Down-regulation of TUFM impairs host cell interaction and virulence by Paracoccidioides brasiliensis

The genus Paracoccidioides consist of dimorphic fungi geographically limited to the subtropical regions of Latin America, which are responsible for causing deep systemic mycosis in humans. However, the molecular mechanisms by which Paracoccidioides spp. causes the disease remain poorly understood. Paracoccidioides spp. harbor genes that encode proteins involved in host cell interaction and mitochondrial function, which together are required for pathogenicity and mediate virulence. Previously, we identified TufM (previously known as EF-Tu) in Paracoccidioides brasiliensis (PbTufM) and suggested that it may be involved in the pathogenicity of this fungus. In this study, we examined the effects of downregulating PbTUFM using a silenced strain with a 55% reduction in PbTUFM expression obtained by antisense-RNA (aRNA) technology. Silencing PbTUFM yielded phenotypic differences, such as altered translation elongation, respiratory defects, increased sensitivity of yeast cells to reactive oxygen stress, survival after macrophage phagocytosis, and reduced interaction with pneumocytes. These results were associated with reduced virulence in Galleria mellonella and murine infection models, emphasizing the importance of PbTufM in the full virulence of P. brasiliensis and its potential as a target for antifungal agents against paracoccidioidomycosis.

During the colonization and invasion of the host, the pathogen uses various mechanisms to subvert host immune defenses 11,12 . After the inhalation of conidia, Paracoccidioides can be phagocytized by several types of cell, including macrophages which have a depleted nutrient environment and produce reactive oxygen species (ROS) and reactive nitrogen species (RNS) with antimicrobial activities. However, the fungus triggers defensive strategies in order to survive under such conditions 13,14 .
Paracoccidioides spp. possesses multiples attributes allowing them to adhere, colonize, disseminate, and adapt to various sites in the human body. Therefore, it is essential that the molecules involved in these processes be identified to improve our understanding of Paracoccidioides spp. virulence and PCM pathogenesis. Moreover, molecules that reduce the virulence of pathogens may prove a promising alternative to conventional antimicrobials 15 .
Previously, Marcos et al. 16 identified the P. brasiliensis TufM protein, and used recombinant PbTufM and its respective polyclonal antibody to demonstrate its surface localization and its importance in interactions with pneumocytes and ECM components, such as fibronectin and plasminogen. These data suggest that PbTufM may be suitable for direct biological analysis by generating silenced or knockout isolates to evaluate its effects on the fitness of P. brasiliensis.
The steps involved in translation are the same in both the mitochondrial and cytosolic systems, however the translation factors involved are different 17 . Unlike cytosolic translation, the processes underlying the protein synthesis in mitochondria still need to be better elucidated. The majority of ribosomal proteins, tRNA synthetases, and factors involved in ribosome assembly and mitochondrial translation, are codified in the nucleus, synthesized in the cytoplasm, and then delivered to the mitochondria 18 . EF-Tu plays an fundamental role in translation conferring charged tRNAs to the ribosomal A-site during peptide chain elongation enabling protein synthesis 19 . Although orthologues of the S. cerevisiae mitochondrial EF-Tu, EF-G, and RF1 have been identified in the P. brasiliensis transcriptome, no sequences have been found corresponding to functional EF-Ts 20,21 .
The absence of protein synthesis due to the deletion of mitochondrial elongation factor (TufM) prevented growth on a respiratory substrate in S. cerevisiae. Alterations to mitochondrial EF-Tu affect multiple cellular networks and pathways, and consequently cell physiology and homeostasis 22 . The vad1∆ Cryptococcus neoformans mutant has decreased TUF1 (an EF-Tu homologue) expression and has been shown to have impaired growth on a respiratory medium with a glycerol carbon source. TUF1 overexpression improves any aspects of this phenotype related to mitochondrial function, such as the ability to grow on a respiratory medium 23 .
To date, the function of PbTufM has not been determined with respect to cell biology and effects of PbTufM on the virulence of P. brasiliensis. Here, we silenced the TUFM gene using antisense-RNA (aRNA), which has a well-established methodology for use in this fungus 24,25 . Additionally, we studied the effects of PbTufM on P. brasiliensis growth and viability, its ability to interact with pneumocytes, and its virulence in alternative invertebrate (Galleria mellonella) and murine models. Our results indicated that the silenced strain, PbTufM aRNA, retained its morphology but had respiratory defects and increased sensitivity to oxidative stress. In relation to fungal fitness, this phenotype likely reflects the reduced ability of P. brasiliensis to interact with host cells.

