Proteomic analysis of serum samples of paracoccidioidomycosis patients with severe pulmonary sequel

Background Pulmonary sequelae (PS) in patients with chronic paracoccidioidomycosis (PCM) typically include pulmonary fibrosis and emphysema. Knowledge of the molecular pathways involved in PS of PCM is required for treatment and biomarker identification. Methodology/Principal findings This non-concurrent cohort study included 29 patients with pulmonary PCM that were followed before and after treatment. From this group, 17 patients evolved to mild/ moderate PS and 12 evolved severe PS. Sera from patients were evaluated before treatment and at clinical cure, serological cure, and apparent cure. A nanoACQUITY UPLC-Xevo QT MS system and PLGS software were used to identify serum differentially expressed proteins, data are available via ProteomeXchange with identifier PXD026906. Serum differentially expressed proteins were then categorized using Cytoscape software and the Reactome pathway database. Seventy-two differentially expressed serum proteins were identified in patients with severe PS compared with patients with mild/moderate PS. Most proteins altered in severe PS were involved in wound healing, inflammatory response, and oxygen transport pathways. Before treatment and at clinical cure, signaling proteins participating in wound healing, complement cascade, cholesterol transport and retinoid metabolism pathways were downregulated in patients with severe PS, whereas signaling proteins in gluconeogenesis and gas exchange pathways were upregulated. At serological cure, the pattern of protein expression reversed. At apparent cure pathways related with tissue repair (fibrosis) became downregulated, and pathway related oxygen transport became upregulated. Additionally, we identified 15 proteins as candidate biomarkers for severe PS. Conclusions/Significance Development of severe PS is related to increased expression of proteins involved in glycolytic pathway and oxygen exchange), indicative of the greater cellular activity and replication associated with early dysregulation of wound healing and aberrant tissue repair. Our findings provide new targets to study mechanisms of PS in PCM, as well as potential biomarkers.

Introduction of collagens in the wall of the bronchial tree, blood vessels, and pulmonary parenchyma. These structural changes lead to a decline in lung function, which may be progressive [10]. Emphysema involves dramatic obliteration of the pulmonary alveoli as a result of an immunological response that leads to recruitment of inflammatory neutrophils, macrophages, and lymphocytes. These cells secrete matrix metalloproteinases, elastase, and other proteases, which destroy the alveolar wall [11].
Thus, the identification of molecular mechanisms for fibrogenesis would be of great value and can be found easily accessible on biological fluids. In addition, blood serum is one of the easiest accessible sources of biomarkers and its proteome presents a significant parcel of metabolism and immune system proteins, due this, the serum proteome analysis can provide not only biomarkers but also biological explanations for observed events [38]. Therefore, we hypothesized that patients with severe PS present different serum proteomic signature, before, during, and after antifungal treatment could help the identification of prognostic markers, molecular mechanisms, and therapeutic targets.
In the present study, we performed the first molecular study of PS in patients with PCM. Using a proteomic approach, we aimed to identify molecules involved in these sequelae by comparing groups of patients with severe and mild/moderate-intensity PS, and determine the significant signaling pathways at various stages of clinical follow-up. In addition, we aimed to identify serum proteins that could function as candidate predictive biomarkers of PS.

Ethics statement
All study participants gave their written informed consent for inclusion before they participated in the study. This study was conducted in accordance with the Brazilian Norms and patients presented with CF PCM and PS. PCM was confirmed by the presence of a suggestive clinical condition, and identification of the typical P. brasiliensis yeast form in one or more clinical materials and/or specific serum antibodies detected by a double immunodiffusion (DID) test at the stage of active disease. Inclusion criteria were PCM pulmonary involvement, blood collection at all four time points of antifungal treatment, and a chest computed tomography (CT) scan at the end of treatment. Exclusion criteria were the presence of other systemic diseases related to infection, inflammation, or neoplasia, pregnancy, and lactation.

