Adjuvant modulation of the cytokine balance in Mycobacterium tuberculosis subunit vaccines; immunity, pathology and protection

Abstract

It is known that protection against tuberculosis is mediated primarily by T helper type 1 (Th1) cells but the influence of the Th1/Th2 balance of a vaccination response on the subsequent protection and pathology during infection has not been studied in detail. We designed a panel of Ag85B-ESAT-6 subunit vaccines based on adjuvants with different Th1/Th2-promoting activities and studied cellular responses, bacterial replication and pathology in the lungs of mice infected with Mycobacterium tuberculosis. All vaccines induced cell-mediated and humoral responses but with markedly different interferon-γ : interleukin-5 (IFN-γ : IL-5) and immunoglobulin G1 (IgG1) : IgG2 ratios. The vaccines promoted different levels of control of bacterial replication with the most efficient protection being exerted by cationic liposomes containing monophosphoryl lipid A and low to completely absent immunity with conventional aluminium. The level of protection correlated with the amount of IFN-γ produced in response to the vaccine whereas there was no inverse correlation with the level of IL-5. Characterizing a protective response was an accelerated recruitment of IL-17 and IFN-γ-producing lymphocytes resulting in the early formation of granulomas containing clustered inducible nitric oxide synthase-activated macrophages. In comparison, non-protected mice exhibited a different inflammatory infiltrate rich in neutrophil granulocytes. This study indicates that the adjuvant component of a tuberculosis vaccine may be crucial in determining the kinetics by which effective granulomas, pivotal in controlling bacterial growth, are formed.

Introduction

Interferon-γ (IFN-γ) is a key effector molecule in tuberculosis (TB), responsible for macrophage activation and mycobacterial killing. In animal models its essential role was clearly demonstrated by infecting mice with a targeted disruption of the gene for IFN-γ.^1,2 Furthermore, humans with a disruption of the IFN-γ signalling pathway show abnormal susceptibility to mycobacteria,³ a condition that is reversed by the administration of synthetic IFN-γ.⁴ The central role of IFN-γ means that most novel TB vaccine strategies have focused on generating strong T helper type 1 (Th1) responses either by using strong, Th1-polarizing adjuvants^5,6 or by heterologous prime-boost approaches.⁷ The relevance of Th1 priming by these vaccines is indisputable and is supported by challenge studies involving experimental vaccines that induce different levels of IFN-γ.⁸ With the first novel vaccines now entering clinical trials, IFN-γ has emerged as a surrogate marker for the protection conferred by the trial vaccine.⁹

The induction of a strong Th2 response in animal models has clearly been demonstrated as detrimental to protective immunity,¹⁰ and switching a Th1 response to a Th2 response by concurrent helminth infection influences the cellular composition and structure of the granulomas.¹¹ Supporting the negative influence of a Th2 response, clinical studies involving patients with latent TB show a clear correlation between the size of a Th2 response and the risk of developing active disease and in particular a direct correlation between the level of interleukin-4 (IL-4) messenger RNA and disease severity.¹² However, that Th2 responses in TB may also have a positive influence was suggested by data from Sugawara et al., who reported an increased bacterial load in the lungs of IL-4-deficient mutant mice.¹³ In this context some level of Th2 response has been suggested as beneficial in TB through counterbalancing a too vigorous Th1 response with its undesirable side effects. Hence, IFN-γ has been associated with the development of mycobacteria-induced caseous necrosis with excessive amounts of this cytokine contributing to the pathology observed in tuberculous patients.¹⁴ Recent in vitro studies have furthermore demonstrated that Th2 cytokines mediate an alternative pathway of macrophage activation not aimed at the elimination of intracellular pathogens but at counter-regulating Th1 responses.¹⁵ It therefore remains unresolved if vaccine-promoted protection against TB correlates with the magnitude of the Th1 response, whether there is an inverse correlation with a counterproductive Th2 response or if a balance between the two arms of the cellular response is optimal.

