Interleukin 8 and Pentaxin (C-Reactive Protein) as Potential New Biomarkers of Bovine Tuberculosis

Bovine tuberculosis (bTB) is caused by Mycobacterium bovis. During the early stage of infection, greater than 15% of M. bovis-infected cattle shed mycobacteria through nasal secretions, which can be detected by nested PCR.

B ovine tuberculosis (bTB) is a zoonotic infectious disease primarily caused by Mycobacterium bovis (1,2). M. bovis is a slow-growing pathogen that may incubate in an infected animal for years before the disease becomes clinically evident (3). M. bovis can infect wild animals and humans via an airborne route, and greater than 15% of M. bovis-infected cattle shed the mycobacteria through nasal secretions, especially during the early stages of infection (4).
Traditional methods of diagnosis of bTB include the tuberculin skin test (TST) and interferon gamma (IFN-␥) release assays (IGRAs). TST, a globally accepted method, is recommended by the World Organization for Animal Health (OIE) as the standard method for bTB diagnosis. IGRA is based on the detection and comparison of cellmediated responses induced by bovine purified protein derivatives (PPD-B) and avian purified protein derivatives (PPD-A) (5). PPD-A testing is used with TST or IGRA to exclude other environmental Mycobacterium infections, but it often fails to detect coinfections of M. bovis and M. avium subsp. paratuberculosis in cattle. Research has focused on identifying new M. bovis-specific antigens to facilitate a higher specificity in bTB diagnoses. Recent work demonstrated that the sensitivity and specificity of the CFP-10/ESAT-6/TB10.4-based skin test (CET-ST) were 92.31% and 97.3%, respectively (6)(7)(8).
Currently, the control of bTB primarily relies on the test-and-slaughter program. Using this strategy, all cattle shown to be positive by TST and/or complementary IGRAs are slaughtered. However, this approach produces a significant economic burden (9). Insufficient government subsidies and a lack of awareness by farmers regarding bTB have led to challenges in implementing this program. Given these challenges, improved diagnostic methods are expected to facilitate control of this disease and reduce the economic burden. Considering that TSTs and IGRAs cannot distinguish the state of M. bovis infection, an mpb70-based nested PCR assay was established for detecting mycobacteria in milk and colostrum (9). Previous studies demonstrated that 23.18% to 87.5% of the M. bovis bacteria shed in the nasal exudates of infected cattle can be detected using nested PCR. We recommend that farm owners prioritize the slaughter of PCR-positive cattle in order to control the transmission of mycobacteria. However, application of nested PCR to detect excreters can be performed only in highly specialized facilities. Therefore, screening for novel biomarkers that correlate with PCR positivity and the establishment of an early and accurate blood-based test for bTB will aid in controlling this disease and reducing the economic burdens in developing countries.
In this study, we explored new biomarkers of bTB by screening for proteins that are associated with different stages of M. bovis infection. First, cattle were divided into three groups: M. bovis-infected cattle that were nested PCR positive (bTB PCR-P ), M. bovis-infected cattle that were nested PCR negative (bTB PCR-N ), and uninfected cattle (NC), as determined by TST, CET-ST, IGRA, a CFP-10 -ESAT-6 (CE)-based IGRA, and nested PCR. Proteins in serum or PPD-B-stimulated plasma that were differentially expressed (DE) between these three groups were identified and screened using iTRAQ labeling coupled with two-dimensional liquid chromatography-tandem mass spectrometry (iTRAQ-2D LC-MS/MS). Second, a total of 15 serum proteins and 15 plasma proteins were selected and validated using parallel reaction monitoring (PRM)-based quantitation. ) were further confirmed by enzymelinked immunosorbent assay (ELISA). The accuracy of new potential biomarkers (PPD-B-stimulated IL-8 and CRP) to differentiate bTB PCR-P and bTB PCR-N or infected cattle and NC was analyzed using receiver operating characteristic (ROC) curves. Finally, the efficiency of the new biomarkers was assessed in a total of 244 cattle, using TST, IGRA, and nested PCR as references (the work flow is shown in Fig. 1).