Procedures
We determined the serum proteome signature of 29 male patients, of whom 12 presented severe PS and 17 presented mild/moderate PS. There were no differences between the two groups in median age, degree of PCM severity, affected organs, type and time of treatment, time to clinical or serological cure, or specific serum antibody titer at admission ( Table 1). Degree of PS was evaluated by CT scan after discontinuation of antifungal treatment. The analyzes of CT scan were performed by a team of experts and the classification was based on the types of lesions frequently associated with less or more extensive pulmonary damage. The presence of fibrotic nodules, bronchial wall thickening and small centrilobular nodules, or signs of focal paracicatricial emphysema with fibroatelectasis and discrete traction bronchiectasis around the foci, were considered signs of mild/moderate PS. More pronounced alterations, with extensive bronchiectasis associated with more extensive emphysema beyond the areas of fibrosis, honeycombing signs, hyper-transparent areas associated with vascular poverty, emphysema blisters, recessed and rectified domes, and pulmonary hyperinflation, were considered signs of severe PS. Patients from both groups were evaluated at four stages during follow-up, as described by Mendes et al. [3]; S0: before antifungal treatment; S1: clinical cure, characterized by the disappearance of the initial symptomatology, reversion of the erythrocyte sedimentation rate (ESR) to normal values, serological serum levels (as determined by DID tests) that are decreasing but usually positive, and ongoing antifungal treatment; S2: serological cure, characterized by clinical cure, a normal ESR, persistent negative DID serology for one year, and ongoing antifungal treatment; S3: apparent cure, characterized by clinical cure, a normal ESR, and a persistently negative DID for two years after the discontinuation of the treatment.

Serum proteomics
Serum was collected after centrifugation of peripheral blood at 300 × g for 15 min, aliquoted, frozen at -80˚C, and thawed once before proteomic analysis. Twelve patients with severe PS (with sera from four patients in each of three pools formed randomly and 17 patients with mild/moderate PS (with sera from five or six patients in each of three pools) were included in the study. Serum albumin and immunoglobulins were depleted using a ProteoPrep Blue Albumin and IgG Depletion Kit (Sigma-Aldrich, St Louis, MO, USA) according to manufacturer's instructions. The expected depletion of these proteins was 80-95%. After depletion, Bradford assays [39] were performed to quantify proteins present in the pooled samples (n = 3 / group) and all samples were standardized to a concentration of 1μg/μL. Samples were submitted to proteomic analysis as previously described [40]. To 50 μL sample, 10 μL of 50 mM ammonium bicarbonate was added, before the following steps. First, 25 μL of 0.2% RapiGest (Waters Co., Manchester, UK) was added and incubated at 80˚C for 15 min. Second, 2.5 μL of 100 mM dithiothreitol was added and incubated at 60˚C for 30 min. Third, 2.5 μL of 300 mM iodoacetamide was added and incubated for 30 min at room temperature (in the dark). Fourth, 10 μL of trypsin (100 ng; Trypsin Gold, Mass Spectrometry Grade; Promega, Madison, WI, USA) was added and digestion was allowed to occur for 14 h at 37˚C. Fifth, after digestion, 10 μl of 5% trichloroacetic acid was added, and the sample was left in an incubation phase for 90 min at 37˚C. The sample was then centrifuged (16,000 g for 30 min). Finally, the supernatant was collected, and 5 μL of alcohol dehydrogenase (1 pmol/μL) plus 85 μL of 3% acetonitrile was added.

LC-MS/MS and bioinformatics analyses
The NanoACQUITY UPLC-Xevo QT MS system (Waters Co., Manchester, UK) was used to separate and identify peptides using the ion count algorithm, exactly as previously described [41]. The software ProteinLynx GlobalServer (PLGS) version 3.0 (Waters, Milford, MA) was used to search the LC-MS continuum data. The identification of serum proteins was performed using Homo sapiens database from the UniProtKB (http://www.uniprot.org) in February 2018. All sample pools were analyzed in triplicate. To determine differences serum proteins expression between PCM patients with severe and mild/moderate PS in each stage of clinical follow-up, ProteinLynx Global Server (PLGS) Expression E software was used, with p < 0.05 and p > 0.95 used to identify downregulation and upregulation of proteins, respectively, as reported earlier [40]. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE [42] partner repository with the dataset identifier PXD026906. Bioinformatics analysis was performed to compare groups, and UniProt protein ID accession numbers were mapped back to their associated encoding Uni-Prot gene entries. Furthermore, the Reactome database of pathways was searched using ClueGo v2.0.7 + CluePedia v1.0.8, a Cytoscape plug-in. UniProt IDs were uploaded separately from Tables 1, 2, 3 and 4 and analyzed with default parameters, which specify an enrichment (right-sided hypergeometric test) statistical test with a Bonferroni step-down correction method, 'single cluster' analysis type, using the genecluster list for Homo sapiens, evidence codes 'All', networking specificity: medium (GO levels 3 to 8), and a kappa score threshold of 0.4. Due to the type of sample used in this study, many immunoglobulins were identified as proteins differentially expressed between the groups. Some abundant proteins such as immunoglobulins are known to mask other protein components that are present in low concentrations [43,44]. We were unable to deplete more than IgG and albumin before the proteomic analysis; therefore, we decided not to include immunoglobulins in the tables and bioinformatic analysis because these were the predominant proteins with altered expression levels. However, these are listed in a supplementary tables (S1-S4 Tables).

