Subjects were prospectively enrolled and recruited by the staff at the Beijing Chest Hospital affiliated to Capital Medical University (Beijing, China) between March and October 2015. The present study was conducted according to the principles of the Declaration of Helsinki and approved by the Ethical Committee of the Beijing Chest Hospital affiliated to Capital Medical University (Beijing, China). All participants were ≥18 years old and provided written informed consent. Individuals receiving immunomodulatory therapy, with extrapulmonary TB, autoimmune disease, chronic disease, malignancies or HIV infection were excluded.

The patients were diagnosed with TB according to combined clinical criteria from the World Health Organization (8). Included patients with pulmonary TB had a positive sputum smear or culture result. The present study included 211 specimens and 161 samples, which were obtained from 161 patients with TB. Patients with pulmonary TB were further stratified according to disease severity, into a severe pulmonary TB (STB) group and a mild pulmonary TB (MTB) group. Patients were considered to be STB if three or more lung lobe alterations were observed in a chest computed tomography (CT) scan, with respiratory failure or hypoxemia (9). Patients were classified as MTB if they had less than three lung lobe alterations observed in a chest CT and normal blood oxygen. In the present study, the 161 patients were divided into 81 STB and 80 MTB cases.

Healthy subjects were recruited from a population who had no exposure to Mycobacterium tuberculosis, a negative tuberculin skin test and T-SPOT TB test result, a normal chest CT and no clinical evidence of any other diseases. A total of 50 age-matched non-Mycobacterium tuberculosis infected healthy control samples (NC) were selected. There was no significant difference in age, sex ratio, body mass index or smoking habits among patients in the STB, MTB and NC groups (all P>0.05; Table I).

All blood samples from each group were collected from venous catheters into tubes containing EDTA. Blood samples were centrifuged at 1,400 × g for 5 min at 4°C and plasma was stored at −80°C.

Albumin and immunoglobulin G (IgG) were removed using an immunodepletion column according to the manufacturer's instructions (Agilent Technologies, Inc., Santa Clara, CA, USA). Briefly, plasma samples (600 µl) were transferred into a 1.5 ml screw cap tube and centrifuged at 10,000 × g for 30 min at 4°C. Following this, 100 µl lysis buffer (7 M urea and 2 M thiourea; pH 7.4) was added into each sample. Proteins were subsequently diluted in 50 mM NH₄HCO₃ solution to a final concentration of 0.5 mg/ml. Finally, trypsin (Promega Corporation, Madison, WI, USA) was added at an enzyme-to-substrate ratio of 1:50 at 37°C for 12 h. Digested supernatant fractions were stored at −80°C without further treatment until liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis.

Digested peptide mixtures were pressure-loaded onto a fused silica capillary column packed with 3 µm dionex C18 reversed phase material (Phenomenex, Inc., Tianjin, China). The reverse phase sections (pore size, 100Å) were 15 cm in length and the column was washed with buffer A (water, 0.1% formic acid) and buffer B (acetonitrile, 0.1% formic acid). Following desalting, a C18 capture tip (5 mm; 300 µm) was placed in line with an Agilent 1,100 quaternary high performance liquid chromatography system (Agilent Technologies, Inc.) and analyzed using a 12-step separation. Tryptic peptides (100 µg) were eluted with a linear gradient from 2–100% buffer B over 70 min at a flow rate of 0.5 µl/min. As peptides were eluted from the micro-capillary column, they were electrosprayed directly into a mass spectrometer with (micrOTOF-Q II; Bruker Scientific Technology Co., Ltd., Beijing, China) with the application of a distal 180°C source temperature of N₂ and a nebulizer pressure of 50 psi. The mass spectrometer was operated in the MS/MS (auto) mode, and the ionization mode used was positive.

Data-independent ion selection was monitored to select the most abundant two ions from an MS scan for analysis. Dynamic exclusion was continued for 5 min, and performed for 70 min in total. Mass spectra were processed with DataAnalysis version 4.1 (Bruker Scientific Technology Co., Ltd.) and the Mascot generic format files were searched against the NCBI human database for tryptic peptides with up to one miscleavage using Mascot software version 2.1 (Matrix Science, Ltd., London, UK).

