Induction of specific antibodies, IgG-secreting plasmablasts and memory B cells following BCG vaccination

Many tuberculosis (TB) vaccine candidates are designed as a boost to BCG; an understanding of the BCG-induced immune response is therefore critical, and the opportunity to relate this to circumstances where BCG does protect may direct the design of more efficacious vaccines. While the T cell response to BCG vaccination has been well-characterised, little is known about the B cell and antibody response. We demonstrate BCG vaccine-mediated induction of specific antibodies in different human populations and macaque species which represent important preclinical models for TB vaccine development. We observe a strong correlation between antibody titres in serum versus plasma with modestly higher titres in serum. We also report for the first time the rapid and transient induction of antibody-secreting plasmablasts following BCG vaccination, together with a robust and durable memory B cell response in humans. Finally, we demonstrate a potential contribution of the antibody response to BCG vaccine-mediated control of mycobacterial growth in vitro. Taken together, our findings indicate that the humoral immune response in the context of BCG vaccination merits further attention to determine whether TB vaccine candidates could benefit from the induction of humoral as well as cellular immunity.


Introduction
Tuberculosis (TB), caused by Mycobacterium tuberculosis (M.tb), remains a major global health threat with 10 million new cases and 1.4 million deaths in 2019 [1]. There are currently no validated correlates of protection from TB. However, M.tb is an intracellular pathogen and the necessity for a T cell response in conferring acquired immunity to TB has been demonstrated in numerous studies [2][3][4][5][6]. This has led to a limited focus on the humoral response to TB, although emerging evidence suggests that antibodies may play a more significant role in protection than previously appreciated [7][8][9][10]. Antibodies could contribute to protection directly through increasing phagocytosis and phagolysosome formation or bacterial neutralisation, and/or indirectly through enhancing T cell-mediated immunity [7]. It has recently been shown that compared to antibodies from patients with active TB disease (ATB), antibodies from individuals with latent TB infection (LTBI) have unique Fc functional profiles, selective binding to FcγRIII and distinct antibody glycosylation patterns, and also that they drive enhanced phagolysosomal maturation, inflammasome activation and macrophage killing of intracellular M.tb [9].
Bacillus Calmette Guérin (BCG) is the only currently available vaccine against TB. BCG confers incomplete and variable protection against pulmonary TB in adolescents and adults, and a new more efficacious TB vaccine is needed [11,12]. However, it is uncertain which aspects of the immune response a candidate vaccine should aim to induce in order to confer protection that is superior to BCG. Due to the important role of BCG vaccination in protecting infants from severe forms of TB disease, and its potential non-specific effects protecting from all-cause mortality, most TB vaccine candidates are designed as a heterologous boost to a BCG prime [13,14]. It is thus critical to understand the immune response to BCG vaccination and which aspects will be, or would ideally be, induced or boosted. Furthermore, because BCG vaccination is partially protective, and can confer superior protection when administered intravenously in macaques [15,16] or as a revaccination in South African adolescents [17], studying the immune response to BCG offers a valuable opportunity to explore immune mechanisms of protection and apply these to inform the design of more efficacious TB vaccine candidates.
Robust Th1 responses to BCG vaccination have been described and are generally considered to be essential, but not sufficient, for protection [18][19][20][21]. A role for trained innate immunity, unconventional T cells and humoral immunity in BCG-mediated protection has been proposed [22][23][24]. Indeed, in a post-hoc correlates of risk analysis, levels of Ag85A-specific IgG were associated with reduced risk of TB disease in BCG-vaccinated South African infants [25]. We have recently comprehensively reviewed what is known about the humoral immune response to BCG vaccination and revaccination across species [26]. In brief, the literature presents inconsistent evidence for the induction of specific antibody responses following BCG vaccination, and the relevance of these responses is unclear, with support both for [25,[27][28][29][30] and against [28,[31][32][33] a protective function.
