Mycobacterium Growth Inhibition Assay of Human Alveolar Macrophages as a Correlate of Immune Protection Following Mycobacterium bovis Bacille Calmette–Guérin Vaccination

Background In order to eliminate tuberculosis (TB), an effective vaccine is urgently needed to prevent infection with Mycobacterium tuberculosis. A key obstacle for the development of novel TB vaccines is the lack of surrogate markers for immune protection against M. tuberculosis. Methods We investigated growth rates of M. tuberculosis in the mycobacterial growth inhibition assay (MGIA) as a marker for mycobacterial growth control of human bronchoalveolar lavage (BALC) and peripheral blood mononuclear cells (PBMC) before and after vaccination with Mycobacterium bovis Bacille Calmette–Guérin (BCG) of healthy adult volunteers. Results Vaccination induced a positive response (p < 0.001) to purified protein derivate (PPD) in 58.8% of the individuals in an interferon-γ release assay-ELISpot. Intraindividual evaluation of the MGIA growth rates before and after M. bovis BCG-vaccination revealed no significant difference in time to culture positivity before and after vaccination in BALC (p = 0.604) and PBMC (p = 0.199). The magnitude of the PPD-response induced by M. bovis BCG-vaccination did not correlate with growth control in BALC and PBMC (correlation = 0.468, 95% CI: −0.016 to 0.775). Conclusion In conclusion, M. bovis BCG-vaccination-induced mycobacterial-specific cytokine immune response does not result in functional immune control against M. tuberculosis in the MGIA.

Background: In order to eliminate tuberculosis (TB), an effective vaccine is urgently needed to prevent infection with Mycobacterium tuberculosis. A key obstacle for the development of novel TB vaccines is the lack of surrogate markers for immune protection against M. tuberculosis.
Methods: We investigated growth rates of M. tuberculosis in the mycobacterial growth inhibition assay (MGIA) as a marker for mycobacterial growth control of human bronchoalveolar lavage (BALC) and peripheral blood mononuclear cells (PBMC) before and after vaccination with Mycobacterium bovis Bacille Calmette-Guérin (BCG) of healthy adult volunteers.
results: Vaccination induced a positive response (p < 0.001) to purified protein derivate (PPD) in 58.8% of the individuals in an interferon-γ release assay-ELISpot. Intraindividual evaluation of the MGIA growth rates before and after M. bovis BCG-vaccination revealed no significant difference in time to culture positivity before and after vaccination in BALC (p = 0.604) and PBMC (p = 0.199). The magnitude of the PPD-response induced by M. bovis BCG-vaccination did not correlate with growth control in BALC and PBMC (correlation = 0.468, 95% CI: −0.016 to 0.775).
conclusion: In conclusion, M. bovis BCG-vaccination-induced mycobacterial-specific cytokine immune response does not result in functional immune control against M. tuberculosis in the MGIA.
Keywords: tuberculosis, vaccine, mycobacterial growth inhibition assay, Bacille calmette-guérin, vitamin D, interferon-γ release assay immune response inTrODUcTiOn Tuberculosis (TB) is a leading cause of morbidity and mortality worldwide (1). In the year 2016, the World Health Organization reported 10.4 million new cases of TB. Elimination of TB appears to be an unrealistic goal in the near future unless TB prevention can be dramatically improved. Prevention of TB can be achieved by vigorous infection control measures, treatment of individuals that are latently infected with Mycobacterium tuberculosis (LTBI) and most effectively by a preventive vaccine. Almost 100 years ago, Albert Calmette and Camille Guérin developed a vaccine based on attenuated Mycobacterium bovis [Bacillus Calmette-Guérin (BCG)], which is still the only available anti-TB vaccine in clinical use today (2,3). Although M. bovis BCG is among the most commonly used of all vaccines worldwide, its effect is largely on the attenuation of severe forms of the disease in children and it does not prevent TB in adults (4,5).
In view of this important limitation, there is an urgent need for the development of a novel TB vaccine, with the property to prevent active TB in adults and children (6)(7)(8). None of the candidates have shown a significant protective effect against TB infection (7)(8)(9)(10)(11). The main obstacle to validate a new vaccine candidate's ability to protect against M. tuberculosis infection is the lack of a robust correlate of protection (12).
The in vitro mycobacterial growth inhibition assay (MGIA) appears to be a promising marker to detect protective immunity of vaccine-induced cells toward M. tuberculosis (13)(14)(15)(16)(17). Improved immune control of mycobacterial growth was demonstrated following primary M. bovis BCG-vaccination when compared to individuals who had been M. bovis BCG-vaccinated as a child and revaccinated (15). In another human challenge trial, improved immune control was reported in individuals with past M. bovis BCG-vaccination, but no additional protective effect was seen after MVA85A-vaccination (16). And H56:CAF01-vaccination revealed an enhanced immune-control in the MGIA corresponding to in vivo protection in mice (17). So far, the MGIA has been evaluated in the following cell types and species: splenocytes (17)(18)(19) and bone-marrow-derived macrophages in mice (20), whole blood of Rhesus macaques (21), whole blood cells, frozen and fresh peripheral blood mononuclear cells (PBMC) (15,16,22,23), and antigen-expanded T-cells (24) in humans.
In order to test the MGIA as a pulmonary correlate of immune protection from mycobacterial infection, we recruited M. bovis BCG-naïve healthy adult volunteers and analyzed mycobacteriaspecific immune responses and growth of M. tuberculosis in an ex vivo growth inhibition model in bronchoalveolar lavage cells (BALCs) and PBMC before and after vaccination with M. bovis BCG.

