Paracoccidioides brasiliensis presents metabolic reprogramming and secretes a serine proteinase during murine infection

ABSTRACT Paracoccidoides brasiliensis and Paracoccidioides lutzii, the etiologic agents of paracoccidioidomycosis, cause disease in healthy and immunocompromised persons in Latin America. We developed a method for harvesting P. brasiliensis yeast cells from infected murine lung to facilitate in vivo transcriptional and proteomic profiling. P. brasiliensis harvested at 6 h post-infection were analyzed using RNAseq and LC-MSE. In vivo yeast cells had 594 differentially expressed transcripts and 350 differentially expressed proteins. Integration of transcriptional and proteomic data indicated that early in infection (6 h), P. brasiliensis yeast cells underwent a shift in metabolism from glycolysis to β-oxidation, upregulated detoxifying enzymes to defend against oxidative stress, and repressed cell wall biosynthesis. Bioinformatics and functional analyses also demonstrated that a serine proteinase was upregulated and secreted in vivo. To our knowledge this is the first study depicting transcriptional and proteomic data of P. brasiliensis yeast cells upon 6 h post-infection of mouse lung.


Introduction
Paracoccidioides brasiliensis and Paracoccidioides lutzii, the etiological agents of paracoccidioidomycosis (PCM), are fungal pathogens that are acquired following the inhalation of mycelia fragments or conidia into the lungs of a human host. 1 A defining feature of Paracoccidioides pathogenesis is the morphologic transition from mycelia or conidia to yeast cells, which is governed by temperature and occurs when mycelia and conidia enter the lungs. Paracoccidioides causes a wide range of clinical manifestations, which range from isolated lung infection to deepseated, life-threating infection, such as disseminated PCM, which occurs through hematogenous spread. 2,3 Our goal is to characterize genes involved with the adaptive response of P. brasiliensis during the early stages of pulmonary infection. Although identification of these genes is crucial for the understanding of fungal pathogenesis, the limited abundance of P. brasiliensis yeast cells at the sites of infection, makes gene and protein assays challenging. To overcome the limited availability of fungal material, in vitro and ex vivo previous approaches have been used to mimic conditions of infection. Prior studies have profiled gene expression in P. lutzii yeast cells incubated in human blood at 36 C, using cDNA representational difference analysis (cDNA-RDA), as well as P. brasiliensis yeast cells internalized by macrophages, using microarray technology. 4,5 These studies demonstrated down regulation of transcripts related to cell wall metabolism and glycolysis, as well as upregulation of putative virulence factors, such as a serine protease. Approaches used to identify genes relevant for infection at the proteomic level, include the use of various in vitro experimental conditions such as iron deprivation, hypoxia, oxidative, nitrosative stress, and fungal-macrophage interactions. 6,7,8,9,10 The later approach demonstrated that during macrophage infection, yeast cells remodel their metabolism to recycle carbon containing molecules and induce gluconeogenesis. This metabolic remodeling may play an important role in the adaptive response to phagocytosis. Additionally, during macrophage infection, P. brasiliensis yeast cells induce the expression of cytochrome C peroxidase (CCP), an anti-oxidant molecule and potential virulence factor. Antisense CCP knockdown strains have reduced survival upon macrophage interaction and during infection in BALB/c mice. Additionally, mutants with low CCP expression were more sensitive to oxidative and nitrosative stress. 9,10 While in vitro and ex vivo studies have provided insight on fungal processes involved with murine pulmonary infection, in vivo analysis of P. brasiliensis yeast cell gene expression during pulmonary infection has not been conducted. To understand how P. brasiliensis adapts to the host, we established a model of intranasal infection coupled with bronchoalveolar lavage for large-scale in vivo gene and protein expression analyses during pulmonary infection. Thus, we captured, for the first time, largescale in vivo gene expression during invasive PCM. Because transcriptome analysis provides a limited understanding of the biology associated with in vivo survival, we also investigated the underlying proteomic changes during infection. For transcriptome and proteome analyses, we focused our attention at 6 hours post-inhalation because our data demonstrated that at this time point, P. brasiliensis invaded mouse lung tissue. Our profiling studies paint a picture of lung invasion in which P. brasiliensis alters the expression of genes related to several functional categories including energy metabolism and cell wall metabolism. Some active processes presumably help the fungus to survive in lung, as here identified, such as the accumulation of detoxifying enzymes. On the basis of our results, we propose that P. brasiliensis remodels cellular lipid metabolism to catabolize its own lipid stores via b oxidation, glyoxylate cycle, which were strongly induced during the first 6 hours of lung infection. Moreover, our approach identified a secreted serine proteinase that may facilitate invasion of lung tissue and dissemination of P. brasiliensis to other organs.

Establishing a method for infection and recovery of yeast cells in vivo
Transcriptomic and proteomic approaches can identify genes and biologic processes important for infection. Although in vivo profiling using mouse models have the potential to uncover gene and protein expression during infection, it has remained a challenge to obtain sufficient quantity of yeast cells for RNA and protein analysis. To overcome this barrier, we sought to establish an infection model and method for recovering sufficient quantity of yeast cells for RNA and protein extraction. For this, mice were infected with P. brasiliensis by intranasal inhalation of yeast cells and after 6 h we performed extensive lung lavage to harvest yeast cells in the bronchoalveolar fluid.
We validated our infection model by analyzing lung sections of infected mice at different post-infection time points. At 6 and 24 h after intranasal exposure, mice were killed and lung tissue collected for histological processing. The experiments were conducted in biologic (3 animals) and experimental (3 time-independent experiments) triplicates. As observed in Fig. 1A and C, lungs of control group, inoculated with saline solution (NaCl 0.9%), had a normal histomorphology including terminal bronchioles, surrounding muscle tissue, and the alveolar duct structure. The lung tissue sections were stained with Gomori-Grocott 11 to highlight the presence of fungal cells (Fig. 1B). Staining with hematoxylineosin 12 was used to evaluate for inflammation (Fig. 1D). Histopathological examination of lung sections at 6 h post-infection showed neutrophilic inflammatory infiltrate associated with circular birefringent fungi cells, ranging from 6 to 20 mm in diameter ( Fig. 1D; Fig. S1). Figure S2 depicts mouse lungs at 24 hours post infection. Histopathology at 24 h post-infection showed a large increase in neutrophilic inflammatory infiltrate associated with circular birefringent fungal cells ( Fig. S2C and 2D). Since we were interested in the initial stages of infection, the 6 h time point was selected for harvest of P. brasiliensis because there were abundant yeast cells in the lungs. Transcriptional and proteomic analysis was performed using high-throughput RNA Illumina sequencing (RNAseq) and NanoUPLC-MS E , respectively.
The procedure for obtaining bronchoalveolar lavage was validated using histopathological analyzes and evaluation of yeast cells number, as depicted in Fig S3. After counting of 50 fields, an average of 106 and 30 yeast cells was observed in lungs that were not submitted to bronchoalveolar lavage and lungs after bronchoalveolar lavage, respectively ( Fig S3). Counting of yeast cells in the bronchoalveolar material provided very similar results (data not shown).