Tandem mass spectra were searched for in the National Centre for Biotechnology Information (NCBI) database (https://www.ncbi.nlm.nih.gov) using Mascot software 2.1 (Matrix Science, Ltd.). The default parameters for the quantification software, ProfileAnalysis software version 2.0, were used throughout the analysis (Bruker Scientific Technology Co., Ltd.). The differentially expressed proteins were identified with the cut-off criteria of false discovery rate <0.01 and |log₂ fold change (FC)|>1. Gene ontology (GO) annotation was carried out to categorize proteins based on biological process (BP), cellular component (CC) and molecular function (MF) using the Protein Analysis Through Evolutionary Relationships database (www.pantherdb.org) (10). Signaling pathway analysis was performed using the Search & Color Pathway tool on the Kyoto Encyclopedia of Genes and Genome (KEGG) database (www.genome.jp/kegg/pathway.html). Pathway enrichment analysis was conducted with the Database for Annotation, Visualization and Integrated Discovery (http://kobas.cbi.pku.edu.cn/anno_iden.php.) (11), with P<0.05 used as the cut-off value. Protein-protein interactions were obtained from the Search Tool for the STRING database (version 9.0; http://string-db.org/cgi/input.pl), containing known and predicted physical and functional protein-protein interactions (12).

Six different serum samples from each group were used for western blot analysis. Proteins were extracted using radioimmunoprecipitation assay lysis buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 1% NP-40, 1% deoxycholic acid sodium, 0.1% SDS, 0.1M phenylmethane sulfonyl fluoride, Roche Complete protease inhibitor cocktail tablets and phosphastase inhibitor cocktail tablets). Following this, total protein was determined using a bicinchoninic acid assay, and protein samples (20 µg/lane) were subsequently separated via 8, 12 and 15% SDS-PAGE gels and transferred onto polyvinylidene fluoride membranes. Following blocking with 3% bovine serum albumin with Tris-buffered saline containing 0.1% Tween-20 (pH 8.0; Sigma-Aldrich; Merck KGaA) at 25°C for 30 min, the membranes were incubated overnight at 4°C with primary antibodies against interleukin-36α (IL-36α; cat. no. ab117925; 1:1,000), α-1-acid glycoprotein (ORM)1 (cat. no. ab134042; 1:2,000), ORM2 (cat. no. ab88869; 1:2,000), S100 calcium binding protein A9 (S100A9; cat. no. ab92057; 1:1,000), superoxide dismutase (SOD)1 (cat. no. ab51254; 1:1,000), probetacellulin precursor (BTC; cat. no. ab10417; 1:1,000), tyrosine-protein kinase Lyn isoform B (LYN; cat. no. ab32398; 1:4,000) and collagen III (cat. no. ab7778; 1:1,000; all Abcam, Cambridge, MA, USA). Membranes were washed with Tris-buffered saline (10 mmol/l Tris-HCl, 150 mmol/l NaCl and 0.1% Tween-20 containing 5% skimmed milk) prior to incubation with horseradish peroxidase-conjugated anti-rabbit IgG (cat. no. S006; 1:5,000) or anti-mouse IgG (cat. no. S002; 1:5,000; both Santa Cruz Biotechnology, Santa Cruz, CA, USA) for 1 h at room temperature. Transferrin (cat. no. YM3527; 1:10,000; Beyotime Institute of Biotechnology, Haimen, China) was used as an internal reference. Proteins were visualized using an electrochemiluminescence kit (Amersham Biosciences, Buckinghamshire, UK) according to the manufacturer's instructions. Proteins levels were normalized via staining with 2% Ponceau S stain for 1 min at 25°C, and transferrin (cat. no. YM3527; 1:10,000; Beyotime Institute of Biotechnology) was used as the loading control. Densitometric analysis was performed using Quant software (version 11.5; TotalLab, Ltd., Newcastle upon Tyne, UK). The experiments were performed in triplicate.

ELISAs were performed to determinate the protein concentrations in plasma samples of STB, MTB and the NC groups. The commercial ELISA kits for four selected proteins were purchased including human IL-36α ELISA kit (cat. no. KA9520; 1:50), human ORM2 ELISA kit (cat. no. KA0480; 1:100), human protein S100-A9 ELISA kit (cat. no. KA1050; 1:50 dilution) and human SOD1 ELISA kit (cat. no. KA0680; 1:50; all Abnova, Taipei, Taiwan). The ELISA was performed according to the manufacturer's protocol. Protein levels of IL-36α, ORM2, S100-A9 and SOD1 in STB (n=72), MTB (n=71), and NC samples (n=41) were measured.

Statistical analysis was performed using SPSS version 18.0 (SPSS, Inc., Chicago, IL, USA). Qualitative variables were analyzed by the Fisher's exact test and Pearson's chi-squared test, while quantitative variables were analyzed by one-way ANOVA followed by the Student-Newman-Keuls post hoc test. P<0.05 was considered to indicate a statistically significant difference. Receiver operating characteristic curves were used to assess the diagnostic value of biomarkers.