Statistical analysis
Clinical data were analyzed as follows. Homogeneity of patient groups was examined by the Chi-square or Fisher's exact test. Mann-Whitney U test was used to analyze the time to serological and apparent cure. Unpaired t-test was used to compare two independent samples. Statistical analyses were performed using SAS Version 9.3 and GraphPad v.5.00 (GraphPad Software Inc., San Diego, CA, USA) software. Significance level was set at p �0.05. For proteomics data, statistical analysis followed the specifications of each algorithm [45] as previously described.

Results
Comprehensive global proteome profiling of serum proteins in severe PS patients compared with the serum of PCM patients with mild/moderate PS During clinical follow-up, 72 proteins were identified as differentially expressed in the serum of PCM patients with severe PS compared with the serum of PCM patients with mild/moderate PS. The biological functions of these 72 proteins, according to the UniProt protein database, and their expression levels in patients with severe PS relative to those in patients with mild/moderate PS, are displayed in a heatmap (Fig 1). Table 2 shows changes in protein expression in patients with severe PS compared to those with mild/moderate PS, before treatment (S0). Fig 2 shows that of 57 differentially expressed proteins, 42 were downregulated and 15 were upregulated in patients with severe PS compared to patients with mild/moderate PS (Fig 2A). Down-and upregulated proteins were separately uploaded into Cytoscape software to evaluate Reactome pathways. As can be observed in Fig  2B, the downregulated proteins were significantly enriched in seven different pathways, whilst the upregulated proteins were enriched in two different pathways.

Reactome pathways analysis of serum proteins in severe PS patients, at clinical cure (S1)
At clinical cure (S1), 38 differentially expressed proteins were identified in patients with severe PS when compared to patients with mild/moderate PS (Table 3). Of these, 32 were downregulated and six were upregulated in patients with severe PS ( Fig 3A). As observed in Fig 3B, Reactome pathway analysis revealed that most downregulated proteins were involved in six different pathways, while the upregulated proteins were involved in two pathways.

Reactome pathways analysis of serum proteins in severe PS patients, at serological cure (S2)
At serological cure (S2), patients with severe PS showed 46 differentially expressed proteins (Table 4) when compared with patients with mild/moderate PS, and of these, 16 were downregulated and 30 were upregulated ( Fig 4A). Reactome pathway analysis revealed that most downregulated proteins were involved in three different pathways, whilst the upregulated proteins were involved in nine pathways ( Fig 4B).

Reactome pathways analysis of serum proteins in severe PS patients, at apparent cure (S3)
Finally, at apparent cure (S3), 44 differentially expressed proteins were identified in patients with severe PS compared to patients with mild/moderate PS (Table 5). Of these, 41 were downregulated and three were upregulated ( Fig 5A) in patients with severe PS. Reactome pathway analysis revealed that most downregulated proteins were involved in five six different pathways, and upregulated proteins were involved in one signaling pathway (Fig 5B).