Investigating the transcriptome of Paracoccidioides brasiliensis yeast cells at 6 h post-infection
On the basis of our pilot experiments, we selected the 6 h post-infection time point for transcriptome analysis. cDNA libraries were generated and the sequenced reads were mapped to the reference Paracoccidioides genome (Pb18), (http://www.broadinstitute.org/annotation/genome/para coccidioides_brasiliensis/MultiHome.html) using the Bowtie 2 tool (doi:10.1038/nmeth.1923) and analyzed by DEGseq package. 13 The list of differentially expressed genes (DEGs) was obtained by comparing RNA levels of yeast cells at 6 h post-infection to the in vitro control; (control fungal cells were obtained by incubation of P. brasiliensis cells in BHI liquid medium at 37 C during 48 h). To determine the up-and downregulated transcripts, a cut-off of 2.0-fold change was applied. Table S1 lists the differentially induced transcripts and Table S2 lists the differentially repressed transcripts.
Five hundred and ninety four differentially expressed transcripts were identified from yeast cells at 6 h post-infection. From this total, 73.40% of transcripts were functionally annotated, corresponding to 436 genes, and 26.60% were unclassified transcripts corresponding to 158 genes.
We used the functional catalog (FunCat) for functional enrichment analysis. We additionally included KEGG terms, when relevant. Subcategories of metabolism related to amino acid, nitrogen/sulfur, C-compound/carbohydrate, lipid/ fatty acid, purines, secondary, and phosphate metabolisms were regulated in host-pathogen interaction (Tables S1 and S2). Categories associated with cell cycle/DNA processing, transcription, protein synthesis and fate and cellular transport were also largely represented in the transcriptome.
The functional classifications and the percentage of up-and downregulated transcripts in each classified category are shown in Fig S4. Functional enrichment analysis demonstrated that at 6 h post-infection of mouse lung, most of the DEGs in P. brasiliensis corresponded to downregulated categories (Table S1 and 2, Fig S4). A total of 430 DEGs (72.6%) were repressed and 154 DEGs (27.4%) were induced at 6 h post-infection. The majority of enriched categories for upregulated DEGs were related to metabolism and cellular transport. The majority of enriched categories for downregulated DEGs were metabolism, transcription and protein synthesis.
Changed expression connected to upregulated transcripts Figure S4 depicts enriched categories for upregulated DEGs. Metabolism and cellular transport constitute the functional categories most highly enriched during fungal infection. In the metabolism category, the oxidation of fatty acids was the most enriched pathway, which includes acyl-CoA dehydrogenases (PADG_07604, PADG_05046, PADG_06335), short chain dehydrogenases/reductases (PADG_06247, PADG_06006), carnitine transport/ metabolism (PADG_05773, PADG_06378), peroxisomal hydratase-dehydrogenase-epimerase (PADG_08651) and a 3-hydroxybutyryl-CoA dehydratase (PADG_05039). These genes demonstrated increase transcript abundance that ranged from 2.29 to 17.00 (Table S1). In addition, ethanol producing genes were induced including alcohol dehydrogenase (PADG_04701) and aldehyde dehydrogenase (PADG_05081) with 4.38 and 5.13-fold change, respectively. As depicted in Table S1, genes involved with cellular transport including amino acid permease (PADG_11786), MFS drug transporters ( PADG_08041, PADG_07610, PADG_04733), multidrug transporter (PADG_12014), ), ABC drug transporter (PADG_06245) and zinc-regulated transporter (PADG_06417), were among the transcripts induced in vivo. Table 1 depicts the most induced transcripts ( 4fold) in yeast cells infecting mouse lungs. The most induced transcripts, depicted in Table 1, include those related to degradation of fatty acids and carbohydrates, nitrogen and ethanol metabolism, proteins related to the defense to stress, among others.
Changed expression connected to downregulated transcripts As described above, the majority of DEGs were downexpressed in vivo compared with the in vitro control. A number of downregulated genes were involved in electron transport and membrane-associated energy conservation (Table S2). This category contains 10 downregulated genes, comprising several subunits of NADH-ubiquinone oxidoreductase (PADG_04699, PADG_08216), ATP synthase complex, and cytochrome complex (Table S2).
The metabolic processes involving the cell wall were downregulated in yeast cells at 6 h post-infection (Table S2). A total of 25 trancripts related to cell wall metabolism were identified that were downregulated.
Three transcripts encoding proteins related to glycan metabolism such as a 1,3 glucosidase (PADG_07523); endo-1,3(4)-b-glucanase (PADG_12370) and glucan synthesis regulatory protein (PADG_05870) were downregulated with expression ratios from 0.37 to 0.1 relative to the in vitro control (Table S2). The activation of the b-1,3 glucan synthase is regulated by Ras/ Rho GTPases cell signaling pathway, that includes protein kinase C and MADS family protein. 14 As illustrated in Fig. 2, this pathway was downregulated at 6 h of P. brasiliensis post-infection infection in mouse lung (PADG_07326: serine/threonine-protein kinase PRKX and PADG_07962: MADS box transcription factor Mcm1), which could be related with lower glycan content in yeast cells upon infection. Also, the proteins pKC and calmodulin are described as activators of chitin synthesis 15 and, interestingly, both signaling pathways can be repressed, putatively leading to repression of chitin synthesis. Transcripts related to chitin synthesis (PADG_06438, PADG_08636, PADG_08120 and PADG_04485) were downregulated in vivo with expression ratios 0.1 -0.45 relative to the in vitro control (Table S2). In addition, chitin activators (PADG_04697 and PADG_05937), responsible for localization of chitin synthases to the cell wall, were negatively regulated, which has been described in other fungal models. 16 In addition, transcripts related to GPI Genes annotation from Paracoccidioides genome database or by homology from NCBI database (http://www.ncbi.nlm.nih.gov/) c Infection/Control means: The level of expression in yeast cells derived from infected lung divided by the level in the control yeast cells. d Biological process of differentially expressed proteins from MIPS (http://mips.helmholtz-muenchen.de/funcatDB/) and Uniprot databases (http://www.uniprot.org/). synthesis (PADG_04786: GPI ethanolamine phosphate transferase) and attachment of GPI anchors into proteins (PADG_08703: GPI transamidase component GPI16) were also downregulated in vivo (Table S2).
The metabolism category includes primary metabolism of amino acids, from which the metabolism of tyrosine and phenylalanine was especially downregulated. The category of mitotic cell cycle and cell cycle control was enriched with 21 downregulated genes including predicted phosphatases and cell division control proteins (Table S2). Regarding DNA processing, genes in this category were downregulated in vivo. In this category there were 11 proteins important for DNA duplication such as ATP-dependent DNA helicase II , replication factor-A protein (PADG_07835), DNA replication licensing factors (PADG_08558, PADG_07901), DNA topoisomerases PADG_07812, PADG_04954, PADG_06339), DNA-directed RNA polymerases (PADG_06985, PADG_08395, PADG_07655), FACT complex subunit pob3 (PADG_07672), nucleosome assembly protein 1like 1 (PADG_05893) , and 5 transcripts encoding proteins/enzymes of DNA repair (Table S2).
The categories of protein synthesis and degradation comprised a large number of genes with reduced expression (Table S2). In the category of protein synthesis, 60 downregulated genes were present including those related to ribosomal biogenesis and translational machinery.
Transcriptional control depicts several downregulated genes (Table S2). Surprisingly, the gene encoding the HapX transcription factor (PADG_07492), involved in the homeostasis of iron is among the most repressed in yeast cells at 6 h post-infection in mouse lung. Other repressed transcription factor in vivo includes APSES MbpA (PADG_07224) involved in the regulation of morphogenesis including filamentation and differentiation in Candida albicans and Aspergillus fumigatus. 17 Signal transduction category had 25 downregulated genes (Table S2) involving several pathways such as calcium-calmodulin, histidine kinase, MAP kinase. Different protein kinases are among the most repressed genes in P. brasiliensis yeast cells at 6 h post-infection in mouse lung (Table S2).