The mass spectra data were searched using the profile analysis search algorithm against Human RefSeq protein database (https://www.ncbi.nlm.nih.gov), which is comprised of 69,002 protein sequences. A total of 3,385 proteins were quantified across all three biological replicates in each group (n=9) using the label-free proteomics analysis method.

Based on the LC-MS/MS data, 1,011 proteins were identified across all three biological replicates in each group (n=9) with a relatively good reproducibility. Compared with the MTB samples, 299 proteins were significantly differentially expressed, including 165 upregulated and 134 downregulated proteins in STB samples. In addition, the expression levels of 512 proteins in STB samples were significantly different (284 upregulated and 228 downregulated proteins) from the NC samples. A total of 153 proteins were found to be significantly differentially expressed in STB samples, of which 82 proteins were upregulated and 71 proteins were downregulated compared with the MTB and NC samples.

The 153 differentially expressed proteins in STB samples were used for GO database annotation in terms of BP, CC and MF (Fig. 1). In BP terms, proteins were mainly involved in ‘cellular process’, ‘responses to stimulus’, ‘apoptotic process’ and ‘immune system process’. In CC terms, the majority of the potential biomarkers were associated with the ‘organelle’, ‘cell junction’, ‘membrane’ and ‘macromolecular complex’. The enrichment analysis in MF terms revealed that ‘catalytic activity’, ‘binding’ and ‘enzyme regulatory activity’.

Pathway enrichment analysis of the 153 differentially expressed proteins in the STB samples obtained 8 upregulated and 1 downregulated signaling pathway (Fig. 2). The 8 upregulated signaling pathways were ‘glycosaminoglycan biosynthesis-chondroitin sulfate/dermatan sulfate’, ‘galactose metabolism’, ‘one carbon pool by folate’, ‘axon guidance’, ‘ECM-receptor interaction’, ‘ErbB signaling pathway’, ‘butirosin and neomycin biosynthesis’ and ‘focal adhesion’. In addition, these proteins were predominantly involved in the downregulated signaling pathway ‘amyotrophic lateral sclerosis’.

Furthermore, a network was constructed from selected protein-protein interactions (Fig. 3). This protein-protein interaction network consisted of 18 proteins, including ORM1, ORM2, S100A9, matrix metallopeptidase 16, copper chaperone for superoxide dismutase, peroxiredoxin (PRDX)1, glutathione peroxidase (GPX)2, SOD1, SOD2, PRDX2, ubiquitin C, catalase, GPX1, IL-36α, S100A8 and TIMP metallopeptidase inhibitor 2.

The expression levels of eight differentially expressed proteins (ORM1, SOD1, IL-36α, ORM2, S100A9, BTC, LYN and collagen III) with the highest values of |log2FC| were detected by western blot analysis. Six proteins among these eight differentially expressed proteins were demonstrated to be differentially expressed. In the plasma samples of the STB group, ORM1, IL-36α, ORM2, S100A9 and collagen III were upregulated, and SOD1 was downregulated, compared with the MTB and NC groups (Fig. 4A). This trend matched with what had been previously observed in the label-free quantitative proteomics analysis. Densitometry demonstrated that the protein expression levels of each protein were significantly different between the STB and NC groups (Fig. 4B; P<0.05) The plasma expression levels of ORM1, ORM2, IL-36α, and collagen III were significantly higher in the STB group compared with those in the MTB and NC groups (Fig. 4B).

The plasma expression of four candidate proteins including ORM2, S100A9, IL-36α and SOD1 were measured by ELISA (Fig. 5). IL-36α and collagen III were not detected due to lack of commercially-available ELISA kits and relatively lower values of |log2FC|. The expression levels of ORM2, S100A9, IL-36α and SOD1 in the STB group were significantly differently compared with the other two groups (P<0.05; Table II). The expression trend of these proteins was consistent with the results obtained from proteomics analysis.

The sensitivity, specificity and accuracy of ORM2 (79.3%, 48.0% and 0.587), S100A9 (86.2%, 90.0% and 0.891), IL-36α (82.4%, 67.4% and 0.791), and SOD1 (79.3%, 31.9% and 0.525) in discriminating between STB and MTB were presented in Fig. 6A. Furthermore, the combination of these four proteins used for discriminating STB from MTB achieved 90.00% sensitivity, 92.16% specificity and 0.965 accuracy.

For discriminating between STB and NC, the sensitivity, specificity, and accuracy of ORM2 was 98.0%, 35.4% and 0.720; S100A9, 62.0%, 84.3% and 0.729; IL-36α, 51.9%, 77.3% and 0.555; and SOD1 36.2%, 94.1% and 0.635, respectively (Fig. 6B). In addition, the combination of these four proteins for STB discrimination from NC achieved a sensitivity of 89.66%, specificity of 98.9% and accuracy of 0.981.