Pattern recognition and cellular immune responses to novel Mycobacterium tuberculosis-antigens in individuals from Belarus

Tuberculosis (TB) is one of the major global health issue, and a serious health concern in Belarus where the prevalence and spread of multidrug-resistant (MDR) TB and extensively drug resistant (XDR) TB has increased during the last few years. About 5000 individuals are newly diagnosed each year in Belarus and the prevalence of TB is 52/100,000 individuals [1]. A recent study identified MDR TB in 35.3% of newly diagnosed patients and in 76.5% of individuals who have previously been treated. XDR TB could be identified in 14.0% among the patients diagnosed with MDR TB [2]; these are alarming levels of resistant TB in Belarus.

Early diagnosis of the disease and the rapid identification of resistance to primary anti-TB drugs are essential for efficient treatment, prevention and control of TB.

The diagnosis of TB in many countries, including Belarus, still relies on the tuberculin skin test (TST) and direct sputum examination by light microscopy. The TST has low specificity due to cross-reactivity to protein purified derivative (PPD) antigens shared by environmental mycobacteria species, and may give false positive responses in Bacillus Calmette-Guerin (BCG) vaccinated individuals. BCG policies vary considerably between countries, primarily depending on the current epidemiological situation. Individuals in Belarus, as well as in Russia may be vaccinated three times with BCG; the pattern of cellular immune responses and antigen recognition after several BCG vaccinations has not been analyzed in detail up to now. BCG vaccination takes place early in life, BCG re-vaccination takes place during school years, a third vaccination is considered based on tuberculin skin tests (TST); BCG re-vaccination may also be postponed in adults (up to 30 years) in areas of low TB prevalence [3, 4].

Interferon-γ release assays (IGRAs) have been designed to overcome the problem of cross-reactive T cell immune responses by measuring immune responses to antigens specific for M. tuberculosis (M.tb). QuantiFERON-TB Gold In-tube (QFT-GIT) [5] measures INF-γ release by sensitized T cells after stimulation with peptides from the early secreted antigenic target 6 (ESAT-6), culture filtrate protein-10 (CFP-10) and TB-7.7 which are absent in BCG and in most environmental mycobacteria [6, 7]. Neither the TST nor the QTF-GIT, however, is able to discriminate between active TB-disease, latent TB-infection (LTBI) and previous TB-infection. Exposure to Mycobacteria other than tuberculosis (MOTT) may lead to false-positive results, and poor specificity of the tests may lead to unnecessary prophylactic treatment with anti-tuberculosis drugs. Thus, the ideal diagnostic test should not only discriminate LTBI from active TB, but also discriminate between TB, MOTT and previous BCG vaccination. This is of particular concern in Belarus since TB treatment can be triggered by a positive TST. In contrast, a negative TST may lead to repeated BCG vaccination according to national guidelines, since TST diameter is also used to gauge treatment responses, in addition to standard clinical and microbiological evaluation.

Although LTBI is clinically silent and not contagious, it can reactivate to cause contagious pulmonary TB [8]. Tubercle bacilli are generally considered to be non-replicating in LTBI, yet it may slowly grow and replicate [9]. LTBI is characterized by highly reduced bacterial metabolism and a significantly altered gene expression [10–12] associated with the different stages of infection [13–15]. Since most cases of active TB arise in people with LTBI, there is an urgent need to identify new potential targets for TB diagnosis and for the development of an improved TB vaccine.

One of the strategies in developing new diagnostic methods and in improving the TB vaccine involves the identification of epitopes in antigens that induce T cell responses. We have previously scanned 61 proteins from M.tb proteome, which facilitated the identification of potential targets to elicit T cell responses, and showed that there is a difference in peptide recognition pattern between M.tb-infected patients and healthy controls [16, 17]. Some of these peptides were also tested for binding to the most frequent Caucasian and African MHC Class II alleles (HLA-DRB1*0101, DRB1*1501 and DRB1*0401) in different populations [17], they may serve as potential strong B and also T cell targets. The aim of this study was to compare T cell responses and INF-γ production to these candidate proteins associated with different stages of M.tb infection, expressed by replicating versus non-replicating bacteria.