Results
tufM is localized in the mitochondria of P. brasiliensis. Previously, Marcos et al. 16 reported the cell surface and cytoplasmic localization of TufM. Using immunoelectron microscopy we verified that TufM is present in the mitochondria, a finding confirmed by the immunogold labeling of this organelle (Fig. 1).

Suppression of pbtUfM expression using an aRnA plasmid and AtMt.
To verify the consequences of TUFM downregulation in P. brasiliensis, we used aRNA and Agrobacterium tumefaciens-mediated transformation (ATMT). aRNA corresponding to the homologous sequences of the PbTUFM exon were constructed for genomic integration by ATMT. PbTUFM downregulated, hygromycin-resistant transformants of P. brasiliensis wild-type Pb18 were selected and successive subcultured in a non-selective and selective medium with hygromycin (HPH) for three consecutive rounds. The presence of the hygromycin cassette in the genomic DNA of a P. brasiliensis transformant, named PbTufM aRNA, was confirmed using PCR (1,000 bp, Fig. 2A) with HPH-specific primers, and the transformant was selected for gene expression analysis. The levels of TUFM mRNA transcript in PbTufM aRNA were measured after 1 month and over 12 months (without statistical differences for the same isolate comparing the evaluated times) of subculture using RT-qPCR and were found to be 64-58% lower than in PbWT (Pb18; wild type-isolate) and 55-53% than in PbEV (empty vector), respectively (Fig. 2B).
PbTufM aRnA has a morphology typical of P. brasiliensis and decreased ability to utilize a nonfermentable carbon source for growth. We analyzed the effects of PbTUFM knockdown by comparing the viability, growth rate, and yeast morphology of the PbTufM aRNA, PbWT, and PbEV strains. Microscopic observation showed that PbTufM aRNA cells were not phenotypically different in fungal morphology than PbWT and PbEV cells. PbTufM aRNA cells had a normal yeast morphology consisting of a single large central cell with multiple buds attached, and no differences between isolates (Fig. 3A).
PbTufM aRNA showed no growth defects under rich fermentable conditions (BHI medium with glucose; Fig. 3B,E), however there was a slight decrease in growth in the other fermentable medium (MMcM with glucose; Fig. 3C,E), and a greater decrease under respiratory conditions (MMcM with glycerol; Fig. 3D   www.nature.com/scientificreports www.nature.com/scientificreports/ tUfM silencing reduces the levels of cell surface and cytosolic tufM in P. brasiliensis. Since it has been reported that TufM is present on the cell surface and in the cytosol of P. brasiliensis 16 , we examined the effects of TUFM silencing on its localization by using permeabilized (total protein) and non-permeabilized (surface protein) yeast cells.
Flow-cytometry showed that both permeabilized and non-permeabilized PbTufM aRNA cells had a lower fluorescence intensity than PbWT and PbEV cells, with reductions of approximately 90.3% and 64.4%, respectively, when compared to PbEV cells (PbEV showed no difference in fluorescence compared to PbWT) (Fig. 5C). The TufM fluorescence intensity measurements were concordant with the TufM fluorescence visualized using microscopy (Fig. 5A,B). Moreover, western blot analysis of the surface protein extracts of PbEV and PbTufM aRNA with anti-PbTufM (normalized using GAPDH as a control), revealed that the level of surface TufM was reduced 3.88-fold in the silenced strain. Taking the level of TufM expression in PbEV to be 100%, the reduction in PbTufM aRNA was approximately 25.7% (Fig. 5D), reinforcing above data which showed that the reduction in TufM protein levels also occurs on the cell surface.

TUFM knockdown influences the ability of P. brasiliensis to interact with pneumocytes. TufM
is expressed on the cell surface of P. brasiliensis, and it has previously been shown that treating P. brasiliensis with anti-TufM specific antibodies reduced its interactions with pneumocytes, demonstrating that cell surface TufM is important for interacting with host cells 16 . We investigated the ability of P. brasiliensis to interact with pneumocytes and cause infection by adhering to and invading host cells 26 . Compared to PbWT and PbEV, PbTufM aRNA cells had reduced adhesion and internalization in pneumocytes at 2, 5, 8 h post-infection by 3.6-fold, 4.5-fold, 3.5-fold respectively (Fig. 6A).