In the present study, we decided to take advantage of recent progress in adjuvant development that has now made it possible to design formulations that induce immune responses with defined differences in the Th1/Th2 balance. We formulated these adjuvants with the Ag85B-ESAT-6 fusion molecule, previously identified as a promising vaccine in various models of TB.¹⁶ We addressed three questions pertinent to current TB vaccine research: (1) does protection against TB correlate with the magnitude of the Th1 response, inversely with the Th2 response or is a balanced immune response induced by vaccination desirable; (2) does vaccine-induced protection correlate with quantitative or qualitative differences in the T-cell population recruited to the site of infection; and (3) given the central role of the TB granuloma in mycobacterial control and pathology, how is the size and morphology of this lesion influenced by the Th1/Th2 balance of the vaccine-induced response?

Materials and methods

Animals

Female BALB/c or C57BL/6 mice, 6–12 weeks old, were obtained from Harlan Scandinavia (Allerod, Denmark). Mice infected with mycobacteria were kept in cages within a BL-3 laminar flow safety enclosure. Experiments were conducted in accordance with the regulations of the Danish Ministry of Justice and animal protection committees and in compliance with European Community Directive 86/609.

Adjuvants and immunization

Mice were immunized subcutaneously (s.c.) at the base of the tail three times with a 2-week interval between each immunization. A panel of four different adjuvants was designed to obtain formulations inducing only Th2 responses [aluminium hydroxide, Al(OH)₃], mixed Th2–Th1 responses [Al(OH)₃/dimethyldioctadecylammonium (DDA)], weak Th1 responses (DDA), or strong Th1 responses [DDA/monophosphoryl lipid A, (MPL)] as described in Table 1. Five hundred microgram Al(OH)₃ (2% Alhydrogel, Brenntag Biosector, Frederikssund, Denmark) was added to the antigen and mixed with saline before immunization. Vaccines containing DDA (250 μg) and DDA/MPL (25 μg) (both Avanti Polar Lipids, Alabaster, AL) were prepared as previously described.¹⁷ All mice were immunized with 2 μg of the vaccine antigen Ag85B-ESAT-6 emulsified in adjuvant in a total volume of 0·2 ml. The Ag85B-ESAT-6 antigen was produced as a recombinant protein as previously described.¹⁶

Table 1.

Panel of T helper type 2 (Th2)/Th1 polarized adjuvants

Adjuvant profile	Adjuvant component1
	Al(OH)3 (μg)	DDA (μg)	MPL (μg)
Th2	500	–	–
Mixed Th2–Th1	500	250	–
Th1	–	250	–
Strong Th1	–	250	25

¹Constituents for a single mouse dose.

Bacteria

Mycobacterium tuberculosis Erdman was grown at 37° in modified Sauton medium enriched with 0·5% glucose and 0·5% sodium pyruvate.

Experimental infections

Immunized mice were challenged 10 weeks after the first immunization by the aerosol route using a Glas-Col inhalation exposure system (Inhalation Exposure System; Glas-Col, Terre-Haute, IN) calibrated to deliver 25 colony-forming units (CFU) of M. tuberculosis Erdman into the lungs. For evaluation of vaccine efficacy, bacterial loads were determined 6 weeks post-infection by plating serial dilutions of lung homogenates onto Middlebrook 7H11 agar (Becton Dickinson, Oxford, UK). Mycobacterial colonies were quantified following 2–3 weeks of incubation at 37° and the numbers were expressed as the log₁₀ values of the geometric mean for six mice.

As a positive control for protective efficacy of experimental subunit vaccines compared to the conventional TB vaccine, bacillus Clamette–Guérin (BCG), a single group of mice received one dose of BCG Danish 1331, 5 × 10⁶ CFU, injected s.c. at the base of the tail.