We used pools of several MHC class II-binding peptides from M.tb for their ability to stimulate IFN-γ production in vitro using whole blood from Belarusian individuals with a history of previous TB, patients with active pulmonary TB and healthy individuals (most likely TB exposed). In addition, 17 non-BCG vaccinated healthy individuals from Sweden were also included in the study. The selected peptides have shown to be expressed in replicating bacteria (Rv0477c, Rv2940c, Rv1690, Rv1085c, Rv1886, Rv2962c, Rv2958c, Rv2957c), non-replicating bacteria (Rv2453c, yet also Rv0477c, Rv2940c) and are suggested to be able to differentiate between M.tb, MOTT and BCG (Rv3019, Rv0066c, Rv3347c). Some peptide pools (glycosyl transferase) are more frequently recognized by T cells from healthy individuals as compared to TB+ patients suggesting that these were sensitized to mycobacteria. In addition, individuals recovered from TB recognized a broader panel of peptide pools compared to the other two groups (healthy or TB+)which indicates that the responses to M.tb specific antigens apparently persist in individuals who had recovered from pulmonary TB (i.e. long-term memory cellular responses). Of note, the analysis presented in the current report is based on IFN-γ production in PBMCs. The simultaneous measurement of additional chemokines/cytokines, including CXCL10 and IL-10 will most likely increase the discriminatory power and biological value of immunological readouts to M.tb antigens [26].

One of the strategies in developing new diagnostic methods improving TB vaccine design involves the identification of biomarkers for protection and novel T cell targets. Of particular interest is glycosyltransferase, involved in cellular metabolism and lipid formation. It has been shown that glycosyltransferase genes contribute to M.tb survival in macrophages [27]. Interestingly, the peptide pools from Rv2957c, Rv2958 and Rv2962c were more commonly recognized by T cells from healthy (most likely TB exposed) individuals as compared to TB-patients (Figure 1, see Additional file 3: Table S3), yet larger clinically well defined cohorts are needed in order to provide biologically relevant biomarkers to study M.tb immune responses and to test for novel M.tb vaccine targets.

Other peptide pools from M.tb proteins involved in lipid formation and synthesis of long fatty acids, Rv0477c (Cyclopropane fatty acyl phospholipid synthase) induced IFN-γ production in the majority of PBMCs from individuals suffering from TB (73%) (Figure 1). Since mycolic acids represent a major constituent of the mycobacterial cell wall complex, they provide the first line of defense against potentially lethal environmental conditions. Slow growing pathogenic mycobacteria such as M.tb modify their mycolic acids by cyclopropanation which is a common membrane modification in bacteria and plants [28, 29] and essential for viability, drug resistance and cell wall integrity [30] playing an important role in M.tb pathogenesis [31, 32].

Chemical inhibition of mycolic acid methyltransferases has shown to be lethal to M.tb and causes alterations in cell envelope structure and drug susceptibility [30]. Thus, Rv0477c may present an interesting target for development of new antibiotics against M.tb. Detectable cellular responses to Rv0477c, although limited, were also found in PBMCs from healthy Swedish controls. It is possible that immune recognition of this antigen is partially primed by exposure to or infection with MOTT since mycolic fatty acids are also present in MOTT. Cyclopropane fatty acids are also present in many other bacterial species expressed in stationary phase cells, including E.coli [29]; structural homologies may be responsible for cross-reactive T-cell responses. Of interest, a recent report showed that mutations in the rpoS-regulated genes in E.coli (i.e. cfa, cyclopropane fatty acid synthetase and osmB, outer membrane lipoprotein) are significantly higher sensitive to (oxidative) stress and impaired in membrane repair mechanism [33].

This notion is supported by our observation that constitutive and antigen-specific IFN-γ, IL-2 and TNFα production can be detected by ICS in CD4+ and CD8+ T-cells from NHPs prior to BCG vaccination, yet that BCG lead to a strong expansion of polyfunctional T cells. Future experiments, using T cells from animals housed under sterile conditions may show the contribution of MOTT and their role in providing an immunological matrix which can be boosted by TB vaccination strategies.