Quantitative proteome profiles of yeast cells infecting mouse lungs
Proteomic data quality analysis Protein quantification was performed using Nano-UPLCMS E as described previously. 18,19,20 The resulting NanoUPLC-MS E protein and peptide data generated by the PLGS process are shown in Figs. S5, S6 and S7. The false positive rates of proteins obtained from yeast cells at 6 h post-infection of mouse lung and from in vitro control were 0.99% and 0.67%, respectively. These . Also, protein kinase C (pKC, PADG_07326) and calmodulin (PADG_07774) dependent signaling pathways can modulate chitin synthesis (indicated by arrows 1 and 3, respectively). Both are negativelly regulated in this condition. Transcripts encoding proteins related to chitin biosynthesis (PADG_06438, PADG_08636, PADG_07802, PADG_08120 and PADG_04485) were repressed; (PADG_12130) was induced. The downregulation of the cell wall synthesis is also evidenced by the negative regulation of transcripts encoding proteins related to the cellular destination of chitin synthase (PADG_04697, PADG_05937)and GPI anchoring (PADG_08703). Transcripts encoding proteins related to GPI synthesis wasrepressed (PADG_04786). Thoese transcripts are supplied in Tables S1 and 2. The MultiExperiment Viewer software V.4.8 (www. tm4.org/mev/) was used to group the compare data of expression ratios and color scale was adapted. experiments resulted in 9,787 and 17,537 identified peptides, where 52 and 58% were obtained from peptide match type data in the first pass and 9% (for both control and test) from the second pass 20 for in vivo and in vitro yeast cells, respectively ( Fig S5). A total of 13% and 21% of total peptides were identified by a missed trypsin cleavage in control and infection conditions, respectively, whereas an in-source fragmentation rate of the same 7% from control and 11% in yeast cells at 6 h post-infection of mouse lung was obtained (Fig. S5). Figure S6 shows the peptide parts per million error (ppm) indicating that the majority, 63.19% and 72.56%, of identified peptides were detected with an error of less than 5 ppm for yeast cells isolated from mouse lung and in vitro control, respectively. For error of less than 10 ppm, 87.75% and 92.22% peptides were identified for yeast cells infecting mouse lung and control, respectively. Fig S7 depicts the results obtained from dynamic range detection indicating that a 3-log range concentration and a good detection distribution of high and low molecular weights were obtained for the both conditions.

The pattern of proteins differentially accumulated upon P. brasiliensis infection of mouse lung
The 3 parallel independent infection assays were performed and equimolar amounts of protein extracts were mixed and used in proteomic analysis. Pooled protein extracts were run in 3 technical replicates and only proteins found in 2 out of the 3 technical replicates, by an identical set of at least 2 unique tryptic peptides, were considered for further analysis. Those criteria have been extensively used in label-free MSE proteomic assays. 21,22,23 Five hundred and 8 proteins were identified, after filtering, using the criterion of minimum repeat rate of 2. A 1.5-fold change was used as a threshold to determine differentially expressed proteins. 147 and 203 proteins down and upregulated, respectively, in vivo, versus in vitro control (Tables S3  and S4) were obtained after filtering using the parameters above. Biological processes and the percentage of down and up-regulated proteins in each classified category are depicted inFig.S8.Theidentifiedproteinswereclassifiedonthebasisof their function according to the Munich Information Center forProteinSequences(MIPS)categorization.

Proteins downregulated in yeast cells during mouse infection
Proteomic analysis of in vivo yeast cells demonstrated that proteins involved with amino acid biosynthesis and protein synthesis were decreased. Enzymes in biosynthetic pathways for threonine, tryptophan, glutamate, and glycine were reduced including threonine synthase (PADG_02777), anthranilate phosphoribosyltransferase (PADG_01789), glutamate decarboxylase (PADG_06319) andserine hydroxymethyl transferase (PADG_05277), respectively. In addition, 30 proteins involved with protein synthesis were decreased(TableS3).