Cellular assays

Blood was obtained by periorbital puncture 7 days after the final vaccination, and the blood lymphocytes were purified as previously described.¹⁸ Lungs were perfused with heparin containing phosphate- buffered saline (PBS; SSI, Copenhagen, Denmark) to minimize contamination of the final lung preparation with blood cells and were subsequently homogenized using a 100-μm nylon cell strainer (BD Biosciences, Bedford, MA). All cell cultures were performed in microtitre plates (Nunc, Roskilde, Denmark) containing 2 × 10⁵ cells in a volume of 200 μl RPMI-1640 supplemented with 5 × 10⁻⁵ m 2-mercaptoethanol, 1 mm glutamine, 1% pyruvate, 1% penicillin–streptomycin, 1% HEPES, and 10% fetal calf serum (all from Gibco Invitrogen, Carlsbad, Denmark). Ag85B-ESAT-6 was used at 5 μg/ml. Wells containing medium only or 5 μg/ml concanavalin A were used as negative and positive controls respectively. Culture supernatants were harvested after 72 hr incubation and the amounts of IFN-γ, tumour necrosis factor-α (TNF-α), IL-17 and IL-5 were determined by enzyme-linked immunosorbent assay (ELISA) as previously described.¹⁹ To investigate the contribution of CD4 and CD8 T cells to IFN-γ production, the T-cell receptors were blocked by adding monoclonal antibodies to the cultures as previously described.²⁰

Evaluation of antibody titres

The presence of Ag85B-ESAT-6 specific antibodies was evaluated as previously described.²⁰ Individual mouse sera from six mice were analysed in duplicate in five-fold dilutions (beginning at a 1/20 dilution). Reagents used were rabbit anti-mouse immunoglobulin G1 (IgG1) and IgG2a/b from Zymed (San Francisco, CA), and 3,3′,5,5′-TetraMethylBenzindin substrate (Kem-En-Tec, Copenhagen, Denmark). The plates were read on an ELISA reader at 450 nm. The absorbance values were plotted as a function of the reciprocal dilution of serum samples. The data were fitted by non-linear regression analysis with a sigmoidal dose–response curve of variable slope by the GraphPad Prism program (version 4.00; GraphPad Software Inc., San Diego, CA). Antibody titres were then defined as the serum dilution that resulted in an absorbance of 1·00 in the parallel portion of the curves as previously described.²¹ As the gene for IgG2a is deleted in C57BL/6 mice,²² titres of IgG2b were evaluated in this strain.

Fluorescence-activated cell sorting analysis

Homogenized lung preparations were purified as described above and 0·5 × 10⁶ to 1·0 × 10⁶ cells were restimulated overnight with 5 μg/ml Ag85B-ESAT-6. Brefeldin A (Sigma, Brondby, Denmark) was added to a final concentration of 2·25 μg/well, and the cultures were incubated for a further 4 hr. Non-specific binding was blocked by a 15-min incubation with the 24G2 clone (CD16/CD32; BD Pharmingen, San Diego, CA) and subsequently stained with surface markers for 20 min. Intracellular cytokine staining was performed using the Cytofix/Cytoperm kit (BD Pharmingen) according to the manufacturer's protocol and using α-phycoerythrin-IFN-γ. Cells were finally washed three times, resuspended in paraformaldehyde and analysed with a FACScan (Becton Dickinson, Mountain View, CA) by collecting 50 000 events. Surface staining was performed with one or more of the following antibodies (all from BD Pharmingen): α-CD4-peridinin chlorophyll protein, α-CD8-fluorescein isothiocyanate, α-CD19-allophycocyanin, α-CD62L-fluorescein isothiocyanate.

Histopathological, histomorphometric and immunohistochemical analyses

Lungs were removed post mortem before and at day 4, 7, 14, 21 and 28 after aerosol infection with M. tuberculosis (six mice per group per time-point). The right lung lobe of each mouse was fixed by immersion in 10% neutral-buffered formalin and processed for histological examination. Cut sections were stained using haematoxylin & eosin and Ziehl–Neelsen techniques and were evaluated without prior knowledge of stage of infection or of treatment group. Lesions were quantified using computer-aided histomorphometry (Palm®robo software, version 1.2.3; Palm Microlaser Technologies AG Ltd, Bernried, Germany). Immunolabelling of inducible nitric oxide synthase (iNOS) within lesions was carried out following dewaxing and rehydration of sections and an epitope retrieval step, where tissue slides were microwaved for 20 min in tri-sodium citrate solution (pH 6·0). Subsequently, treatment of sections with normal goat serum blocking solution (Vectastain ABC kit; Vector Laboratories, Burlingame, CA) was performed to block endogenous peroxidase. The primary antibody (rabbit polyconal anti-mouse iNOS/NOSII, Upstate, Lake Placid, NY) was then applied for 1 hr at 1/1000 dilution followed by sequential application of biotinylated goat anti-rabbit IgG (Vector Laboratories) and ABC solution for 30 min respectively at room temperature (Vectastain ABC kit, Vector Laboratories). Visualization of target cells followed application of diaminobenzidine-tetrahydrochloride solution (Sigma-Aldrich, Steinheim, Germany) and a haematoxylin counterstain. The modified Ziehl–Neelsen staining technique was then carried out on these sections before dehydration and the application of coverslips.

