, which is also the case for several others 1-Cys Prx. As other Prx enzymes, Prx1 and PrxC probably display a ping-pong mechanism and the determination of true K M and k cat parameters required the determination of rates at distinct concentrations of the reducing and oxidizing substrates.
Factors such as the regulation of gene expression, sub-cellular location and substrate affinity should determine specific roles for each 1-Cys Prx in the cellular response to H 2 O 2 . Nevertheless, in most of the conditions analyzed herein, Prx1 appears to play major roles in the response of A. fumigatus to oxidative challenge. Noteworthy, recently we developed a device to electrochemically monitor the degradation of H 2 O 2 in the microenvironment of A. fumigatus and indeed the efficiency of H 2 O 2 degradation for the Δprx1 mutant was much lower (8%), in comparison to the wild-type strain (71%) 54 .
The finding that A. fumigatus contains no typical 2-Cys Prx and three 1-Cys Prx enzymes is very unusual, since the majority of organisms possess only one functional copy of 1-Cys Prx. Therefore, we identified their sub-cellular localization to gain insights into their individual roles. In a previous proteomic study of both crude mycelial and mitochondrial protein extracts, Prx1 was detected only in the mycelial extract, PrxC was detected in both preparations, while no information was recovered for PrxB 55 . These results were confirmed here as we further advanced in determining the specific sub-cellular localization of PrxB and PrxC compartmentalization (Fig. 5).
The mitochondrial repertoire of antioxidant enzymes is quite diverse, but generally Prx enzymes play central roles in the removal of H 2 O 2 in this organelle. For instance, mitochondria of S. cerevisiae contains only one 1-Cys Prx 56 that is located in both matrix and intermembrane space 56 , whereas mammalian organisms contain two 2-Cys Prx (Prx3 and Prx5). Noteworthy, human Prx3 decomposes 90% of the H 2 O 2 generated within the mitochondrial matrix 21,57,58 . N-termini cleavable leader sequences of a preprotein synthesized by cytosolic ribosomes is the classical type of mitochondrial targeting signals, as we observed for yeast Prx1 and A. fumigatus PrxC. However, additional non-cleavable targeting and sorting signals, which are located in the mature regions of mitochondrial proteins, have also been described [59][60][61] . This could be the case of A. fumigatus PrxB since we could not detect a canonical cleavable N-terminal presequence for this protein.
ROS production derived from the leaking of electrons from the mitochondrial transport chain is a well-established phenomenon 38,62,63 . However, these mechanisms are still scarcely understood in fungi, but there are evidences that the canonical mitochondrial respiratory pathway via cytochrome c or cyanide-resistant respiration (via AOX) impact fungal virulence and the host immune response 41,46,[64][65][66] . Intriguingly, cytosolic Prx1 is involved in the maintenance of mitochondrial homeostasis (Fig. 8). AOX is another system involved in the decreased accumulation of mitochondrial ROS in the response of plant and fungal cells to oxidative challenge 41,42,44,67 . Curiously, we observed decreased levels of AoxA in the prx1 deletion mutant (Fig. 8B,C) what can partially accounts for the more pronounced cell susceptibility to oxidants in this mutant (Figs 4 and 8). A potential reason for this phenotype is if AoxA is depleted, H 2 O 2 could accumulate in mitochondria and then diffuses to the cytosol 63 . Thus, AoxA would act in parallel with Prx to counteract oxidative stress. Cytosolic Prx is important for Yap1 activation through nuclear localization in yeast 68,69 . Loss of Prx1 in A. fumigatus might impact the general response to oxidative challenge. So, in the absence of prx1, response of H 2 O 2 would be decreased and levels of this oxidant should be higher in deletion mutant for aoxA. Confirmation of this hypothesis awaits further experimentation. Survival of the mutant strain was compared to that of the wild-type and complemented strain. Infection experiments were repeated two times with similar results. Survival curves between the Δprx1 mutant and wild-type strain and mutant and reconstituted strains were statistically significant (*p ≤ 0.0132 and p ≤ 0.0074, respectively). No significant difference was observed between the wild-type and ΔprxB or ΔprxC deletion strains. Phagocytosis index (C) and killing activity (D) is increased in the A. fumigatus Δprx1 mutant strain. The results are expressed as means ± SD for three independent experiments. *p ≤ 0.05 compared to wild-type; # p ≤ 0.05 compared to the complemented strain.
