All the experiments were performed in strict accordance with the Brazilian Federal Law 11,794 establishing procedures for the scientific use of animals, and the State Law establishing the Animal Protection Code of the State of São Paulo. All efforts were made to minimize animal suffering. The procedures were approved by the Ethics Committee on Animal Experiments of the Federal University of São Paulo (N° 9596270919).

8 to 12 weeks old male C57BL/6 mice obtained from the Center for the Development of Experimental Models for Biology and Medicine (CEDEME-UNIFESP) and bred as specific pathogen-free mice in the animal facility of the Institute of Science and Technology at UNIFESP were used throughout this study.

The virulent strain Pb18 of the fungus P. brasiliensis was used and maintained by weekly sub-cultivation in a semi-solid medium of Fava Netto (Netto, 1955) at 37°C. The fungus was collected and washed with phosphate-buffered saline (PBS, pH 7.4). The number of yeasts was evaluated using a hemocytometer and their viability was determined using the Janus-Green vital dye. All procedures were performed with fungal suspension with viability greater than 95%.

Groups of C57BL/6 WT mice were anesthetized with intraperitoneal (ip.) injection of ketamine (90 mg/kg) and xylazine (10 mg/kg) and subjected to intratracheal (IT.) infection as previously described (Cano et al., 1995). Briefly, after anesthesia the animals were infected with 1 x 10⁶ yeasts of P. brasiliensis contained in 50 μL of PBS by surgical (IT.) inoculation, allowing the direct dispense of the yeasts into the lungs. After the infection the experiments were performed as showed in Figure 1 .

To analyze the histopathology of the infected lungs, groups of 3-4 animals with two, six, eight and twelve weeks of infection were euthanized with xylazine (30 mg/kg) and ketamine (180 mg/kg) and the removed lungs were stored in 10% formaldehyde at 4 °C. Sections of 5 µm were stained with Hematoxylin-Eosin (H&E) for analysis of lesions and stained with Grocott for fungal evaluation. The anatomy of the lesion was analyzed according to size, morphology, and presence of fungi. The measurement of the area of the lesions was performed using the Leica DM750 microscope and the Leica Application Suite V4.8 software, thus allowing the observation of the establishment of the granuloma through time. To evaluate the fungal load a colony-forming unit (CFU) assay was performed. Briefly, the granulomatous lesions were separated from the apparently healthy regions with the help of a scalpel, and the tissue was macerated with a Doucer. Then, the cells were centrifuged at 1,200 x g for 10 minutes and the supernatant was discarded. The pellet was resuspended in 1 mL before cultivation of the yeasts in BHI. The colonies were counted daily.

After eight and twelve weeks of infection, 3 independent groups with 2-4 animals were euthanized, the lungs were removed, and the granulomatous lesions were separated from the apparently healthy lung regions with the help of a scalpel. For the transcriptomics analysis the entire granulomas were frozen before the RNA extraction. For the proteomic analyzes of yeasts present in the granulomas, the lesions were enzymatically digested with collagenase (2 mg/mL) for 40 minutes at 37°C, as previously described (Lindell et al., 2006). The digested tissue was gently macerated with a Doucer and then centrifuged at 1,200 x g for 10 minutes. The supernatant was discarded, and the pellet resuspended in 2 mL of distilled water. Then 1 mL of this suspension was dispensed in two different Petri dishes, and the volume completed with 4 mL of distilled water. The dishes were incubated at 37 °C for 1 h to lyse the mouse cells. The suspension was then centrifuged (1,200 x g, 10 minutes) and the supernatant discarded, remaining only fungi cells in the tube. To evaluate the virulence and the proteins expressed by yeasts recovered from granulomas, these lesions were macerated with a Doucer, centrifuged (1,200 x g, 10 minutes) and then pellet resuspended in Phosphate-buffered saline (PBS). The recovered yeasts were cultivated in BHI for 5 days before cultivation in Fava Netto medium for 7 days (recovered yeast). The cultivation of the yeasts in BHI before cultivation in Fava Netto medium is necessary because there is growth factor in the BHI composition, which is essential for fungal growth after recovery from the lesions.