Proteins upregulated in yeast cells during mouse infection
P. brasiliensis proteins upregulated in vivo were involved with multiple pathways including fatty acid catabolism (beta and omega oxidation), glyoxylate cycle, pentosephosphate pathway, and ethanol production (Table S4). Proteins involved with fatty acid breakdown, acyl-CoAbinding protein (PADG_01363), enoyl CoA hydratase (PADG_01209), and isocitrate lyase (PADG_01483) in the glyoxylate cycle was induced 1.53-fold in vivo. Enzymes involved with ethanol production are upregulated; alcohol dehydrogenase (PADG_04701) and pyruvate decarboxylase (PADG_00714) increased 13.73 and 1.58 fold compared with in vitro control, respectively.
In the functional class of protein synthesis, 39 molcules were increased (Table S4). They belong to the subcategories of ribosome biogenesis, translation, translational control and translational initiation. Thirty-two were induced in the functional class of protein degradation (Table S4), including components of the proteasome complex, ubiquitin system and proteases. The up regulation of enzymes involved with protein degradation, may supply anaplerotic precursors such as pyruvate, which can be metabolized by alcohol dehydrogenase (PADG_04701) and pyruvate decarboxylase (PADG_00714), as cited above.

Overview of metabolic changes in P. brasiliensis at 6 h post-infection in mouse lung
We have integrated transcriptional and proteomic data to summarize the fungal response after 6 h in mouse lung. Our data suggest that yeast cells reprogram pathways for energy metabolism (Fig. 4) including using lipids as an energy source during infection. Most of the enzymes involved in oxidation of even and odd chain fatty acids were upregulated. Similarly, enzyme in pentose phosphate pathway was increased, which can supply the redox equivalents for the oxidative stress response.  Glyoxylase was also induced, which detoxifies methylglyoxal generated from lipid peroxidation. The glyoxylate cycle was upregulated, which further supports lipid catabolism during infection. Interestingly, transcripts and proteins related to aerobic metabolism, which supports fatty acid derived energy production were downregulated. This metabolic contradiction may be explained by the fact that yeast cells were grown in a rich medium with abundant nutrients and then transferred to a hostile in vivo environment. Thus, the high metabolic activity that is required to support the exponential growth phase in vitro is no longer required during early phases of infection, when the yeast cells are adapting to the in vivo environment. Rather, during infection P. brasiliensis utilizes acetyl-and propionyl-CoA, which are derived from fatty acid catabolism, to fuel glyoxylate cycle for anaplerosis of oxaloacetate, which provide building blocks for biomolecule synthesis.
Immunohistochemical analysis of infected mice lungs with polyclonal antibodies to serine protease (PADG_07422) During pulmonary infection, P. brasiliensis yeast cells upregulated 32 proteins involved with protein fate and degradation (Table S4). In this class of proteins we prioritized a serine proteinase (PADG_07422), which was 2.4fold upregulated, for in vivo analysis by immunostaining.
Previous studies in our laboratory demonstrated that PADG_07422 is a secreted serine proteinase and is induced during nitrogen deprivation. 24 After 6 h of intranasal infection, polymorphonuclear inflammatory infiltrate surrounding P. brasiliensis yeast cells stained positive for the serine proteinase (Fig. 5B). In addition, the serine protease was also detected on the surface and in the cytoplasm of murine cells (Fig. 5B). The detection of enolase, 25 which was used as a control, only stained yeast cells (Fig. 5A). Thus, the labeling of the host tissue supports that the serine protease is secreted by P. brasiliensis yeast in vivo.
Additional controls were performed using different inducers of immune response to demonstrate lack of cross-reactivity of the anti-serine proteinase polyclonal antibodies to mouse proteins. Considering the inflammatory infiltrate caused by the inoculation of zymosan 26 , carrageenan 27 or heat killed fungi, serine proteinase was not detected by immunohistochemistry assays, as shown in Fig. 6, panel A to C, respectively.

Treatment of mouse with anti-serine proteinase polyclonal antibodies induces decreased fungal burden
We performed an additional experiment to evaluate the potential role of serine proteinase (PADG_07422) in mouse infection by P. brasiliensis. After 6 hours of infection, the lungs of animals pretreated with anti-serine proteinase polyclonal antibodies and the control group were assessed using histopathological and quantitative analyses. The amount of fungus in the pre-treated animals was significantly reduced (Fig. 7 A), corresponding to approximately 23% of control, as depicted in Fig. 7 B. We also analyzed the CFU after bronchooalveolar lavage in infected, pre-treated mice (vs. control) with the polyclonal antibodies (Fig. 7 C). A 90.5% reduction in CFU was observed in the pre-treated animals.

Increased production of ethanol and thiols by P. brasiliensis yeast in vivo
At 6 h post-infection, 2 genes involved with ethanol production, pyruvate decarboxylase (PADG_00714) and alcohol dehydrogenase (PADG_04701) (Tables S1 and S4) were upregulated. Ethanol concentration was approximately 13-fold higher in yeast cells at 6 h of infection in mouse lung compared with controls (Fig. 8A).
Enzymes in the thioredoxin system are used to defend against oxidative stress. In vivo proteomics analyses demonstrated that two thioredoxins (PADG_03161 and PADG_05504) and a thioredoxin reductase (PADG_01551) were induced (Fig. 3, Table 4). To functionally assess fungal thioredoxin activity, we quantified  thiol levels in yeast cells of P. brasiliensis at 6 h post-infection of mouse lung. Thiol levels were approximately 2.5-fold higher in yeast cells during infection (vs. in vitro control) (Fig. 8B).

