Mycobacterium tuberculosis infection is associated with increased B cell responses to unrelated pathogens

Antigens from Mycobacterium tuberculosis (M.tb), have been shown to stimulate human B cell responses to unrelated recall antigens in vitro. However, it is not known whether natural M.tb infection or whether vaccination with, Mycobacterium bovis BCG, has a similar effect. This study investigated the effects of M.tb infection and BCG vaccination on B cell responses to heterologous pathogen recall antigens. Antibodies against several bacterial and viral pathogens were quantified by ELISA in 68 uninfected controls, 62 individuals with latent TB infection (LTBI) and 107 active pulmonary TB (APTB) cases, and 24 recently BCG-vaccinated adolescents and naive controls. Antibody avidity was investigated using surface plasmon resonance and B cell ELISPOTs were used to measure plasmablast and memory B cell responses (MBC) in APTB cases and healthy donor controls. APTB was associated with higher levels of antibodies to respiratory syncytial virus and measles virus, compared to uninfected controls. BCG vaccination did not alter levels of antibodies against heterologous pathogens. Tetanus toxoid (TT)-specific antibody avidity was increased in APTB cases in comparison to uninfected individuals and the ratio of TT-specific plasmablasts to MBCs in the APTB cases was 7:1. M.tb infection is associated with increased antibody responses to heterologous pathogens in human subjects.

www.nature.com/scientificreports/ from infectious diseases among low birth weight infants who were given BCG vaccine at birth 5,6 . These effects are thought to be as a result of improved function of innate immune cells brought about by cellular epigenetic modifications induced by BCG 7 . There may also be non-specific effects on the adaptive immune system. Ota et al. 8 described higher hepatitis B virus-specific antibody responses in infants who were given BCG in addition to hepatitis B vaccination at birth, compared to those who were only vaccinated with hepatitis B. A more recent study by Ritz et al. 9 reported that the level of antibodies elicited by pneumococcal vaccination of infants at 2, 4 and 6 months of age were higher in those infants who were given BCG at birth compared to those who were not. BCG may therefore improve immune responses to vaccines given at the same time as BCG or those given following BCG vaccination, in a non-specific manner.
The studies highlighted above describe the effect of mycobacteria on responses to concurrently administered vaccine antigens or those given after vaccination with mycobacterial preparations. However, studies in the past have shown that purified protein derivative (PPD) from M.tb can stimulate secretion of antibodies against measles, rubella and herpes simplex viruses from human peripheral blood mononuclear cells (PBMCs) in vitro 10 . This finding suggests that M.tb may be able to enhance antibody/B-cell memory responses generated from previous exposure to unrelated pathogen-derived antigens. The exact immunological mechanism underlying this observation is not yet known; however, it is possible that Mycobacterium antigens could non-specifically activate pathogen-specific memory B cells (MBCs), resulting in the expansion of antibody-secreting cells and a subsequent rise in antibody levels. Human MBCs are prone to activation by polyclonal stimulation; studies by Bernasconi et al. 11 have shown that stimulation of these cells by bacterial CpG (cytosine-phosphate-guanine) DNA or by T cell cytokines can lead to their proliferation and expansion into antibody-secreting cells. However, this hypothesis has not been investigated in regard to Mycobacterium tuberculosis exposure. Furthermore, it is not known whether natural M.tb infection has the same non-specific stimulatory effect on antibody responses to recall antigens from heterologous pathogens. We also do not know whether recent vaccination with the related species, M. bovis BCG has a similar effect on serological recall responses to unrelated antigens.
This study characterised antibody responses to heterologous pathogen recall antigens in uninfected controls, individuals with a latent TB infection (LTBI) or active pulmonary TB cases (APTB) cases participating in a TB household contact study in Uganda, as well as adolescent recipients of the BCG vaccine and their age-matched naïve controls from the United Kingdom (UK). Additionally, polyclonal activation of MBCs was explored as a possible mechanism by studying frequencies of tetanus toxoid (TT)-specific plasmablasts and MBCs in the APTB cases and healthy donors.

Results
Demographics. The characteristics of the Ugandan TB household contact study participants have been reported previously 12 and are shown in Table 1 below. In summary, APTB cases and individuals with LTBI were older than the uninfected individuals and there was a larger proportion of males in the APTB group in comparison to the other two groups. There were also more HIV positive individuals and a larger proportion of people of low socioeconomic class among the APTB group.