Unlike the core components of fungal respiration, the fungal AOX appears to have complex mechanisms for regulation of expression and enzyme activity which are remarkably different from plants 42,67 . Thus, a further understanding of how AOX and 1-Cys Prx activities are regulated across fungi can yield insights about the consequences for the fungal cell under accumulation of hydroperoxides.
In addition, all three 1-Cys Prxs are regulated by the HOG pathway and appear to cooperate in the response of the fungi to the crop fungicides (Fig. 9). Hyperactivation of the HOG pathway is the mechanism that explains cell death in yeast and C. neoformans while perturbations in any of the three components of the HOG pathway render C. neoformans highly resistant to fludioxonil 50,51 . Similarly, Hog1 mutants in S. cerevisiae and other filamentous fungi are also resistant to these drugs [70][71][72][73] . Although in A. fumigatus mutants for both sakA HOG1 and mpkC were slightly more resistant to fludioxonil 49 , we observed increased activation of SakA in the Δprx1 mutant ( Supplementary Fig. S14). Our data indicated that SakA, and not MpkC, is necessary for the expression of 1-Cys Prxs during exposure to fludioxonil. Expression of prx1 and prxC is also regulated by Yap1 (Fig. 9A,B), a transcriptional regulator that is redox regulated in yeast 74 . Accordingly, here we also identified in silico a conserved binding motif for Yap1 in the promoter regions of the three A. fumigatus 1-Cys Prxs ( Figure S16). Functional validation of these predicted binding motifs in the transcriptional activation of the 1-Cys Prxs awaits further experimentation. A previous genome-wide study also observed that PrxC is down-regulated in a Δyap1 mutant exposed to H 2 O 2 30 . How the distinct signaling pathways (HOG and Yap1) interact to integrate the responses of this pathogenic fungus to stressful conditions awaits further investigation.
Importantly, we observed that Prx1 is required for A. fumigatus virulence in our mouse model (Fig. 10). Based on data presented by this study and others 14,75 , Prx-enzymes emerge playing leading roles in A. fumigatus protection against host defenses. For instance, neither a triple sod mutant nor individual deletion of catalases significantly causes virulence attenuation in A. fumigatus 39,40,53 . In addition, since mammalian 1-Cys Prx (Prx6) presents acidic Ca 2+ -independent phospholipase A2 activity 24,25 , is possible that such putative activity in the A. fumigatus 1-Cys Prxs might also contribute to the fitness of the pathogen inside the host. Altogether, we cannot rule out the possibility that PrxB and PrxC are also important for virulence under other experimental conditions or additional animal models. Remarkably, BMDMs stimulated with Δprx1 showed enhanced phagocytosis and killing levels. Although this finding is consistent with the survival experiments, the verified cytokines levels were not significant among the analyzed mutants ( Supplementary Fig. S15). Probably the possible modifications in the tolerance of A. fumigatus to oxidative damage could promote an unbalance at the phagolysosome acidification during phagocytosis and killing, but were not enough to increase the recognition of the mutated fungus by dectin-1 receptor, which would promote an exacerbated production of pro-inflammatory cytokines. Reactive oxidant intermediates have been demonstrated involved in the inhibited killing of A. fumigatus without affecting the uptake rate 76 . Also, the mechanisms of conidia killing by oxidation are still elusive since the available data in the literature for killing of A. fumigatus by macrophages are very unrelated 76 .
Taken together, our studies uncovered that the three 1-Cys Prxs act together in the A. fumigatus response to oxidative challenge with Prx1 playing a major role in virulence. We argue that further characterization of additional biochemical properties of the 1-Cys Prxs, the identification of their potential interacting proteins and understanding the mechanisms whereby they are reduced in vivo will help us to elucidate how A. fumigatus relies on 1-Cys Prxs to cope with ROS arising from its own metabolism and from the host during the course of infection.

Methods
Cloning the 1-Cys Prx genes. Sequence analysis of the A. fumigatus 1-Cys Prx were performed as described in Supplementary Methods. Prx cDNA sequences were obtained from a paraquat-induced culture of CEA17 wildtype strain. RT-PCR was run using the primer sets described in Supplementary Table S1. The PCR products were cloned into the pTZ57R/T easy vector (Thermo Scientific) and subcloned into the pET15b expression vector at the NdeI and BamHI sites using standard protocols 77 . The pET15b/prx1, pET15b/prxB and pET15b/prxC clones were validated by fully sequencing.