MATERIALS AND METHODS
Animals and sample collection. bTB in Holstein cattle (Exact 1211) was detected by TST and CET-ST (6). Heparinized whole blood was collected from each cow and dispensed into 24-well tissue culture trays (1.5 ml/well, four wells per animal). The blood samples were stimulated with 0.1 ml PPD-B (300 mg/ml; Prionics AG, Schlieren, Switzerland), PPD-A (300 mg/ml; Prionics AG, Schlieren, Switzerland), phosphatebuffered saline (PBS), or CFP-10 -ESAT-6 (CE; 20 g/ml; expressed and purified in IAS-CAAS with a Trx-His-S tag at the N terminus and with an endotoxin concentration of Ͻ10 endotoxin units [EU]/mg) in a 24-well tissue culture tray. Plasma was collected from each well after incubation with antigens for 20 to 24 h at 37°C in 5% CO 2 . The collected plasma was used for IGRA and CE-based IGRA, as summarized in a previous study (6). Blood samples were collected and clotted in vacuum tubes without anticoagulant and then centrifuged at 1,500 ϫ g at 4°C for 15 min to collect the serum. The serum collected from each animal was used for M. avium subsp. paratuberculosis and brucellosis antibody tests. All samples were aliquoted in sterile microcentrifuge tubes and frozen at Ϫ80°C until further use. Nasal swab specimens were collected from cattle infected with M. bovis, and infection was further confirmed by nested PCR, as previously reported (7).
Pipeline software) algorithm. Only proteins with at least one unique peptide and an unused score of Ͼ1.3 were considered for further analysis. Proteins with expression ratios of Ͼ1.5-fold or Ͻ0.67-fold and with P values of Ͻ0.05 were considered significant.
Bioinformatics analysis. The functional annotations of proteins, including their cellular component, molecular function, and biological process, were analyzed using the Gene Ontology (GO) database (http://geneontology.org/). Pathway analysis of DE proteins was conducted using the Kyoto Encyclopedia of Genes and Genomes (KEGG) database.
PRM-based quantification. iTRAQ results were validated by targeted MS analysis using parallel reaction monitoring (PRM), performed on a TripleTOF 5600ϩ LC-MS/MS system (AB Sciex). Proteins were prepared in a manner similar to that used to prepare the iTRAQ-labeled samples. MS data acquisition was first performed in the data-dependent acquisition (DDA) mode to obtain MS/MS spectra for the 40 most abundant precursor ions following each survey MS1 scan in each cycle. Protein Pilot software was used to identify proteins, and spectrum library building was accomplished by importing the database search results into Skyline software. Target proteins for PRM validation were imported into Skyline software, and the peptides for protein quantification were selected according to the ion signals in the spectrum library. A list of associated peptides containing m/z values and retention times was exported from Skyline software and then imported into Analyst MS control software for PRM acquisition method construction. Data collection from each sample was performed using the final PRM acquisition method on a QqTOF mass spectrometer, where each precursor ion was selected by the quadrupole and fragmented and then all fragment ions were quantified in the time of flight (TOF) mass analyzer. To eliminate protein carryover, a blank was run between adjacent samples to wash the column. Data processing was performed using Skyline software, and the quantification results were manually inspected for each peptide of the targeted proteins and normalized by the total peak area.
ELISAs. Candidate biomarkers for bTB were validated using a bovine IL To assess the specificity of the IL-8 and CRP responses in M. bovis-infected cattle, bTB PCR-P (n ϭ 5), bTB PCR-N (n ϭ 5), and NC (n ϭ 5) were identified by TST, IGRA, CE-based IGRA, and nested PCR. Heparinized whole blood was collected from each cow and dispensed into 24-well tissue culture trays (1.5 ml/well, five wells per animal). Blood samples were stimulated with 0.1 ml of PPD-B (3,000 IU/ml; Prionics AG, Schlieren, Switzerland), CE (20 g/ml), PET (Trx-His-S tag protein with an endotoxin concentration of Ͻ10 EU/mg), bovine serum albumin (BSA; 20 g/ml, Sigma), or PBS in 24-well tissue culture trays. Plasma was collected as mentioned above. The concentrations of IL-8 and CRP were tested using ELISA kits.
PPD-B-stimulated IL-8 and CRP tests used in clinical testing. To evaluate the efficiency of the PPD-B-stimulated IL-8 and CRP tests, a total of 244 cattle (42 males and 202 females, including the data used for ROC analysis) from four dairy farms were double-blind tested using TST, IGRA, nested PCR, and PPD-B-stimulated IL-8 and CRP tests.
Statistical analysis. Data were analyzed by analysis of variance (ANOVA) followed by a Kruskal-Wallis or Spearman correlation test using GraphPad Prism (version 5) software (San Diego, CA). A P value (two-tailed) of Ͻ0.05 was considered significant. Agreement between tests was evaluated using the coefficient.