The adolescents studied in the UK were all aged between 12 and 13 years, had no previous BCG scar and were tuberculin skin test negative 13 . M.tb infection state is associated with increased antibody responses to unrelated pathogens. Antibody responses were compared across groups of uninfected individuals, individuals with LTBI and APTB cases from Uganda to determine the impact of M.tb infection state on antibody responses to heterologous pathogens. Since we were interested in recall responses, we studied antibodies against antigens contained in childhood vaccines, such as TT, diphtheria toxoid (DT) and measles virus (MV). We also measured antibodies specific for infections that our participants may have had a high chance of encountering in childhood in this setting. These included respiratory syncytial virus (RSV) 14 , adenovirus 15 , Kaposi's sarcoma herpes virus (KSHV) 16 , cytomegalovirus (CMV) 17 and Epstein-Barr virus (EBV) 18 . M.tb PPD was included as a positive control indicative of mycobacterial exposure. Initial crude analyses revealed evidence of differences in the levels of antibodies specific to PPD, DT, RSV, MV, KSHV ORF73, KSHV 8.1 and CMV antigens across the groups (Fig. 1). There were also differences in total IgG antibody concentrations ( Supplementary Fig. S1). There was no evidence of differences in EBV and adenovirus-specific antibodies across the groups ( Supplementary Fig. S1).
In the final analyses, associations between M.tb infection status and antibody responses were determined by means of linear regression analysis using the uninfected controls as the baseline comparison group while www.nature.com/scientificreports/ adjusting for the effects of age, gender, HIV infection status and socioeconomic status. APTB was associated with higher levels of anti-PPD, anti-RSV and anti-MV antibodies compared to the uninfected individuals (Table 2). We observed evidence of increased antibodies against TT, DT and KSHV K8.1 antigens ( Table 2) and total IgG levels in APTB (Supplementary Table S1), however this was not statistically significant after correcting for multiple testing. There were no associations between M.tb infection status and anti-EBV or anti-adenovirus antibodies, in agreement with findings from the crude analyses (Supplementary Table S1). We also observed positive associations between LTBI and anti-DT, anti-TT and anti-PPD antibodies but these did not reach the threshold of statistical significance.
BCG vaccination is not associated with increased antibody responses to unrelated pathogens. Next, we determined whether recent vaccination with the related species, Mycobacterium bovis BCG, was also associated with elevated antibody responses to unrelated recall pathogen antigens. Antibody responses were evaluated in a UK cohort of adolescents comprising a test group of 12 individuals who received BCG, and 13 age-matched BCG naïve controls, and the two compared at timepoints before and 3 weeks after vaccination of the test group. This cohort allowed us to compare BCG vaccinated individuals to a BCG naïve group, which we could not do in the Ugandan participants because BCG is routinely given to infants as part of the Uganda National Expanded Programme on Immunisation 19 . We observed no differences in antibody responses to any antigen, including PPD, between the BCG vaccinated individuals and naïve controls at the 3 week time point (Fig. 2b). These results mirrored non-significant differences observed between the two groups before BCG vaccination (Fig. 2a). The fold change in antibody over the three week period was also compared between BCG vaccinated individuals and naïve controls and no statistically significant differences were observed ( Supplementary Fig. S2). These findings indicate that BCG vaccination may not be able to boost antibody responses to heterologous pathogens, or at least not within the 3 week time frame.
Avidity of tetanus toxoid-specific antibodies is increased in APTB. We determined whether the observed increases in antibody responses to unrelated recall antigens in M.tb infected individuals was accompanied by changes in their avidity or overall binding strength. An increase in avidity could point to MBCs being the source of heterologous pathogen antigen-specific antibodies because these cells would have already undergone affinity maturation. To this end, the avidity of antibodies against TT and measles virus haemagglutinin (MVHA) antigens were investigated using surface plasmon resonance (SPR). Due to limitations in sample volumes, only 88 of the 237 KTB study participants' samples were tested.