Protein expression and purification. Single colonies of E. coli BL21(DE3) harboring pET15b/prx1 or pET15b/prxC were cultured in LB (100 μg/ml ampicillin) and grown for 16 hours (37 °C) in an orbital shaker (250 rpm). The culture was diluted in 1 L of LB (100 μg/ml ampicillin) and cultured at 37 °C until the OD 600 reached 0.6-0.8. IPTG was then added to a final concentration of 0.3 mM and the culture was incubated at 37 °C for an additional 3 h (250 rpm). Cells were harvested by centrifugation, and the pellet was washed and suspended in the start buffer (20 mM sodium phosphate buffer, pH 7.4; 500 mM NaCl; 20 mM imidazole; 625 μM PMSF). Twenty cycles of 15 s of sonication (30% amplitude) following 40 s in ice were applied to cell suspensions. Cell extracts were kept on ice during streptomycin sulfate 1% treatment for 15 min and the suspension was centrifuged at 15,000 rpm for 45 min to remove nucleic acid precipitates and the insoluble fraction. Finally, cell extracts were filtered and purified by immobilized metal ion affinity chromatography (IMAC; HiTrap column GE Healthcare) using ÄKTA FPLC (GE Healthcare).
The conditions of protein purification were optimized using the gradient procedure for imidazole concentration as described by the manufacturer. Imidazole was removed from purified proteins by gel filtration using a PD10 column (GE Healthcare). Purified proteins were stocked in 20 mM sodium phosphate buffer (pH 7.4) containing 500 mM NaCl. Recombinant protein concentrations were determined spectrophotometrically at 280 nm. The extinction coefficients for reduced Prx1 (ε 280 = 20,970 M −1 .cm −1 ) and PrxC (ε 280 = 22460 M −1 .cm −1 ) were determined using the ProtParam tool (http://www.expasy.ch/tools/protparam.html).
SCIEntIFIC REPORts | (2018) 8:12314 | DOI:10.1038/s41598-018-30108-2 Determination of the peroxidase activity by competitive kinetics. The second-order rate constant for the reaction between Prx1 and PrxC with H 2 O 2 was determined by competition with horseradish peroxidase (HRP) 33,78 . The Prxs were first reduced with 20 mM DTT during 16 hours at 4 °C and desalted twice using a PD10 column. Following incubation, the amount of free thiols was determined using DTNB [5,5′-dithiobis(2-nitrobenzoic acid); D8130 Sigma-Aldrich] (10 DTNB: 1 protein) under denaturing conditions (ε 412 = 13,600 M −1 .cm −1 ) in order to obtain the percentage of reduced protein, which was kept at above 90%. The concentration of HRP (P6782 Sigma-Aldrich) was determined spectrophotometrically using ε 403 = 102,000 M −1 . Cell fractionation. For the extraction of mitochondrial proteins, mitochondria were isolated from A. fumigatus cells based on the previous methodology 55 . Mycelia (10 g wet weight) were obtained by growing 1 × 10 8 conidia of prx1::gfp, prxB::gfp or prxC::gfp strains for 16 h at 37 °C in 100 ml of YG. Cells were harvested, frozen in liquid nitrogen and subjected to cell fractionation procedures. 20 µg of protein from each sample were loaded on 12% SDS-PAGE and electroblotted to PVDF (polyvinylidene difluoride) membranes, which were then analyzed by Western blot using anti-GFP antibody.
Strains and media. The A. fumigatus strains used in this study are described in Supplementary Table S2.
Strains were grown in either complete medium (YG; 2% [wt/vol] glucose, 0.5% [wt/vol] yeast extract, trace elements) or minimal medium (MM; 1% [wt/vol] glucose, 1 × original high-nitrate salts, trace elements, pH 6.5). Trace elements, vitamins, and nitrate salt compositions were as described previously 79,80 . Solid YG and MM were the same as described above except that 2% (wt/vol) agar was added. When necessary, uridine and uracil (1.2 g/ liter) were added. For all the phenotypic and expression analyses, MM was used to avoid the side reactions of peroxides with extracellular components present in a rich medium 81,82 . To induce oxidative stress, 1 × 10 7 conidia from wild-type and mutant strains were incubated in 50 ml of liquid MM for 24 hours, at 37 °C. Following incubation, strains were challenged to increasing concentrations of H 2 O 2 or paraquat for 30 min. Alternatively, paraquat (10 mM) or fludioxonil (1 µg/ml) was added to the cultures, and they were incubated for an additional 15, 30 or 60 min. The control was left untreated. Mycelia were collected and stored as described previously 83 .