To evaluate the virulence of cultivated yeasts recovered from granulomas (experimental group) in comparison to the yeasts used in the original inoculation (control group), a group of six C57BL/6 WT mice IT. infected with the recovered fungus and another group of 6 mice was infected with the control fungus. Deaths were recorded daily. The Log-rank test (Mantel-Cox) was used for the survival curve.

Frozen granulomatous lesions were first macerated with the help of a pestle and pot. To proceed with total RNA extration, TRIzol LS (Thermo Scientific) was used according to the protocol described by Pisu et al., 2020. For the control lung was used uninfected healthy lungs, while for the control yeasts was used the original yeast suspension used in the infection. The control lung and control fungus were mechanical lysed with pestle and pot and zirconium beads, respectively, followed by RNA extraction with TRIzol LS (Thermo Scientific) according to the manufacturer’s protocol. After RNA extraction, the DNA was digested with Rnase-free Dnase (RQ1 Rnase-Free Dnase, Promega Corporation) in the presence of the ribonuclease inhibitor Rnasin® (Promega Corporation). The purification was done with the addition of 1 volume of phenol:chloroform (1:1), followed by 100% ethanol:3M sodium acetate (10:1) RNA precipitation. RNA integrity was verified with a 2100 Bioanalyzer (Agilent) and RNA 6000 Nano Kit (Agilent).

First, rRNA was depleted using the RiboMinus™ Eukaryote Kit for RNA-Seq (Thermo Scientific). Then, cDNA library was synthesized using the NEBNext® Ultra II Directional RNA Library Prep Kit for Illumina® (New England BioLabs) and the NEBNext Multiplex oligos E7335 and E7500 (New England BioLabs). The cDNA library was sequenced in the NGS equipment Brazilian Biorenewables National Laboratory (LNBR), part of the Brazilian Centre for Research in Energy and Materials (CNPEM), a private non-profit organization under the supervision of the Brazilian Ministry for Science, Technology, and Innovations (MCTI). Because there is genetic material from two different organisms in the sample, these reads refer to two different genomes. Therefore, we chose to align against a chimeric genome of P. brasiliensis (Refseq accession number GCF_000150735.1) and Mus musculus (Refseq accession number GCF_000001635. 27). Bowtie2 was used in the default configuration for paired reads, but as it only returns the best pairing by default, the output mode was changed to report the best 5 alignments.

Proteins from the yeasts were extracted using the Yeast Buster Protein Extraction Reagent (MerckMillipore) according to the manufacturer’s specifications. Benzonase Nuclease (MerckMillipore), protease inhibitors (Protease Inhibitor Cocktail, Sigma Aldrich) and phosphatase inhibitors (PhoSTOP, Sigma Aldrich) were also added to the buffer. In-solution trypsin digestion was performed according to Kleifeld et al., 2011 with modifications (Kleifeld et al., 2011). Briefly, a solution of 6 M guanidine hydrochloride (GuHCl) was added to a sample of 100 μg of protein from each cell lysate to a final concentration of 3 M GuHCl, followed by the addition of 5 mM dithiothreitol (DTT) (final concentration). The mixture was incubated at 37°C for 1 h. Iodoacetamide (IAA) was then added to a final concentration of 15 mM and the samples were incubated for 30 min at room temperature, in the dark. To quench the excess of IAA, DTT was added to a final concentration of 15 mM followed by the incubation of samples for 20 min at room temperature. Clean-up of samples was performed by the addition of ice-cold acetone (8 volumes) and methanol (1 volume), followed by the incubation of samples for 3 h at -80° C. After centrifugation at 14,000 x g for 10 min, protein pellets were washed twice with one volume of ice cold methanol and then resolubilized with NaOH solution (final concentration of 10 mM), followed by the addition of HEPES buffer (final concentration of 50 mM), pH 7.5, to a final volume of 100 μL. Trypsin (Proteomics grade; Sigma, USA) was added at 1:100 ratio (enzyme/substrate) and protein samples were incubated at 37°C for 18 h.

Tryptic peptides were desalted using C-18 cartridges SPE Extraction disks (3M EmporeTM), resuspended in 50 μL of 0.1% formic acid, quantified with Pierce™ Quantitative Colorimetric Peptide Assay (Thermo Scientific) and stored at -80° C.

