PPE 59 immunoreactivity in tuberculosis patients: IgA immunodominance to that IgG and IgM CURRENT STATUS: UNDER REVIEW

Background: The role of protein families PE/PPE remains relatively enlightening. The gene rv3429, coding the PPE59 protein, is one of the 12 members of the PPE subfamily, and it is absent in all BCG strains. Although elicits low T cell response, there are no clues for the ability to induce a humoral response. Methods: We cloned, expressed, and analyzed by ELISA IgG, IgM and IgA anti PPE59 in 212 sera. Of them, 69 were from tuberculosis patients’ residents in Italy (TB IT , 12 native and 67 immigrants), the remaining 133 are Brazilians citizens (BR). Pulmonary (pTB IT ) or extrapulmonary (eTB IT ) clinical forms were 54 or 25. The pTB BR were 52 and 81 non-TB patients, including 10 non-tuberculous mycobacteria infected (NTM). Results: Keeping the specificity at 97%, IgA sensitivity decreased from pTB BR (53%) to pTB IT (38%) and eTB IT (28%), with an overall sensitivity of 42.7% and moderate accuracy (61.8 %), while IgG showed significant lower (p<0.0001) performance, 21%, 3%, 0%, 9% and 37%, respectively, at specificity of 83%. Groups were not discriminate by IgM. False-positive NTM IgG was high. Combination 16kDa IgG, previously performed only on Bazilians’ sera, plus PPE 59 IgA results increased sensitivity to 71% at 87.4% specificity. The overall smear microscopy (SM) diagnosis was 70.8% and a combination of SM/IgA increased sensitivity to 74% (p=0.01), mainly among SM- cases. Among pTB BR , all these rapid tests, including SM, sensitivity reach 86.5% (p=0.001). Positive IgA polarization was significant in extensive lung disease (p=0.001) and alcohol users (p=0.04). Conclusion: Although PPE59 IgA independently has moderate accuracy on TB diagnosis, together with other biomarkers contributes to improving its detection. Moreover, the polarized reactivity deserves further investigation.

laboratory confirmation [2]. Delay in TB diagnosis prevents early treatment, playing a role in the maintenance of TB transmission in the community [3]. The oldest immune-based tests for TB diagnosis, i.e, tuberculin skin test (TST) and serology, although relatively cheap, have low specificity because of cross-reactivity with antigens both present on vaccine bacilli Calmette Guérin (BCG) and non-tuberculous mycobacteria (NTM) species [4]. Respect to TST, serology has the potential of being used at point-of-care and can be performed at work field without onsite microscopy services, particularly useful when dealing with special populations, such as immigrants, indigenous, homeless [5; 6].
The protein members of PPE (Pro-Pro-Glu) family have not been fully explored; recent studies have suggested the involvement of PPE family in many aspects of pathogenesis, including bacterial attachment to host cells, immunomodulation, and ability to persist in granuloma [7]. The importance of the PPE proteins in the generation of antigenic variations has been shown. The PPE41 was reported to induce immune responses of B cells using a panel of human sera from patients with active TB and relapsing patients with extrapulmonary TB [8; 9]. The PPE Rv1168c was described as having strong specific immunoreactivity in sera of patients with active TB compared to that PPD, ESAT-6, and hsp60 [10]. On the other hand, PPE55 protein is differentially recognized by antibodies elicited during subclinical TB infection in animal models, as well as their peptide PPE-C in humans with clinical TB-HIV negative and months prior TB manifestation on HIV positive cases, suggesting been useful in different status of TB [11]. Therefore, the PPE proteins have the potential to be specific markers for the diagnosis of tuberculosis.
The rv3429, encoding the protein PPE59, has low homology in the C-terminal region with other mycobacterial species and was described to elicit low T cell immune response [12], but there are no clues about its ability to induce a humoral response. The PPE 59 bioinformatics' analysis revealed seven B cell epitopes, belonging to the different region 11 that is absent in M. bovis BCG strains and in other mycobacterial species [12; 13; 14]. The aim of this study was to clone and express the rv3429 gene coding PPE59 and to screen immunoreactivity to IgG, IgM, and IgA in sera of TB patients using an in-house ELISA method.