RESULTS
Sample collection. Based on an initial screening of serum proteins related to M. bovis infection, 82 naturally M. bovis-infected cattle and 39 NC were identified by TST, CET-ST, IGRA, and CE-based IGRA. In addition, M. bovis-infected cattle were subdivided into the bTB PCR-P (n ϭ 41) and bTB PCR-N (n ϭ 41) groups using nested PCR. Of these samples, 20 bTB PCR-P and 20 bTB PCR-N and 20 NC were utilized for iTRAQ and PRM analyses, and the remaining 21 bTB PCR-P , 21 bTB PCR-N , and 19 NC were used for ELISA validation.
Identification and relative quantification of serum proteins. We compared three groups of mixed serum samples with two biological replicates in each group (i.e., the bTB PCR-P group [labeled with iTRAQ tags 113 and 114], the bTB PCR-N group [labeled with iTRAQ tags 115 and 116], and the NC group [labeled with iTRAQ tags 117 and 118]), using iTRAQ-2D LC-MS/MS analysis. We identified a total of 845 proteins; of these, 681 had two or more peptides. A total of 223 proteins were found to be significantly DE (fold change, Ն1.5 or Յ0.67; P Յ 0.05) between M. bovis-infected cattle (bTB PCR-P and bTB PCR-N ) and NC (n ϭ 20), including 112 upregulated proteins and 111 downregulated proteins (see Table S1 in the supplemental material). A total of 74 DE proteins, including 46 upregulated and 28 downregulated proteins (Table S1), were identified between bTB PCR-P and bTB PCR-N . Gene ontology analysis indicated that most of the identified DE proteins are involved in cellular processes, biological regulation, and the response to stimuli; are located in the extracellular region and organelles; and possess binding, catalytic, and enzyme regulator activities (Table S1). By KEGG analysis, the largest proportion of these proteins was associated with complement and coagulation cascades ( Fig. S1A-1 and A-2). Venn diagram analysis revealed 35 DE proteins shared among the bTB PCR-P , bTB PCR-N , and NC groups (Fig. S1B-1 and Table S2). Based on the bioinformatic and Venn diagram analyses, coupled with the fold changes in protein expression, 15 proteins were selected for preliminary verification (Table S3).
Identification and relative quantification of PPD-B-stimulated plasma proteins. Although all animals in this study were free from paratuberculosis and brucellosis and their infections were confirmed by TST, IGRA, and PCR, it is difficult to exclude the possibility of infection with additional pathogens. To screen for proteins in the blood that are associated with M. bovis infection, PPD-B-stimulated plasma from M. bovisinfected cattle and NC were compared by iTRAQ-2D LC-MS/MS analysis. We identified 719 proteins; of these, 531 proteins had two or more peptides. In total, we identified 207 DE proteins (fold change, Ն1.5 or Յ0.67; P Յ 0.05), including 151 upregulated and 56 downregulated proteins, between M. bovis-infected cattle (bTB PCR-P and bTB PCR-N ) and NC (Table S4). A total of 107 DE proteins, including 75 upregulated and 32 downregulated proteins, were identified between bTB PCR-P and bTB PCR-N (Table S4). Gene ontology analyses indicated that the DE proteins enriched in PPD-B-stimulated plasma have biological processes, cellular components, and molecular functions similar to those of the DE proteins identified to be enriched in serum (Table S4). Similarly, KEGG analysis revealed that the largest proportion of those proteins was also associated with complement and coagulation cascades (Fig. S1C-1 and C-2). We observed 37 DE proteins shared between the bTB PCR-P , bTB PCR-N , and NC groups by Venn diagram analysis (Fig. S1B-2 and Table S2). Based on the bioinformatics and Venn diagram analyses coupled with the fold changes, 15 were selected for further verification (Table S3).