Discussion
In this study, we used transcriptional, proteomic and bioinformatics analyses to identify and characterize gene expression in P. brasiliensis during pulmonary infection. The proteomic and transcriptional experiments were designed to capture the fungal response early in the infection process. Studies focusing on host-microbe interactions at molecular level have traditionally been challenging due to the limited number of reliable methods for harvesting fungal cells from infected organs. Herein, we describe a straightforward method for recovering P. brasiliensis yeast cells during the early stages of pulmonary infection for in vivo transcriptional and proteomic analyses. Additionally, we integrated transcriptional and proteomic data to assess how P. brasiliensis adapts to the host during the initial stages of infection in mouse lung. The changes in proteins profiles showed low congruency with the transcripts levels. We attribute this fact to post transcriptional processes playing a critical role in the protein levels, as described in Aspergillus flavus. 28 Although we have used a 2-fold threshold in the transcriptional analysis, the presence of excess of host RNAs in the samples, could impact the results.
Adaptive responses within 6 h of infection include inhibition of cell wall metabolism, and increase in pathways linked to protective responses against reactive oxygen species, as well as remodeling metabolism toward lipid degradation, ethanol production and glyoxylate cycle. In addition, a serine proteinase, potentially linked to pathogenesis was characterized.
Analysis of transcriptional data identified 1761 transcripts with 34% (594 genes) that were differentially expressed (164 upregulated, 430 downregulated) at 6 h post-infection. Analysis of the downregulated transcripts in vivo demonstrated a strong effect on cell wall metabolism. A striking pattern was the repression of several genes involved with the biosynthesis of glycan, chitin and genes that activate chitin synthases. Also, glycosyl hydrolases and chitin metabolism responsive genes were repressed. The cell wall is dynamic and responds to changes in the external environment. Multiple cellular functions, including cell wall remodeling, negatively regulated by the Zcf21 transcription factor, contribute to the ability of C. albicans to promote systemic infection. 29,30 Zcf21 mutants have higher content of cell wall polysaccharides, increased susceptibility to phagocytosis, and impaired intracellular survival in macrophages. 30 If inhibition of cell wall genes in P. brasiliensis influences the host recognition remains to be elucidated.
We identified 350 proteins differentially expressed in vivo (203 upregulated, 147 downregulated). Among the upregulated proteins, an arsenal of ROS defense enzymes were found. Host phagocytic cells use reactive oxygen species as part of their armory to neutralize foreign microorganisms. 31 The phagocytic cells of the innate immune system produce several oxidants that are presumed to play roles in microbial killing. Proteomic studies, during the treatment of Paracoccidioides sp. with hydrogen peroxide, depicted the upregulation of ROS detoxification enzymes. These enzymes include catalases, superoxide dismutases, cytochrome c peroxidase and thioredoxin. 8 Analysis of detoxification mechanisms of P. brasiliensis at 6 h post-infection of mouse lung, strongly support the concept that the fungus is exposed to a significant oxidative stress, since cytochrome c peroxidase, superoxide dismutases, thioredoxins and thioredoxin reductase are induced. Histopathology analysis demonstrated that polymorphonuclear cells were in direct contact with P. brasiliensis yeast cells, which might account for the fungal response to oxidative stress. In agreement with in vitro studies, proteins such as thioredoxin and Cu/Zn superoxide dismutases, accumulated in P. brasiliensis yeast cells infecting macrophages. 10 Enzymes related to detoxification, such as superoxide dismutases, SOD1 (PADG_07418) and SOD2 (PADG_01755) were also accumulated in P. brasiliensis yeast at 6 h post-infection. Knockdown mutants for SOD1, although able to infect mouse lung, were unable to disseminate to other organs, suggesting its involvement in intracellular fungus activity. 32 Cytochrome c peroxidase (PADG_03163) and thioredoxins (PADG_03161, PADG_05504 and PADG_01551) were induced at proteomic levels. Previous studies have described a possible relationship between ROS detoxification and virulence. Cytochrome c peroxidase (CCP) is a potential virulence factor, as demonstrated in studies using the silenced mutant for this gene. CCP silencing reduces P. brasiliensis survival during macrophage infection and in yeast cells subjected to in vitro nitrosative stress. 9,10 To maintain cellular redox homeostasis, yeast cells may provide potential reducing for most antioxidant and regulatory enzymes. 33,34 In this way, activation of the pentose phosphate pathway, which is a source of cellular reducing power in the form of NADPH, reinforces the concept that P. brasiliensis yeast cells respond to oxidative stress at 6 h post-infection. Neutralization of free radicals largely depends on NADPH.
Changes in nutrient availability early in the infection process could play a role in the repression of glycolytic genes and upregulation of the glyoxylate cycle. Paracoccidioides sp co-cultured with murine macrophages upregulated the transcripts for the glyoxylate cycle-specific enzymes isocitrate lyase and malate synthase, suggested by RT-PCR analyses. 35 In contrast, in P. brasiliensis yeast cells co-cultivated with macrophages, 10 gluconeogenesis did not appear to play a major role during lung infection, since genes encoding key enzymes as well as cognate proteins in this pathway were not upregulated as determined by proteomics analysis. This finding is in agreement with our results. The stage and site of infection can influence metabolism. Studies on C. albicans mutants blocked in various core carbon metabolic pathways demonstrated that the glyoxylate cycle and gluconeogenesis were essential for fungal survival during initial interactions with host immune cells. Later, as the yeast cells colonize, they altered carbon metabolism from non-fermentative to fermentative (i.e., glycolysis). 36 Similarly, research in C. neoformans demonstrated that deletion of pyruvate kinase (PYK1) and hexokinases (HXK1, HXK2) genes, involved with glycolysis, resulted in impaired fungal survival in the cerebral spinal fluid of rabbits, while displaying persistence in the lungs of mice. C. neoformans required gluconeogenesis when it resides in the lung, since the deletion of phosphoenolpyruvate carboxykinase (the pck1D mutant), resulted in attenuated virulence in a murine inhalation model. 37 P. brasiliensis yeast cells exposed to the lung environment seem to face a carbohydrate-poor environment, which is suggested by increased accumulation of isocytrate lyase. Activation of the glyoxylate cycle may compensate for the lack of carbohydrates in the milieu, as demonstrated for C. albicans after phagocytosis by macrophages. 38 P. brasiliensis probably uses multiple carbon sources including lipids, amino acids, and ethanol during infection. At 6 h post-infection, yeast cells upregulated genes and proteins involved with lipid catabolism (e.g., b-oxidation). Similarly, proteomic analysis of P. brasiliensis during macrophage infection suggested that yeast cells use fatty acids to survive inside phagocytes. 10 In the context of metabolic adaptation to the host, we also observed an increase in alcohol dehydrogenase and pyruvate decarboxylase. Pyruvate is most likely supplied by degradation of amino acids. Transcriptional studies from Paracoccidioides sp yeast cells recovered from mouse liver suggested that ethanol is produced during infection. 39 Similarly, we were able to confirm ethanol production by P. brasiliensis yeast cells during pulmonary infection. Transcriptome and enzymatic analyses of Aspergillus fumigatus demonstrated that this fungus produces ethanol under hypoxic conditions. 40,41 Ethanol deficient A. fumigatus strains promoted an increase in the inflammatory response in infected mice that corresponded to a reduction in fungal burden. 41 During lung infection, the production and secretion of a serine proteinase was induced in P. brasiliensis yeast cells. Immunostaining experiments, demonstrated that, this protein is located intracellularly and in the extracellular milieu. Serine proteinase was not detected in the murine host, after induction of inflammatory response by zymozan, carrageenan and heat-killed fungal cells. The results clearly demonstrated that the polyclonal antibodies do not react with hostt cell proteases. Additionally polyclonal antibodies against this protein reduced the fungal burden of infected animals. There is potential that this serine protease could function as a virulence factor. In Mycobacterium tuberculosis, a serine proteinase homolog, mycosin-1, is secreted and contributes to virulence during acute infection. 42 Additional potential virulence factors can be inferred from our data. Transcriptome data depicted the induction of the Laccase IV gene (PADG_05994), which can be related to increase in fungal virulence. Pathogenic fungi, such as Paracoccidioides spp., C. neoformans, H. capsulatum and C. albicans produce DOPA-melanin via laccase. Melanization appears to contribute to virulence by reducing the susceptibility of melanized fungi to host defense mechanisms and antifungal drugs. 43 In conclusion, we propose that P. brasiliensis, in the initial stages of murine pulmonary infection remodels its carbon metabolism, using preferentially lipids, to survive the harsh host environment. Also, the synthesis of the main components of cell wall can be repressed, putatively enabling the fungus to assemble a cell wall conformation suitable to colonize the host and escape the host immune system. Of special note, an induced serine proteinase is secreted in the murine lung tissue at 6 h post-infection, probably enabling the fungus entering and spreading the lung tissue. To our knowledge, up to know, this is the first report of transcriptional and proteomic data of P. brasiliensis upon host infection. We believe that this research provides a successful strategy for recovering P. brasiliensis yeast cells from infected lung and opens new perspectives in the study of fungus-host interaction. Additionally this research provides an arsenal of molecules that are expressed in vivo that can be relevant to the fungus-host interaction. This study opens new perspectives in the study of fungushost interaction. Future experiments focusing on kinetics of infection will be performed to establish the metabolic fungus adaptation to the host at different infection times.