Cloning, expression and purification of recombinant PPE 59 protein
The rv3429 gene sequence encoding the PPE 59 protein was generated by PCR from the genomic DNA of H37Rv using upstream 5´GCG^GATCCATGCATCCAATGATA3´ and downstream 5 CA^AGCTTCTACCCGCCCCCGCCCCCGTA3´ primers, comprising BamHI and HindIII restriction endonucleases, respectively. The standard conditions for thermal cycle amplification was 94°C for 5 min; 25 cycles of 1 min each of 94°C, 60. 6°C, and 72°C; final step of at 72°C for 5 min. The cloning, expression and purification were performed as described previously with minor modifications on protein purification [15]. Briefly, the transformed expression system containing the pQE80L-rv3429 were better isopropyl-beta-D-thiogalactopyranoside (IPTG) 1mM induced at 37°C for 4h. The recombinant proteins were mainly present in the insoluble fraction, and therefore, the purification of His tag fusion protein was carried out under denaturing condition with 8M urea lyses buffer. The over expressed recombinant proteins were confirmed by probing with anti-His tag monoclonal antibody and SDS-PAGE analyzed. Solubilized protein was loaded into HisLink Protein Purification Resin (Promega, USA) and eluted at different concentrations (5mM, 20mM, 100mM and 250mM) of imidazole (Quiagen, EUA) buffer and of urea (8M e 6M). to 2012. All individuals were ≥18 years and HIV negative. TB diagnosis were based on clinical, radiographic and microbiological methods. Sputum microscopy (SM) performed using the hot Ziehl-Neelsen method and semi quantitative acid-fast bacilli analyzed as positive or negative. Chest X ray (CXRs) changes classified as low (normal or suggestive of pulmonary disease), intermediate (infiltrates) and high probability for TB (infiltrates and/or cavities) [17]. Patients who had negative microbiological tests but clinical and laboratory response to anti-TB treatment were diagnosed with probable TB. TB IT patients were tested by interferon-gamma release assay (IGRA) using Quantiferon Gold Tube® for M. tuberculosis infection detection.

Enzyme Linked Immunosorbent Assay (ELISA)
In order to analyze the individual immune response of IgG, IgM and IgA to the PPE59 protein, ELISA was standardized for each isotype using a pool of sera from 10 pTB BR and 16 HD BR , half-positive or negative TST. The tests were performed in microplates with 96 wells-flat bottom (Nunc, USA) coated with antigen at the 0.5 and 1.0µg/ml prepared in coating buffer carbonate/bicarbonate, pH=9.6 and incubated at 37°C for 2h. The test was performed as described previously [15] with minor modification. The 1:1000 diluted HRP-conjugate goat anti-human IgG and -IgM (Thermo Fisher Pierce, USA) and the 1:2500 diluted anti-human IgA were used. Sera aliquots used did not experience the freezer-thaw cycle before testing. Individual response of all samples was blind tested (the operator had no knowledge on samples characteristics). Taking advantage of the Brazilian sera previously were IgG tested with other mycobacterial antigens [16], we analyzed the results combining with those obtained for PPE59 IgG, IgM or IgA.