Initial crude comparisons showed marginal evidence of a difference in dissociation rate (the measure of avidity) in TT-specific antibodies across the three groups ( Supplementary Fig. S3). For the final analysis, linear regression was used to evaluate associations between M.tb infection state and antibody avidity using the www.nature.com/scientificreports/ uninfected controls as the baseline comparison group while adjusting for the effects of age, gender, HIV infection status and socioeconomic status (Table 3). APTB was associated with the presence of TT-specific antibodies of a slower dissociation rate, a sign of higher avidity. A similar positive association was observed with LTBI but this was not statistically significant after correcting for multiple testing. We also observed similarly weak evidence to suggest that APTB was associated with MVHA specific antibodies of a slower dissociation rate (Table 3). These results lend credence to the hypothesis that the increased antibody responses against heterologous pathogen recall antigens observed in M.tb infection arise from antigen-specific MBCs.
Tetanus toxoid-specific plasmablasts are higher than tetanus toxoid-specific memory B cells in APTB. In order to explore the possibility that M.tb infection may be driving differentiation of MBCs specific to unrelated recall antigens into plasmablasts in our study participants, we examined the frequency of TTspecific MBCs and plasmablasts in 48 APTB cases and 115 healthy donors using B cell ELISPOT assays. We had limited PBMC numbers from the APTB cases and so MBC frequencies were ascertained in 18 of the APTB cases while plasmablast frequencies were evaluated in the rest (n = 30). The ratio of the median frequencies of TT-specific plasmablasts to MBC in the APTB cases was 7:1 (8.75:1.25; Fig. 3). The high abundance of TT-specific plasmablasts relative to MBCs in the APTB cases provides evidence of polyclonal activation of Table 2. Associations between Mycobacterium tuberculosis infection status and concentrations of IgG antibodies to heterologous pathogen antigens. Linear regression analysis of antibody data from 67 uninfected controls, 62 individuals with LTBI and 89 APTB cases. Statistically significant associations following adjusting for multiple testing are shown in a bold font. GMR geometric mean ratio, LTBI latent tuberculosis infection, APTB active pulmonary tuberculosis, PPD purified protein derivative, TT tetanus toxoid, DT diphtheria toxoid, RSV respiratory syncytial virus, MV measles virus, KSHV Kaposi's sarcoma herpesvirus, CMV cytomegalovirus. § Adjusted for age, gender, socioeconomic status and HIV infection status. ƒ q values were calculated using the Bonferroni method for comparisons between M.tb infection states and uninfected individuals.

Antibody optical density
Adjusted GMR (95%CI) § p value Q value ƒ Anti-PPD www.nature.com/scientificreports/ antigen-specific MBCs in APTB. This process may be driving their differentiation into plasmablasts and could explain the increased levels of antibodies against heterologous pathogen recall antigens observed in active TB disease. Efforts to determine TT-specific plasmablast to MBC ratio in the healthy donors for comparison were not possible because TT-specific plasmablasts were not readily detected during preliminary analyses. However, high frequencies of TT-specific MBC were detected in these individuals at levels above those seen in the APTB  www.nature.com/scientificreports/ cases (Fig. 4). Although TT-specific MBCs have been shown to remain in circulation years after vaccination, TT-specific plasmablasts are short-lived, disappearing within 2 weeks after antigen exposure 20 . This may explain the lack of detectable TT-specific plasmablasts in healthy donors.