Moreover, the inhibition of the PbTufM aRNA -pneumocyte interaction was considerably more pronounced when cell surface PbTufM was blocked using specific anti-PbTufM antibodies, with reductions of 1.3-fold, 1.4-fold, 2.9-fold at 2, 5, 8 h post-infection, respectively (Fig. 6B). These data suggest that PbTufM is important in the pathogenic host cell interactions of P. brasiliensis. pbtUfM silencing increases sensitivity to oxidative stress and decreases the survival of P. brasiliensis against macrophages. The sensitivity of the silenced PbTufM aRNA strain to oxidative stress was evaluated by challenged the PbWT, PbEV, and PbTufM aRNA cells with hydrogen peroxide (H 2 O 2 ). All concentrations of H 2 O 2 used reduced the viability of PbTufM aRNA. When compared to the PbWT, the exposition to 10 mM, 50 mM and 100 mM reduced the viability of PbTufM aRNA in 45.6%, 50.5% and 51.7%, respectively; and when compared to PbEV, the viability reduction was 44.4%, 59% and 53.3% respectively (Fig. 7A); indicating that the down expression of PbTUFM resulted in an increase in the sensitive to hydrogen peroxide.
The survival of PbTufM aRNA adhered and internalized during the infection of RAW 264.7 macrophages was determined by evaluating the number of colony-forming units (CFUs) 2 h post-infection. PbWT and PbEV showed no significant difference, whereas a reduced number of CFUs were recovered from PbTufM aRNA to both conditions, yeasts adhered and internalized (Fig. 7B). This suggests that PbTUFM downexpression resulted in yeasts less able to survive the macrophage environment, maybe due the fact to be more sensitive to oxidative stress, a condition found in the phagocyte environment.
Knockdown isolate PbTufM aRnA has attenuated virulence in Galleria mellonella and murine models of paracoccidioidomycosis. G. mellonella model. G. mellonella larvae were injected with PbWT, PbTufM aRNA, or PbEV isolates and incubated at 37 °C for 7 days. As shown in Fig. 8, the PbTufM aRNA isolate had a higher rate of larval survival than the PbWT and PbEV isolates. The percentage of surviving larvae 7 days after infection with PbTufM aRNA was 52.7%, whilst at the end of the fifth day all larvae infected with PbWT and PbEV were dead. No mortality was observed in PBS-injected or heat-killed yeast cells. These results demonstrate that PbTufM aRNA causes significantly less mortality of the larvae, suggesting that the depletion of PbTufM reduces virulence and is important in fungal pathogenesis.
Murine model. PbTufM aRNA virulence was also evaluated in a murine model of PCM. The level of disease in the lungs of mice infected with PbTufM aRNA, PbWT, or PbEV was evaluated by counting the number of CFUs recovered and by histopathological analysis (Fig. 9A). The fungal burden was significantly lower in the lungs of mice infected with PbTufM aRNA 15 days post-infection. A 75.6% and 74.5% reduction in fungal burden was observed compared with PbWT and PbEV, respectively (Fig. 9B). The mice inoculated with PbWT, PbEV, or PbTufM aRNA cells showed inflammatory responses in the lung tissue that were directly related to fungal spread and proliferation, characterized by the infiltration of bronchovascular bundles by leukocytes and granuloma formation by mononuclear cells surrounded by yeast cells. However, the mice infected with PbTufM aRNA had lower numbers of yeast cells and giant cells, and reduced granuloma formation compared to mice infected with www.nature.com/scientificreports www.nature.com/scientificreports/ PbWT or PbEV, likely due to the reduced number of viable cells ( Table 2). The number of granulomas (Table 2) reflected the number of CFUs (Fig. 9B).

Discussion
Previously, we suggested that the P. brasiliensis surface protein TufM is involved in mediating interactions with pneumocytes, host ECM components, and the fibrinolytic system 16 . Here, we constructed a silenced strain using aRNA in order to evaluate the effects of PbTUFM downregulation on P. brasiliensis phenotype, fitness, and virulence.
The silenced strain had reduced levels of TufM both on the cell surface and in intracellular compartments. Furthermore, we verified the presence of TufM in the mitochondria of P. brasiliensis yeast cells. It has been shown that the EF-Tu protein is localized in the mitochondria of P. lutzii yeast cells 27 by using nanoUPLC-MS E on the mitochondrion-enriched fractions. However, the mechanisms by which this protein is translocated to the cell  www.nature.com/scientificreports www.nature.com/scientificreports/ surface are unclear, especially as the conventional secretion or anchoring signals are unknown. TufM may be translocated non-conventionally by vesicles like other P. brasiliensis components 28 .