Strains and culture conditions
Paracoccidioides brasiliensis, Pb18 (ATCC 32069, phylogenetic species S1), 44 was used in this study. The yeast phase was maintained in vitro at 36 C in semisolid BHI medium containing 4% (w/v) glucose. For experiments, the cells were grown during 48 h at 36 C in liquid BHI medium containing 4% (w/v) glucose.

Ethics statement
Animals were cared according to the Brazilian National Council for Animal Experimentation Control (CON-CEA) guidelines. Male BALB/c mice were maintained under controlled environmental conditions, with a temperature of 23-24 C, and a light/dark cycle of 12 h, and provided with food and water ad libitum. Male mice aged between 6 to 8 weeks were used for infection with P. brasiliensis yeast cells. Animal experimentation was approved by the Ethics Commission on the use of animals at the University Federal of Goi as (CEUA / UFGprotocol numbers 036/2015 and 086/2015). All efforts to minimize animal suffering were made.

Intranasal infection of mice with P. brasiliensis yeast cells
The animals were anesthetized by intraperitoneal injection with a solution containing ketamine hydrochloride (100 mg/kg) (Park, Davis & Company,) and xylazine (10 mg/kg) (Bayer). 45 Infection was performed by intranasal inoculation of 10 5 P. brasiliensis yeast cells in 100 mL of 0.9% (w/v) NaCl saline solution. For histopathological and immunohistochemical experiments, control group was intranasally 46 injected with 100ul of saline solution. For transcriptional and proteome experiments, control was obtained with yeast cells growing in liquid medium.

Histopathological analysis
Histopathological analyses were performed using 3 mice infected via intranasal inoculation and 3 uninfected controls, intranasally inoculate with NaCl saline solution, for 6 and 24 h. Mice were killed in CO 2 chamber, lungs were harvested, and fixed in 4% (v/v) paraformaldehyde during 24 h. Histological sections of lung were stained with: (a) hematoxylin and eosin (H&E) 12 and (b) silvermethenamine, in accordance with Gomori-Grocott, 11 for visualization of yeast cells in the lung tissue. Tissues sections were examined and photographed using Axio Scope A1 microscope. The images were acquired by using software AxioVision (Carl Zeiss AG, Germany).

Standardization of bronchoalveolar lavage fluid
To confirm that the bronchoalveolar lavage protocol was efficient to remove fungal cells from lung tissue, a histopathological study of the lung was performed after the procedure. This assay was performed with biologic triplicates. Tissues sections were examined and photographed using Axio Scope A1 microscope. The images were acquired by using software AxioVision (Carl Zeiss AG, Germany). The number of yeast cells were counted in 50 different regions of the lung section and compared with lung sections of mouse that were not washed for P. brasiliensis removal.

Obtaining yeast cells from mouse lung
Infected mice were killed in CO 2 chamber and the procedure of bronchoalveolar lavage was performed by an incision at trachea and insertion of catheter. A total of 3 washes were performed with 1 mL of 0.9% (w/v) NaCl solution. Cells were immediately used for RNA or protein extractions. Control cells were obtained by incubation of P. brasiliensis cells in BHI liquid medium at 37 C during 48 h.
For RNA-seq analysis, 9 animals were infected in each experiment, in a total of 3, to obtain each replicate of RNA. An average of 1.5 mg of total RNA was obtained for each triplicate. For proteomic analysis, 12 animals were infected in 3 different assays and equimolar amounts of protein extracts were mixed and used in proteomic analysis. In synthesis, for obtaining the replicates for RNAseq, a total of 27 animals were infected. Considering proteomic analyzes, 36 animals were infected in 3 independent assays.

RNA-seq analysis
Following recovery of P. brasiliensis from lungs, as described above, cell were washed by centrifugation at 4000 g, for 10 min and immediately added of TRIzol reagent (Invitrogen, Carlsbad) to obtain RNA molecules, from 3 independent series of experiments, each one comprising 9 animals. Control cells obtained as described before, were also used to perform RNA extraction. RNA integrity and concentration were assessed by electrophoresis on a 1% agarose gel and by Qubit analyses. The mRNAs from each experimental replicate (9 animals per replicate) and from controls (3 independent culture growth) were purified from total RNA samples, processed and sequenced using Illumina HiSeq 2500 Platform (Tufts Medical School, Boston ).

