Mycobacterium tuberculosis PPD-induced immune biomarkers measurable in vitro following BCG vaccination of UK adolescents by multiplex bead array and intracellular cytokine staining

That the cytokine interferon gamma (IFNγ) plays an important role in the protective immune response against tuberculosis (TB) is indicated by the susceptibility of mice and humans with IFNγ signalling pathway deficiencies to TB disease [1–3]. Its detection in isolation however is not a sufficient indicator of a protective immune phenotype as those with latent infection and a positive IFNγ release assay status can progress to active disease and IFNγ secretion can also be detected in samples from patients with active disease [4].

The Th-1-type immune response that is most effective against TB and of which IFNγ is a component is likely to include other cytokines such as tumour necrosis factor alpha (TNFα), interleukin (IL) -2 and IL-12. A role for the more recently identified Th-17 phenotype involving IL-17 has also been described [5, 6]. Furthermore, the presence of cells that secrete such cytokines as well as other immune effector molecules may be included as a component of a protective biomarker profile. For example, CD4⁺ T-cells play an important role in TB immunity, however CD8⁺, NKT and γδ T-cells may also be necessary [7–10]. A protective biomarker signature may also be defined by the absence of particular biomarkers as certain immune states may subvert the response to TB, allowing bacterial infection to persist and for disease to eventually progress. Cytokines such as IL-4 or other immunoregulatory cytokines such as IL-10 derived from Th-2 biased T-cells or regulatory T-cells respectively may indicate such a subversion if detected [11, 12].

BCG vaccination has previously demonstrated a protective efficacy of 77% against pulmonary tuberculosis when administered to UK adolescents [13]. We have used this setting to investigate the nature of the immunity induced by BCG vaccination in representative cohorts of UK schoolchildren (age range 12-15). Diluted whole blood assays on samples from such a cohort, in which responses to antigen during 6 day cultures were measured by quantifying IFNγ in assay supernatants, revealed increased IFNγ after vaccination compared to that measured in pre-vaccination samples [14]. The status of IFNγ as a cytokine that is necessary but not sufficient for protection against TB is illustrated by parallel experiments carried out in Malawi where high concentrations of IFNγ were detected both prior to and following BCG vaccination in a setting where BCG is much less protective than in the UK [14]. Furthermore, studies have described other biomarkers that can differentiate latent infection from active disease [15] and have highlighted the importance of cytokines such as TNFα [5, 16]. Such observations emphasise the need to measure a greater diversity of potential biomarkers in order to develop a more detailed representation of the BCG-induced immune response in different settings. In South Africa, studies on samples from BCG-vaccinated infants to look at responses comprising the Th-1 cytokines IFNγ, TNFα and IL-2 revealed multiple T-cell phenotypes with distinct cytokine secretion profiles [17]. Our group has also recently reported extensive and complex cytokine responses measurable in Mycobacterium tuberculosis purified protein derivative (Mtb PPD)-stimulated blood from BCG-vaccinated, UK infants [18]. New candidate TB vaccines that are thought to represent more efficacious alternatives to BCG have demonstrated the ability to induce populations of cells with polyfunctional cytokine activity. The BCG/modified Vaccinia Ankara-Ag85A vaccine regime, for example, can generate CD4⁺ T-cells that secrete up to 4 cytokines including IFNγ, TNFα, IL-2 and MIP-1β [19].

In the present study, we have used multiplex bead array to determine the concentration of 20 biomarkers (including cytokines and chemokines which together represent a comprehensive coverage of possible immune phenotypes) in the supernatants of Mtb PPD-stimulated, diluted whole blood assays on samples from UK adolescents taken prior to, and after BCG vaccination. A smaller panel of biomarkers were further investigated by flow cytometry. Cells from blood samples taken 12 months after BCG vaccination were stimulated and stained with antibodies to these biomarkers and other markers of T-cell phenotype.