Overview of serum proteins differentially expressed between patients with severe PS, at various stages of clinical follow-up
Analysis of the pathways implicated across all PCM stages showed that the 72 differentially expressed proteins identified in this study were involved in 12 different pathways participating in pulmonary tissue wound healing (Fig 6). At S0, pathways related with the initial steps of wound healing were downregulated in patients with severe PS compared to patients with mild/moderate PS, while the 'erythrocytes take up oxygen and release carbon dioxide', and gluconeogenesis pathways were upregulated. The expression pattern of these pathways changed

PLOS NEGLECTED TROPICAL DISEASES
Molecular aspects of severe paracoccidioidomycosis pulmonary sequel after the introduction of antifungals. The expression of proteins involved in these pathways reversed slowly; at serological cure (S2), an upregulation of the pathways related to the initial phases of wound healing was observed, whilst there was no change in the 'erythrocytes take up oxygen and release carbon dioxide' pathway. Conversely, at apparent cure (S3), alterations in these pathways were now observed in the opposite direction: pathways related with tissue repair (fibrosis) became downregulated, and pathways related with gas exchange processes, e.g. scavenging of heme from plasma, became upregulated.

Serum biomarkers of inflammatory and fibrotic processes before treatment (S0)
In order to to validate the findings of proteomic data, serum levels of SPD, MIP-1α, IL-10, TNFα, IL-1 β, TGFβ, FGF, VEGF, and PDGF were measured in chronic PCM patients with mild/moderate or severe PS. In accordance with our proteomic findings, we found that at moment before treatment (S0) serum concentration of MIP-1α and VEGF were lower and in patients with severe PS compared to patients with mild/moderate PS (Fig 7). No differences were found in serum levels of SPD, IL-10, TNFα, IL-1 β, TGFβ, FGF, and PDGF between the two groups.

Discussion
Wound healing is a dynamic and highly regulated process consisting of cellular, humoral, and molecular mechanisms [46], with many opportunities for dysregulation, and thus the potential to lead to numerous pulmonary disorders [47]. Most of the studies on pulmonary wound healing have been carried out in patients with inflammatory, non-infectious diseases. The present