LC-MS/MS was made using the Thermo Fisher Q-Exactive mass spectrometer coupled to a Dionex RSLC nano. LC gradients ran from 4% to 35% B over 2 h, and data were collected using a Top15 method for MS/MS scans (O'Keeffe et al., 2014; Dolan et al., 2017). Mass spectrometric (RAW) data were analyzed with MaxQuant software (version 2.0.3.0). A False Discovery Rate (FDR) of 1% was required for both protein and peptide-to-spectrum match identifications. Raw data were searched against a target database restricted to the taxonomy ‘Paracoccidioides brasiliensis’ (UniProt/Proteomes - UP000001628; 8,399 entries). This database was also combined with the sequences of 245 common contaminants and concatenated with the reversed versions of all sequences. Enzyme specificity was set to trypsin and up to two missed cleavages were allowed; cysteine carbamidomethylation was selected as fixed modification whereas methionine oxidation, glutamine/asparagine deamidation and protein N-terminal acetylation were selected as variable modifications. Peptide identification was based on a search with an initial mass deviation of the precursor ion of 4.5 ppm and the fragment mass tolerance was set to 20 ppm. Label-free quantitation was performed using the MaxLFQ algorithm (Cox and Mann, 2008; Cox et al., 2014) with the ‘re-quantify’ function of MaxQuant software enabled. As is observed from complex proteomes such as those of vertebrates, peptides can be shared between homologous proteins or splice variants, leading to “protein groups”. For each protein group in the MaxQuant’s ‘proteinGroups.txt’ file, the first protein entry was selected as representative.

Transcriptomic data were normalized by log2 and Z-score. For proteomics data, all the protein intensity values were log2-transformed and quantile-normalized using the ‘preprocessCore’ library in R scripting and statistical environment (Ihaka and Gentleman, 1996; Bolstad et al., 2003). Statistical analyses were performed using the ‘limma’ package in R/Bioconductor, using eBayes for granuloma data and treat for the recovered yeast (Gentleman et al., 2004; Smyth, 2004; Ritchie et al., 2015). Differentially expressed genes (DEG) and proteins analyzed by eBayes were delimited by an adjusted p-value < 0,05 and a cutoff of |fold change| > 1, where the p-values were adjusted for multiple testing with the Benjamini-Hochberg method. The data analyzed by treat don’t need a delimitation, as specified in the limma package user’s guide. Correlations between replicates were also calculated in R, using Pearson correlation coefficient. Genes and proteins that were differentially expressed in both infection groups in comparison to the control group were reanalyzed by pooling the expression values for 8 and 12 weeks, generating a new log (Fold change) and adjusted P-value for the disease for both experimental time points (disease).

For function prediction, Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) annotations were taken, as well as enrichment analyzes (Ashburner et al., 2000; Kanehisa and Goto, 2000; Kanehisa, 2019; Gene Ontology Consortium, 2021; Kanehisa et al., 2023). The P. brasiliensis Gene Onthology functions were taken from FungiDB. Gene Onthology enrichment was performed by Enrichr for Mus musculus, using default settings (Chen et al., 2013; Kuleshov et al., 2016; Xie et al., 2021). The P. brasiliensis metabolic pathway was obtained from KEGG.

Fungi obtained from 8 weeks granulomatous lesions, fungi recovered from granulomas and control fungi were stained with Oil Red O (Merck Millipore) using the manufacturer’s protocol with minor modifications. Briefly, a 0.5% solution of Oil Red O in isopropanol was diluted in water at a 3:2 ratio and incubated at room temperature (~ 25°C) for 1 hour. After that, the diluted solution was filtered using a 0.22 μm filter. The cells, previously fixed in 10% paraformaldehyde, were centrifuged and the supernatant discarded; then, the filtered Oil Red O solution was added in sufficient quantity to cover the fungi. Staining took place for 2 hours at room temperature. Subsequently, the cells were washed twice with ultrapure water and then resuspended in PBS for analysis on a microscope slide. The photos were taken using the Leica DM750 microscope and the Leica Application Suite V4.8 software, thus allowing the observation of stained neutral lipids inside the cells.