Antibodies are produced by antibody-secreting B cells (ASCs). Upon activation by recognition of their cognate antigen during infection or vaccination, B cells undergo clonal expansion and differentiate into plasmablasts and memory B cells (mBCs) [34]. While plasmablasts and plasma cells secrete antibody, mBCs can survive quiescently for decades, poised to rapidly respond to antigen restimulation by differentiating into short-and long-lived ASCs that can prolong the duration of high serum antibody levels [35,36]. It is generally accepted that mBCderived ASCs are central to the long-term protection mediated by most licenced vaccines [34,37], and the dynamics, magnitude and specificity of the ASC response to vaccination against other infections have been described [38][39][40]. While one study has reported a significantly higher frequency of purified protein derivative (PPD)-specific mBCs in peripheral blood mononuclear cells (PBMC) from historically BCG-vaccinated compared with unvaccinated volunteers [41], the longitudinal ASC response following BCG vaccination has not been described. Evaluating these cells may provide a more dynamic, discriminative measure of humoral immunity than detecting stable serum antibodies by conventional serology.
The hitherto poorly-defined humoral response to BCG vaccination merits further investigation.
We set out to explore the BCG vaccine-induced specific antibody response in serum and plasma collected from different cohorts of human volunteers representing TB-endemic and non-endemic populations. As the macaque model is widely used in TB studies [42], we included macaque samples to better understand the translatability of antibody responses to humans. We also sought to determine the frequency and kinetic of BCG-induced antigenspecific ASCs (both plasmablasts and mBCs), and their association with serum antibody levels. Finally, we employed the direct PBMC mycobacterial growth inhibition assay (MGIA) to explore a potential functional contribution of BCG-induced antibodies to the control of mycobacterial growth. Our findings offer proof-of-concept and provide some logistical foundations for a more comprehensive characterisation of the BCG vaccine-induced antibody response across different populations and vaccine regimens for which different levels of protection are observed. ELISAs were performed as previously described [44]. The antigens used were purified protein   6 PBMC and also as the percentage of total IgG-secreting cells. Responses were considered positive if the count was two or more spots in each replicate well, and the total number of spots in the antigen-coated wells twice that observed in the blank control wells.

Memory B cell (mBC) ELISpot
For each volunteer, 500µl of thawed PBMC at a concentration of 2x10 6 cells/ml in R10 media was added to six wells of a 24-well flat-bottom tissue culture plate. 500µl of Mitogen stimulation mix containing Staphylococcus aureus Cowan (SAC, 1:2400 dilution), CpG (5µg/ml) and pokeweed mitogen (PWM, 1:6000 dilution from a 1mg/ml stock), was added to five wells per volunteer. 500µl R10 was added to an unstimulated well as a control. Plates were incubated at 37°C, 5% CO2, for 6 days. On day 5, ELISpot plates were coated overnight at described above. On day 6, plates were washed and blocked as described. The cultured cells set up on day 0 for each volunteer were harvested by gentle resuspension and the stimulated cells and unstimulated cells were each pooled, washed twice by centrifugation at 700g for 5 minutes at room temperature, resuspended in R10 and counted. Cells were then resuspended at 2x10 6 cells/ml in R10, and 100µl was added to the negative control wells and the first three antigencoated wells. For the total IgG wells, dilutions of 1:1, 1:50 and 1:100 were added to each well in duplicate. 100µl of unstimulated cells were added to duplicate IgG wells. Plates were then incubated at 37°C, 5% CO2, for 16-18 hours and developed and counted as described above.
Frequencies of mBC were shown as the number of BCG-specific mBC-derived ASC per million cultured PBMC and as the proportion of mBC-derived ASC of total IgG-secreting ASC.

Statistical analysis
Following normality testing (Shapiro-Wilk), longitudinal data was analysed using a one-way ANOVA with Dunnett's correction for multiple comparisons (for normally-distributed data) or a Friedman test with Dunn's correction for multiple comparisons (for non-normally distributed data). For studies with random missing values due to sample unavailability ( Figure 1), data was logged and analysed using a mixed-effects model with Dunnett's correction for multiple comparisons. When comparing two conditions (eg. serum vs. plasma), a paired t-test (for normally-distributed data) or Wilcoxon signed-rank test (for non-normally distributed data) was conducted. Associations were determined using a Spearman's rank correlation.

