Screening and identification of five serum proteins as novel potential biomarkers for cured pulmonary tuberculosis

Rapid and efficient methods for the determination of cured tuberculosis (TB) are lacking. A total of 85 differentially expressed serum proteins were identified by iTRAQ labeling coupled with two-dimensional liquid chromatography-tandem mass spectrometry (2D LC-MS/MS) analysis (fold change >1.50 or <0.60, P < 0.05). We validated albumin (ALB), Rho GDP-dissociation inhibitor 2 (ARHGDIB), complement 3 (C3), ficolin-2 (FCN2), and apolipoprotein (a) (LPA) using the enzyme-linked immunosorbent assay (ELISA) method. Significantly increased ALB and LPA levels (P = 0.036 and P = 0.012, respectively) and significantly reduced ARHGDIB, C3, and FCN2 levels (P < 0.001, P = 0.035, and P = 0.018, respectively) were observed in cured TB patients compared with untreated TB patients. In addition, changes in ALB and FCN2 levels occurred after 2 months of treatment (P < 0.001 and P = 0.030, respectively). We established a cured TB model with 87.10% sensitivity, 79.49% specificity, and an area under the curve (AUC) of 0.876. The results indicated that ALB, ARHGDIB, C3, FCN2, and LPA levels might serve as potential biomarkers for cured TB. Our study provides experimental data for establishing objective indicators of cured TB and also proposes potential markers for evaluating the efficacy of anti-TB drugs.

Scientific RepoRts | 5:15615 | DOi: 10.1038/srep15615 bacteria (sensitivity = 19.6%, specificity = 87.0%) 5 , and sputum cultures require 4 to 8 weeks of incubation time (sensitivity = 91.8%, specificity = 57.8%) 6 . Therefore, there is an urgent demand to establish a rapid and efficient method to determine cured TB 7,8 . Nahid et al. 9 used the SOMAscan proteomics method and found differential changes in serum extracellular matrix protein 1, the tyrosine-protein kinase Yes, insulin-like growth factor-binding protein 1, cathepsin Z, coagulation factor V and serum amyloid A proteins between TB patients with 2 months treatment and untreated TB patients. Moraes et al. 10 found that serum copper levels and the C-reactive protein/serum albumin ratio might be important indicators of TB treatment. IFN-γ -inducible protein 10 and pentraxin 3 are also significantly decreased in cured TB patients compared with untreated patients (P < 0.0001) 11 , indicating that these proteins could serve as biomarkers for cured TB. The above studies aimed at identifying biomarkers for treated TB, but they failed to establish a cured TB model.
In this study, differential serum proteins were screened using iTRAQ labeling coupled with two-dimensional liquid chromatography-tandem mass spectrometry (2D LCMS/MS) to identify potential biomarkers for cured TB. The candidate protein biomarkers were validated as predictors of anti-TB treatment outcomes. Our study provides experimental data for establishing an objective indicator of cured TB. It also provides potential markers for evaluating the efficacy of anti-TB drugs.

Results
Identification and relative quantification of serum proteins. We identified 85 differentially expressed proteins by iTRAQ labeling coupled with 2D-LC MS/MS in cured patients and untreated smear-positive TB patients. Further screening revealed that 51 differentially expressed proteins were up-regulated (> 1.50-fold, P < 0.05), whereas 34 proteins were down-regulated (< 0.60-fold, P < 0.05) (Supplementary Table S1). Most of the differentially expressed proteins were involved in the metabolic process (30,5.57%), cellular component organization or biogenesis (28, 5.19%), and response to chemical stimulus (28, 5.19%). In addition, these proteins were located in the extracellular region (38,21.71%) and the organelles (34,19.43%) and possessed catalytic activity (12,27.91%) and enzyme regulator activity (11,25.58%) (Fig. 1A). In addition, the proteins were also assembled in the following pathways: focal adhesion (7 proteins), extracellular matrix (ECM)-receptor interaction (7 proteins), complement and coagulation cascades (6 proteins), and phagosomes (6 proteins) (Fig. 1B). Protein-protein interactions were noted among all the proteins (Fig. 1C). We narrowed these 85 proteins to 35 proteins by adding a control group. Proteins down-regulated in untreated TB patients and up-regulated in the cured TB patients (Table 1), and proteins up-regulated in untreated TB patients and down-regulated in the cured TB patients (Table 2) were chosen.