study approval
This study complied with the declaration of Helsinki (2008) and ethics approval was obtained from the Ethics Committee of the University of Lübeck (14-091). Reporting follows the "Standards for Reporting of Diagnostic Accuracy" (STARD).

inclusion and exclusion criteria
Participants had no history of TB or M. bovis BCG-vaccination and negative IGRA results in PBMC to exclude LTBI at the beginning of the study.
To reduce potential adverse events of M. bovis BCGvaccination, volunteers were investigated for their health-status and immune-competence by testing for HIV, Hepatitis B or C-infection, analysis of the CD4/CD8 T-cell quotient and the differential blood cell-count and immunoglobulin status by serum electrophoresis. Pregnancy was excluded in female participants. Before bronchoscopy, the differential blood cell-count and coagulation parameters were analyzed repeatedly and symptom screen for any on-going viral or bacterial infection and physical examination were performed. The study flow diagram is shown in Figure 1.

cell sampling and Preparation
Flexible bronchoscopy including bronchoalveolar lavage (BAL) of the middle lobe was performed with 300 mL saline by a pulmonologist according to German national recommendations (25). Sedation was achieved by the application of intravenous midazolam and propofol.
Bronchoalveolar lavage was processed immediately to isolate the BALC as previously described (26,27). In brief, after centrifugation (10 min, 578 g, 4°C) and washing steps, cells were counted and adjusted to 10 Mio/mL in cell culture medium (Rosewell Park Memorial Institute; RPMI with 5 µg/mL Amphotericin B, 100 U/mL Penicillin G and 5% human serum-Sigma-Aldrich Corp., St. Louis, MO, USA). Lavage differential cell counts were performed on cytocentrifuged preparations stained by "hemacolor rapid staining of blood smear" (May-Grünwald-Giemsa; Merck, Germany). Thousand cells per sample were counted.
Peripheral blood mononuclear cells were isolated from venous lithium-heparin blood by ficoll density centrifugation, washed, counted, and adjusted for 10 Mio/mL in cell culture medium as previously described (28,29).
ELISpot assay results were considered conclusive if the number of SFCs in the positive control well was more than 20 SFCs after subtracting the number of spots in the negative control well and had less than twice the number of spots of the negative control well, positive if more than five spots were counted in the ESAT-6 or CFP-10 well after subtracting the number of the SFC in the negative control and if the total number of SFC was at least twice the number of the negative control, and negative in any other case.
A new vial was thawed for every experiment and dilution series of the bacterial suspension were prepared in MGIA assay medium for M. tuberculosis. Viable M. tuberculosis counts were determined by plating serial dilutions of the cultures on Middlebrook 7H10 agar plates. This enabled us to calculate the absolute number of CFU as inoculum per experiment. Furthermore, the infection inoculum was directly suspended into MGIT and tubes were placed in a BACTEC 960 instrument until tubes were detected positive. For each test series, the TTP and the concentration of M. tuberculosis measured by CFU were paired and the coefficient of variation was calculated between the experiments.
To determine the necessary bacterial inoculum to reach 156 h in TTP, which fits the midlog phase of the mycobacterial growth and is said to be the widest window of growth inhibition (19), several pre-experiments with increasing concentrations of M. tuberculosis and different cell numbers were performed ( Figure S1 in Supplementary Material).