PbTUFM downregulation did not cause morphological alterations in the yeast cells, which grew normally in a rich fermentative medium in vitro, and did not change fungal viability when compared to wild-type and control yeasts. In a chemically defined fermentation medium, PbTufM aRNA showed a subtle decrease in growth and viability. A greater difference was observed under respiratory conditions with glycerol as a sole carbon source, suggesting that the silenced strain has impaired respiratory abilities.
Genetic alterations to translation on mito-ribosomes can result in a cell proliferation detriment, probably due to defects associated with aerobic metabolism even though proliferating cells require glycolisys as the main source to their metabolic demands 29 . Mithocondrial EF-Tu null mutants in other species, such as S. cerevisiae, are defective for mithochondrial translation, accumulate deleted mtDNA molecules or have total loss of mtDNA 18,[30][31][32] , and have phenotypes with small cell size and slow growth 33,34 . In our study PbTUFM was still expressed, hence the strain construct was not a null mutant, however respiratory-deficient phenotypes similar to those reported  www.nature.com/scientificreports www.nature.com/scientificreports/ previously were observed. The Cryptococcus neoformans vad1Δ mutant identified in an insertional mutagenesis screen exhibited decreased expression of TUF1, which encodes an EF-Tu homologue. This mutant had a moderate melanization deficiency due to reduced laccase expression, severe virulence attenuation, and fluconazole resistance 23 . In the vad1Δ mutant, TUF1 downregulation was associated with impaired growth on a respiratory medium with a glycerol carbon source, suggesting defective mitochondrial function 35 .  The amount and quality of the proteins produced by translation can be affected by the degree of translation elongation. The rate of codon translation is inversely proportional to the occupancy of elongating ribosomes with mRNAs featuring this codon 36 . The silenced strain showed higher puromycin polypeptide incorporation, which may be due to the downregulation of TUFM in P. brasiliensis triggering the upregulation of cytosolic proteins. This effect was demonstrated in Trypanosoma brucei with mitochondrial EF-TU silencing, where the downregulation of many nuclear-encoded cytochrome oxidase subunits and bc1-complex components resulted in the upregulation of most cytosolic ribosomal proteins 37 . Conversely, the elongation defect in the silenced strain may be due to slower elongation leading to a higher ribosome occupancy which favors the incorporation of puromycin, as there are more nascent polypeptides to incorporate puromycin or more time for this process to occur 38 .
When comparing the interaction (adhesion/invasion) capabilities of the different isolates, TUFM-silenced P. brasiliensis had significantly reduced interactions with pneumocytes due to the inhibition of TufM on its surface 16 . This effect was also observed when cell surface TufM was blocked in the silenced strain, with a significant reduction in pneumocyte interactions.
As for the phenotypic differences of PbTufM aRNA, when exposed to oxidative stress the silenced strain showed reduced viability, as observed in S. cerevisiae 39 . Besides generating energy via oxidative phosphorylation, mitochondria also coordinate several aspects of cellular physiology and functional responses to different stressors 40,41 . Additionally, the capacity of pathogenic fungi to cause disease has been associated with mitochondrial activity, as shown by mitochondrial mutants with attenuated virulence in animal models 42 . The consequences of dysfunctional mitochondria may be related to decreased fitness, metabolic variation, and sensitivity to oxidative stress. Moreover, the environment pressure that operates on respiration-deficient mutants in the host and in vitro may be different, can even trigger a compensatory pathways mainly in the host despite the loss of mitochondrial function that contribute to virulence 42 .
The silenced strain also showed reduced survival following interaction with macrophages, as indicated by the lower number of CFUs. During infection, thermally dimorphic fungi such as Histoplasma capsulatum, Blastomyces dermatitidis, and P. brasiliensis are frequently subjected to stresses, including ROS and RNS which are produced by host cells [43][44][45] . Phagocytosed pathogens find a hostile envinroment in the macrophage, through exposure to ROS and RNS. In order to protect themselves from the harmful effects of ROS, fungal cells have developed strategies to reduce ROS levels 46,47 . Using a proteomics approach, Chaves et al. 48 demonstrated that TufM was upregulated during the infection of alveolar macrophages by P. brasiliensis. PbTufM may therefore play a role during the phagocytosis of P. brasiliensis by macrophages in response to the stressful of the internal environment of the macrophages.