Data analysis
Diagnostic codes were open at the end of all tests, and healthy donors and ORD BR were used as reference control groups for receiver operating characteristic curve analysis (ROC), defining the cut off value for sensitivity and specificity calculation. Antibodies levels were analyzed as a function of medium absorbance and minimum and maximum. GraphPad Prism 7 software (San Diego, CA, USA) was applied for statistical analysis. As the data was not normally distributed, non-parametric Kruskal Wallis was performed to analyze significant differences in all groups and Mann-Whitney test was used for pairwise comparisons, while the Fisher exact test and Chi-Square test with Yates's correction was used to evaluate the significance of the difference between the numbers of positives of different serology tests. Accuracy was calculated over the total of truly positive and truly negative, relative to the sample studied, formulae: (true positive + true negative) / total number of populations studied.
Having as indicators of accuracy the values between 0 -40 % said as low accuracy, 50% -70% moderate and 80% -100% high accuracy. Odds ratios (OR) and logistic regression analyzes was used in the uni and multivariable analyzes, p values ≤ 0.05 were considered significant with their respective 95% confidence intervals. SPSS Statistics 18 (IBM Corporation) was applied for data analyze.

Characteristics of study subjects
The clinical, epidemiological and laboratory characteristics of the adults' participants are shown in table 1. The group I, HD BR participants was younger than the other patients were (p=0.01) and most of the patients were males, for both groups. Opposite to group II, the group I was significantly alcohol users (p=0.004). Contrary to the group I (65.3%), the majority of group II (88.6%) patients were diagnosed through positive SM and/or culture; only seven pTB BR and five pTB IT had probable TB.
Approximately 60% of the pTB BR showed CXR findings of high probability for TB, while in the Italians/immigrants' cases intermediary changes predominated (p=0.03). In group I, the presence of BCG scar was high among HD BR (100%), followed by ORD BR (49%) and pTB BR (38%), while the frequency of positive TST performed in the way back: 50.0%, 61.8%, and 78.8%, respectively.
Information on TST was not available to study group II, but Quantiferon® was performed in 53% of them, of which 44% eTB IT and 48% pTB IT were positive. Most of TB IT patients (84.8%) were already on treatment with first-line TB drugs, while TB BR were free of treatment, at blood withdrawn.

Cloning, expression, and purification of PPE 59 protein
The 534pb of rv3429 gene was successfully cloned into pQE80L and most of the expressed PPE 59 protein was present in the inclusion bodies of the pellet being purified under denaturing condition with urea 8M. The release of the target protein from the Ni-resin column with expected molecular weight (21 kDa) occurred in the eluate containing 6M urea and 100mM imidazole, and the concentration of the expressed protein was high, 700 µg/ml (figure 2). The protein proved to be stable under stored condition (4°C or 8°C), and the same batch was used for all immunological tests.
Purification process under denaturing conditions, incubated with mouse IgG anti-His. Lane 1, pre-

Detection of humoral responses by quantitative IgA, IgG and IgM ELISA PPE59:
The pTB IgM class serum levels were not significantly different from HD BR with positive or negative TST (p=0.17) and the results remained inconclusive even after testing the pools of sera individually, then it was not further analyzed. There were, however, significant differences for IgG and IgA responses (p=0.02) ( Figure 3). For further analysis, 1:50 serum dilution was applied in individuals' tests.
Was used pools of sera from 20 Brazilians individuals with pulmonary tuberculosis (•) and 8 healthy donors positive (■) or negative (▲) TST, respectively.
After compilation of all ELISA results, patients' clinical forms were open and, in general, sera of TB participants showed IgA medians higher that IgG in all groups studied (p< 0.0001). The PPE59 specific IgA medians in pTB sera was highly significant related to all controls (ORD/HD, p<0.0001). Stratifying the population by geographic area of origin, medians of absorbance slightly decreased from Brazilians  Table 2).