Lung granuloma, control lung, control yeasts and recovered yeasts were frozen, and 100 mg of frozen material was macerated with the help of a pestle and pot. The secondary metabolites extraction was performed with 1 mL of methanol (100%; HPLC grade), followed by sonication bath for 40 minutes. For sample preparation, 500 L of each sample extract were filtered in a 0.22 μm PTFE and diluted to 1 mL volume with HPLC-grade methanol. LC-HRMS/MS analysis were performed in positive ionization mode in a Thermo Scientific QExactive Hybrid Quadrupole-Orbitrap Mass Spectrometer, with m/z range of 100–1500, capillary voltage at 3.5 kV, source temperature at 300°C and S-lens 50 V. The stationary phase was a Thermo Scientific Accucore C₁₈ 2.6 μm (2.1 mm x 100 mm) column. Mobile phase composition was a gradient of formic acid (A) and acetonitrile (B). Eluent profile (A/B %): 95/5 for 5 minutes, up to 60/40 within 5 minutes, up to 55/45 within 2 minutes, up to 2/98 for 6 minutes, maintaning 2/98 for 2 minutes, down to 95/5 within 2 minutes, maintaning 95/5 for 2 minutes. Total run time was 24 minutes for each sample run and flow rate was 400 µL/min. Injection volume was 5 μL. MS spectra were acquired with m/z ranges from 100 to 1500, with 70000 mass resolution at 200 Da. MS² spectra were acquired in data-dependent acquisition (DDA) mode. Normalized collision energy was applied stepwise (20, 30 and 40), and the 5 most intense precursors per cycle were fragmented with 17500 resolutions at 200 Da. Spectra data were processed with Xcalibur software (version 3.0.63) developed by Thermo Fisher Scientific. Compound Ferricrocin was annotated based on its exact mass and MS/MS fragmentation pattern.

The total RNA extracted from granulomas was subjected to reverse transcriptase using ImProm-II (Promega) according to the manufacturer’s instructions. The synthesized cDNA was used for real-time PCR analysis using the SYBR green PCR master mix kit (Applied Biosystems) on the ABI 7500 Fast real-time PCR system (Applied Biosystems, Foster City, CA, USA). For the M. musculus, Usp9x was used as normalizer, while for the P. brasiliensis, pyruvate decarboxylase (PADG_00714) was employed. The genes and primers used for RNA-seq validation are listed in Supplementary Table 1 .

For statistical analyses, the mean ± standard deviation/standard error was compared. For comparison between two groups, an unpaired Student’s t-test was performed, while for multiple analyses, a one-way ANOVA was conducted with Bonferroni correction. The level of significance was set at p < 0.05. The Log-rank test (Mantel-Cox) was employed for survival curve analysis, for comparison between two groups.

C57BL/6 mice were infected with 1x10⁶ P. brasiliensis yeasts via IT. route. Histological sections of two, six, eight and, twelve weeks of infection indicated that eight and twelve weeks of infection are the most adequate time points to obtain chronic granulomatous lesions considering the lesions organization and lung area occupied ( Figures 2A, B ).

To evaluate the fungal load a CFU assay was performed with the granulomatous lesions and the apparently healthy lung tissue. As it can be seen, approximately 90% of the yeasts present in the lung are within the granulomatous lesions, whereas 10% are present in the apparently healthy lung tissues ( Supplementary Figure 1 ). Thus, all results from the granulomatous lesions comprise 90% of yeasts present in the lungs.