In the G. mellonella in vivo model, the PbTufM aRNA strain had significantly reduced virulence compared to the wild-type and empty vector controls. The relationship between greater adhesion to pneumocytes and virulence in G. mellonella and murine models of infection, and high levels of adhesin expression in Paracoccidioides spp. 49 have been reported previously.
PbTufM aRNA was able to infect the lungs of mice, however PbTUFM downregulation significantly reduced granuloma formation and the presence of fungal cells, whilst giant cells also lowered the fungal burden in the lungs after intratracheal infection.
Mansano et al. 50 showed that the number of P. brasiliensis in the lung lesions decreased during the later periods of infection in a manner directly proportional to the lowest number of granulomas. This suggests that granuloma formation at the beginning of the infection could be an initial attempt by the host to prevent or decrease the multiplication and spread of the fungi. Host factors as susceptibility and resistance to infection modulate the granuloma formation in experimental models 51 . However, disease establishment and progression rely on virulence, antigenic profile, environmental conditions, and host immune responses 52 . Macrophages constitute part of the granuloma structure 53 in the typical inflammatory tissue response to Paracoccidioides spp. 54 , which promotes the recognition and processing of specific antigens and plays an essential role in antigen-specific T-cell activation. Isolated macrophages form giant cells with plasmocytes and neutrophils 55,56 . Giant cell formation modulates the secretory and synthetic functions of macrophages, and may occur due to the presence of poorly digestible or indigestible foreign material, or persistent pathogens that are not eliminated for various reasons 57 . We demonstrated that PbTUFM downregulation resulted in more efficient fungal clearance from mouse lungs post-infection, whilst a lower number of granulomas and giant cells indicated reduced inflammation levels, possibly due to the reduced fitness of the strain in the hostile host environment.
Pleiotropy refers to the occurrence of several phenotypes occasioned by a single gene mutation which may occur in yeasts as well as in humans 58,59 . The downexpression of PbTUFM assembling into this strain multiple phenotypes, suggesting that PbTUFM has pleiotropic effects on many biological characteristics of P. brasiliensis.
In this study to PbTufM aRNA, various phenotypes may be related to the impairment of translation rate, respiratory metabolism and ability to interact with pneumocytes; greater susceptibility to oxidative stress and  macrophage killing activity, indicating a link between the PbTUFM expression and virulence with consequences to resistance in the mice. However, we did not know if these multiples observed phenotypes are related specifically to a single or multiple roles of PbTufM, which can also be influenced by the assessed environmental condition. Further studies as transcriptome analysis can help to investigate the pleiotropic effect of PbTufM.
Understanding the virulence factors exploited by human fungal pathogens during infection is of great importance, as their targeted inhibition may in turn reduce fungal pathogenicity. Finally, we suggest that the TufM protein is required for full P. brasiliensis virulence, and that future studies should explore the possibility of its use as a target for controlling PCM.

Methods
Microorganisms and culture conditions. P. brasiliensis strain Pb18 (PbWT, wild type) was cultured in liquid BHI medium with 1% glucose under agitation (150 rpm) at 37 °C, and yeast cells were harvested during the exponential phase (72 h). The Pb18 strain was used for all genetic manipulations 60 . All experiments were performed in the class III biosafety cabinet.
The Agrobacterium tumefaciens strain LBA1100 61 , which contains the binary vector pUR5750 62 , was used as a recipient for binary vectors containing cassettes, and was maintained at 28 °C in Luria-Bertani (LB) medium containing kanamycin (100 μg mL −1 ) for selection purposes 63 .
The E. coli DH5α strain was used to cloning and propagation of vectors; the bacteria grown at 37 °C on LB medium with kanamycin (50 μg mL −1 ) 64 .
constructs for pbtUfM gene silencing. To obtain an isolate with reduced PbTUFM gene expression, we used aRNA and ATMT according to the methods described by Almeida et al. 65 . DNA from Pb18 was isolated and the fragment referring to aRNA oligonucleotide targeting the exon coding sequence of PbTUFM (PADG_01949.1; Broad Institute) amplified using a high-fidelity proofreading DNA polymerase 63 . The oligonucleotides used are described in Table 3.