Discussion
This study highlights for the first time a possible role for natural M.tb infection in the boosting of antibody immunity to heterologous pathogens that individuals may have been previously exposed to, through infection or immunisation, and shows that vaccination with the related bacterium, Mycobacterium bovis BCG does not have a similar effect. The increase in the levels and avidity of these antibodies in M.tb infected individuals may mean that MBCs are their likely source. Additionally, the abundance of TT-specific plasmablasts relative to TT-specific MBCs in active TB disease provides evidence to support the hypothesis that M.tb infection drives polyclonal activation of MBCs generated against unrelated recall antigens. If this property can be exploited, it could lead to the production of immunological therapies than can be used to boost serological memory to a wide variety of infections, particularly in people with waning immunity such as the elderly. The fact that the increase in antibody responses to heterologous pathogens was most evident in APTB disease may mean that this non-specific effect is most pronounced in an environment of actively replicating tubercle bacilli. Indeed, previous research has shown that active TB disease is associated with increases in CMV-specific  www.nature.com/scientificreports/ antibodies 21,22 , however, these studies attributed the rise in antibodies to concurrent infection with CMV rather than as a marker of serological memory. A study by de Paus et al. 23 observed higher levels of influenza-specific antibodies in individuals with TB and recognized that in addition to concurrent infection there was a possibility that these may have arisen as a result of mycobacterial driven non-specific increases in antibodies. The observation that BCG vaccination did not have a similar effect may mean that proteins unique to M.tb may mediate the increase in antibody responses to these heterologous pathogens observed in APTB. Alternatively, this may point to the effect of BCG mycobacterial load at the site of inoculation, which may fall short of the M.tb bacillary load observed in the lungs during active TB disease. The lack of significant stimulation of anti-mycobacterial antibodies in the BCG vaccinated recipients, in contrast to the more than threefold rise in the same in the APTB cases (Table 1) supports this reasoning. The lack of antibody response may also simply be due to the timing after vaccination. A later time point may have allowed the accrual of BCG-specific antibodies; we were not able to assess this in our study.
Antibody responses to recall antigens from heterologous pathogens were increased in both level and avidity in M.tb infection. This gave further evidence that the source of these antibodies are MBCs because these cells would have already undergone affinity maturation, and as a consequence produce antibodies of high avidity. A study by Zimmerman et al. 24 argues that the primary source of anti-mycobacterial antibodies in TB disease are MBCs, as demonstrated by the high level of somatic hypermutations in antibody-secreting cells isolated in this disease state. Tubercle bacilli may drive the activation of MBCs specific to a wide variety of pathogen antigens including those specific for M.tb. A similar phenomenon has been demonstrated for other infections such as malaria 25 , hepatitis C 26 and hantavirus pulmonary disease 27 . Bernasconi et al. (2006) have previously shown that MBCs are easily activated by polyclonal stimuli such as bacterial DNA and T cell cytokines produced following vaccination. Additionally, innate immune cytokines such as IFN-α and IL-6 promote the activation of B cells, their differentiating into plasmablasts and their secretion of antibodies 29 . It is, therefore, possible that mycobacterial products and the cytokine milieu present during human M.tb infection may activate MBCs leading to their differentiation to antibody-secreting cells. Our B cell ELISPOT data illustrated an abundance of TT-specific plasmablasts relative to TT-specific MBCs in APTB. Frequencies of TT-specific MBCs were lower in APTB cases in comparison to healthy donors while TT-specific plasmablasts were not readily detected in the latter. These observations further support the hypothesis that M.tb drives MBC polyclonal activation. It may be important to assess the impact of anti-mycobacterial drug treatment in APTB cases on antibody responses to heterologous pathogens in future studies. A reduction in antibody responses following treatment may further support our hypothesis. It is not clear how M.tb may benefit from increasing recall antibody responses to other pathogens. However, we speculate that this process may aid in the survival of M.tb within the host by limiting host death as a result of infection by unrelated pathogens.
It is unclear as to why APTB was only significantly associated with antibody responses to some pathogen recall antigens and not others. This may relate to varied rates of exposure to the different infections over time. Multiple recent exposures to some of the pathogens investigated could generate new antibody responses which may mask any "benefit" gleaned from mycobacterial driven MBC activation. Certain pathogens such as EBV has been shown to establish latency in MBCs 30 . This virus has also been shown to increase proliferation and activation of B cells and add to the effect of known TLR agonists such as CpG 31 . This would support the idea of increased EBV antibody responses in M.tb infected individuals, however, this was not observed in our study.
This study has a few limitations. We assumed that the antibody responses detected against the chosen heterologous pathogen antigens in our study participants arose as a result of past exposure via infection or vaccination. However, recent or current exposure to these multiple pathogens, although unlikely, was possible. The individuals included in the study were of varied, age, gender, socioeconomic and HIV infection status reflecting common demographic characteristics in Ugandan populations. HIV infection is known to significantly alter the immune responses and was strongly associated with M.tb infection status (Table 1). We were able to adjust for the effects of HIV, as well as the other measured potential confounders in the regression analyses to reduce their impact on the associations. Further analyses performed in only HIV negative individuals revealed similar associations as those observed in the total study population (Supplementary Table S2 and S3). Our findings are mostly based on IgG responses in our study population, however, it may be important in the future to look at other relevant antibody isotypes such as IgM. Individuals able to resist M.tb infection despite exposure to TB have been shown to have IgM in addition to IgG anti-mycobacterial antibodies 32 . The presence of IgM antibodies in sera has also been associated with increased inhibition of infection of epithelial by M.tb 33 .