Statistical analyses

Correlation between IFN-γ levels and CFU were evaluated by Pearson correlation. For comparison of cellular recruitment and cytokine secretion, analyses of variance were used followed by Tukey test. For evaluation of vaccine efficacy, differences in the number of bacterial colonies between immunized and unimmunized mice were tested by analyses of variance. When significant differences were indicated, mean CFU of vaccinated mice was compared to mean CFU of unimmunized mice using Dunnett's test. A similar statistical analysis was carried out to compare the number of lesions and total pulmonary lesion area between the treatment groups.

Results

Protection against TB correlates with the magnitude of the vaccine-induced Th1 response

A panel of four adjuvant formulations that induced immune responses with different degrees of Th1/Th2 balance was established (Table 1). The experiment was performed using two mouse strains (BALB/c and C57BL/6) with different intrinsic abilities to generate a Th1 response.²³ Analysis of the post-immunization peripheral immune responses was measured to evaluate Th1 (IFN-γ) and Th2 (IL-5) cytokine production as well as the secretion of antigen-specific antibodies. As shown in Fig. 1(a), the adjuvant panel represented a range of Th1/Th2-inducing responses in C57BL/6 mice with alum [Al(OH)₃] giving rise to a Th2 response, the combination of Al(OH)₃ and DDA resulting in a mixed Th2/Th1 response characterized by the simultaneous secretion of both IL-5 and IFN-γ, and DDA and DDA/MPL giving rise to Th1 responses with DDA/MPL being the strongest inducer of IFN-γ. Antigen-specific IL-17 production was also induced and followed the same pattern (DDA/MPL 2907 ± 474 pg/ml; DDA 1120 ± 393 pg/ml; Al(OH)₃/DDA 11 ± 6 pg/ml; Al(OH)₃ 0 pg/ml). Similar results were obtained in BALB/c mice although the responses in general were less prominent in this mouse strain (Fig. 1b). The different adjuvants did not alter the specificity of the T-cell response because in all groups this was directed against ESAT-6_1–15 and Ag85B_241–255, previously identified as the major CD4 T-cell epitopes of these antigens (results not shown),^17,24 whereas no CD8 T-cell response was observed. Analysis of the antigen-specific antibodies generated supported the cytokine data with the Al(OH)₃-based vaccine exclusively inducing antibodies of the IgG1 isotype (Fig. 1c,d). In the C57BL/6 strain this difference was very pronounced with DDA/MPL eliciting more than five-fold higher levels of IgG2b compared to IgG1.

Figure 1.

Polarized vaccine-induced immune responses generated by T helper type 1 (Th1)/Th2 panel of adjuvants. Release of interleukin-5 (IL-5) and interferon-γ (IFN-γ) from blood lymphocytes isolated from C57BL/6 (a) or BALB/c (b) mice immunized with 2 μg Ag85B-ESAT-6 in Al(OH)₃, Al(OH)₃/DDA, DDA and DDA/MPL. Blood lymphocytes were isolated 5 weeks after the first immunization and restimulated in vitro with 5 μg Ag85B-ESAT-6. Antigen-specific antibody midpoint titres in serum of C57BL/6 mice (c) measured as IgG1 and IgG2b or BALB/c mice (d) measured as IgG1 and IgG2a. Antibodies were measured in the serum 7 weeks after the first immunization.