We further performed ELISA to clarify the changes in the ALB, ARHGDIB, C3, FCN2, and LPA levels during anti-TB treatment in 57 untreated patients (0 month), 53 2-month treated patients (2 months), and 59 cured patients (> 6 months). Significant differences were observed at the three time points for the five proteins (P < 0.001, P < 0.001, P = 0.011, P = 0.030, and P = 0.014, respectively). Additionally, linear trends were observed in the five proteins at the three time points (P = 0.008, P < 0.001, P = 0.008, P = 0.004, and P = 0.005, respectively). Serum levels of ALB and FCN2 were altered after 2 months of treatment (P < 0.001 and P = 0.030, respectively) (Fig. 3). All untreated TB patients were separated by age, gender, sputum smear results, lung lesions, and chest X-ray results. Significantly increased ARHGDIB levels and reduced FCN2 levels were observed in untreated smear-negative TB patients compared with smear-positive TB patients (P < 0.001 and P = 0.016) ( Table 3). However, no such differences were observed in cured TB patients.

Discussion
China has the world's second largest TB epidemic 1 with 4.99 million active TB cases, including 720,000 smear-positive cases and 1.29 million culture-positive cases 12 . However, due to the lack of rapid and efficient methods for determining cured TB, 14% of the patients are discharged without being fully cured     Table 4. Clinical data of the pulmonary tuberculosis patients and controls. All data are presented as the mean ± SD. A/G: albumin/globulin ratio; HDL-C: high-density lipoprotein cholesterol; LDL-C: low-density lipoprotein cholesterol; APOA1: apolipoprotein A1; APOB: apolipoprotein B; CRP: C-reactive protein.
a P-value between two groups using the t-test. * P < 0.05, ** P < 0.01, *** P < 0.001. and are vulnerable to relapse. The relapse rate for TB is 5.3% globally 1 , whereas the rate is 11.8% in China 12 . In addition, 20.5% of previously treated TB cases are estimated to develop multi-drug resistant TB, which is greater than newly treated cases (3.5%) 3 , and may increase transmission potential. Previously reported 2-month TB treatment biomarkers, such as coagulation factor V, thrombospondin-4, and alpha-1-antitrypsin 10,13 , and previously reported diagnostic biomarkers, such as apolipoprotein A-II 14 and fibrinogen beta chain 15 , were also identified by iTRAQ-2DLC-MS/MS analysis in cured and untreated TB patients in our study. Although coronin-1A (fold change = 0.60, P < 0.05), proteasome subunit alpha type-5 (fold change = 0.32, P < 0.05), and MAP/microtubule affinity-regulating kinase 4 (fold change = 4.00, P < 0.05) function in T-cell activation 16 , apoptosis 17 , and the positive regulation of programmed cell death 18 , commercial ELISA kits for these proteins are not available. Further studies on these unselected proteins may help us identify more effective diagnostic strategies for the determination of cured TB. ELISA revealed significantly increased serum ALB and LPA levels and significantly reduced ARHGDIB, C3, and FCN2 levels in cured TB patients compared with untreated ones, including both smear-positive (Fig. 2) and smear-negative patients.
ALB is reduced during inflammatory states. In this study, ALB levels were reduced in untreated TB patients compared with controls (P < 0.001). An epidemiological follow-up study found that people with low serum ALB levels were susceptible to TB (P = 0.006) 19 , whereas TB patients with low serum ALB levels were vulnerable to death (P < 0.001) 20 , indicating that ALB levels of patients could influence the pathogenesis and prognosis of TB. LPA is a hydrophilic glycoprotein that can be assembled with LDL 21 and potentially exhibits coagulated function 22 . Apolipoproteins serve as biomarkers for the diagnosis and treatment of TB 14,23-25 . We also found that the total cholesterol, triglyceride, HDL-C, LDL-C, lipoprotein (a), APOA1, and APOB levels were significantly altered in TB patients ( Table 4), indicating that lipid metabolism biomarkers might be used as indicators for anti-TB treatment.