Humoral responses by qualitative IgA and IgG ELISA PPE59
Based on the differences observed in the quantitative analysis of the serologic test, we further performed a ROC curve analysis to evaluate its potential diagnosis. The value of the area under the curve (AUC) was higher for IgA (0.823) than IgG (0.467) as well as the cut-off of 1.030 (with LR of 21.81) and 0.732 (with LR of 1.043), respectively ( Figure 4). The predicted sensitivity at the high specificity was variable, which denote heterogeneity of antibodies response according to population tested (table 3). The pTB BR ELISA IgA sensitivity was 53.3%, decreasing toward pTB IT (38.2%) and eTB IT (28%), with an overall detection of 42.7%. The IgG sensitivity was 21.3%, 3%, and 0%, respectively, with an overall detection of 9%, at a lower specificity (83.1%) ( Table 3). IgG false positive was high among NTM and rCt ( Figure 5). The overall accuracy of ELISA PPE59 IgA to detect TB was moderate (61.8%) with higher positive predictive valor (PPV=96.5%) but low negative one (NPV=47.9%), while the low for IgG. A combination of positive scores obtained by both isotypes tested did not improve the accuracy, because only slightly increase the sensitivity jeopardizing the specificity (Figure 4).

TB Detection value of IgA and IgG PPE59 combined with results of five other antigens previously tested
Taking advantage of the Brazilians sera previously tested with other mycobacterial antigens [16], we analyzed the results combining with those for the present PPE59 IgA and/or IgG (Table 3). In the former study lower ELISA sensitivity for the single antigens ESAT-6 (22.5%), 16kDa (37.1%), MT10.3 (23.6%), MPT-64 (36%) or 38kDa (33.7%) IgG, with specificity ≤ 92%. However, the combination of positive results of PPE59 IgA plus ESAT-6 or 38 kDa IgG increased the sensitivity (55.8% or 59.6%), without jeopardizing the specificity (90.2% or 94.4%). The combination of PPE59 IgA with 16kDa-IgG reach the best sensitivity (71.1%), maintaining good specificity (87.4%). The other results of IgA and/or IgG combinations did not perform better (Table 4).

ELISA performance compared to smear microscopy results
Typically, although microscopy is not powerful enough when used solely as the sole screening and / or diagnostic method for TB it performed easily, rapidly and at a low cost. Thus, we combine both methodologies to improve the detection of patients with active TB. The SM plus ELISA PPE59-IgA results increased overall TB detection to 74%, while SM detects 70.8% (p=0.01). ELISA PPE59 IgG showed no improvement, performed lower to that SM alone (p=0.69). Stratifying the TB groups, the increase was significant in the Groups I for positive SM/IgA, improving sensitivity to 80.7% (p=0.01), once serology detects 12 negative SM pTB BR patients or test not applied (n=3). While in Group II slightly increased sensitivity detecting four of the negative SM eTB IT (p=0.26) and one pTB IT (p=0.82). Combining SM, PPE59 IgA and 16kDa IgG ELISA results (available only to Brazilians), there was an improvement in sensitivity, as those not detected by PPE59 IgA (n=10, of which 6 SM-and 4 SM not done) 16kDa IgG recognized 3 subjects (two SM-and one not done), bringing the sensitivity to 86.5% (p=0.001) ( Table 5).

Uni and multivariate analysis based on IgA and IgG ELISA PPE59
In tables 6 and 7, we present the association of the PPE59 reactivity and clinical and demographic