The PbTUFM antisense fragment was inserted into a pCR35 plasmid under the control of the calcium-binding protein gene (CBP-1) promoter region from Histoplasma capsulatum 66 , and was propagated in E. coli DH5α. Each CBP-1 promoter-AS cassette was subcloned into a pUR5750 plasmid, the parental binary vector used to harbor the transferred DNA (T-DNA) and an antisense molecule 67 . A recombinant control was generated by transforming Pb18 with an empty vector (PbEV) 65,68 . After transformation, the cells were spread on BHI medium supplemented with hygromycin (100 μg mL −1 ) to select for transformants. Colony formation was monitored at 36 °C for 15-20 days. The phenotypic stability of the transformants was assessed by consecutive (at least three times) subculturing in BHI medium nonselective and then again in medium non-selective and selective medium with hygromycin B (150 μg mL −1 ) for three consecutive times at 36 °C 63,69 .

Molecular detection of the hygromycin resistance gene (hph). The isolation of genomic DNA from
PbWT, PbEV, and PbTufM aRNA was performed as described by Van Burik et al. 70 . To confirm the transference of the hph cassette, PCR amplification was performed to detect the hph fragment (~1000 bp) using the primers shown in Table 3. PCR cycling was carried out as described by Marcos et al. 71 , and the PCR products analyzed by electrophoresis in 1% agarose gels and visualized using GelRed under UV light.
Gene expression analysis by real-time pcR. Both, Trizol ® Reagent (Invitrogen, Waltham, Massachusetts, USA) and mechanical disruption by maceration with glass beads were employed to extract the total from PbWT, PbEV, and PbTufM aRNA cells 60,72 . The first cDNA strand was reverse transcribed using the RevertAid TM H Minus Reverse Transcriptase (Fermentas, Waltham, Massachusetts, USA) with 1 μg of total RNA per reaction. Real-time PCR reactions were carried out in triplicate using specific PbTUFM primers (Table 3)  The dissociation curves were analyzed for all reactions to verify single peaks/products 73 . Relative quantification values were calculated using the 2 −ΔΔCt method with the housekeeping β-tubulin gene (primers in Table 3) as an endogenous control and the target gene expression levels in PbEV as calibrators 74  Growth ability using different carbon sources. The growth rates of PbWT, PbEV, and PbTufM aRNA cells in liquid BHI (1% glucose, amino acids, and polypeptides as carbon sources), MMcM (1% glucose as a   75 , and MMcM (1% glycerol as a carbon source) media (100 mL) were incubated at 37 °C and measured at different times until the stationary phase of growth was reached. Yeast cells were counted every 24 h using a hemocytometer with optical microscopy. All cultures started at an initial concentration of 10 6 cells mL −1 , as described by Torres et al. 72 .
The length of time required for each strain to double the number of cells (doubling time) during log phase was calculated using the formula of Rieg et al. 76 : ln 2 × t/(ln b − ln a) (t = time period in hours; a = initial cell concentration; b = final cell concentration).
Yeast cells were generally diluted to concentrations of 10 5 , 10 6 , and 10 7 cells in PBS, plated on the culture media described above, and incubated for 7 days at 37 °C. puromycin-labbeling western blotting. Firstly, 4-fold serial dilutions of PbWT, PbEV, and PbTufM aRNA were prepared, and 5 µL of each dilution (10 5 to 10 2 ) was plated onto BHI plates with different concentrations of puromycin (0, 40, 200, or 1000 µM; Sigma-Aldrich, St. Louis, Missouri, USA). The plates were incubated at 37 °C for 3 days to determine sensitivity to puromycin. Once puromycin sensitivity had been verified, all the strains were grown in a liquid medium until the exponential phase of growth was reached (3 days at 37 °C and 150 rpm). To evaluate the translation rate, the yeast cells were treated with 200 µM of puromycin for 12 h under the conditions described above. Protein extracts were then obtained by alternating freezing in liquid nitrogen and macerating with glass beads according to the methods described by Marcos et al. 16 . Protein concentrations were estimated using the Bradford method. Equivalent amounts of protein (50 µg) isolated from all strains after the puromycin treatment were used for western blotting with the anti-puromycin antibody, and Coomassie staining and Ponceau red loading dye as controls. Nascent polypeptides containing puromycin were detected using a secondary antibody conjugated to alkaline phosphatase according to the manufacturer's instructions. Translation rate was estimated based on the intensity of immunoreactive bands using Image-J (Bethesda, Maryland, USA).

Mitochondrial location of tufM in P. brasiliensis.