Rho GDP dissociation inhibitor 2 (ARHGDIB, LyGDI) not only plays a role in inflammation and immunity [26][27][28] , but also affects cell invasion by regulating the extracellular matrix [29][30][31] . In the present study, ARHGDIB increased significantly in untreated TB patients compared with controls (P = 0.003) and decreased in cured TB patients (P < 0.001). No significant difference was noted between cured TB patients and controls. Therefore, we suspect that ARHGDIB might affect the formation of TB granulomas through the regulation of extracellular matrix proteins, resulting in its increase in untreated TB patients followed by a reduction in cured TB patients. Moreover, ARHGDIB levels were significantly reduced in smear-positive patients compared with smear-negative patients (P < 0.001) ( Table 3). We suggest that smear-negative patients infected with lower amounts of bacteria 32 , which might exhibit more proliferative pathological changes, and lead to a higher level of ARHGDIB.
The complement system has been seriously studied in TB studies 23,[33][34][35] . As a part of the complement system, C3 is potentially related to TB pathogenesis 34 and anti-TB treatment 35 . Our study revealed a C3 level reduction in cured TB patients (P = 0.035). FCN2 is involved in the immune defense. This protein binds to pathogen-associated molecular patterns (PAMPs) on the pathogen surface and initiates the complement lectin cascade, thereby clearing the pathogens 36,37 . Eisen et al. 38 proposed that FCN2 is associated with inflammatory changes in the respiratory system, which is supported by our findings. Faik et al. 39 found that FCN2 levels decreased significantly after malaria treatment (P < 0.0001). Our study also revealed decreased FCN2 levels after anti-TB treatment. Therefore, we suspect that the lower levels of C3 and FCN2 in cured TB patients (P = 0.035, P = 0.018, Fig. 3) might be due to the alleviation of complement system activation. In addition, FCN2 levels were significantly increased in smear-positive patients compared with smear-negative patients (P = 0.016, Table 3). We assumed that smear-negative patients had a higher proportion of latent infections 40 , resulting in limited complement system activation and reduced FCN2 levels.
In summary, ALB and LPA levels increased significantly in cured TB patients, whereas ARHGDIB, C3, and FCN2 levels decreased significantly. This finding might be due to the improved inflammation and lipid metabolism status as well as eased immune system and complement system activation status. Therefore, ALB, LPA, ARHGDIB, C3, and FCN2 levels might serve as potential biomarkers for cured TB. In addition, significant linear trends and increased levels of ALB and FCN2 were observed in 2-month treated TB patients (P < 0.001 and P = 0.030, Fig. 3), indicating the predictive value of ALB and FCN2 levels for treatment outcomes.
Studies have demonstrated that a diagnostic model established by a combination of markers exhibits increased sensitivity and specificity compared with a single marker model 14,41 . Our study established a model with a sensitivity of 87.10%, a specificity of 79.49%, and an AUC of 0.876. The sensitivity and specificity were not only higher than those for single proteins but were also higher than those of the sputum smear and sputum culture; thus, the model was more accurate for cured TB determination. The lack of efficacy evaluation markers largely hinders the development of new anti-TB drugs and therapies 8,42 . Our study established a combination model to provide the experimental basis for evaluating the efficacy of anti-TB drugs.

Methods
Patients and Control Subjects. The complete details of the entire study design and procedures involved were in accordance with the Declaration of Helsinki. This study was approved by the Ethics Committee of the Faculty of Medicine Zhejiang University, China. Written informed consent was obtained from all subjects before blood sampling. The methods used were carried out in accordance with approved guidelines and regulations.
Blood was drawn into regular bottles in the morning from newly diagnosed TB patient at three time points (before anti-TB therapy, after the intensive phase, and upon cure) between November 2013 and November 2014 at the Sixth Hospital of Shaoxing and the First Hospital of Jiaxing. Pulmonary TB patients were diagnosed according to the diagnostic criteria of the Ministry of Health, China 43 . All patients met one of the following pulmonary TB diagnostic criteria: (1) a positive sputum examination (smear or culture); (2) a negative sputum examination and a chest X-ray and CT revealing evidence typical of active TB; (3) a pathological diagnosis of TB in lung specimens; (4) suspected pulmonary TB after clinical follow-up and X-ray observations after excluding other lung diseases; and (5) clinical elimination of other causes of pleural effusion and a diagnosis of tuberculosis pleurisy. We recruited both smear positive and negative patients. Standard TB therapy consists of rifampin, isoniazid, pyrazinamide and ethambutol for the first 2 months followed by rifampin and isoniazid for an additional 4 to 6 months 2 . Patients with extra-pulmonary TB, malignancies, chronic disease, autoimmune diseases, or HIV infection were not included in the study. Fasting blood samples were drawn from healthy controls at the Zhejiang Hospital. The samples were stored at − 80 °C for further analysis. Data including age, gender and clinical examination findings of TB patients and controls were collated into databases separately by different time points. In total, 122 untreated TB patients, 91 2-month treated TB patients, 59 cured TB patients, and 122 healthy controls were enrolled in the study.