Discussion
The rv3429c gene, which encodes the PPE59 protein, is located in the RD11, a region absent from all BCG strains [18; 15] and corresponds to 5 ORFs in the Mycobacterium tuberculosis genome, coding also for the TB 7.7 antigen [18], which is used in the commercial interferon-gamma release assay for TB/latent infection identification [19; 20]. Moreover, the predicted location is the cellular periplasm, between the inner membrane and the cytoplasmic, and the successful cloning and high expression of the PPE59 protein described here justify its evaluation.
Does IgA PPE59 is a potential biomarker to compose a TB diagnostic tool? Over the decades, several mycobacterial antigens studied have revealed the potential for TB diagnostic testing including the use of multiple antigens, as recently 57 field-based multiplexed serological assays of M. tuberculosis, and reference assay based on 132 antigens applied to screen patients with suspected TB symptoms on two continents [26]. As a result, IgG performed similarly for three or multiple antigens, with a sensitivity of 35% at 90% specificity. However, our study, in response to a single antigen, achieved an overall result of 42.7%, performing better for the IgA than IgG, with higher specificity (97.5%) even involving sera of TB patients from different geographic areas and clinical forms of TB.
Combination of IgA PPE59 with other biomarkers improves sensitivity? It is noteworthy the sensitivity improvement in ELISA IgA results when combinations of PPE59 IgA plus ESAT-6 or 38 kDa IgG were analyzed, maintaining better or similar specificity to Broger et al (2017) study. Variability in the median levels of reactivity found by these authors, among Vietnamese and Peruvian samples, was also observed in our study, where the absorbance was higher in the Brazilian cohort, decreasing toward Europeans but without achieve statistical significance, except for IgG. However, it is important to note that the lack of proper control from a non-endemic area may limit the interpretation of our results. Our findings confirm that due to heterogeneity in the recognition of antigens by serum antibodies in TB, multi-cocktails of antigens or their peptides remain the best strategy to potentiate a diagnostic test accuracy; the major challenge is to find the best antigens and isotypes combination [13; 27]. Note that combination of PPE 59 IgA plus 16kda IgG bring the sensitivity to 74%, and adding the results of other rapid tests (SM = 66%) the sensitivity reaches 86.5% (Sp = 87.4%), making this combination a potential target product profile triage test in TB high burden area. However, this data based only on the Brazilian population requires further evaluation. In addition, the sensitivity of SM was significantly variable (53-78.2%, p = 0.002), however despite ELISA PPE59 was not powerful enough to equally diagnosing the TB, serology improved case finding, mainly among negative SM cases of pulmonary or extrapulmonary clinical form, based on IgA PPE 59 single antigen or multiple antigens and isotypes. Thus, could be explained by Ab production that may not dependent on bacillary load and produced in the body site of infection. Therefore, serology may be an adjuvant tool detecting TB paucibacillary cases and PPE 59 IgA advantage is the high specificity.
Does the antibodies' isotypes choice important in a serological test? Increased IgA levels were mainly found among patients' control group with COPD, smokers, and rCt. COPD in some studies is considered a risk factor for the development of pulmonary TB, and both diseases have common risk factors, such as smoking and low socioeconomic status [28]. Moreover, the two ORD individuals who were positive for PPE59 IgA had a history of recent contact with active TB patients; therefore, these participants could have been infected by M. tuberculosis and potentially progress for active TB.
Contrary to PPE59 IgA, NTM IgG cross-reactivity was higher; corroborating that beside antigens, the isotypes choice can be a big deal as TB biomarker.
To date, the actual function of PPE59 protein in metabolism and its evolutionary role in M.
tuberculosis is still unknown. The role of PPE59 protein as an inducer of cell-mediated response by IFN-γ and interleukin-1 was previously described [14; 15], but as a result of this present study, the dominance of IgA reactivity and lower-IgG in the sera of patients with TB was uncovered. In a study carried out in China and India, the PPE 57 (rv3425) and PPE 17 (rv1168) proteins, respectively, were recognized by IgG antibodies with high sensitivity and specificity (100%), even in extrapulmonary TB cases [11; 29]. However, opposite to our study, the control group was smaller (n = 20) and did not comprise ORD patients. These results may also suggest that PPE59 and PPE57 bacteria expression may changes under different host conditions since recognition occurs by different antibodies isotypes.
Comparing with other serological biomarkers associated with risk factor for TB. Studies with different antigens showed that IgG ESAT6/16 kDa/HBHA was increased in active TB (declining after treatment), and in latently infected contact, suggesting that specific antibodies were associated with the bacillary load [30]. While IgA-HrpA levels correlated significantly with active TB, the IgA-16 kDa/ESAT6 was significantly associated with the nutritional status of the participants. Others reported IgA antimycobacterial antigens HrpA and MDP1 higher response among controls than in TB patients suggesting that different stress on bacterium and host modulates proteomic expression, and the IgA immune response may be protective [31]. Williams et al (2004) reported IgA against the 16 kDa αcristalin mediated some protection against TB, as evidenced by reduced mycobacterial burden in the lungs [32]. However, IgA PPE 59 reactivity was associated with the presence of chest radiographic alteration of high probability of TB. Considering that in Brazil TB patients are most detected in the spontaneous demand of health care units and are usually diagnosed when the disease is more advanced, it is possible that PPE59 is preferentially expressed in later phases of the clinical evolution of TB disease [33]. On the other hand, TB patients alcohol user (both Brazilian and Italian/immigrants) are ≥ 2.7 times more likely to be positive to PPE59 IgA or IgG. Similar results were observed for 38 kDa and MPT-64 IgG responses tested previously in the same Brazilian population [16]. Louw et al. However, our study was not designed to specifically address the effect of alcohol in humoral response of TB patients, more studies are needed to elucidate these serological findings. Moreover, lower positive IgA PPE 59 frequency in IT patients may be related to treatment, as recently described for LppZ IgA low response after 2 months of anti-TB treatment [38], however, our study was not designed to follow up the treated patients. No other demographic or clinical characteristic was associated with Ab positivity among Italian/immigrants, especially for IgG, which is mainly due to the low number of individuals who were positive for the test.