Mycobacterial growth inhibition assay
One million of each of PBMC or BALC were infected with the dosage multiplicity of infection (MOI) of 0.058 (58,000 M. tuberculosis in CFU with 1 Mio cells) per well on a 24-well plate (Nunclon™ Delta surface, Apogent, Roskilde, Denmark) in 600 µL RPMI without HEPES and 5% (0.05 mL/mL) human serum. Since lavage is not physiologically sterile (33), all cells were cultured in the presence of antibiotics (100 U/mL penicillin G, and 5 µg/mL amphotericin B, both Biochrom AG, Berlin, Germany). As control group, M. tuberculosis was cultured for 96 h in cell culture medium without cells. Infection experiments in the MGIA were performed in duplicates, plates were incubated at 37°C and 5% CO2 atmosphere. After 96 h, the mycobacterial growth in PBMC or BALC was stopped by cell lysis with hypotonic aqua destillata. The mycobacterial suspension was transferred into the two different bacterial growth systems: MGIT and agar plates.

statistical analysis
General linear mixed-effects models were used to describe and graph the estimated effects of all predictors (except eosinophils and PPD outliners, described later) and provided p-values for differences between groups, while adjusting for the possible confounding effects of the other predictors. Because multiple observations for each individual in the study were present, the models were corrected for the correlation between observations on the same individual by including the individual identities in the model as random effects. The predictors were several groupings (such as M. bovis BCG vaccination: after/before and vitamin D: no/yes), and the numerical predictor, dilution of the outcome. The CFU counts and the fluorescence data were log-transformed to approximate normality for all analyses. As each outcome was modeled once only, all estimates, including p-values, for a specific outcome, will be adjusted for each other. McNemar tests were used to estimate the p-values for differences in eosinophils and PPD outliners in Table 1. Notched boxplots are used to illustrate the distribution of outcomes in groups. The notches represent approximate 95% confidence intervals for the medians. If the notches do not overlap, the medians of the groups probably differ significantly, and vice versa. All statistics was done in base R: a language and environment for statistical computing, and R packages lmer and lmerTest (https:// www.r-project.org/).

Demographics and Patient characteristics
Nineteen healthy volunteers (nine male, ten female) with a median age of 25 years (range 19-56 years) were recruited for this intervention study. After the first bronchoscopy, one study participant withdrew her consent and one participant was excluded due to hypotension side effects following the first bronchoscopy. Seventeen participants were vaccinated with M. bovis BCG and received a follow-up bronchoscopy 8 weeks after vaccination

infection assays
The growth of M. tuberculosis was tested in PBMC and BALC, before and after M. bovis BCG-vaccination ( Figure 4A). Com paring the capability of cells to restrict M. tuberculosis-growth before and after vaccination, no significant difference in TTP in PBMC (afterbefore = 2.2 h, p = 0.1990) nor in BALC (after-before = 0.9 h, p = 0.6040) was shown. Irrespective of M. bovis BCG-vaccination, growth (TTP) in PBMC was an estimated 7.6 h longer than in BALC before vaccination (95% CI: 4.3-10.9 h; p < 0.0001) and 9.0 h longer after vaccination (95% CI: 5.6-12.4 h; p < 0.0001).
In PBMC, the CFU counts did not differ between before and after M. bovis BCG-vaccination (estimate 0.01 log10, p = 0.7512) (Figure 4B). After M. bovis BCG-vaccination, the BALC had an estimated 0.1 lower log10 CFU count (95% CI: 0 to 0.2 h p = 0.0231) than BALC before vaccination. But this effect had to be relativized by the deviation of the infection inoculum, when each TTP of the M. tuberculosis inoculum per experiment was plotted against the corresponding TTP of the M. tuberculosis infected cell type ( Figure 5A shows TTP data, Figure 5B CFU data, respectively). M. tuberculosis growth measured as TTP or CFU was not affected by BCG-vaccination.