PLOS NEGLECTED TROPICAL DISEASES
study was performed in patients with PCM, a chronic granulomatous infectious disease, classified into two groups by the severity of their pulmonary sequelae: mild/moderate, or severe. The serum proteome signature of these two groups were compared at different stages of the disease, from the active stage until the apparent cure. In this study, 72 proteins were found to have altered expression across different PCM stages in patients with severe PS compared to patients with mild/moderate PS. These proteins were identified as participating in pathways important in the wound healing process. The physiological response to wounds can be characterized by key stages, including hemorrhage and fibrinclot formation, inflammatory responses, re-epithelialization, granulation tissue formation, angiogenic responses, connective tissue contraction, and remodeling [48]. In healthy tissue, this process is likely to happen continuously at a background level to maintain homeostasis. However, in chronic lung disease, repair processes are not able to adequately offset the injurious process, and aberrant repair fails to restore normal epithelial integrity, leading to loss of lung function [47].
At S0 and S1, the pathways related to the initial steps of wound healing, including coagulation, such as fibrin clot formation [49][50][51][52] and platelet degranulation plugs [53]; pro-inflammatory responses including complement cascade and syndecan interactions [54,55]; release of growth factors, including insulin-like growth factor [56][57][58]; and essential steps in the healing process, such as non-integrin membrane-extracellular matrix (ECM) interactions [59][60][61], were less active in patients with severe PS than in patients with mild/moderate PS. In addition, pathways involved in the protective tissue repair processes, including high-density lipoprotein assembly [62][63][64] and retinoid metabolism [65] pathways, were also downregulated in patients with severe PS. These data suggest lower wound healing activity in patients with severe PS than in patients with mild/moderate PS, in this active phase of PCM. In accordance with our proteomic findings, at S0, levels of a pro-inflammatory cytokine, macrophage inflammatory protein 1 alpha (MIP-1α), and a grown factor mediator, vascular endothelial growth factor (VEGF), were lower in patients with severe PS when compared with patients with mild/moderate, as measured by enzyme-linked immunosorbent assay. MIP-1α is a member of the C-C subfamily of chemokines, inducible proteins that exhibit various proinflammatory activities in vitro including leukocyte chemotaxis [66] whilst VEGF has an important role in wound healing through angiogenesis [67,68]. Furthermore, it is known that a less effective Paracoccidioidesspecific T-cell mediated response leads to a chronic infection, in which causes a dysregulated wound response, and inducing the development of pulmonary sequelae. The lower levels of these pro-inflammatory and growth factor mediators, allied to downregulation of wound healing pathways, in the sera of patients with severe PS indicate a dysregulation in tissue repair during the active phase of PCM.
On the other hand, at S0, the erythrocytes take up oxygen and release carbon dioxide, and enzymes alpha, gamma and beta enolases that are participating of gluconeogenesis and glycolytic pathways were upregulated in patients with severe PS compared with those with mild/ moderate PS. It is well-known that during ECM production, fibroblasts have high glycolytic flux and biosynthetic activity even when they are not growing [69]. To prevent fibrosis, ECM anabolism and catabolism need to be aligned and tightly controlled. It was observed that a consistent downregulation in fatty acid oxidation and upregulation of glycolysis in fibrotic skin and in normal skin with abundant ECM [70]. In addition to the an increased glycolytic activity, the higher activity of erythrocytes take up oxygen, indicating increased levels of oxygen in wound healing process. In the chronic wound microenvironment, there inevitably exists a substantial imbalance between the supply of oxygen and the high energy demand of the healing tissue [71]. From a molecular standpoint, the key factors that propagate this imbalance include the following: (1) the increased utilization of oxygen by the hypermetabolic regenerating tissue, (2) the sustained and increased production of ROS by phagocytes (respiratory burst), and (3) reduction-oxidation (redox) signaling [71]. Furthermore, oxygen is needed in the later steps of collagen synthesis for proline and lysine hydroxylation and cross-linking, which is the step required for collagen to be released from cells [72]. The upregulation of these two pathways at the same time in severe PS patients when compared with mild/moderate PS patients in during the active phase of PCM indicating higher ECM deposition [59,60,69,70,73], which is related with the development of more fibrosis [47]. At the same way, an excess of ECM deposition is a hallmark of chronic progressive scarring conditions that fall under the fibrotic interstitial lung disease umbrella, including IPF [47].
Corroborating our findings, Tobón and collaborators [30], analyzing clinical records and chest radiographs from 47 itraconazole-treated patients with PCM undergoing prolonged post-therapy follow-up, found fibrotic lesions in 31.8% of patients at diagnosis (S0), and, at the end of the study, fibrosis persisted in these patients as sequelae. Unfortunately, the study did not classify sequelae severity at the end of follow-up. Taken together, our data suggest that patients who progressed to severe PS showed, at active disease (S0), more delayed wound healing responses than patients progressing to mild/moderate PS, characterized by lower activity of pathways important for tissue repair, and associated with overproduction of ECM, causing early fibrosis.