To determine whether TufM was also localized in the mitochondria of yeast cells grown in Fava-Netto medium 77 , immunochemistry was carried out at the ultra-structural level by immunogold labeling, as described by Marcos et al. 16 , using a polyclonal anti-PbTufM antibody previously obtained by our group. tufM quantification by flow-cytometry and western blot analysis of P. brasiliensis cell-wall associated proteins. To determine whether silencing affected the cytosolic and cell surface levels ofTufM, we performed the assay described by Silva et al. 78 with modifications. PbWT, PbEV, and PbTufM aRNA cells were grown in BHI medium as already described. Two aliquots of each strain (10 7 cells) were collected. The first aliquot (non-permeabilized cells) was fixed in 3.5% paraformaldehyde for 20 min at room temperature (25 °C), washed twice with PBS, and incubated with blocking solution (BSA-bovine serum albumin 2%) for 1 h. After washing, the aliquot was incubated with polyclonal anti-PbTufM (diluted 1:10 in blocking solution) at 4 °C overnight. The second aliquot (permeabilized cells) was fixed as described above, washed, and permeabilized with Triton-X 100 (diluted 1:100 in PBS) for 45 min at room temperature. After washing, the aliquot was incubated with polyclonal anti-PbTufM as described above. After multiple washes in PBS, the cells of both aliquots were incubated with anti-rabbit IgG labeled with phycoerythrin (PE) (diluted 1:10 in PBS) (Sigma-Aldrich) for 2 h at room temperature. After two washes, the fluorescence intensity was estimated by cytometry (FACSCanto TM , Becton Dickinson, East Rutherford, New Jersey, USA), counting 10 4 cells. Both aliquots were also incubated with pre-immune rabbit serum and with the secondary antibody only as controls and were observed using an IN Cell Analyzer 2000 (GE Healthcare, Chicago, Illinois, USA).
The cell surface associated proteins were extracted using dithiothreitol (DTT), a reducing agent 79 . According to Longo et al. 80 , the yeast cells were harvested by centrifugation (6,000 rpm, 15 min), washed with ice-cold buffer containing 25 mM Trish-HCl, pH 8.5 (the same buffer used in the subsequent steps), vigorously vortexing for 1 min between each wash, and then treated with 2 mM DTT-buffer solution containing 5 mM ethylenediamine tetraacetic acid (EDTA) and 1 mM phenyl methyl sulphonyl (PMSF). The samples were incubated at 4 °C with gentle agitation. By centrifugation the cells were pelleted and the supernatant containing the DTT-extractable proteins were filtered using a 0.22-µ filters followed by dialysis in Amicon (Millipore) with 0.1 M ammonium acetate. After the sample lyophilization, the surface proteins extracts were suspended in deionized water and precipitated with acetone at −70 °C; the precipitated proteins separated by centrifugation (10,000 rpm a 4 °C) were washed with acetone and dried in the air. Protein concentration was estimated using the Bradford method. Equivalent amounts of PbEV and PbTufM aRNA proteins were separated by gel electrophoresis using SDS-PAGE 81 and were transferred onto nitrocellulose membranes (Hybond-C extra) 64 . The membranes were blocked with 5% skim milk and 1% bovine serum (BSA) for 4 h at room temperature and washed three times with PBST (PBS with 0.05% Tween-20). The membranes were then incubated with primary polyclonal antibodies, anti-GAPDH (diluted 1:1000; Invitrogen) and anti-PbTufM 16 (diluted 1:250; in blocking buffer overnight at room temperature and were washed as described above. Blots were developed using a horseradish peroxidase-labelled second antibody against rabbit IgG (diluted 1:200 in blocking buffer; Sigma ® ) with 3,3′ diaminobenzidine as the substrate. Relative levels of PbTufM were determined by band quantification using ImageJ software (NIH), and were normalized to GAPDH, a protein abundantly founded in the P. brasiliensis cell-wall 82 , which was used as an internal control. www.nature.com/scientificreports www.nature.com/scientificreports/ medium. Triplicate wells were arranged on each plate and four plates were set up to perform a time course analysis of infection (at 2, 5 and 8 h) under the same conditions as described above. After infection, unbound yeast cells were removed by washing three times with PBS. The bound cells were detached from the plastic with cold PBS and a rubber cell scraper, put on ice, fixed with 4% paraformaldehyde, and analyzed by flow-cytometry using a FACSCantoTM (Becton Dickinson). Uninfected host cells and unlabeled yeast cells were used as internal controls to define the gates and to evaluate autofluorescence. The excitation wavelength was 488 nm, and emitted light was collected using the 530/30 nm band-pass filter. Data were processed and analyzed using FACs-Diva software. A549 cells were gated and the mean CFSE fluorescence (corresponding to labeled yeast cells) of this gate was used to quantify the association between P. brasiliensis and host cells 60 .