Transcriptomic bioinformatics analysis
Approximately 40 million of reads of 100 bp paired-end sequences were obtained for each sample. The resulting FastQ files were analyzed for quality using FastQC (http://www.bioinformatics.babraham.ac.uk./projects/ fastqc/). Reads were aligned to the mouse genome reference (GRCm38) to remove any possible host contamination, using Bowtie2 2.2.4 (https://doi.org/10.1038/ nmeth.1923). After filtered, the reads were aligned to the reference Paracoccidioides genome, Pb18, obtained in NCBI under ABKI00000000 identifier. The alignment was done by Bowtie2 program, and each gene was quantified by counting of reads by feature using the function multicov from bedtools package. 47 Candidates for differentially expressed genes were identified by statistical package DEGSeq, from the Bioconductor repository, using Fisher's test implemented in the package. 13 Gene features presenting 2.0-fold change cut-off, and p-value < 0.001 were considered regulated and classified.
The biologic processes were obtained using the Pedant on MIPS (http://pedant.helmholtz-muenchen.de) which provides a tool to browse and search the Functional Categories (FunCat) of proteins. The heat maps were generated using Multiexperimental Viewer 4.8 (www.tm4.org/ mev).

Preparation of protein extracts
P. brasiliensis yeast cells from infected mouse lung or control cells were collected by centrifugation at 10,000 x g during 5 min and washed by addition of 100 mL of RapiGEST (0.2%v/v) (Waters Corp), followed by 3 washes using ultrapure water, and by the addition of extraction buffer (20 mM Tris-HCl pH 8.8; 2 mM CaCl 2 ). This suspension was distributed in tubes containing glass beads in equal volume of the cell pellet and the suspension was processed on ice in bead beater equipment (BioSpec) during 5 cycles of 30 sec. The cell lysate was centrifuged at 10,000 x g during 15 min at 4 C and the supernatant was quantified for protein content, using the Bradford reagent (Sigma Aldrich, Co.), and bovine serum albumin (BSA) was used as a standard. 48

Digestion of protein extracts for nano-ESI-UPLC-MS E acquisition
Enzymatic digestion of proteins was processed, as described previously 10,19,20,24,49 with some modifications. Briefly, approximately 200 mg of protein was added to 10 mL of 50 mM ammonium bicarbonate pH 8.5, in a microcentrifuge tube. Then 100 mL of RapiGEST TM SF Surfactante (0.2% v/v) (Waters Corporation) was added, and the sample was vortexed and incubated in a dry bath at 80 C for 15 min. Following it was performed the reduction of disulfide bonds with 2.5 mL of 100 mM dithiothreitol (DTT) (GE Healthcare) at 60 C for 30 min, and alkylation of cysteine with 2.5 mL of 300 mM iodocetamide (GE Healthcare) for 30 min at room temperature in the dark. The proteins were subsequently digested with 40 mL of trypsin 0.05 mg/mL (Promega) at 37 C in dry bath for 16h. Trypsin digestion stop and precipitation of RapiGEST were done by acidifying samples whit 40 mL of 5% (v/v) trifluoracetic acid (Sigma-Aldrich). The mixture was incubated for 90 min at 37 C in a dry bath, and centrifuged at 18,000 g at 4 C for 30 min. The supernatants were dried in speed vaccum (Eppendorf). All obtained peptides were suspended in 80 mL of a solution containing 20 mM of ammonium formiate and 150 fmol/mL of PHB (Rabbit Phosphorylase B) (Waters Corporation) (MassPREP TM protein) as internal standard. Samples were transferred to a Waters Total Recovery vial (Waters Corporation).
Nanoscale LC separation of tryptic peptides was performed using a nanoACQUITY TM system (Waters Corporation) equipped with a nanoEase TM 5 mm x Bridge TM BEH130 C18 300 mm x 50 mm precolumn; trap column 5 mm, 180 mm x 20 mm and BEH130 C18 1.7 mm, 100 mm x 100 mm analytical reversed-phase column (Waters Corporation). The peptides were separated in 10 fractions; the gradient elution was performed as follows: 8.7, 11.4, 13.2, 14.7, 16, 17.4, 18.9, 20.7, 23.4 and 65% of acetonitrile/0.1% (v/v) formic acid, with a flow rate of 2000 mL/min. The source was operated in positive ionization mode nano-ESI (C). The lock mass was used for calibration of the apparatus, using a constant flow rate of 0.5 mL/min at concentration of 200 fmol protein GFP [Glu] 1 -Fibrinopeptide B human ([MC2H] 2C D 785.8426) (Sigma-Aldrich). Mass spectrometry analysis was performed on a Synapt G1 MS TM (Waters) equipped with a NanoElectronSpray source and 2 mass analyzers: a quadrupole and a time-of-flight (TOF) operating in Vmode. The mass spectrometer was programmed in the data-dependent acquisition mode, in which a full scan in the m/z region of 50-2000 was used. Data were obtained using the instrument in the MS E mode, which switches the low energy (6 V) and elevated energy (40 V) acquisition modes every 0.4 s. Samples were analyzed from 3 replicates.

Data processing and protein identification
The acquired MS raw data were processed using the Pro-teinLynx Global Server version 2.4 (PLGS) (Waters Corporation). The data were subjected to automatic background subtraction, deisotoping and charge state deconvolution. After processing, each ion comprised an exact mass-retention time (EMRT) that contained the retention time, intensity-weighted average charge, inferred molecular weight based on charge and m/z. Then, the processed spectra were searched against P. brasiliensis Pb18 protein sequences (Broad Institute; http:// www.broadinstitute.org/annotation/genome/paracocci dioides_brasiliensis/Multiome.html) together with reverse sequences. The mass error tolerance for peptide identification was under 50 ppm. The parameters for protein identification included: (i) the detection of at least 2 fragment ions per peptide, (ii) 5 fragments per protein, (iii) the determination of at least 1 peptide per protein, (iv) carbamidomethylation of cysteine as a fixed modification, (v) phosphorylation of serine, threonine and tyrosine, and oxidation of methionine were considered as variable modifications, (vi) maximum protein mass (600 kDa), (vii) one missed cleavage site was allowed for trypsin, (viii) maximum false positive ratio (FDR) of 4% was allowed. For the analysis of the protein quantification level, the observed intensity measurements were normalized with a protein that showed a variance coefficient of 0.025 and that was detected in all replicates (accession number: PADG_04710).