A total of 57 untreated TB subjects ( iTRAQ-2DLC-MS/MS. To increase the precision and accuracy of the data in the proteomics study 44 , equal amounts of 10 different samples were mixed to produce a sample group. In addition, 10 samples were randomly divided into two pools as biological replicates. Then, iTRAQ-labeled sample pools were generated. High-abundance serum proteins, such as albumin, IgG, and haptoglobin, were removed using the Human 14 Multiple Affinity Removal System (Agilent Technologies, Santa Clara, CA, USA). Next, the proteins were concentrated and desalted 14 . A total of 100 μ g of protein from each group were soaked in ice-cold acetone and then centrifuged. The subsidence samples were reduced and blocked. The samples were then digested with trypsin at 37°C overnight. Finally, iTRAQ reagents (Applied Biosystems, Foster city, CA, USA) were labeled as follows: the control group, iTRAQ reagent 113, 117; the untreated smear-positive pulmonary TB group, iTRAQ reagent 114, 118; and the cured pulmonary TB group, iTRAQ reagent 115, 119. The six sample groups were mixed, desalted, and dried. The iTRAQ-labeled peptides were separated using a polysulfoethyl column (2.1 × 100 mm, 5 μ m, 200 Ǻ; Nest Group, Inc., Southborough, MA, USA) with strong cation exchange (SCX) chromatography 45 . A total of ten SCX components were collected, concentrated, and dissolved. Samples were subsequently loaded onto the ZORBAX 300SB-C18 column (5 μ m, 200 Ǻ, 0.1 × 150 mm; Microm, Auburn, CA, USA). The components produced by SCX chromatography were subjected to MS analysis twice. The ratio of the peak area of the iTRAQ reporter ion reflected the relative abundance of the peptide and protein 46,47 . Protein identification and quantification were performed using the ProteinPilotTM version 4.2 software (Applied Biosystems). The ProGroup algorithm was used to identify peptides. MS/MS data were searched against the Human International Protein Index database (version 3.87) with parameter settings as described previously 14,46,47 . To reduce false positive results, a strict cutoff for protein identification was applied with the unused ProtScore > 1.3 and at least one peptide with a 95% confidence limit 48,49 . The protein expression ratio between the two groups (< 0.60 or > 1.50) was considered significant. The cellular component, molecular function, and biological process were analyzed by the Gene Ontology database, whereas KEGG pathway analysis was performed using the KEGG database (false discovery rate < 1.00%). The protein-protein network was analyzed by STRING software (http://string-db.org/). , and the human Ficolin-2 (FCN2) ELISA kit (Cusabio Biotech. Co., LTD, China; SwissProt: Q15485) were used to detect protein levels in the serum. The protein concentration of 57 untreated TB patients, 53 2-month treated TB patients, 59 cured TB patients and 60 healthy controls were measured according to the manufacturer's instructions. Serum samples were diluted with dilution factors of 1:10,000, 1:800, 1:4,000, and 1:20,000 for ALB, C3, FCN2, and LPA, respectively. ELISAs were performed according to the instructions of each kit. Statistical Analysis. Parametric data are presented as the mean ± standard deviation (SD), whereas nonparametric data are presented as the median ± interquartile range (IQR). P-values < 0.05 are considered statistically significant by the SPSS software (version 16.0, Chicago, IL). The parametric data were tested using the chi-square test for the composition ratios and t-tests and one-way analysis of variance (ANOVA) for means. Nonparametric analysis was performed using the Mann-Whitney U-test for two groups and the Kruskal-Wallis H-test for three or more groups. The test for linear trends was performed to examine the trend of protein expression during treatment, whereas Spearman's correlation method was performed to determine the association between two different parameters. For each protein, a ROC curve was generated. During model construction, the diagnostic score of untreated TB patients was set as 0, whereas that of cured TB patients was set as 1. In the other model, the score of sputum-positive patients were 1, and the score of sputum-negative patients were 0. To increase the diagnostic accuracy of the combined serum proteins, multiple logistic regression analysis was performed. ROC curves and logistic regression models were calculated using MedCalc Software (Version 12.4.2.0, Belgium).