Analysis of hosts’ lung transcriptional profile comparing 8-weeks infected lungs to non-infected lungs show 563 transcripts upregulated and 131 with downregulated, while lungs from 12-weeks infection display 509 up- and 136 downregulated genes ( Figures 3A–C ). Functional analysis of the host’s transcripts whose abundance increased during P. brasiliensis infection using KEGG, GO and Panther databases identified transcripts encondig Pattern Recognition Receptors (PRRs) and other receptors that are related to antigen presentation, such as CD68, Ptx3 (Pentraxin-related protein PTX3), Clec4a2 (C-type lectin domain family 4 member A2), Clec4d (C-type lectin domain family 4 member D), Clec4n (C-type lectin domain family 4 member N), Clec7a (C-type lectin domain family 7 member A), TLR13 (toll-like receptor 13), TLR2 (toll-like receptor 2), and H2-M2 gene (Histocompatibility 2, M region locus 2) ( Figure 3D ; Table 1 ). Likewise, transcripts of some cytokines such as IL-1α, IL-1β, TNF and IFN-γ and chemokines (CXCL2, CXCL9, CXCL5, CCL8, CCL7, CCL9, CCL20, CXCL1, CC20, CXCL10, CCL6, CCL4 and CXCL13) were also increased in granulomatous host cells. The abundance of transcripts related to cell activation and chemotaxis functions (which are not chemokines) were also found in the disease, such as Cd177 and Trem1 (triggering receptor expressed on myeloid cells 1), Trem3 (triggering receptor expressed on myeloid cells 3) and Gpr15 (G protein-coupled receptor 15). Transcripts for pro-inflammatory proteins were also identified as more abundant in infected lungs, such as Nos2 (nitric-oxide synthase), Gbp2b (Guanylate-binding protein 2b), Cd300ld (CMRF35-like molecule 5), Cfb, C1qb, C3ar1, and C1qc. Transcripts for immune regulatory proteins, such as Arg1 (arginase-1), Il1rn (IL-1 receptor antagonist), CTLA4 (cytotoxic T-lymphocyte-associated protein 4) and, Pdcd1 (programmed cell death protein 1), as well as genes related to tissue repair (matrix metalloproteinase-12 (Mmp12) and collagenase 3 (Mmp13) were also found in higher levels when compared to transcripts in naïve lung control. Additionally, a high number of transcripts related to antibody chains were also identified as more expressed in the fungal granuloma. Taken together, these data point to a prominent and highly regulated Th1 response in addition to the production of antibodies by the host immune system.

Transcripts to IL-17 and IL-21 were found exclusively in granulomas from infected animals, as well as some transcripts related to PRRs (the C-type lectin domain family 6 member A and theCD209 antigen-like protein E), several variable portions of T cell receptors, myeloperoxidase and genes related to the regulation of the immune response and tissue repair (Matrix metalloproteinase-1, Matrix metalloproteinase-7) ( Supplementary Figure 2 ), indicating a Th17 response in the granuloma. Together with the enrichment analysis where a Th17-related response was present in higher abundance in the lung granuloma when compared to uninfected lungs, the transcripts profile exclusively found in our analysis ( Supplementary Table 5 ) support that the susceptibility of the mice linage used in this work might be a Th17 response. Not only transcripts for immune response proteins were upregulated within the granuloma, but also transcripts for enzymes related to lipid metabolism such as Pla2g7 (platelet-activating factor acetylhydrolase), Gla (alpha-galactosidase) and Pla1a (phosphatidylserine sn-1 acylhydrolase), as well as genes related to protein regulation (Ctss, Ctsk, Stlfa2l1, stfa2, stfa1, Cstdc6, Csta2, Cstb, Serpina3g, Serpina3h and Serpina3f) and cell cycle/cell death regulation (Bcl2a1b, Bcl2a1c, Bcl2a1d and Bcl2a1a).

In contrast, genes involved in muscle organization and contraction, such as myosin and troponin, nervous system, action potential and organogenesis (Shank3, Slc7a10, Slc17a7, Kcna1, Kcnh2 and Sox11) were down-regulated ( Figure 3E ; Table 1 ). The repression of these genes may be related to impaired lung function within the granuloma since immune cells are expected to be located there, and as seen from Figure 2A there are less normal lung tissue present. Interestingly, lipid metabolism was also identified among repressed genes, as for example, [Cyp1a1 (cytochrome P450 family 1 subfamily A1), Plin1 (perilipin-1), Thrsp (thyroid hormone responsive)]. Curiously, Hepcidin (Hamp), a gene related to the negative regulation of iron availability was also downregulated in the granulomatous lesions.