Vitamin D supplementation
A test series with vitamin D proved that mycobacterial growth control by human immune cells can be measured by the MGIA: in PBMC and BALC, M. tuberculosis grew slower in the presence of vitamin D, which is reflected by the estimated 17.2 h higher TTP

Facs analysis
Phenotype characterization (   we would not overestimate any effects seen, the expression of Granulysin on CD4+ T-cells was not detectable.

DiscUssiOn
We evaluated the performance of the MGIA as a marker for immune protection following M. bovis BCG-vaccination of healthy adult volunteers. M. bovis BCG-vaccination did not have a significant impact on the mycobacterial growth rate in BALC or PBMC and M. bovis BCG-induced T-cell immune response did not correlate with the mycobacterial growth rate in pulmonary or systemic mononuclear cells.
Due to the absence of an effective vaccine, M. bovis BCG is still applied in new-borns from high-incidence countries to reduce the severity of disease (4,34). Although a promising vaccine candidate MVA85A induced T-cell mediated immune responses, these signals failed to correlate with protective immunity in a clinical trial (12). In the present study, 58.  is calculated as TTP ratio after vaccination to before vaccination. The dotted line on the y-axis depicts a ratio of 1, indicating the same TTP for M. tuberculosis-growth before and after vaccination. A ratio >1 represents a reduced mycobacterial growth (increase in TTP in the second MGIA after BCG-vaccination) and a ratio <1 represents an increased mycobacterial growth (decrease in TTP in the second MGIA after BCG vaccination). TTP in bronchoalveolar lavage cells (BALC) is given in black squares, TTP in peripheral blood mononuclear cells (PBMC) in red triangles. Before vaccination, all participants had a negative interferon-γ release assay (IGRA) enzyme-linked immunospot (ELISpot) response to purified protein derivate from M. bovis BCG. On the x-axis, a change in the immune response before and after M. bovis BCG-vaccination is given as increase of netto spots per 1 million PBMC in the IGRA-ELISpot 8 weeks after vaccination. The dotted line depicts an increase of 25 netto spots per 1 million PBMC. An increase of less than 25 netto spots in the IGRA means that the immune response after BCG-vaccination stayed negative in that individual, whereas BCG-vaccination had induced an IGRA-response in these individuals, who had an increase >25 netto SFC in the second IGRA. Patients presented in the upper right quadrant have gained a successful innate immune response following vaccination (better growth control of M. tuberculosis and a new positive cytokine immune response), whereas patients situated in the right lower quadrant are having a new positive cytokine response but miss an enhanced growth control. The upper left quadrant shows better growth control without any adaptive immune response. This overall spread depicts the lack of correlation of immune responses and growth control measured by the MGIA. indicates a lack of protective immunity following M. bovis BCG vaccination (4,5). The missing immune response after M. bovis BCG-vaccination fits to epidemiological data, showing that M. bovis BCG-vaccination does not protect from tuberculosis infection and can only reduce the rates of miliary tuberculosis and tuberculosis meningitis in children (4,35).
Our results support previous findings that a positive vaccineantigen-specific IGRA immune response does not correlate with functional immune control for mycobacterial infection in vivo and is, therefore, an inappropriate marker to predict immune protection (12).
The MGIA was presented as a promising surrogate for immune protection in previous studies (16,18). Vitamin D3 [1α,25(OH)2-Cholecalciferol] is known for its immune-modu lating effects on M. tuberculosis-control (36)(37)(38) and for inducing an antimicrobial activity of human monocytes/macrophages toward M. tuberculosis. In a vulnerable population with genetic variations in the vitamin D pathway, an adjuvant treatment with high-dose vitamin D was shown to accelerate the time to sputum culture conversion (39). Possible cellular mechanisms were observed to be associated with expression of cathelicidin (40). In our study, this vitamin D-effect is used as a model to monitor improved immune control as expected in an effective vaccine. In the presence of vitamin D, mycobacterial growth rates decreased (measured as increase in TTP or lower CFU counts) confirming the functionality of the MGIA. Our findings corroborate previous studies with regard to the antimycobacterial activity of vitamin D (38,41).