Validation of iTRAQ analysis results by PRM. A total of 15 serum proteins and 15 PPD-B-stimulated proteins from the bTB PCR-P , bTB PCR-N , and NC groups were selected and subjected to PRM-based relative quantification (Table S5). The relative abundances of peptides from those individual proteins were acquired and normalized by the corresponding total peak area. By PRM, eight proteins in serum (AHSG, CGN1, OMD, SAA, C4, C7, APOE, and TF) and nine proteins in PPD-B-stimulated plasma (TF, IL-8, CD14, LRG1, C6, AGP, EFEMP2, AMBP, and RSU1) exhibited fold changes in expression similar to those identified by iTRAQ (Table S3). Although similarly pooled samples were used to conduct the iTRAQ and PRM analyses, the PRM results were not completely consistent with those obtained by iTRAQ. Beta-2-microglobulin and vitamin D-binding protein (serum), as well as PPD IL1RN, MBL, OMD, and protein HP-25 homolog 1 (PPD-B-stimulated plasma), were not detected by PRM (Table S3). Furthermore, five proteins in serum (AGP, AMBP, asporin, SERPINA1, and C6) and CRP in plasma exhibited contrasting results between PRM and iTRAQ analyses (Table S3). The removal of high-abundance proteins from serum and plasma samples may affect the detection of other low-abundance proteins. Considering our iTRAQ and PRM results, we further validated the levels of four serum proteins (AGP, AHSG, SAA, and TF) and five plasma proteins (CRP, TF, IL-8, C6, and AGP) using commercial ELISA kits.
Validation of potential bTB biomarkers by ELISA. To assess whether these four serum proteins (AGP, AHSG, SAA, and TF) and five plasma proteins (CRP, TF, IL-8, C6, and AGP) could be used to detect M. bovis infection and determine the stage of M. bovis infection, serum and antigen-stimulated plasma was harvested from 42 M. bovis-infected cattle (including 21 bTB PCR-P and 21 bTB PCR-N ) and 19 NC, and the proteins were detected using ELISA kits.
The concentrations of SAA in the serum of M. bovis-infected cattle (including bTB PCR-P and bTB PCR-N ) were significantly less than those in NC. SAA levels were significantly higher in bTB PCR-P than in bTB PCR-N ( Fig. 2A). CRP concentrations in PPD-B-stimulated plasma were significantly higher in bTB PCR-P than in bTB PCR-N and NC (Fig. 2B). Serum AGP levels were significantly higher in NC than in M. bovis-infected cattle. However, following PPD-B stimulation, AGP levels increased in M. bovis-infected cattle, and the AGP level was significantly higher in bTB PCR-P than in bTB PCR-N ( Fig. 2C and D). For serum TF levels, no significant differences were observed between M. Serum and plasma were harvested to measure the levels of SAA, CRP, AGP, and TF using commercial kits. Significant differences in protein levels between bTB PCR-P , bTB PCR-N , and NC were determined using a one-way ANOVA (Kruskal-Wallis test) followed by Dunn's multiple-comparison test. *, P Ͻ 0.05; **, P Ͻ 0.001; ***, P Ͻ 0.0001.
bovis-infected cattle and NC. TF levels decreased following stimulation with PPD-B, and TF was significantly higher in bTB PCR-P than in NC ( Fig. 2E and F). The concentrations of AHSG (serum) and C6 (plasma) were similar in M. bovis-infected cattle and NC (data not shown).
PPD-B-or CE-induced IL-8 levels were increased in bTB PCR-P and bTB PCR-N and were significantly higher than those in NC. However, the levels of unstimulated IL-8 (treated with PBS) were significantly higher in bTB PCR-N than in bTB PCR-P and NC (Fig. 3A to C). To further analyze the diagnostic potential of IL-8 for bTB, the correlations among IL-8, our previously published potential biomarkers (IL-17A, IP-10), and PPD-B-induced IFN-␥, CE, or PBS were calculated. This comparison included all individuals and all treatments (n ϭ 61 [21 bTB PCR-P ϩ 21 bTB PCR-N ϩ 19 NC]). As shown in Table 1, PPD-B-or CE-induced IL-8, IL-17A, IP-10, and IFN-␥ levels showed significant correlations with one another, and the IL-8 level was moderately correlated with the IFN-␥ level. The PBS-induced IL-8 level exhibited a weak correlation with the IP-10 and IL-17 levels but no correlation with the IFN-␥ level. Moreover, IL-8 concentrations in PPD-B-stimulated or unstimulated plasma were significantly greater than those of IFN-␥, IP-10, or IL-17A ( Fig. 3D and E). These data indicate that PPD-B-stimulated IL-8, IP-10, and IL-17A have the potential to differentiate M. bovis-infected cattle from NC, while PPD-B-stimulated CRP, PPD-B-stimulated AGP, unstimulated IL-8, and serum SAA could potentially differentiate bTB PCR-P from bTB PCR-N . ROC analysis. ROC analysis was used to evaluate the diagnostic value of our proteins of interest. The area under the curve (AUC) indicated that PPD-B-stimulated IL-8 displayed a diagnostic ability superior to that of IP-10 and IL-17A in discriminating  Table 2 and Fig. S2).