On the whole, these data point to a hitherto unknown role for M.tb in the maintenance of serological memory generated against past infections and vaccinations. It has been hypothesised that exposure to infections in later life could help maintain antibody responses to a broad set of pathogens in the absence of their specific antigens. This process is proposed to be due to polyclonal stimulation of antigen MBCs 11,28 and the data presented here supports this. However, further experimentation is required to confirm the mechanism involved and identify the specific M.tb antigens mediating this process. This may involve in vitro cell culture assays using various mycobacterial stimulants and B cell ELISPOTs to measure antibody secreting cells. This research may aid in the development of therapeutics to boost antibody immunity to heterologous pathogens. It is envisaged that this treatment could be useful in ageing populations in whom antibody responses to specific infections are known to wane.

Materials and methods
Study populations and design. A cross-sectional study design was used to determine whether there were associations between M.tb infection status and antibody responses to heterologous pathogen recall antigens in study participants from a TB household contact study based in Uganda investigating the effect of coinfections in humans on their susceptibility to infection with M.tb 34  www.nature.com/scientificreports/ (HHCs) were recruited from the suburbs of Kampala. The APTB cases were all adults over 18 years of age and had either only recently began anti-TB treatment or received treatment for less than 4 weeks. APTB was ascertained using acid fast bacilli sputum microscopy. Individuals with LTBI were identified from among healthy HHCs using the tuberculin skin test and QuantiFERON-TB Gold in-tube (QFT-GIT) test. The HHCs that tested positive on both these tests were classified as having LTBI whereas those that tested negative on both tests were classified as uninfected controls. A total of 68 uninfected controls, 62 individuals with LTBI and 107 APTB cases were included in our study. QFT-GIT nil supernatant samples from these individuals were used for the antibody analyses. The choice of QFT-GIT supernatants was because serum samples were not available for all the individuals studied. We have previously shown that there are no differences between antibody concentrations obtained from assaying serum and those from assaying QFT-GIT supernatants 35 . Stored plasma samples from adolescents participating in a UK based prospective study on BCG responses were used to determine whether BCG vaccination was associated with non-specific increases in antibody responses to heterologous pathogen recall antigens. These individuals were part of a UK government programme on BCG vaccination in schools in the year 2005 and were invited to take part in a research study that sought to identify immune responses involved in BCG mediated protection against TB. The adolescents studied included 12 subjects who were given BCG and then followed up after three weeks as well as 13 age-matched controls who did not receive BCG and were followed up for the same period of time. All the subjects had no previous BCG scar (a sign of earlier BCG vaccination) and had a negative TST prior to immunisation 13 . Samples obtained from these individuals before BCG vaccination and at the 3 week timepoint were analysed. This cohort was chosen because it gave us the unique opportunity to compare BCG vaccinated and unvaccinated controls, something which could not be easily done in Ugandan cohorts considering BCG is part of the immunisation schedule in the country 19 .
In order to investigate MBC polyclonal activation as a possible cause of the elevated antibody responses to heterologous pathogens in people with TB, cryopreserved PBMCs from 48 APTB cases from the TB household contact study and fresh PBMCs from 115 healthy donor controls were analysed to determine the effect of an active TB infection on the frequencies of TT-specific plasmablasts and MBCs. The healthy donors were recruited from a voluntary HIV counselling and testing clinic at the Uganda Virus Research Institute. They were all HIV negative, aged 18-55 years and found to be healthy by a clinician.
These studies were all exploratory and the sample size was determined by the availability of sufficient samples and reagents for laboratory analyses.