BCG vaccination in humans and macaques induces higher levels of PPD-specific antibodies compared with baseline
Using samples taken from UK adults enrolled into human Study 1, we observed a significant increase in PPD-specific IgG at 28 and 84 days post-BCG vaccination in serum (p=0.01 and p=0.0001 respectively, Figure 1A) and plasma (p=0.0004 and p=0.0002 respectively, Figure   1B) compared with baseline. PPD-specific IgA increased at 28 days post-BCG vaccination in serum (p=0.003, Figure 1D) and at days 28 and 84 in plasma (p=0.004 and p=0.0005 respectively, Figure 1E). As IgM is the first isotype to be produced following B cell stimulation, PPD-specific IgM was measured at earlier time-points (7, 14 and 21 days post-BCG vaccination), and we saw a significant increase at 14 and 21 days in plasma only (p=0.001 and p=0.0001 respectively, Figure 1G).
Similarly, in rhesus macaques enrolled into macaque Study 1, we observed a significant increase in PPD-specific IgG at 56 days post-BCG vaccination in serum and plasma (p=0.01 and p=0.004 respectively, Figure 2A-B) compared with baseline. PPD-specific IgA increased at 28 and 56 days post-BCG vaccination in plasma only (p=0.02 and p=0.02 respectively, Figure 2E), although there was a similar non-significant trend in serum ( Figure 2D). There was a trend towards increased PPD-specific IgM taken at the same time-points but this was not significant; earlier time-points were not available for this study. No differences were observed over time in the control group for any of the isotypes measured for either species. There was no difference in fold change in PPD-specific IgG or IgA levels at 28 days following BCG vaccination (the only directly comparable time-point) in serum or plasma between humans and macaques ( Figure S1A-B). We observed several associations between fold change following BCG vaccination in levels of different PPD-specific isotypes in both humans and macaques, as summarised in Tables S1-S4. 3.2 PPD-specific antibody levels, but not fold-change following vaccination, are higher in serum compared with plasma with a strong correlation between responses In humans, we observed significantly higher levels of PPD-specific IgG in serum compared with plasma at all time-points (p=0.02, p=0.04 and p=0.007 at baseline, day 28 and day 84 respectively, Figure 1C). There were also higher levels of PPD-specific IgA in serum compared with plasma at all time-points measured (p=0.002, p=0.004 and p=0.01 at baseline, day 28 and day 84 respectively, Figure 1F), and higher levels of PPD-specific IgM in serum compared with plasma at all time-points measured (p=0.007, p=0.0001, p=0.005 and p=0.003 at baseline, day 7, day 14 and day 28 respectively, Figure 1I). Similarly, in macaques, there were significantly higher levels of PPD-specific IgG in serum compared with plasma at all timepoints measured (p=0.0009, p=0.02 and p=0.04 at baseline, day 28 and day 56 respectively, respectively, Figure S3D), IgA at day 28 (p=0.001, Figure S3E) and IgM at day 28 (p=0.03, Figure S3F).
IgG levels specific to PPD and whole BCG were measured in serum taken at baseline and 28 days in human Study 1. Following BCG vaccination, there was a significant increase in IgG specific to whole BCG and PPD (p=0.006 and p=0.04 respectively, paired t-test, Figure S4A-B). There was no change in the control group over time. There was a significant correlation between BCG-and PPD-specific IgG levels at 28 days post-vaccination (p=0.005, Spearman's correlation, Figure S4C), but not at baseline. Similarly, IgG levels specific to PPD and whole BCG were measured in serum taken at baseline and 56 days in macaque Study 2. There was a significant increase in both BCG-specific and PPD-specific IgG following BCG vaccination (p=0.04 and p=0.04 respectively, paired t-test, Figure S4D-E). There was a significant correlation between the two measures at baseline (p=0.04, Spearman's correlation, Figure   S4F). There was no change in PPD-specific or BCG-specific IgG levels in the control group over time in either study.