Specificity of IL-8 and CRP responses in M. bovis-infected cattle.
To access the specificity of the IL-8 and CRP responses in M. bovis-infected cattle, we stimulated whole blood from infected and uninfected cattle with PPD-B, CE, and antigens unrelated to M. bovis. We found that the unrelated antigens and PBS induced similar levels of IL-8 or CRP in infected and uninfected cattle. The levels of PPD-B-or CE-induced IL-8 were significantly higher in M. bovis-infected cattle than in NC, and the levels of PPD-Bor CE-induced CRP were significantly higher in bTB PCR-P than in bTB PCR-N or NC (Fig. 4).
Clinical testing. To assess the application of PPD-B-stimulated IL-8 and CRP tests to the diagnosis of bTB, 244 cattle were double-blind tested using TST, IGRA, nested PCR, and PPD-B-stimulated IL-8 and CRP tests. Strong correlations were observed between the concentrations of PPD-B-stimulated proteins (for IL-8 versus IFN-␥, Spearman r ϭ 0.75 and P Ͻ 0.001; for CRP versus IFN-␥, Spearman r ϭ 0.339 and P Ͻ 0.001; for CRP   (Table 3). When the PPD-B-stimulated CRP test was used to discriminate between bTB PCR-P and bTB PCR-N among 115 M. bovis-infected cattle (the same as those used for the PPD-Bstimulated IL-8 test), it exhibited good agreement with nested PCR ( ϭ 0.9117). Compared with nested PCR, the relative sensitivity and specificity of the PPD-Bstimulated CRP test were 94% and 97%, respectively (Table 4). Taken together, our clinical testing demonstrated that the PPD-B-stimulated IL-8 and CRP tests can be used to detect bTB and to distinguish bTB PCR-P from bTB PCR-N , respectively.  IL-17A transcripts in peripheral blood mononuclear cells (PBMC) are associated with pathology in M. bovis-infected cattle (10-13). However, a recent study found that PPD-B-induced IP-10 and IL-17A levels are lower than those of IFN-␥ and that CEinduced IL-17A levels were similar in bTB PCR-P and NC (14). Our study found that, with a target specificity of 95%, the sensitivities of IP-10 and IL-17A were only 52.38% and 28.57%, respectively. Thus, IP-10 and IL-17A are not suitable for bTB diagnosis, but IP-10 is considered a promising biomarker for human TB and can detect TB in children, HIV-positive patients, and those undergoing therapy for tuberculosis (10,(15)(16)(17). Using iTRAQ analysis, Seth et al. confirmed that serum alpha-1-microglobulin/bikunin precursor (AMBP) protein, alpha-1-acid glycoprotein (AGP), fetuin, and alpha-1B glycoprotein levels were significantly increased in M. bovis-infected cattle, and dot analysis revealed a significant increase in the vitamin D-binding protein precursor (DBP) in Mycobacteriuminfected cattle, but the study lacked validation by ELISA (18). The combined application of untargeted proteomics (iTRAQ, data-dependent acquisition [DDA], and data-independent acquisition [DIA]) and targeted proteomics (PRM and selected/multiple reaction monitoring [SRM/MRM]) to screen and validate biomarkers has been reported in several studies (19)(20)(21). This approach dramatically improved the identification of several putative biomarkers in the discovery phase and simplified the work flow using relatively high-throughput validation. These advantages make this combined strategy a powerful tool for the characterization of biomarkers in a wide range of infectious diseases. The aim of the present study was to identify biomarkers of bTB and to assess differences between bTB PCR-P and bTB PCR-N . Serum biomarkers facilitate testing but may not be specific for bTB. Thus, we compared both serum and PPD-B-stimulated plasma proteomes in M. bovis-infected cattle and NC using iTRAQ analysis and validated our results using relatively high-throughput PRM and commercial ELISA kits. We identified a total of 845 serum proteins and 719 plasma proteins, significantly more than were identified in previous studies (18). However, cytokines related to M. bovis infection, such as IFN-␥ and IL-17A, were not previously identified by iTRAQ analysis (18). Interestingly, in our study, we identified differences in PPD-B-stimulated IL-8 levels between M. bovis-infected cattle and NC by both iTRAQ and PRM analyses. PPD-B-and CE-stimulated IL-8 levels were significantly increased in M. bovis-infected cattle compared to NC, while unstimulated IL-8 levels were significantly higher in bTB PCR-N than in bTB PCR-P and NC. Furthermore, PPD-B-and CE-induced IL-8 levels exhibited a good correlation with IFN-␥ levels, and the concentration of IL-8 was higher than that of IFN-␥, IP-10, or IL-17A. Our results suggest that PPD-B-or CE-stimulated