This profile persisted in the evaluation performed at S1, at which time patients were receiving antifungal treatment and had presented clinical cure. These findings indicate that the pathways responsible for the initial stages of tissue repair remain downregulated despite the clinical improvement followed by clinical cure. These pathways changed at the stage of serological cure (S2); the heatmap shown in Fig 1 illustrates an upregulation of the proteins involved in pathways related to wound healing [49-53, 55-58, 62-64]. Interestingly, the evaluation carried out at apparent cure (S3) showed a further alteration to these pathways, but in the opposite direction, i.e., the pathways related with tissue repair (fibrosis) were downregulated, and the pathway related to the scavenging of heme from plasma was upregulated, favoring the activities related to gas exchange [74]. As the pulmonary sequelae are also deleterious for patients due to its interference in respiratory function [5], the alterations observed at S3 may develop as a means of avoiding a worsening of respiratory function.
Our data suggest that, in PCM patients with severe PS, wound healing develops with higher intensity in the S1-S2 period, i.e., between the clinical and the serological cure. The specific causes underlying the dysregulated, and consequently delayed, wound healing observed in patients with severe PS should be explored in further studies.
Of five unique serum proteins in severe PS patients at S0, angiotensinogen and complement C5 have been showed an important pro-fibrotic role by activation of myofibroblasts cells. Angiotensinogen (AGT) is an precursor in tissue renin-angiotensin system (RAS) that plays an important role in promoting the development of hepatic fibrogenesis [85], renal interstitial fibrosis [86], and idiopathic pulmonary fibrosis [87]. Interesting, it was observed that a direct inhibition of AGT in pro-fibrotic cells could attenuate the progression of hepatic fibrosis in the early stage [85]. In addition, preliminary clinical studies on chronic hepatitis C and nonalcoholic steatohepatitis suggest that RAS blocking agents may have beneficial effects on progression of fibrosis [88,89]. Therefore, the study of angiotensinogen in the context of chronic PCM should be better investigate. Complement C5 seems to be a druggable mediator of pancreatic fibrosis that directly activates pancreatic stellate cells and whose deletion or inhibition greatly reduces fibrogenesis after pancreatic necrosis [90]. In addition, the pro-fibrotic role of C5 have been observed also in renal [91] and liver fibrosis [92]. Unique 60S ribosomal protein L7-like 1, a putative uncharacterized protein MYH16, and the shieldin complex subunit 3 had never been associated with wound healing process, its role in fibrosis of PCM should be investigate in future studies.
In the same context, the up-regulated proteins in severe PS patients at S0 are also related with higher activity of pro-fibrotic cells and collagen production as enolase isoforms, hemoglo-bin´s family and transthyretin. The enolase is an enzyme participating of glycolytic functions and its upregulation have been contributing with diverse pathological process including hepatic fibrosis [93]. In addition, drugs blocking enolase activity has already been investigated [94,95], but needs to be evaluated in context of chronic PCM. Although hemoglobin family and transthyretin are not usually associated with fibrosis process, a recent study demonstrated that transthyretin affects cardiac fibroblasts, contributing to heart fibrosis [96].
Concerning the downregulated proteins in severe PS patients compared with mild/moderate patients at S0, some of them have also been shown to be altered in fibrotic diseases, such as serum β2-glycoprotein I in Chagas disease [97], and hepatitis C [98]. Vitamin D binding protein in chronic obstructive pulmonary disease [99] and, similar to our findings, plasminogen has been found down-regulated in the plasma of patients with IPF [23]. Also, decreased expression of vitamin D binding and apolipoprotein A-I are indicative of liver fibrosis in patients with hepatitis C [100].
As severe PS leads to a compromised respiratory function in PCM patients, and time points immediately before treatment and at clinical cure are key moments at which PS can progress to severe, we suggest that the prognoses for severe PS should be considered as soon as possible and as early as diagnosis of PCM. For this, we have identified 15 proteins that were unique or most highly upregulated at the moment of diagnosis (before treatment, S0) as predictive biomarkers of severe PS development in PCM, as following: 60S ribosomal protein L7-like 1, angiotensinogen, complement C5, putative uncharacterized protein MYH16, shieldin complex subunit 3, hemoglobin subunit beta, hemoglobin subunit gamma-2, hemoglobin subunit delta, hemoglobin subunit epsilon, hemoglobin subunit gamma-1, hemoglobin subunit alpha, alphaenolase, gamma-enolase, beta-enolase, transthyretin.
The limitations of this study was our inability to validate the proteins identified by the proteomic approach, the study did not quantifyed the extension of sequelae severity. Others challenges of this study were 1) low number of new cases/year in one center of research; 2) low adherence to visiting during the whole long-term of follow-up; 3) frequent co-morbidities that exclude a high number of patients; 4) poor adherence to the antifungal treatment. Indeed, the present study is a result of more than two decades of hard work that evaluated and followedup the enrolled patients carefully. In addition, the answer to the question why some patients develop more fibrosis than others when faced same infection remains open and should be further investigated. The strengths of our study were the thorough analysis of differentially expressed proteins, the prospective design, including evaluation at the four clinical follow-up stages, and the analytically useful comparison with patients with mild/moderate PS.
We conclude that severe PS as a PCM outcome results from a dysregulation in important stages of wound healing, especially before treatment and at clinical cure. In addition, we identified the 15 most highly upregulated proteins in patients with severe versus mild/moderate PS immediately before treatment as candidates for predictive severe PS biomarker in PCM. Our findings provide new insights into pulmonary fibrogenesis in PCM patients and a guide for further studies on antifibrotic treatments in combination with antifungal therapies.
Supporting information S1 Table. Proteins with expression significantly altered in the serum of paracoccidioidomycosis patients with severe and mild/moderate pulmonary sequel (PS) as outcome in the moment of before treatment (S0). (DOCX) S2 Table. Proteins with expression significantly altered in the serum of paracoccidioidomycosis patients with severe and mild/moderate pulmonary sequel (PS) as outcome in the moment of clinical cure (S1).