Laboratory assays. In-house antigen-specific IgG ELISA. Immulon 4 HBX microtiter plates (Thermo Scientific, USA) were coated with 50 µl/well of 5 µg/ml DT, 5 µg/ml TT (both from National Institute for Biological Standards and Control), 5 µg/ml M.tb PPD (Serum Statens Institute, Denmark), 5 µg/ml RSV antigen, 5 µg/ml measles grade 2 antigen, 5 µg/ml adenovirus grade 2 antigen, 1.25 µg/ml Epstein-Barr viral capsid (all from Microbix, Biosystems, Canada) or 0.1% (w/v) skimmed milk (control for nonspecific binding) in carbonatebicarbonate buffer (pH 9.6) at 4 °C overnight. Following overnight incubation, the plates were washed four times with 1 × PBS (pH 7.4) containing 0.05% Tween 20 (PBS-T). The plates were then blocked with 150 µl/well 1% (w/v) skimmed milk in PBS-T for 2 h at room temperature. A dilution of 1 in 100 of each sample in 0.1% skimmed milk PBS-T (assay buffer) was made and 50 µl was then added to the antigen-coated and control wells. After an incubation of 2 h at room temperature, the wells were washed as before and incubated with 50 µl/well polyclonal rabbit anti-human IgG conjugated with horseradish peroxidase (Dako, Denmark) at 0.5 µg/ml for another hour at room temperature. Plates were then washed four times and enzyme activity detected by incubation with 100 µl/well o-phenylenediamine (Sigma) containing hydrogen peroxide for 15 min at room temperature in the dark. The reaction was stopped by addition of 25 µl/well 2 M sulphuric acid and thereafter the optical density (OD) measured at a test wavelength of 490 nm and a reference wavelength of 630 nm in an ELISA plate reader (Biotek). The ODs from the control wells were subtracted from the test antigen wells to eliminate background antibody levels.
The plates were thawed at 37 °C in preparation for the assay and washed three times with PBST. A 1 in 100 dilution of the sample or controls was prepared and then 100 µl was added per well in duplicate. Each plate had 3 positive and 2 negative control samples for quality control. After an incubation of 90 min at 37 °C, the wells were washed five times and incubated with 100 µl/well goat anti-human IgG conjugated with alkaline phosphatase (Roche Diagnostics) diluted 1 in 5,000 for 30 min at 37 °C. The plates were washed five times and enzyme activity detected by incubation with 100 µl/well 1-step p-nitrophenyl phosphate substrate solution (Thermo Scientific Pierce, USA) for 30 min at room temperature in the dark. This reaction was stopped by the addition of 50 µl/ well of 3 N NaOH and then the plates were read at a wavelength of 405 nm.
CMV antibody ELISA. Platelia CMV IgG ELISA was used to measure CMV-specific antibodies as per the manufacturer's instructions 36  www.nature.com/scientificreports/ dase enzyme bound onto anti-human IgG monoclonal antibody was then added to each well and incubated for 45 min at 37 °C. The plates were washed and an enzyme substrate was added to each well followed by a 15-min incubation at room temperature. After this, a stop solution was added and the OD read at 450 nm wavelength.
Total IgG antibody ELISA. Human IgG total Ready-SET-Go (Affymetrix ebioscience) was used to measure total IgG following the manufacturer's instructions 37 . In the procedure, Immulon 4 HBX microtiter plates were coated with 100 µl/well purified anti-human IgG monoclonal antibody overnight at 4 °C. The plates were then washed twice with wash buffer and incubated with 250 µl/well blocking buffer for 2 h at room temperature. The plates were washed twice and eight, recombinant human IgG controls were prepared and the samples were diluted 1 in 100,000. A volume of 100 µl/well of the prepared standards and samples was added to the plates and then incubated for 2 h at room temperature. In the next step, the plates were washed four times and incubated with 100 µl/well HRP-conjugated anti-human IgG monoclonal detection antibody for 1 h at room temperature. The plates were washed four times and incubated with 100 µl/well tetramethylbenzidine (TMB) substrate solution for 15 min at room temperature in the dark. After this duration, the reaction was stopped by addition of 100 µl/well stop solution and the plates read at 450 nm wavelength.