IL-8 could be used to differentiate between M. bovis-infected cattle and NC, while unstimulated IL-8 could differentiate between bTB PCR-N and bTB PCR-P . Similarly, another study indicated that CE-stimulated and unstimulated IL-8 levels are significantly higher in active TB patients than latent TB patients and heathy controls, and IL-8 concentrations may help differentiate between active TB and latent M. tuberculosis infection (LTBI) (22). In addition, the levels of IL-8 were higher in patients who died from TB than in survivors;  (23,24). Although the exact role of IL-8 in the pathogenesis of TB is not fully understood, recent studies have shown that IL-8 can bind to tubercle bacilli, and the IL-8 -pathogen interaction contributes to the increased mycobactericidal properties of macrophages and neutrophils (25). Higher levels of IL-8 contribute to the recruitment of monocytes and T cells to the pulmonary compartment, attract neutrophils to the infection site, and are required for granuloma formation (25,26). However, TB bacilli avoid the neutrophil response by downregulating IL-8 secretion by infected monocytes through an IL-10mediated autocrine loop (27). In cattle, although the levels of IL-8 and the role of IL-8 in bTB pathogenesis have been relatively unstudied, our results indicate that IL-8 also plays an important role in bTB. Furthermore, the ROC analysis indicated that PPD-Bstimulated IL-8 is a more suitable biomarker of M. bovis infection than IP-10 or IL-17A. Clinical testing indicated a high relative sensitivity and specificity for the PPD-Bstimulated IL-8 test compared to those of TST or IGRA. Establishing the role of IL-8 as a biomarker for bTB will require verification with a larger cohort of both experimentally and naturally M. bovis-infected cattle and NC.

DISCUSSION
PPD-B-stimulated CRP facilitates differentiation of bTB PCR-P and bTB PCR-N . The acute-phase response is a prominent systemic reaction to an organism to local or systemic disturbances in homeostasis due to infection, tissue injury, trauma or surgery, neoplastic growth, or immunological disorders (28). In this study, we identified several positive acute-phase proteins, including SAA, CRP, AGP, complement factors, haptoglobin, and alpha-2-microglobulin, as well as one negative acute-phase protein, TF. Upon validation by ELISA, we found that AGP and SAA levels were significantly decreased in M. bovis-infected cattle and that the serum SAA level was significantly higher in bTB PCR-P than in bTB PCR-N . Similarly, the level of serum SAA has been found to be increased in smear-positive compared to smear-negative TB patients (29). However, the variations in the levels of these four acute-phase proteins in TB patients differed from those in M. bovis-infected cattle. In human TB, the SAA level was significantly higher in TB patients than in healthy controls, and it was also significantly increased in TB patients with lung lesions and tended to decline in patients undergoing TB treatment (29)(30)(31). The AGP level was found to be increased in active TB patients and may be a possible marker of a slow response to anti-TB treatment (32). Jensen et al. first identified an association between LTBI and elevated AGP levels (33). TF levels in serum and sputum specimens were significantly higher in active TB patients than in latent TB patients (34). It is difficult to explain the higher levels of SAA and AGP in uninfected cattle. Although 39 paratuberculosis-and brucellosis-free NC were randomly selected from two bTB-free dairy farms, the possibility of infection or immunization with other viruses or bacteria, which may lead to elevations in the levels of SAA and AGP, cannot be excluded. Furthermore, infection with M. bovis may inhibit host immune responses, and SAA and AGP levels decreased significantly in M. bovis-infected cattle, but their effect on M. bovis infection requires further investigation. We also detected TF and AGP in PPD-stimulated plasma, and the TF level decreased in both M. bovis-infected cattle and NC following PPD-B stimulation, while the AGP level was increased in M. bovisinfected cattle. PPD-B may activate acute-phase proteins, leading to increased AGP and decreased TF. As variations in serum SAA and AGP are not specific to M. bovis infection, they are likely not suitable bTB biomarkers. However, PPD-B-stimulated AGP can discriminate between bTB PCR-P and bTB PCR-N with a target specificity of 95% but a sensitivity of only 52.38%. Thus, we did not find PPD-B-stimulated AGP to be a suitable biomarker for bTB.