While there are no precisely defined correlates of protection against mycobacterial infection at present, IFN-γ producing Th1 cells are believed to be essential in protection against TB [34, 35]. In this study we focused on INF-γ and showed that there is a difference in IFN-γ production between the study groups. At this point, we cannot determine if the IFN-γ production observed in this study was due several BCG vaccinations, due to contact with M.tb in Belarus, or - mutually not exclusive - to related target structures present in other bacterial species.

Rv3804c, Rv1886c and Rv0288 present in M.tb and BCG have been extensively used in different vaccine trials. In general, Rv3804c was most commonly recognized in each group tested. Also several Swedish subjects responded to this antigen. Furthermore, IFN-γ production was higher in response to proteins compared to the peptides (Figures 1, 2, 3, 4). Differences may be due to i) peptide instability, ii) differential antigen processing and presentation and iii) the fact that the peptide pools did not cover the entire protein antigen. These differences may also, in part, explain the strong recognition of the recombinant proteins from the PPE family members Rv0978, Rv0754 and Rv1971c in patients with TB [18–21] in PBMCs from individuals who recovered from TB and also in healthy individuals with no record with TB (yet most likely exposed, since these individuals are health care workers) (Figure 2). The same was true for target recognition from T cells obtained from individuals (TST-, non-BCG vaccinated) from Sweden (Figure 4). These antigens have been reported to provide epitopes differentiating humoral immune responses in individuals with TB. Yet more detailed experiments showed that Rv0978 and Rv0754 recognize TLR2 and induce maturation and activation of human DCs [19]. They enhance the ability of DCs to stimulate CD4+ T-cell responses and activate the ERK1/2, p38 MAPK, and NF-kappaB signaling pathways in DCs and may be instrumental to shape the quality of the innate cellular immune response in exposure to mycobacterial species. This may, in part, explain the strong IFN-γ production in response to the M.tb proteins; corresponding peptide cocktails may not provide these signals leading to DC maturation.

Differences in M.tb antigen recognition may be associated with i) the 'genetic makeup' of the test population, ii) alternate M.tb target molecules in circulating strains, iii) extent of TB disease, previous exposures to M.tb and iv) cross-reactive T-cell responses directed against closely related targets from other bacterial species: only a clinically well defined study population, along with genetic M.tb strain [36, 37] and host immunological marker analysis will aid to visualize a more realistic pattern of anti-M.tb directed immune responses.

We also compared in the current study the cellular immune responses to M.tb-antigens in healthy individuals from Sweden and from Belarus. The number of responders to each antigen was higher and the IFN-γ production was stronger in the group of individuals from Belarus. The difference seen between the groups is most likely a reflection of the environment since Sweden is a low TB endemic country in contrast to Belarus where TB burden is high [2]. Furthermore, individuals from Belarus received two to three BCG vaccinations, which may alter the immune responses to M.tb proteins. None of the Swedish controls were BCG vaccinated and had no identified risk of exposure to TB.

TST is the primary method for diagnosis of latent TB infection in many countries but is not able to distinguish between BCG vaccination and reactions caused by M.tb infection itself. In the present study, viable M.tb bacilli could be detected in some patients with a positive AFS, while the immunological assays used (TST and QFT-GIT) provided negative results (Table 2). In three culture confirmed TB cases, the IGRA tested negative. All the TST positive individuals were positive for QFT-GIT within TB-positive cases. The discrepancy seen between the immunological and bacteriological assays is most likely due to not fully functional T cells from patients with TB. We did not assess the role of different HLA backgrounds in response to selected peptides in this study. Although most of the people in Belarus are Caucasian, it is still likely that some individuals may not have responded to selected target peptide pools.

We show that PBMCs from healthy individuals from Belarus and those recovered from TB recognized a broader panel of M.tb-antigens as compared to TB-patients, defined by IFN-γ production. Individual M.tb antigens were strongly recognized by polyfunctional T cells in PBMCs from NHPs housed in a controlled environment after BCG vaccination. The data show that screening with selected peptide pools aids to identify biologically relevant M.tb targets, e.g. cyclopropane fatty-acid synthase and glycosyl transferase, which may have potential as targets for TB-diagnosis and to measure BCG vaccine take.