Surface plasmon resonance. The avidity of antibodies directed against TT and MVHA antigens was evaluated. These antigens were immobilised onto CM5 sensor chips (Biacore, GE Healthcare, Amersham) by amine coupling to a level of 2000 RU. Samples were diluted 1 in 3 in HBSPE (10 mM HEPES pH 7.4 containing 0.15 M NaCl, 3 mM EDTA and 0.005% (w/v) Surfactant P20; GE Healthcare) and run through Bio-Gel P-30 (Bio-Rad, UK) polyacrylamide gel spin columns to minimise non-specific binding. They were further diluted 1 in 8 in HBSPE running buffer containing 1% (w/v) carboxymethyl-dextran sodium salt (Sigma) and analysed in the Biacore 3000 instrument at a temperature of 25 °C. A 90 µl volume of sample was injected over the chip surface at a rate of 15 µl/min followed by a dissociation time of 8 min. Prior to analysis of the next sample, the chips were regenerated with 50 mM HEPES containing 3 M MgCl2 and 25% (v/v) ethylene glycol, followed by 20 mM glycine pH 1.5 and re-equilibration in HBSPE.
BIA evaluation software version 4.1.1 was used for data analysis and control flow-cell traces with immobilised alpha-1 antitrypsin background were subtracted from test flow cell data. A Langmuir 1:1 dissociation model was used to determine the dissociation rate between 10 and 300 s post sample injection.
B cell ELISPOT. We used in-house in vitro and ex vivo B cell ELISPOTS to determine the frequency of TTspecific plasmablasts and MBCs respectively in APTB cases and healthy donors following methods described by Sebina et al. 38,39 . Cryopreserved PBMCs were thawed, resuspended in RPMI (Gibco by Life Technologies) containing 10% (v/v) fetal bovine serum, 2 mM l-glutamine (both from Sigma-Aldrich, UK) and 1 × penicillin-streptomycin (Invitrogen). The cells were then rested for 4-6 h at 37 °C and 5% CO 2 .
In the in vitro ELISPOT, PBMCs at a cell density of 1 × 10 6 cells/ml were stimulated with a mixture of 25 ng/ ml human IL-10 (R&D Systems, UK), 0.5 µg/ml of pokeweed mitogen (PWM), 6 µg/ml of CpG oligodeoxynucleotide, 1.2 mg/ml Staphylococcus aureus Cowan, 50 µM ß-mercaptoethanol at 37 °C and 5% CO 2 for six days. Approximately 4 × 10 5 of the cultured cells were then transferred to each well of 96-well filter (Merck Milipore) pre-coated with 2 µg/ml TT in 1 × PBS or only 1 × PBS. The plates were sealed and incubated for 6 h at 37 °C and 5% CO 2 after which biotin-SP conjugated affinipure fragment donkey antihuman IgG (Jackson ImmunoResearch) was added to the plate wells to aid in the detection of anti-TT specific antibody secreting cells. After overnight incubation at 4 °C, streptavidin-AKP (BD biosciences) was added and the plates incubated for an hour at room temperature followed by AP-conjugate substrate (Bio-Rad, USA) to develop the spots. After 10 min the plate wells were rinsed with water and left to dry before they were read using an AID ELISPOT reader (AID Diagnostika, Germany). The number of MBCs was then expressed as the total number of spots per million cells.
The ex vivo ELISPOT involved the direct transfer of the rested PBMCs to the 96-well filter plates at a cell density of 4 × 10 5 cells/well. All the subsequent steps were similar to those described for the in vitro ELISPOT. Data analysis. Study data were analysed using Stata Release statistical package and GraphPad Prism software. The initial analyses were made using the Kruskal-Wallis test to compare antibody responses across the groups of uninfected controls, individuals with LTBI and APTB cases. Linear regression analysis with bootstrap confidence intervals was then be used to determine associations between M.tb infection status and log transformed antibody responses 12,17 . Adjusting was done for the effects of potential confounders such as HIV infection, age, socioeconomic status and gender. In order to account for multiple testing Q values were computed for every comparison between the M.tb infected groups and uninfected individuals, using the Bonferroni method. The Wilcoxon rank sum test was used to compare antibody responses between BCG vaccinated individuals and BCG naïve controls as well as frequencies of MBCs and PBs in APTB cases. www.nature.com/scientificreports/

Data availability
All data used to generate the figures and tables presented in this manuscript can be found in the S1 file in the supporting information. There are no restrictions to the availability of these data.