The protective efficacy of the various adjuvant combinations was measured 6 weeks after the final immunization by the administration of an aerosol challenge with M. tuberculosis Erdman (Table 2). There was a high degree of correlation (r = 0·93 and 0·80 for BALB/c and C57BL/6, respectively) between the IFN-γ level and protection with DDA/MPL inducing the highest level of protection (P < 0·05) in both mouse strains. In contrast, there was no inverse correlation between protection and the level of IL-5 induced by the adjuvant combinations in either of the two strains (r = 0·47 and 0·52). Although in general IL-5 production was observed in the minimally protected vaccine groups, the levels observed did not mirror the differences seen in protection. This was most pronounced in C57BL/6 mice, in which the combined Al(OH)₃/DDA adjuvant was as effective as the DDA adjuvant despite producing seven times higher IL-5 levels. These results suggest that vaccine efficacy correlates with the induction of a Th1 response and not with the absence of a Th2 response.

Table 2.

Vaccine-induced protection against an aerosol infection with Mycobacterium tuberculosis

Vaccine1	Log10 CFU ± SEM2
	BALB/c	C57BL/6
Al(OH)3	5·69 ± 0·11	4·95 ± 0·06
Al(OH)3/DDA	5·74 ± 0·17	4·56 ± 0·17
DDA	5·38 ± 0·12*	4·79 ± 0·12
DDA/MPL	4·87 ± 0·15*	4·30 ± 0·06*
BCG	N.D.	4·17 ± 0·12*
Naive	5·79 ± 0·06	4·90 ± 0·07

N.D., not determined.

¹Mice (n = 6) were immunized with 2 μg Ag85B-ESAT-6 in the indicated adjuvant or 5 × 10⁶ CFU of BCG.

²Log₁₀ CFU in the lungs 6 weeks after aerosol challenge.

^*P < 0·05 compared to naïve.

Protection against TB correlates with accelerated recruitment of IFN-γ-positive CD4 effector cells

Detailed analysis of the cells entering the lungs was carried out to investigate whether protection correlated with quantitative or qualitative differences in lymphocyte populations recruited to the site of infection. In both strains of mice the most polarized responses were mediated by Al(OH)₃ and DDA/MPL. Lungs from these groups were harvested before infection and at various time-points after aerosol infection with M. tuberculosis. Lungs were perfused with heparinized PBS to remove contaminating blood and the lung tissue was disrupted to facilitate purification and analysis of the infiltrating lymphoid cells. Hence, the phenotypic characterization was performed by flow cytometry and lung cells were restimulated in vitro with Ag85B-ESAT-6 for cytokine analysis (Fig. 2). Events in mice immunized with the DDA/MPL-adjuvanted vaccine were characterized by early pulmonary recruitment of CD4 T cells. Significantly higher levels of CD4 T cells were found in this group compared to unvaccinated mice at day 14 (P < 0·001) and levels reached a 10-fold increase by day 28. In both unvaccinated mice and in the Al(OH)₃ group there was little CD4 T-cell recruitment at the early time-points and more limited increases in the pulmonary levels at day 28, than in the DDA/MPL-adjuvanted vaccine group. There was no significant difference in CD4 T-cell recruitment between the naïve and Al(OH)₃ groups at any time-point during infection. Earlier mobilization of CD8 T cells was also noted in the lungs of the DDA/MPL group; however, by day 28 all groups exhibited comparable pulmonary CD8 T-cell numbers. Although B cells were recruited to the lungs of the DDA/MPL group to a lesser extent than CD4 T cells, significant differences in pulmonary B-cell levels were noted between DDA/MPL-immunized mice and the other groups from day 14. Analysis of in vitro cytokine responses after restimulation with vaccine antigen showed that concurrent with the pulmonary recruitment of CD4 T cells in the DDA/MPL group, large amounts of IL-17 and IFN-γ were released in response to Ag85B-ESAT-6 antigen. Following the same kinetics, antigen-specific TNF-α secretion was found in mice receiving DDA/MPL whereas only negligible cytokine production was noted in the Al(OH)₃ group and in naïve mice with levels only detectable at day 28. Both the DDA/MPL- and the Al(OH)₃-adjuvanted vaccine groups had detectable levels of antigen-specific IL-5 secretion before infection. This was followed by an early transient increase in IL-5 production at day 7 and a subsequent reduction to a low but detectable level at the remaining sampling points. This was in contrast to the naïve group, in which no IL-5 was detectable. No IL-10 was observed at any time in any group (results not shown). Groups of mice receiving either DDA/MPL or Al(OH)₃ without Ag85B-ESAT-6 followed the pattern in cytokine release and cellular recruitment seen in unvaccinated control animals (results not shown).