Analysis of granuloma yeasts DEGs comparing 8- and 12-weeks infected lungs to control yeasts show 37 up- and 139 downregulated genes and 28 up- and 157 downregulated genes, respectively ( Figures 4A–C ). This large number of repressed genes when compared to induced genes may indicate a possible fungal metabolism repression within the granuloma. Functional analysis of the corresponding gene products was also carried out by using the KEGG, GO and FungiDB databases. A large number of the transcripts augmented in the fungus within the granuloma are related to the synthesis and transport of siderophores, such as: sidA (L-ornithine-N5-oxygenase; PADG_00097), sidI (Acyl-CoA ligase; PADG_00099), sidF (Hydroxyornithine transacylase; PADG_00100), fusarinin C synthase (sidD; PADG_00102), and OXR1 oxidoreductase (PADG_00104) ( Figure 4D ; Table 2 ). Siderophores are molecules responsible for the capture and storage of iron, being considered an important virulence factor in several pathogens (Schrettl et al., 2007; Blatzer et al., 2011; Silva et al., 2020).

In addition, genes that have already been described as virulence factors such as superoxide dismutase (PADG_02842), alcohol dehydrogenase (PADG_04701) (two non-iron enzymes that also manage antioxidant defenses) and polysaccharide synthase Cps1 (PADG_04274) were also found as upregulated. Superoxide dismutase was identified as upregulated in both 8- and 12-weeks infection groups. Transcripts which products are related to amino acid metabolism such as amino acid permease (PADG_07440), O-acetylhomoserine/O-acetylserine sulfhydrylase (PADG_08406) and aromatic-L-amino-acid/L-tryptophan decarboxylase (PADG_00215) were induced in both infection groups, while 4-aminobutyrate aminotransferase/(S)-3-amino-2-methylpropionate transaminase (PADG_02214) was induced only in the 8-weeks infection group.

Transcripts for gene and protein expression/degradation regulators are the most downregulated in the transcriptome of the fungus within the granuloma, in addition to transcripts related to the control of cell cycle, energy metabolism and virulence factors ( Figure 4E ; Table 2 ). Among the repressed genes, we identified the transcription factors (TFs) C6 finger domain TF nscR (PADG_06413), GATA-binding protein (PADG_05497), Zn (2)-C6 fungal-type domain-containing protein (PADG_04314), C6 TF (Ctf1B) (PADG_00411) and Zn (2)-C6 fungal-type domain-containing protein (PADG_04889). In addition to the repressed TFs, there are also two genes related to DNA repair that are downregulated, the DNA repair protein Swi5/Sae3 (PADG_02195) and the structure-specific endonuclease subunit SLX4 (PADG_04429). Furthermore, transcripts which products are related to energy metabolism were also repressed in the granulomatous yeasts, such as glucan 1,3-beta-glucosidase (PADG_07615), hexokinase (PADG_00393), acyl-CoA dehydrogenase (PADG_03602) and aconitate hydratase (PADG_11844). We also identified the downregulation of genes related to oxidative phosphorylation, such as cytochrome b5 (PADG_05290) and NADH dehydrogenase (ubiquinone) 1 alpha subcomplex subunit 2 (PADG_04699). Furthermore, some genes related to fungal virulence were also found at lower expression levels when compared to the control fungus, such as fungal nitric oxide reductase (PADG_05660), alcohol dehydrogenase (PADG_01716 and PADG_01987), superoxide dismutase (PADG_01954) and sidL (N5- hydroxyornithine acetyltransferase (PADG_07849).

We identified 157 and 97 proteins after 8- and 12-weeks of infection, respectively, and 1199 in the control group. When comparing the yeasts present in the granulomatous lesions at 8-weeks of infection with the control group, 1 upregulated protein and 124 downregulated proteins were identified, while the comparison of the 12-weeks infection group with the control group showed a total of 2 more abundant and 74 less abundant proteins ( Figures 5A–C ).

Increased abundance was observed in only three proteins, glycerol-3-phosphate dehydrogenase (C1G2K1), which is related to lipid metabolism; replication factor C subunit 3 (C1G5E7), involved in DNA replication, DNA repair and regulation of cell cycle progression in yeast; and GTP-binding protein ypt5 (C1G785), involved in vesicles transport ( Figure 5D ; Table 3 ). The six different proteins related to siderophore biosynthesis and transport ( Table 2 ) which transcripts have accumulated in granulomatous yeasts, have not been identified among the proteins revealed by the proteome, nor any other iron-capture protein have been identified among the accumulated proteins.

