Historical BCG vaccination combined with drug treatment enhances inhibition of mycobacterial growth ex vivo in human peripheral blood cells

Abstract

Tuberculosis (TB) is a leading infectious cause of death globally. Drug treatment and vaccination, in particular with Bacillus Calmette-Guérin (BCG), remain the main strategies to control TB. With the emergence of drug resistance, it has been proposed that a combination of TB vaccination with pharmacological treatment may provide a greater therapeutic value. We implemented an ex vivo mycobacterial growth inhibition assay (MGIA) to discriminate vaccine responses in historically BCG-vaccinated human volunteers and to assess the contribution of vaccine-mediated immune response towards the killing effect of mycobacteria in the presence of the antibiotics isoniazid (INH) and rifampicin (RIF), in an attempt to develop the assay as a screening tool for therapeutic TB vaccines. BCG vaccination significantly enhanced the ability of INH to control mycobacterial growth ex vivo. The BCG-vaccinated group displayed a higher production of IFN-γ and IP-10 when peripheral blood mononuclear cells (PBMC) were co-cultured with INH, with a similar trend during co-culture with RIF. A higher frequency of IFN-γ⁺ and TNF-α⁺ CD3^- CD4^- CD8^- cells was observed, suggesting the contribution of Natural Killer (NK) cells in the combined effect between BCG vaccination and INH. Taken together, our data indicate the efficacy of INH can be augmented following historical BCG vaccination, which support findings from previous observational and animal studies.

Figure 1

Immune response (A) and growth inhibition (B) following historical BCG vaccination. Assessment was performed from 21 BCG-naïve and 29 BCG-vaccinated participants. (A) IFN-γ production following stimulation with PPD was compared (Mann-Whitney test). Numbers above each group represent median (range). SFC, spot forming cells. (B) Growth inhibition was compared using BCG input ∼100 Colony Forming Unit (CFU) as immune target (unpaired t-test). Data are presented as total number of log CFUs per sample, which was determined by use of a standard curve. Dots and squares represent individual data points, and the central lines indicate the median response with inter-quartile range (IQR).

Figure 2

Growth inhibition in the absence and presence of INH (A) and RIF (B). Mycobacterial growth was assessed in titration curves. INH inhibited mycobacterial growth in a dose-dependent manner in the naïve and vaccinated groups (A,B), as well as RIF (D,E). Data from both groups was compiled in dose-response box plots to identify the BCG effect in addition to the INH- and RIF-mediated killing (C,F). Dots and squares in the titration curves (A,B,D,E) represent individual data points from the participants and the central lines indicate the median response with IQR. Each group is represented in a single box plot with range in the dose-response analysis (C and F). Samples size is indicated in the figure (for INH: n = 29 BCG-vaccinated and n = 21 BCG-naïve participants; for RIF: n = 25 BCG-vaccinated and n = 18 BCG-naïve participants). Statistical significances were tested using one-way ANOVA (A,B,D,E) and unpaired t-test (C,F).

Figure 3

Cytokine responses from co-culture with INH (A) and RIF (B). MGIA supernatants were analysed for the released cytokines IFN-γ, IP-10 and IL-10. For INH, dark blue and dark green lines and symbols indicate BCG-naïve and BCG-vaccinated groups, respectively. For RIF, these were represented by dark brown (BCG-naïve) and dark red (BCG-vaccinated). Lines indicate mean response and shadings indicate range (A,B). Comparison of responses between BCG-vaccinated and BCG-naïve groups at different drug concentrations were performed using unpaired t-test. Correlation between mycobacterial growth at INH concentration of 1 µg/ml and the production of IFN-γ (C), IP-10 (D) and IL-10 (E) were assessed using Spearman’s correlation. Correlation between mycobacterial growth at RIF concentration of 0.01 µg/ml (based on the significant difference in the MGIA assay) and the production of IFN-γ (F), IP-10 (G) and IL-10 (H) were also assessed (Spearman’s). Note: for the correlations, non-responders were excluded, defined as responses below the following cut-off of the ELISA assays: 7.5 pg/ml (IFN-γ), 5 pg/ml (IP-10) and 3.5 pg/ml (IL-10). Refer to Supplementary Table S1 for comparison and correlation of other cytokines responses.

Figure 4

Frequencies of Th1 cytokine-expressing lymphocytes. Expressions were measured from PBMCs after stimulation with BCG, with and without 1 μg/ml of INH, for 4 days. The grey dot symbols represent stimulation with BCG only, while the black squares represent BCG + INH. Comparisons were made between the naïve (blue) and historically BCG-vaccinated (red) groups. Data is presented as percentages of cell population of interest, and displayed as bar graphs with error bars represent mean ± SD. For this experiment, PBMCs from an alternate cohort of participants were used (Table 1), consisted of 16 naïve and 34 historically BCG-vaccinated participants, with a comparable demographics profile and immune responses as the main cohort. The Mann-Whitney U test was used to determine significance. The P value < 0.05 is considered statistically significant.

Figure 5

NK cells correlations. The frequency of CD3⁻ CD4⁻ CD8⁻ cells was correlated with NK cells (CD3⁻ CD56⁺) (A) and the ex vivo mycobacterial growth was associated with the NK cells frequency (Spearman’s correlation) (B). Surface-staining flow cytometry was performed to characterise NK cells. PBMCs from 16 naïve and 34 historically BCG-vaccinated participants were used.