Figure 2.

Accelerated emergence of cell subsets and vaccine-specific cytokine responses during infection with Mycobacterium tuberculosis in mice vaccinated with Ag85B-ESAT-6 in DDA/MPL. Lung tissue of unvaccinated C57BL/6 mice or mice immunized with Ag85B-ESAT-6 in Al(OH)₃ or DDA/MPL (n = 3) were disrupted and analysed for the presence of CD4 T cells, CD8 T cells, and B cells (CD19⁺) by flow cytometry before infection and at different time-points after infection with Mycobacterium tuberculosis Erdman. Interferon-γ (IFN-γ), tumour necrosis factor-α (TNF-α), interleukin-17 (IL-17) and IL-5 responses were measured by ELISA after restimulation in vitro with 5 μg Ag85B-ESAT-6. *P < 0·05, **P < 0·01 and ***P < 0·001 compared to the corresponding values for naïve mice.

We continued by investigating the phenotype of the IFN-γ-producing cell subset in the DDA/MPL group using flow cytometry followed by intracellular staining of antigen restimulated cells isolated at day 14 post-infection. The majority of the IFN-γ-positive cells were localized in the CD4 T-cell population (5·98% after stimulation with Ag85B-ESAT-6 compared to 1·52% in unstimulated control wells), whereas a minor proportion of CD8 T cells were IFN-γ-positive (Fig. 3a) indicating that the CD4 T cell response induced by vaccination was boosted during infection. The same pattern was observed when using peptides for restimulation (results not shown). Further confirming that CD4 T cells formed the major subset of cells responsible for the high levels of IFN-γ was the almost complete abrogation of response by the addition of anti-CD4 antibodies in the culture (Fig. 3b). Assessment using other phenotypic markers revealed a highly activated profile for the IFN-γ-positive CD4 T cells with 80% of the cells being within the CD62L^low effector population and with upregulated levels of the activation marker CD69 (results not shown).

Figure 3.

Lung interferon-γ (IFN-γ) responses predominant in the CD4 T-cell subsets. (a) Lung tissue was obtained from C57BL/6 mice immunized with Ag85B-ESAT in DDA/MPL 2 weeks after infection and the IFN-γ responses analysed in CD8 and CD4 T-cell subsets using flowcytometry. Percentage of IFN-γ-positive cells after restimulation in vitro with 5 μg Ag85B-ESAT-6 within the two subsets has been indicated with values for unstimulated control wells shown in parenthesis. (b) Lung tissue obtained from mice immunized with Ag85B-ESAT-6 in DDA/MPL 4 weeks after infection was cultured in the presence or absence of anti-CD4 or anti-CD8 antibodies and the IFN-γ response was measured by ELISA.

TB protection is associated with rapidly emerging lymphocyte-rich granulomas

Histopathological, histomorphometric and immunohistochemical examinations were carried out to assess the influence of the Th1/Th2 balance on lesion number, size and morphology at various time-points after aerosol infection. In all animals small, multifocal, well-demarcated, unencapsulated pulmonary granulomas were observed, frequently adjacent to bronchioles, from day 14 onwards. At day 14, granulomas in mice receiving DDA/MPL were more numerous and covered a larger total pulmonary surface area than those of both the Al(OH)₃-immunized and naïve groups (Fig. 4). Following this initial, more rapid, granuloma development in the DDA/MPL group, further development was arrested with the number of granulomas maintained at relatively constant levels.