Proteome bioinformatics analysis
For analysis comparing the ratios between samples recovered from infected lung and control, was used as mathematical model part of the Expression algorithm inside the PLGS software from the Waters Corporation. 50 The minimum repeat rate for each protein in all replicates was 2. Proteins that presented 50% differences in expression ratios compared with the control were considered to be regulated. Protein tables generated by PLGS were merged, and the dynamic range of the experiment was calculated using the software MassPivot v1.0.1 and FBAT. 51 The peptide and protein tables were compared using the Spotfire Ò v8.0 software, and suitable graphics were generated for all data. Microsoft Excel (Microsoft Ò ) was used for table managements. The Uniprot (http:// www.uniprot.org) and Pedant on MIPS (http://mips. helmholtz-muenchen.de/funcatDB/) database were used for functional classification. NCBI database was used for annotation of uncharacterized proteins (https://www. ncbi.nlm.nih.gov/). The heat maps were generated using Multiexperimental Viewer 4.8 (www.tm4.org/mev).

Immunohistochemistry studies
For immunohistochemistry studies, 3 animals were used of test group and 3 animals of control group. Three-mmsections were obtained from formalin-fixed-paraffinembedded (FFPE) tissue blocks distended on charged slides (Starfrost White, Sakura Adhesion microscope slide with Cut Edges, ready to use -Dako 9545-1) for hematoxylin-eosin (HE) staining and for immunohistochemistry for microscopic examination. Immunohistochemistry was performed in duplicate, in all slides, which were removed of paraffin by treatment, rehydrated by treatment and washed in distilled water. Endogenous peroxidase was blocked with 8% (v/v) hydrogen peroxide incubation for 20 min. To block non-specific protein, rodent block M (Biocare Medical, #RBM961G) was used, by incubation for 30 min at 37 C.
Antigen retrieval was performed with Rodent Decloaker 1X (Biocare Medical, #RD913L) pH 6.0, for 3 min, at 121 C in pressure cooker (Pascal, Dako Citomation). The mouse polyclonal antibodies anti serine proteinase (anti-PbSAP) 24 and anti enolase, 25 were added to the slides at 1:150 dilutions and incubated at 37 C for 2 h, in wet chamber. After, the slides were incubated with Mouse-on Mouse HRP-Polymer (Biocare Medical, #MM620G) for 30 min at 37 C for signal detection, followed by addition of Diaminobenzidine (DAB, Dako, #K3468-1). Sections were counterstained with Mayer's hematoxylin, washed, dehydrated, cleared, coverslipped, and examined by light microscopy. For negative control, the primary antibody was replaced by phosphate buffered saline (PBS 1X), pH 7.4 in slides with P. brasiliensis.

Treatment of animals with immune response inducers
Controls for cross-reactivity of the anti-serine protease antibodies with mouse proteins were performed. Mice were inoculated intranasally, with zymozan, carrageenan or P. brasiliensis killed by heating, to induce host immune response. Zymozan (Sigma-Aldrich, Co.), was used to intranasally inoculate mouse at a dose of 0,6mg/g of body weight, as described previously. 26 Carrageenan (Sigma-Aldrich, Co.) was intranasally inoculated at a dose of 0,05mg/g of mouse body weight, as described previously. 27 The yeast cells of P. brasiliensis were submitted to autoclaving and 10 5 cells in 100 mL of 0.9% (w/v) NaCl were used to inoculate mouse by intranasal route.
Mice were killed in CO 2 chamber after 6 hours, lungs were harvested and processed for immunohistochemical analysis, that was performed using polyclonal antibodies to serine proteinase. The assays were performed in biologic triplicates.

Treatment of mouse with polyclonal antibodies to serine proteinase
Mice were treated with 1mg of polyclonal antibodies to serine proteinase and infected 24 h later with P. brasiliensis, with the same protocol described above. After 6 h of intranasal infection, the mice were killed and the lungs were obtained, for further histological and immunohistochemistry analysis.

Ethanol measurement assay
The concentration of ethanol was quantified by using an enzymatic detection kit according to the manufacturer's instruction (UV-test for ethanol, RBiopharm, Darmstadt, Germany). In the presence of nicotinamide-adenine dinucleotide (NAD), ethanol is oxidized to acetaldehyde by the enzyme alcohol dehydrogenase. This acetaldehyde is quantitatively oxidized to acetic acid in the presence of aldehyde dehydrogenase, releasing NADH, which is determined by means of its absorbance at 340 nm. A total of 10 6 P. brasiliensis yeast cells, obtained from lungs of 15 infected animals, were used to perform the assay. Control condition was also performed by using 10 6 P. brasiliensis yeast cells obtained after incubation in BHI medium, as described previously. Briefly, the cells were lysed using glass beads and bead beater apparatus (Bio-Spec) in 5 cycles of 30 sec, keeping the samples on ice. The cell lysates were centrifuged at 10,000 x g during 15 min at 4 C and the supernatant was used for enzymatic assay according to the manufacturer's instructions. Concentrations of ethanol were obtained in triplicates.

Evaluation of reduced thiol level
Thioredoxin (Trx) is reduced to dithiol T(SH) 2 by thioredoxin reductase (TR), in the thioredoxin system. The inhibition of TR decreases total reduced thiol. 52 Free thiol levels were determined using Ellman's reagent, 5, 5 0dithio-bis-(2-nitrobenzoic acid), DTNB (Sigma Aldrich, Co). A total of 10 6 P. brasiliensis yeast cells, as described above, were obtained from infected mice lung and from control. The cells were centrifuged for 5 min at 8,000 x g and lysed by adding 0.5 ml of lysis buffer (50 mM Tris-Cl, 150 mM NaCl, 50 mM ethylenediamine tetraacetic acid [EDTA], pH 7.2), and glass beads in equal volume of the cell pellet. Lysis was performed by vortexing for 3 mixing cycles of 3 min with 1 min intervals, on ice. Cellular debris were removed by centrifugation, and 100 mL of the supernatant was added to 100 mL of 500 mM phosphate buffer, pH 7.5, in each microtiter well, followed by the addition of 20 ml of 1 mM DTNB to each well. Absorbance was measured at 412 nm using a plate reader. Triplicates were obtained for each condition.

Disclosure of potential conflicts of interest
No potential conflicts of interest were disclosed.

Funding
This work at Universidade Federal de Goi as was supported by grants from Conselho Nacional de Desenvolvimento Cient ıfico e Tecnol ogico (CNPq) and Fundação de Amparo a Pesquisa do Estado de Goi as (FAPEG). This work is part of the INCT program of Strategies of Host Pathogen Interaction (HPI).