The cytokine response measured following BCG vaccination of adolescents in this study was broad. Such a varied response was also observed by us and others in infant studies of BCG vaccination [17, 18]. Of 20 analytes measured, 10 were significantly raised in Mtb PPD stimulated, diluted whole blood samples taken 1 month after BCG vaccination compared to pre-vaccination samples. IFNγ was the most markedly increased analyte (fold increase of 240 in post-vaccination samples), which bears out this cytokine's long-standing association with mycobacterial immunity [2, 20]. TNFα and IL-2 have also both been described by others as cytokines that are central to effective immunity to TB [5, 16] and indeed we found them to be raised following BCG (increased by a factor of 4.6 and 5.0 respectively). The protection that has been afforded against TB by BCG vaccination in the UK is likely to rely on the actions of these cytokines. IFNγ and TNFα can activate macrophages to better destroy phagocytosed Mtb [21] and TNFα helps induce maturation of dendritic cells for more effective activation of mycobacteria-specific T-cells [22]. IL-2 is also responsible for the promotion of T-cell responses and memory [23]. Such immune functions rely on the targeted release of these cytokines by antigen experienced effector cells. That BCG vaccination is priming such cells is indicated by our observation that these biomarkers are detected at increased concentrations in post-vaccination samples after stimulation with Mtb PPD. The many other pro-inflammatory biomarkers detected in post-vaccination assay supernatants may represent a combined response originating from Th-1 cells, blood monocytes or neutrophils that are responding to IFNγ and/or TNFα. It should be noted that the inflammatory biomarkers IL-6, IL-8 and IP-10, although increased in post-vaccination samples were also present in relatively large amounts in pre-vaccination samples following Mtb PPD stimulation. This suggests an antigen-non-specific response of the innate immune system, possibly as a result of stimulation of monocytes or neutrophils via toll-like receptors by an undefined component of the Mtb PPD protein cocktail. This hypothesis requires further investigation. In addition, IL-8 and IP-10 were present in unstimulated wells to a greater extent than other biomarkers suggesting that these are released, to some degree in these conditions, spontaneously.

IL-17 is a pro-inflammatory cytokine that has been linked to a number of autoimmune and chronic inflammatory conditions such as rheumatoid arthritis [24] and asthma [25]. In mice, γδ T-cell-derived IL-17 has been demonstrated during Mtb and BCG infection and in the latter, shown to contribute to effective Th-1 responses and granuloma formation [26, 27]. In our current study, IL-17 was strongly increased in post-BCG whole blood assay supernatants (28.3 fold increase) and was also significantly associated with concurrent increases in IFNγ (r = 0.769). As Mtb PPD comprises a number of undefined components these results do not determine the precise nature of the IL-17 response detected. Possibilities include a CD4⁺ Th-17 response to protein antigen(s) in Mtb PPD that are processed and presented by MHC class II molecules. Alternatively, as there is evidence from other studies that γδ T-cells may produce IL-17, it may be that in our assay such cells are recognising an unknown component of Mtb PPD.

If BCG vaccination of UK adolescents is taken as a setting where there has been a degree of success in promoting immunity to TB, it is surprising to see the anti-inflammatory cytokine IL-10 also raised in Mtb PPD-stimulated, post-BCG assay supernatants. It is possible that the 6 day duration of the whole blood assay may allow time for a complex network of physiological homeostatic interactions between the cells present and the cytokines released on stimulation with Mtb PPD. One outcome of this may be the initiation of regulatory negative feedback loops that are characterised by the production of IL-10.

Any biomarkers released upon an encounter with Mtb antigen may originate from specific lymphocyte populations activated by vaccination. Although multiplex bead array assays are a powerful tool, they cannot identify the cellular source of any biomarkers measured. For this reason we selected a smaller panel of biomarkers (IFNγ, TNFα, IL-2, IL-17, IL-10) that could be assessed by flow cytometry using specific monoclonal antibodies. When combined with antibodies to phenotypic markers it was possible to identify the cells that produced certain cytokines. Due to availability, only PBMC from 12 month post-BCG samples were available for flow cytometric analysis. However, we used the multiplex assay to demonstrate that many of the biomarker responses detectable at one month remained detectable at 12 months.