Suggested PPE 59 IgA function
The IgA main function is to exert an effector action on the mucosal immune system and is an important defense against pathogens that invade the host via mucosa, such as M. tuberculosis [13; 36]. IgA exerts a variety of protective functions between which it interferes with pathogen adherence in mucosal cells (immune exclusion), promotes intracellular neutralization and excretion of the pathogen. Its monomeric form, present in the serum, exerts a complementary action, eliminating pathogens that have crossed the mucosal barrier and still interact with the Fc receptors of cytotoxic cells-dependent antibody [37]. Therefore, the presence of detectable PPE59 IgA in TB patients may be exerting a protective action, containing the bacilli that escaped the mucosal barrier, whose immune response best controls bacillary multiplication.
Limitations. Limitation of this study includes heterogeneity of specific treatment between study groups, no availability of proper control in the group II, not enough TB IT sera to test the other antigens and lack of follow-up immune response of treated patients. Further studies to evaluate the dynamic of PPE59 IgA during TB disease/treatment must be accomplished.

Conclusion
This is the first study showing PPE59-IgA immunodominance than IgG and no IgM among TB patients of different geographic areas at high specificity and slightly increased sensitivity when in combination with ESAT-6, or 38 kDa or 16 kDa IgG assays lowering specificity. Notably, the combination of both rapid teste SM and ELISA PPE59 IgA (and 16kDA IgG) sensitivity and specificity, although imperfect, fit for triage test. Moreover, significant PPE59 IgA reactivity was strongly associated with the presence of CXR alterations of high TB probability (p = 0.001) and with alcohol use (p = 0.04). Although PPE59 IgA is not enough accurate to be used independently, this biomarker may be an adjuvant tool for detect symptomatic TB suspects and together with other antigens and/or isotypes and SM tests may be prospectively evaluated as a potential target in high TB burden areas. Additionally, it deserves further investigation on the positive polarization of immune response.

Ethics approval and consent to participate
The study was approved by the Fiocruz Ethics Committee in Research (No. 560-10), Rio de Janeiro, Brazil and by the Ethical Committee of Azienda Ospedaliera Spedali Civili di Brescia, Italy (N. 196/03 -11). Before recruiting patients, a detailed explanation was given and written informed consent was obtained from them.

Consent for publication
Not applicable.

Availability of data and materials
All data generated or analyzed during this study are included in this published article.

Competing Interest
The authors declare not have competing interests.        Figure 1 Flowchart of patients groups enrolled in the present study.

Figure 2
Western blot of recombinant PP59 protein.

Figure 3
The in house IgA, IgG and IgM-ELISA standardization with the recombinant PPE 59.