Using serum collected from rhesus macaques (macaque Studies 3 and 4) and cynomologus macaques (macaque Study 5), we compared responses between the two species at the same time-points following BCG vaccination with the standard dose. We observed similarly significant induction of PPD-specific IgG at 56 and 140 days post-BCG vaccination in the two species (p=0.0003 and p=0.001 respectively in rhesus macaques; p=0.001 and p=0.0007 respectively in Mauritian cynomolgus macaques; RM ANOVA with Dunnett's post-test; Figure   S5A). There was no difference in fold change in IgG response following BCG vaccination between the species at either time-point ( Figure S5B).
In There was a trend towards a correlation between ASC responses at 7 days and BCG-specific IgG levels at 7 days ( Figure 5E), which was significant at 70 days (p=0.003, r=0.83, Spearman's rank correlation, Figure 5F). There were no associations between mBC-derived ASC responses and BCG-specific IgG levels at either time-point studied.

Antibodies may play a functional role in the control of mycobacterial growth in vitro
The direct PBMC MGIA was conducted using cryopreserved cells and autologous serum  Figure 6B).

Discussion
We have observed significant induction of specific IgG following BCG vaccination in two independent human cohorts (one from a TB-endemic region and one from a non-endemic region it should be noted that animals received 3 weeks of daily oral isoniazid/rifadin at 8 weeks [33]. Rhesus macaques are more susceptible to acute progressive TB disease and BCG confers lower efficacy compared with cynomolgus macaques [33,57]; our finding of similar total specific IgG responses between the species may therefore suggest that these antibodies are not protective, or that different attributes (such as antigen-specificity, subtype, posttranslational modifications, affinity or avidity) are more relevant. Importantly, we observed similar fold change in IgG and IgA levels relative to baseline in humans and macaques at the interchangeably to test for antibody responses to mycobacterial antigens even in the same assay [58]. However, we observed PPD-specific antibody responses that were modestly but significantly higher in serum than plasma across isotypes and species. Other studies have similarly noted lower levels of proteins in plasma than serum [59,60], potentially associated with the addition of anticoagulant (such as heparin) to blood prior to obtaining plasma, or the presence of fibrinogen which may influence antibody-antigen binding [61]. However, fold change in BCG vaccine-induced antibody responses was comparable between serum and plasma indicating that sensitivity to detect a vaccine-induced response does not differ.
To our knowledge, the specific antibody-secreting B cell response to BCG vaccination has not been previously described. Using an overnight ELISpot assay, we observed induction of BCGspecific IgG-secreting plasmablasts at 7 days post-BCG vaccination, returning to baseline by day 70. This is consistent with the kinetic of response to other primary vaccinations including those against rabies and influenza [38,39]. The poor correlation between ASC responses and early IgG antibody induction is perhaps to be expected; at 7 days following vaccination, BCGspecific IgG would still be in the very early stages of production. While a significant association was noted between ASC responses and IgG levels at 70 days post-vaccination, it may be that levels have started to wane by this time and that determining responses at an earlier timepoint would provide an even stronger association. Strong correlations of IgG plasma cell responses with IgG antibody responses after primary rabies immunisation by area under the curve analysis has been reported elsewhere, but our limited time-points did not permit such analysis [38].
BCG-specific mBC responses were detectable in the historically BCG-vaccinated group, who had received BCG up to 20 years previously, at all time-points. This is in concordance with a previous study of historic BCG vaccinees assessed 13-45 years post-vaccination [41]. The responses noted here are higher than previously seen, which may be due to the use of different antigens (whole BCG vs. PPD) and different study populations. In the previously BCG-naïve group, mBCs induced by BCG vaccination were apparent at 7 days and had further increased by 70 days, consistent with the detection of mBCs in peripheral blood by 7 days post-immunisation in a murine model [62]. A study assessing longitudinal mBC responses following exposure to candidate malaria vaccines showed a gradual increase in specific mBC after immunisation which peaked at 84 days before declining by 140 days [40]. One limitation of our study is the inability to define the peak or longevity of mBC responses due to logistical restrictions on collecting later follow-up samples, but one would similarly expect contraction to a persisting residual level over time [63].