In this study of Mtb-specific memory lymphocytes induced by BCG, we detected predominantly single cytokine producing cells when PBMC were stimulated for 24 hours with Mtb PPD. The largest responder phenotype in each PBMC sub-population (CD4 T-cells, CD8 T-cells and NK cells) was IFNγ single positive cells. NK cells and CD4 T-cells also contained populations of TNFα single positive responding cells, but double positive cells were less common. For the CD8 T-cell population, single positive IFNγ cytokine secreting cells represented the only notable responding cells. In a study of infant responses to BCG vaccination from a TB-endemic country, analysis of live BCG-stimulated whole blood samples showed that, as in our study, within the CD8⁺ T-cell population, the dominant population expressed IFNγ only [17]. However, three other secretion patterns were also detected within the CD8⁺ T-cells in that study; IL-2 only, IFNγ/TNFα/IL-2 triple positive and IFNγ/IL-2 double positive cells although these latter three patterns were at levels only slightly above background. Such responses may be indicative of the superior ability of live BCG over Mtb PPD to stimulate CD8 T-cells as previously reported [28]. Also, in contrast to our study, Soares et al detected 7 distinct patterns of cytokine secretion within the CD4⁺ T-cell population including triple, double and single positive cytokine producers whereas we only observed 2 (IFNγ and TNFα single positives). Using a PBMC stimulation method similar to that described here, polyfunctional T-cells (secreting up to four cytokines including IFNγ, TNFα and IL-2) have been demonstrated following Modified Vaccinia Ankara-85A administration after previous BCG vaccination of a UK cohort [19]. The lack of multiple cytokine secreting cells in our study may be a result of the late timing of these samples (12 months post-vaccination). Any polyfunctional, memory lymphocytes present may exist only at very low frequencies that would require the acquisition of many more events than were possible in this study due to the amount of sample available. Another possibility is that the exposure of responding lymphocytes to antigen is minimal or absent 12 months after BCG vaccination. Any multifunctional T-cells may, as a result, convert to a resting state out of which the short incubation time we used in flow cytometry assays is insufficient to stimulate them. It should also be considered however, that the failure of BCG to deliver long-lasting (i.e. lifetime) immunity against pulmonary TB in many settings may be predicted by the lack of induction of long term polyfunctional T-cell responses as shown here. Unfortunately, the lack of pre-vaccination or 1 month post-vaccination comparator samples for the flow cytometry aspect of this study is a significant limitation and renders us unable to comment on the absolute effect of BCG vaccination on intracellular cytokine profiles. We were unable to detect any expression of IL-2, IL-17 or IL-10 by flow cytometry after Mtb PPD stimulation. Again, this may be explained by the timing of the samples and the need to acquire a large number of events to detect very rare, cytokine producing cells. Soares et al. also reported low percentages of IL-10 expressing T-cells that were, in most cases, below their cutoff for a positive response [17]. The stimulation period used in that study (12 hours) and in ours (24 hours) may be sub-optimal for the detection of IL-10. Although we did not detect IL-17 after Mtb PPD stimulation, others have observed CD3⁺CD4⁺IL-17⁺ cells following BCG and PPD stimulation of whole blood from mycobacteria-exposed individuals [29]. IL-17 induced by Mtb PPD was reported to be low by Scriba et al and our use of purified PBMC as opposed to their use of whole blood and the inherent antigen processing differences involved may explain the absence of IL-17⁺ cells in the current study. Others have also found that purified T-cells with specificity for Mtb PPD produced IFNγ but failed to produce IL-17 in response to stimulation [30]. It should be noted that at 72 pg/ml, the median concentration of IL-17 measured by multiplex bead array is relatively low in 1 month post-BCG whole blood assay supernatants compared to IFNγ (1273 pg/ml) and the increase over the median concentration in pre-vaccination assays is only detectable due to the sensitivity of the multiplex assay. Such marginal secretion kinetics may be less easy to observe using a 24 hour flow cytometry assay at 12 months post BCG.

In summary, we have shown that the Mtb PPD-specific response 1 and 12 months after BCG vaccination as measured in a 6 day, diluted whole blood assay was broad in terms of cytokines released and encompassed Th-1 and inflammatory mediators as well as immunoregulatory elements. Such a multi-faceted response indicates the complex nature of the immune response to BCG as well as the ability to detect such responses in longer duration assays. When Mtb PPD-stimulated PBMC were analysed by flow cytometry, polyfunctional CD8 or CD4 T-cells were not detected. The majority of responding cells in these populations and especially in the NK cell population were single cytokine (either IFNγ or TNFα) producers. Although these findings may relate to sample sizes, timings and assay kinetics they may also represent an inability of BCG to induce the polyfunctional T-cells that are thought to be important mediators of immunity against infection [31]. Whether such cells are detectable earlier following vaccination and are lost at later time points remains to be seen.