Figure 4.

Computer-aided lesion morphometry indicated more numerous granulomas (a) covering a larger total pulmonary surface area (b) in C57BL/6 mice (n = 6) vaccinated with Ag85B-ESAT-6 in DDA/MPL adjuvant relative to those vaccinated using Al(OH)₃ adjuvant or unvaccinated at different time-points after challenge with 25 CFUs Mycobacterium tuberculosis Erdman by the aerosol route. *P < 0·05 and ***P < 0·001 compared to the corresponding values for naive mice.

The histopathological appearance of granulomas in the DDA/MPL group at day 14 differed from the other two groups, consisting of closely apposed macrophages including many with abundant, finely microvesiculated foamy cytoplasm and admixed lymphocytes (Fig. 5b). In comparison, granulomas of Al(OH)₃-immunized and naïve mice contained relatively large numbers of intact and degenerate neutrophils and relatively few lymphocytes and macrophages (Fig. 5a,c). The location and number of acid-fast organisms, as well as the distribution of iNOS-positive macrophages, also differed in granulomas in the DDA/MPL-treated mice relative to those of Al(OH)₃-immunized and naïve animals. Few acid-fast bacteria were observed in the lesions of the DDA/MPL-treated mice and they were rarely identified outside iNOS-positive macrophages (Fig. 5b). This contrasted with lesions in Al(OH)₃-treated and naive mice where larger numbers of acid-fast bacteria were observed both within and outside iNOS-positive macrophages (Fig. 5a,c). This finding suggests that vaccination with DDA/MPL enhances the killing capacity of the macrophages through more rapid upregulation of iNOS activity. The distribution pattern of iNOS-positive macrophages within granulomas also differed between the treatment groups at day 14, in that clusters of up to 40–50 iNOS-positive macrophages were observed in the lesions of the DDA/MPL-immunized mice whereas in the two other groups, iNOS-labelled macrophages were frequently observed in isolation or in small clusters of three to four cells. By day 28, however, similar large clusters of iNOS-positive macrophages were observed in the granulomas of all three treatment groups and the total surface area covered by granulomas in all three groups had increased in size relative to the day 14 time-point (Fig. 4). Also at this late time-point the lesions in the DDA/MPL group contained greater numbers of lymphocytes and acid-fast bacilli remained much less numerous within the contained large clusters of iNOS-positive macrophages (Fig. 6b).

Figure 5.

Photomicrographs with insets of lung tissue 14 days after challenge with 25 CFUs Mycobacterium tuberculosis Erdman by the aerosol route illustrating reduced numbers of acid-fast bacteria in large tight clusters of inducible nitric oxide synthase (iNOS)-positive macrophages and reduced numbers of neutrophils in C57BL/6 mice vaccinated with Ag85B-ESAT-6 in DDA/MPL adjuvant (b), relative to those vaccinated using Al(OH)₃ adjuvant (a) or unvaccinated control animals (c). Images in the left-hand panels are stained using haematoxylin & eosin and in right-hand panels are double-stained using immunohistochemistry for iNOS followed by the modified Ziehl–Neelsen method for acid-fast bacteria. The images shown are from one representative mouse of six in each group. Magnification × 400.

Figure 6.

Photomicrographs with insets of lung tissue 28 days post-challenge with 25 CFUs of Mycobacterium tuberculosis Erdman by the aerosol route illustrating reduced numbers of acid-fast bacteria within inducible nitric oxide (iNOS)-positive macrophages and increased numbers of lymphocytes in C57BL/6 mice vaccinated with Ag85B-ESAT-6 in DDA/MPL adjuvant (b), relative to those vaccinated using Al(OH)₃ adjuvant (a) or unvaccinated control animals (c). Images in left-hand panels are stained using haematoxylin & eosin (magnification × 200) and in right panels double stained using immunohistochemistry for iNOS followed by the modified Ziehl–Neelsen method for acid-fast bacteria (magnification × 400). The images shown are from one representative mouse of six in each group.