Generation of 1-Cys Prx mutant strains and Southern blot analyses.
The gene replacement cassettes used in this study were constructed by in vivo recombination in S. cerevisiae as reported previously 80 and are described in detail in the Supplementary Methods. Primers used for each mutant construction are described in Supplementary Table S3.
Susceptibility assays against oxidative damaging agents and antifungals. To monitor growth of the mutant strains under drugs that cause oxidative stress 1 × 10 4 conidia of each strain were grown in 1 ml of liquid MM in 24-well plates that were supplemented with varying concentrations of paraquat, menadione, iprodione and fludioxonil. The plates were incubated for 72 hours at 37 °C and photographed. The same assay was used to test the tolerance to pharmacological inhibitors of the electron transport chain, however using 1 × 10 4 conidia of each strain grown in 200 µl of liquid MM (96-well plates) that were supplemented with varying concentrations of each compound. In these experiments, we used 10% Alamar blue (Thermo Scientific) as viability indicator, according to the procedures described previously 84 . The plates were incubated for 72 hours at 37 °C and photographed.
To generate the survival curves of the wild-type and mutant strains in the presence of oxidants (paraquat and menadione) and the electron transport chain inhibitors (malonate, SHAM and oligomycin), 1 × 10 4 conidia of each strain were grown in 200 µl of liquid MM in black 96-well plates (CellStar 650086). Fluorescence intensities (λ = 590 nm) of treated and untreated samples for each strain were obtained using a Spectra Max SpectraMax i3 (Molecular Devices) after 24, 48 and 72 hours of incubation at 37 °C and used to calculate the percentage of growth, according to the manufacturer's protocol (Thermo Fisher #DAL1100).
Sensitivity to the oxidative damage generated by H 2 O 2 was evaluated by an inhibition zone assay on MM agar plates, as described previously 30 . Alternatively, to test the susceptibility of the strains to antifungal drugs, serial dilutions of wild-type and null mutant conidia were spotted onto agar plates.
RNA extraction and gene expression analysis. The total RNA obtained upon cell treatments with paraquat or fludioxonil was extracted and processed for cDNA synthesis as previously described 83 . The primers for the individual genes are listed in Supplementary Table S4. Three independent biological replicates were used and the relative fold change in mRNA quantity was calculated using comparative cycle threshold (Ct) (ΔΔCt) analysis 85 . All values were normalized to the expression of the A. fumigatus tubA gene. Statistical analysis was performed using one-way ANOVA with Tukey's post hoc test to assess differences in the mutant strains compared to the same growth condition in the wild-type strain (p ≤ 0.05).
Immunoblot analysis. The GFP-tagged Prx strains either non-stressed or exposed to H 2 O 2 or paraquat were used to assess the sub-cellular localization and protein abundance. Protein extraction and quantification was performed as described previously 83  Nordic-Immunology), anti-phospho-p38 MAPK or anti-p38 MAPK (9211 and 9212, respectively; Cell Signaling Technologies) and anti-alternative oxidase (AoxA) (a gift from Drs. Sergio Akira Uyemura and Taisa Magnani Dinamarco, USP, Brazil; unpublished) are described in Supplementary Methods. Microscopy. Prx1::GFP, PrxB::GFP and PrxC::GFP conidia were cultivated on 35 mm Glass-Botton dishes (MatTek Corporation) in 2 ml of MM for 16 h at 30 °C. For co-localization experiments, germlings were co-stained with MitoTracker DeepRed (Thermo Scientific) as described previously 86 . Coverslips were inspected on Observer Z1 fluorescence microscope (Carl Zeiss). GFP and Mitotracker DeepRed were visualized using 38 HE and 63 HE filter sets (Carl Zeiss), respectively. DIC (differential interference contrast) and fluorescent images were captured with an AxioCam camera (Carl Zeiss) and processed using AxioVision software.