The plasma level of CRP, an activator of the classical complement pathway, increases during an inflammatory state and is widely used as a biomarker for pulmonary infections (35). Previous studies have shown that TB patients display higher CRP levels than healthy controls (36, 37), but in our study, no significant differences in serum CRP levels were observed between M. bovis-infected cattle and NC by iTRAQ analysis. In contrast, CRP levels in the PPD-B-stimulated plasma of bTB PCR-P were significantly higher than those in bTB PCR-N and NC. Similarly, patients with smear-or culture-positive TB exhibited higher CRP levels than smear-or culture-negative patients (38). Considering that host factors (ethnicity, HIV infection) and mycobacterial factors (M. tuberculosis strain type, the site of disease) were strongly associated with the baseline CRP response in TB (39), we assessed the diagnostic efficiency of PPD-B-stimulated CRP using ROC analysis and found that CRP produced a better diagnostic outcome (AUC ϭ 1) than PPD-B-stimulated AGP, unstimulated IL-8, and serum SAA in discriminating between bTB PCR-P and bTB PCR-N . Furthermore, clinical testing revealed that, compared to the results of nested PCR, the sensitivity and specificity of the PPD-Bstimulated CRP test were 94% and 97%, respectively, and the agreement between the two tests was also high ( ϭ 0.9117). Therefore, PPD-B-stimulated CRP displays the potential to differentiate between different stages of M. bovis infection.
Stages of M. bovis infection can classified by nested PCR. The serum and PPD-B-stimulated plasma proteomes of bTB PCR-P and bTB PCR-N were compared by iTRAQ analysis, and 74 serum and 201 plasma proteins, respectively, were found to be significantly differentially expressed. These DE proteins showed broad functional distributions by GO and KEGG analyses, and a total of 16 DE serum proteins and 27 DE plasma proteins were found to be involved in the complement and coagulation cascades. Significant differences in serum SAA, serum IL-8, plasma AGP, and plasma CRP levels between bTB PCR-P and bTB PCR-N were also observed by ELISA. A previous study demonstrated that CE-induced or uninduced IL-17A levels were significantly higher in bTB PCR-N than in bTB PCR-P and NC (14). We speculate that bTB PCR-P and bTB PCR-N may be at different stages in the progression of bTB and may produce different immune responses to M. bovis-specific antigens.
Limitations of this study. First, our iTRAQ, PRM, and ELISA results did not display complete agreement. The lack of a commercial kit specific for the depletion of highabundance proteins in bovine samples and the depletion bias produced by the kit used may explain this phenomenon. Second, the iTRAQ, PRM, and ELISA analyses were not conducted at the same time, and thus, some proteins may have degraded during storage. Third, there are few commercial ELISA kits for bovine serum proteins, and some DE proteins, such as CD14 and C1q, displayed strong signals by iTRAQ or PRM analysis but were not validated by ELISA. Finally, it was impossible to slaughter all of the infected cattle used in this study to confirm the stages of bTB progression, including those in the PCR-positive and PCR-negative groups. Thus, we are unable to discern whether our putative biomarkers could be used to differentiate between M. bovisinfected cattle and NC and cattle with active or latent bTB. Therefore, further effort is required to validate these DE proteins as biomarkers for bTB diagnosis.
Conclusion. PPD-B-stimulated IL-8 can differentiate between M. bovis-infected cattle and NC, and PPD-B-stimulated CRP can differentiate between bTB PCR-P and bTB PCR-N . SAA, AGP, IP-10, and IL-17A are closely related to M. bovis infection but are not preferred markers for bTB diagnosis. The proteome of serum or PPD-Bstimulated plasma in bTB PCR-P was significantly different from that of serum or PPD-B-stimulated plasma in bTB PCR-N , indicating that bTB PCR-P and bTB PCR-N are at different stages of bTB progression and display different immune responses to M. bovis-specific antigens.
Ethics approval. All animals used in this study were treated with care and with the approval of the Animal Care and Use Committee of the China Institute of Veterinary Drug Control (SYXK 2005-0021).