Instead, yeast cells sharply down regulate two key iron-dependent enzymes required for energy production by mitochondria and for antioxidant defence, respectively, the Cytochrome b-c1 complex subunit 2 (C1GM03), decreased by 4,11-fold; and cytochrome c peroxidase [EC:1.11.1.5], decreased by 3,10035 fold; that will be further discussed in this work.

Along with these iron-dependent enzymes, a large number of reduced abundance proteins were found ( Figure 5E ; Table 3 ), which mainly are related to redox potential, energy production, protein/amino acid and lipid metabolism have also been significantly down regulated in the imprisoned yeasts. Those are Superoxide dismutase [Cu-Zn], that converts superoxide radicals (O2-) into molecular oxygen (O2) and hydrogen peroxide (H2O2), which is further converted to water (H2O) by catalase, an enzyme that contains four iron-containing heme groups; the allosteric enzyme Pyruvate kinase (C1G2W2), that provide pyruvate to be converted to acetyl-CoA in mitochondria, to gluconeogenesis and to amino acid anabolism; succinyl-CoA synthetase alpha subunit (C1G294), which is part of an enzyme that regulates citric acid cycle by controlling the interconversion between succinyl CoA and succinate and the GTP/ATP production at substrate level, and to provide precursors for porphyrin and heme biosynthesis; enoyl-CoA hydratase (C1G2P3) for beta-oxidation, that provide acetyl-CoA to citric acid cycle, but also intermediates for the metabolism of lipids, cholesterol and ketone body; and, Aspartate aminotransferase (C1G3V5), that catalyses the interconversion of aspartate and α-ketoglutarate to oxaloacetate and glutamate, important for both amino acid synthesis and for providing intermediates for the citric acid cycle/energy production; and enzymes in amino acid metabolism, such as aspartate aminotransferase, 3-oxoacid CoA-transferase (C1GBD8) and 1 -pyrroline-5-carboxylate dehydrogenase (C1GBT4). Protein synthesis, degradation and folding have also been sharply affected in granulomatous yeasts as seen by the down regulation of two different peptidases, the Carboxypeptidase Y homolog A (EC 3.4.16.5) and the Vacuolar aminopeptidase I; the molecular chaperone DNAK (C1G0P0); the 40S ribosomal protein S14 (C1G0E5); and the pre-mRNA-splicing helicase BRR2 (C1G5P0). Finally, proteins involved in vesicle transport, such as coatomer subunit alpha (C1GAF5), importin subunit alpha-6/7 (C1FZ88) and phosphatidylinositol transfer protein SFH5 (C1GJS2) have also been decrease in abundance.

We hypothesize that the fungi inside the granulomas avoid energy loss by lowering protein synthesis, since these are one of the most consuming energy processes in a cell. At the same time there is a decrease of protein degradation in an attempt to tune the protein content to adapt the general metabolism to the constraint of the granulomatous environment.

We should keep in mind that although granulomatous yeasts have lowered their protein content in general, with 1037 proteins that have not been detected (either because they were not present or were below the detection threshold of the proteome analysis), the protein that have been detected were those that the yeast maintained because somehow they are important to cope with the constraint during the host-parasite interaction, are required to resist to/modulate of immune response, to keep yeast viability and competence for proliferation and dissemination within host. These assumptions make the proteins identified in this work as promising target for drug design and screening for inhibitors.

Additionally, the absent proteins in the infection groups were also related to gene/protein regulation, amino acid metabolism, energy metabolism, and cell cycle and transport ( Figure 5F ), which support a reprogramming of these biochemical processes by the imprisoned yeasts, indicating a possible fungal low activity within the granuloma, but with their virulence preserved, as shown in the next session.

The analysis of differentially expressed proteins shows that there is a large number of upregulated proteins. When yeasts from 8-week granulomatous lesions were cultivated and compared with control yeasts, 382 proteins were found to be induced and only 1 repressed. Similarly, when comparing the fungal proteins from granulomatous lesions at 12 weeks of infection, 319 proteins were found to be more abundant ( Supplementary Figures 3A–C ).

Opposing to what was observed for RNAseq and the proteome of the yeasts present in the lesions, a greater number of differentially abundant proteins were observed in the post-infection groups when compared to control group. Those abundant proteins have functions related to gene/protein regulation, energy metabolism, lipid and amino acid metabolism ( Supplementary Figure 3D ; Supplementary Table 11 ).