Interestingly, we saw relatively high antibody responses at baseline and in the naïve unvaccinated controls across isotypes and studies, which may indicate cross-reactive antibodies induced by environmental exposure to non-tuberculous mycobacteria (NTM) [64,65]. Others have also noted pre-existing specific antibodies in serum from PPD-negative individuals with no known exposure to M.tb [27,66]. While NTM are prevalent in Nepal, Western countries are generally considered to have low circulating levels, but there are indications of a steep rise in incidence in the UK in recent decades [67,68]. The macaques included here were housed partially outdoors, and as such may also have had environmental exposure. It is possible that an element of the ASC response observed in the Nepalese cohort was due to the generation of plasmablasts from pre-existing mycobacteria-specific mBCs.
Indeed, the rapidity and magnitude of the ASC response was comparable to recall responses observed in secondary vaccination studies [38,40]; although this hypothesis is not supported by the lack of mBC responses at baseline in unvaccinated volunteers, they may have been present at frequencies below the limit of detection of the ELISpot assay. It remains to be determined whether baseline reactivity interferes with induction of a humoral response to BCG vaccination in the way it appears to affect IFN-γ responses [69], particularly given the widelyaccepted masking/blocking hypotheses for the geographical variation in BCG efficacy [12].
In order to explore a potential functional role for the antibodies detected, we applied an unbiased sum-of-the-parts mycobacterial growth inhibition assay (MGIA) that we have previously optimised, standardised and harmonised as a potential surrogate of TB vaccineinduced protection [45,70]. Using this assay, we (and others) have demonstrated ability to detect a BCG vaccine-induced response and an association between the outcome of the direct MGIA and protection from in vivo mycobacterial infection [44,[71][72][73][74]. Various immune mechanisms have been proposed to contribute to mycobacterial control in the direct MGIA including polyfunctional CD4+ T cells and trained innate immunity [44,72,73]. In two independent cohorts, we confirmed previous findings of enhanced control of mycobacterial growth in the direct MGIA following BCG vaccination, and importantly showed that this effect was cancelled out by exchanging pre-for post-vaccination serum and vice versa.
Further work is required to determine the serum component/s responsible, but other studies have reported that post-BCG vaccination serum can enhance the control of mycobacterial growth in vitro via antibody-mediated mechanisms [27,28], and we hypothesise that antibodies are likewise playing a role in our assay. Indeed, we have previously shown that IgG1 responses to M.tb-specific antigens correlated with improved mycobacterial control in a study of individuals with ATB or LTBI [43], and that specific IgG responses following in vivo mycobacterial infection in historically BCG vaccinated volunteers were associated with control of mycobacterial growth following in vitro infection in the direct MGIA [44]. A role for antibodies may explain in part why we saw no association between outcome of the direct whole blood and PBMC MGIAs when pooled human serum rather than autologous serum was used in the latter [75]. In our second human cohort, we observed enhanced mycobacterial control when baseline cells were cultured with post-vaccination serum compared with baseline cells cultured with baseline serum, although the effect was less pronounced than between baseline cells cultured with baseline serum and post-vaccination cells cultured with post-vaccination serum. This suggests that control of mycobacterial growth in the direct MGIA is mediated by a combination of cellular and humoral immunity, and supports the recommendation to use autologous serum in the assay as a more representative ex vivo approach.
In conclusion, we have demonstrated BCG vaccine-mediated induction of antibodies specific to PPD and whole BCG that are modestly higher in serum than plasma and comparable across humans and macaques, as well as serum responses to a range of mycobacterial fractions in humans. We have also shown for the first time the rapid and transient induction of antibody-       and rhesus macaques enrolled into macaque Study 1 (red). Fold change in PPD-specific IgG following BCG vaccination was compared between adults from the UK (non-TB endemic) enrolled in human Study 1 at day 84 (circles) and adults from Nepal (TB-endemic) human