Our findings differ from previously published results, where M. bovis BCG-vaccination was associated with improved mycobacterial growth control in humans (15,16). We avoided the potent antituberculosis drug streptomycin and used freshly isolated BALC or PBMC instead of cryopreserved human PBMC or whole blood, third, in our setting, the infection was performed with M. tuberculosis and not with M. bovis BCG. So far, M. tuberculosis has only been used in murine splenocytes (17)(18)(19) or murine bone-marrow-derived macrophages (20) to investigate vaccine-induced changes of immune control against M. tuberculosis. With regard to vaccine-induced growth inhibition in human cells, the MGIA was successfully performed with M. bovis BCG in either PBMC and/or whole blood (15,16,22). To our knowledge, the current study represents the first description of the MGIA as a surrogate marker for vaccine-induced antimycobacterial activity using M. tuberculosis in human cells (in PBMC and especially in human lung cells), an aspect that has not been addressed so far.
Compared to the mentioned previous studies, which have used MOIs between 0.0001 and 0.0002, it needs to be stated that the MOI of 0.058 used in the current study is significantly higher. In order to make sure that this MOI would allow the measurement of an enhanced antituberculosis response, we included vitamin D to our experiments. As we observed a vitamin D-dependent average TTP increase by 17.2 h in PBMC and 20.6 h in BALC (Figure 6A), we concluded that even at an MOI of 0.058, a given vaccination-dependent antituberculosis directed response leading to prolonged TTP values should have been detectable. This was not the case. Nevertheless, we cannot exclude, that different experimental conditions such as the use of a lower MOI could have led to different results. The current study may, however, indicate that the type of bacteria used in the MGIA needs to be considered. Several authors have shown that M. bovis BCG-vaccination of human individuals induces growth inhibition of M. bovis BCG, whereas our experimental data suggest that M. bovis BCG-vaccination does not lead to growth control of M. tuberculosis by human PBMC and BALC.
The TTP in BALC was significantly lower than TTP in PBMC, implicating higher growth rate in BALC. This effect was not seen in CFU read-outs, which may indicate that CFU and TTP measure different conditions. The MGIT-device detects fluorescence, if actively respiring micro-organisms consume the oxygen. In comparison, CFU documents mycobacteria in colonies on agar plates, where M. tuberculosis grows typically eugone in firmly attached, beige, dry crumbles, but may differ in colony size. Latter observation may indicate a substantially reduced replication status of a given tiny/small colony. Microbiologists have described these small cell populations as persisters or dormant, non-growing bacteria with the capacity to regrow or viable but not-culturable (42). Regarding to M. tuberculosis, it was shown that the replication rate is reduced in small persister-type M. tuberculosis in comparison to a population of larger, cultureviable mycobacteria (43). In vivo replicating and dormant bac terial populations exist within the same patient (43)(44)(45). We assume that small, slow-replicating colonies will be counted in the CFU analysis, but may be left underestimated in the MGIA assay, which documents rates of metabolism and growth by oxygen consumption and explain the observed differences between the TTP and CFU analysis.
Another reason to explain this difference in growth behavior might be cell differences between the lung and the blood compartment (46). Lavage consisted of significantly less lymphocytes than PBMC and additionally a lower number of cytotoxic CD8 + T-cells expressing Perforin was found in BALC in comparison with PBMC. These results once more demonstrate that in vitro experiments with PBMC may not imitate the cell-based immune responses and control in human TB (47)(48)(49).
Our study is limited by the number of participants and by the lack of an alternative and more promising vaccine candidate for the immune intervention. Although other candidate markers such as mRNA transcripts (50,51) or cytokines (52) may also appear promising as correlates for protection from M. tuberculosis infection, our findings may serve as valuable proof-ofconcept for future vaccine evaluations.
In conclusion, vaccination with M. bovis BCG did not result in ex vivo growth inhibition of M. tuberculosis in human alveolar macrophages from healthy volunteers despite the development of antimycobacterial immune responses in peripheral blood.

eThics sTaTeMenT
This study was carried out in accordance with the recommendations of the Declaration of Helsinki (2008) and the Ethics Committee of the University of Lübeck. The protocol was Immune Responses After BCG-Vaccination Frontiers in Immunology | www.frontiersin.org July 2018 | Volume 9 | Article 1708 approved by the Ethics Committee of the University Lübeck (14-091). All subjects gave written informed consent in accordance with the Declaration of Helsinki.