To better illustrate the yeast metabolism present in the granuloma when compared to the recovered yeasts (cultivated in Fava Netto medium), the Figure 6A shows the KEGG metabolic pathway of the P. brasiliensis in a transcriptional level, while Figure 6B shows the metabolic pathway at a proteomic level. This data indicates that there is a repression of the metabolism, especially the energy metabolism, both at transcriptional and proteomic levels in granulomatous yeasts. Instead, when we analyze the metabolic pathway of the recovered yeasts ( Figure 7 ) we can see that the metabolism that was once repressed it is now induced, evidenced by the upregulation of lipid, nucleotide, energy, carbohydrate, and amino acid metabolisms after yeast cultivation.

RT-qPCR analysis was performed to validate the dual-RNAseq results for both mouse ( Supplementary Figure 4 ) and P. brasiliensis genes ( Supplementary Figure 5 ). For the host validation, we used Cxcl9, Tnf, Gbp2b, Pdcd1, Tnnc1, Lcn2, and Hamp. For the pathogen validation, we used PADG_00104 (Oxidoreductase OXR1), PADG_00097 (L-ornithine N5-oxygenase SidA) and PADG_00099 (Acyl-CoA ligase SID4), PADG_05660 (fungal nitric oxide reductase), and PADG_01954 (superoxide dismutase, Fe-Mn family). Both datasets were validated by RT-qPCR data, supporting the findings in the RNAseq with a Pearson statistical correlation between 0.8981 and 0.9679 for the mice dataset and between 0.8906 and 0.906 for the P. brasiliensis dataset.

To verify the expression of siderophores by P. brasiliensis, LC-HRMS/MS analyses were performed with the granuloma fungus extracts, control fungus and recovered fungus. The siderophore Ferricrocin (m/z 718.3370) was annotated based on its exact mass and MS/MS fragmentation pattern typical of this siderophore class. Ferricrocin levels were lower in the yeasts present in the granuloma and the recovered ones when compared to the control yeasts ( Figure 8 ). This result corroborates the RNAseq data considering a lower expression of SidL observed in the granuloma yeasts ( Figure 4E ; Table 2 ).

Furthermore, a OilRed O lipid staining was performed aiming to evaluate the neutral lipid content in the three yeasts groups (control, recovered and present in the lesions). Oil Red O dye binds to neutral lipids, as those found in lipid droplets, but not to phospholipids in membranes, revealing the amount and size of lipid droplets by staining them with a red colour. As showed in the Figure 8B , the lipid droplets in the infection yeasts are less abundant when compared to control yeasts, supporting both the transcript and proteomic data, where genes and proteins related to beta-oxidation were significantly less abundant in the fungus inside the granuloma. Because lipid droplets are in greater number and size in the recovered and cultured yeasts compared to the yeasts present in the lesions, our results indicate 1) neutral lipid stocks decrease during infection and 2) that lipid biosynthesis is resumed by subsequent cultivation of yeast isolated from the lesions in rich medium.

Aiming to compare the virulence of the yeasts obtained from the granuloma after 8 weeks of infection with the non-granuloma yeasts (control yeasts, i.e, yeasts used for mice infection), a histopathology analysis of the lung was made. The mice infected with the yeasts isolated from granulomatous lesions had a less compact granuloma with a larger lesion area when compared to the granulomas size formed in mice infected with the control yeasts ( Figures 9A, B ). Furthermore, a survival analysis of infected C57BL/6 wild-type (WT) mice was performed. All mice infected with granulomatous lesion-yeasts died within 79 days; in contrast, mice infected with control yeasts start to die on day 91 and 100 days after infection only two mice were dead ( Figure 9C ). These results demonstrate that yeasts recovered from granulomatous lesions are more virulent than the non-granuloma control yeasts. To evaluate the fungal load a CFU assay was performed with lungs, liver, and spleen of mice. As it can be seen in the Figure 9D , the mice infected with the recovered yeast had a higher fungal load in the lungs and liver when compared to animals infected with control yeasts, showing that the yeasts from granulomatous lesions can disseminate more easily than control yeasts.