The strains were also treated with anti-PbTufM as described previousl 16 and labelled with CFSE before A549 infection to verify whether blocking cell surface PbTufM altered the association between A549 and PbTufM aRNA cells. Analysis was performed as above.
Viability assay after oxidative stress and P. brasiliensis macrophage infection. Individual inoculums of PbWT, PbEV, and PbTufM aRNA cells (100 μL, 10 6 cells mL −1 ) were treated with different H 2 O 2 concentrations (10, 50, and 100 mM) for 1 h. The cells were then pelleted by centrifugation, washed twice, resuspended with PBS and plated on solid BHI to evaluate the viability counting the colony forming units (CFU) after 7 days of incubation at 37 °C.
RAW 264.7 macrophages were used in the P. brasiliensis survival assay, cultured in Dulbecco´s Modified Eagle Medium (DMEM) containing 10% bovine fetal serum at 37 °C and 5% CO 2 until completely confluent. A total of 2.5 × 10 4 RAW macrophages was added to each well of 96-well polypropylene plates with DMEM medium and incubated for 2 h at 37 °C and 5% CO 2 . Then 2.5 × 10 4 PbWT, PbEV, and PbTufM aRNA cells were added to each well with the macrophages, resulting in a yeast:macrophage cell ratio of 1:2, since the doubling time for RAW cells is around 15 h. The cells were incubated for 2 h at 36 °C and 5% CO 2 , then washed three times with PBS to remove non-adherent yeast cells. To kill only extracellular fungi, the cells were treated with 15 µg mL −1 of ketoconazole solution in DMEM medium for 1 h at 36 °C 10,83 . The macrophages were lysed with 100 µL of MilliQ-water, the fungal cells were recovered and plated onto BHI with 1.5% glucose, 5% P. brasiliensis (strain 192)-spent culture medium, 4% bovine fetal serum and gentamycin (40 mg L −1 ) 84 ; the same was done without ketoconazole treatment to determine the total rate of infection (adhered/internalized yeasts); the CFUs were counted after 14 days of incubation at 37 °C. Galleria mellonella infection model. Sixteen G. mellonella larvae (0.1-0.2 g) were selected for inoculation. Each larva was cleaned with 70% ethanol and, using a Hamilton syringe (Hamilton, Nevada, USA), was injected through the last left proleg with 10 μL (5 × 10 6 cells) of PbWT, PbEV, or PbTufM aRNA cell suspensions prepared in PBS without cell clumps. Each experiment was repeated independently three times. Control groups were inoculated with a) PBS, to monitor the effects of physical injury on survival; b) heat-killed P. brasiliensis yeast cells, to evaluate the effects of osmotic stress due to the inoculum size on survival, and c) non-infected larvae. The larvae were then incubated in Petri dishes at 37 °C for 7 days and checked daily for mortality 85  intratracheal infection of BALB/c mice. We used male BALB/c mice (6-8 weeks-old), which were anesthetized intraperitoneally with 200 μL of a xylazine (10 mg kg −1 ) and ketamine (80 mg kg −1 ) solution. Viable PbWT, PbEV, or PbTufM aRNA cells (3 × 10 5 cells) were administered to each mouse by intratracheal inoculation with a 26-gauge needle after hyperextension of the neck and exposition of the trachea at the level of the thyroid. Incisions were sutured with 5-0 silk 86 . fungal burden in lungs of infected mice. The lungs were removed, weighed, and homogenized in 1 mL of PBS at 15 days post-infection. Samples (100 μL) were then plated on solid BHI medium supplemented as described above. Petri dishes were incubated at 37 °C and colonies counted after 20 days.
Histopathology. Formalin (10%)-fixed murine lungs were embedded in paraffin, sectioned, and stained with hematoxylin and eosin (H&E). Microscopic analysis was performed and evaluated blind by a pathologist using a ZEISS Primo Star (Zeiss, Oberkochen, Germany) microscope with 4 and 40× objectives. The number of fungi and granulomas, and the percentage of lung tissues with giant cells were recorded Statistical analysis. Statistical comparisons were made using analysis of variance (one-way ANOVA) followed by Tukey's test, and Student's t-tests. For the survival curve, statistical analyses of the results of independent experiments were performed using the Long-rank test. p-values of <0.05 were considered statistically significant.