Virulence assay. The virulence of A. fumigatus strains was analyzed using a murine model for invasive aspergillosis, as detailed previously 87 . Outbreed female mice (BALB/c strain; body weight, 20 to 22 g) were housed in vented cages containing five animals. The mice were immunosuppressed with cyclophosphamide at 150 mg/ kg body weight, which was administered intraperitoneally on days -4, -1, and 2 prior to and post-infection. Hydrocortisone acetate (200 mg/kg) was injected subcutaneously on day -3. The A. fumigatus conidia that were used for inoculations were grown on solid YG medium for 2 days prior to infection. The conidia were freshly harvested in PBS and filtered using Miracloth (EMD Millipore). Conidial suspensions were spun for 5 min at 3,000 x g, washed three times with PBS, counted using a hemocytometer, and resuspended at a concentration of 5 × 10 6 conidia/ml. Viable counts of the administered inocula were determined, following serial dilution, by plating on solid YG medium, and the conidia were grown at 37 °C. The mice (10 animals) were anesthetized by halothane inhalation and infected through intranasal instillations of 1 × 10 5 conidia in 20 µl of PBS. For the negative control, a group of 10 mice received sterile PBS only. Mice were weighed every 24 hours starting from the day of infection and visually inspected twice daily. The statistical significance of comparative survival values was calculated using the Prism statistical analysis package by using Log-rank (Mantel-Cox method) and the Gehan-Breslow-Wilcoxon test.
Ethics statement. This study and all protocols involving animal care were approved by the local Ethics Committee for Animal Experiments (Campus of Ribeirão Preto, Universidade de São Paulo; Permit Number: 08.1.1277.53.6). All animals were housed within individually ventilated cages and were cared for in strict accordance with the principles outlined by the Brazilian College of Animal Experimentation (Colégio Brasileiro de Experimentação Animal, COBEA) and Guiding Principles for Research Involving Animals and Human Beings, American Physiological Society. The principles that guided our studies were based on the Declaration of Animal Rights ratified by the UNESCO in January 27, 1978 (8 th and 14 th articles). All efforts were made to minimize suffering. The animals were humanely sacrificed if moribund (as defined by lethargy, dyspnea, hypothermia and weight loss). All stressed animals were sacrificed by cervical dislocation.
Bone marrow-derived macrophages (BMDMs) preparation, phagocytosis and killing assay. All experiments with macrophages in this study were performed with BMDMs. The BMDMs preparation was performed according to 88 with slight modifications. BMDMs were prepared from C57BL/6 adult mice femurs and tibias after flushing with RPMI medium to release bone marrow cells. These cells were cultured for 6 days in RPMI 1640 medium supplemented with 20% fetal cow serum (FCS) and 30% L-929 cell conditioned medium. Non-adherent cells were removed and the adherent cells (majority macrophages) were removed and washed twice with cold PBS. Cell concentration was determined using a Neubauer chamber.
The phagocytic assay was performed as previously described 89 with slight modifications. In 24-well microplates at 37 °C with 5% CO 2 , 1 ml of RPMI-FCS containing 1 × 10 5 conidia (1:5 macrophage/conidia ratio) was added and incubated for 90 min. The supernatant was removed and added 0.5 ml of 3.7% formaldehyde-PBS. The samples were washed with ultrapure water and incubated for 20 min with 495 μl of water and 5 μl of CFW (10 mg/ml). The samples were washed and mounted on slides with 50% glycerol. The images were acquired on a fluorescence microscope Eclipse E800 (Nikon Instruments), and the phagocytosis index calculated counting at least 100 conidia per sample. The experiments were repeated in triplicate.
To assess conidial killing, the phagocytic cells were obtained as described above. Subsequently, 1×10 5 conidia (1:5 macrophage/conidia ratio) were incubated at 37 °C with 5% CO 2 for 4 h. As positive control, conidia without BMDMs were used. The 24-well microplates were centrifuged for 10 min at 3,500 rpm, and the culture supernatant removed. Then, 100 μl of 1% Triton X-100 were added. After 10 min at room temperature, samples were removed from the microplate and serially diluted in PBS. The dilutions were plated and incubated at 37 °C for 48 h. The percentage of conidial killing was calculated by the measurement of colony forming units (CFU) numbers carried out after macrophage lysis, compared from samples incubated with macrophages to CFU numbers from those incubated without macrophages. The experiments were repeated three times, each performed in triplicate.
Cytokine measurements. Cytokines were measured by capture enzyme-linked immunosorbent assay (ELISA) as previously described 90 with antibody pairs purchased from BD Biosciences. Supernatants were used to quantify the levels of TNF-α, IL-1β, and IL-12p40. The amount of cytokine was determined from standard curves, using murine recombinant cytokines as standards. Data are the average of at least three independent experiments, each performed in triplicate. Data Availability. All data generated or analyzed during this study are included in this published article (and its Supplementary Information files).