Discussion

An increasing number of studies suggest that efficient immunity against TB is associated not only with elevated Th1 responses but also with tightly controlled Th2 responses.²⁵ In human cohorts the monitoring of IL-4 and its antagonistic splice variant IL-4d2 as well as IFN-γ has demonstrated that exposed individuals who are prone to developing disease, at very early time-points exhibit an immune response that is skewed away from a protective Th1 response and that this Th2 profile is maintained as these individuals later progress to active TB.^12,26 In addition, an association between systemic IL-5 and progression to active TB was found in a study involving more than 600 human immunodeficiency virus-negative Ugandan adults with the trend being most pronounced in the group of BCG-vaccinated individuals.²⁷ This was interpreted as the existence of a Th2 bias at the time of vaccination resulting in increased susceptibility to TB. Although at present no experimental data support the hypothesis, it may have significant relevance for vaccine design and development. Hence, preventive TB vaccines with a Th2 component could theoretically result in increased susceptibility to TB or even exacerbate disease, as observed by the negative influence of the Th2 adjuvant aluminium used in early studies of adjuvanted subunit vaccines.¹⁰

In the current study, we designed a panel of Th1/Th2-inducing adjuvants to generate polarized immune responses in vivo. We used the aerosol route of infection to model natural infection and analysed the cellular recruitment at the site of infection in the lung. Taken together, our data indicate that vaccine protection correlates with the magnitude of the IFN-γ response induced by the vaccine resulting in an accelerated pulmonary recruitment of IFN-γ-producing T cells during disease. In agreement with recent data, the IFN-γ recall response was associated with the appearance of IL-17 secretion.²⁸ The data do not provide evidence of an inverse correlation between protection and the Th2 response because animals vaccinated with adjuvant formulations inducing a Th2 bias were still protected as long as the IFN-γ response was substantial. A certain degree of Th2 stimulation may theoretically be beneficial as Th2 cells, in addition to their direct effector function, also downregulate the Th1 response.¹⁵ In particular, IL-17 and IFN-γ are known to cause severe immunopathology if not tightly regulated²⁹ and Th2 cytokines therefore may counterbalance the inflammatory environment, often resulting in granuloma necrosis.¹⁴ However, based on the protection in the lung our data do not support the theory that a balanced immune response containing both Th1 and Th2 components is advantageous because the most efficient vaccine in both strains of mice was the most Th1-polarized DDA/MPL vaccine. Therefore, a polarized response with a complete absence of a Th2 response may be necessary for the manifestation of a maximal Th1 response.

Investigation of the early cellular events leading to the establishment of protective lesions showed an accelerated accumulation of lymphocytes (both CD4 and CD8 T cells and B cells) at day 14 whereas the influx of all lymphocyte subsets in the Al(OH)₃-treated and naïve groups was delayed. The delayed recruitment of cells in the Al(OH)₃ group could reflect suboptimal expansion of the T-cell population as a result of limited antigen availability, as demonstrated recently in the case of CD8 T cells.³⁰ While such reduced antigen availability is considered a problem for some non-replicating vaccines³¹ it is less likely to occur with adjuvant preparations such as Al(OH)₃, which create a depot at the site of injection that results in prolonged exposure to the antigen.³² Indeed, use of the Al(OH)₃-based vaccine led to maintenance of both detectable antigen-specific cytokine and very high antibody responses (at the same level as in the DDA/MPL group) as late as 5 months after immunization (data not shown). Alternatively, as different subpopulations of T cells exhibit distinct migratory capacities the lack of cellular recruitment in the Al(OH)₃ group could reflect an inherent poor homing capacity of Th2 cells to the lung.³³ In this context, Iezzi et al. compared the accumulation of Th1 and Th2 cells generated in vitro in various organs and found lower numbers of Th2 cells in bronchoalveolar lavage fluid³⁴ presumably as a consequence of differential expression of adhesion and chemokine receptors.^35,36 Paradoxically, Th2 cells clearly have the potential to migrate into lung tissue in the presence of a strong inflammatory signal, as suggested by the tendency for respiratory infections to exacerbate asthma.³⁷ With the recent discovery of a third T-cell subset (Th-17 cells) regulated by IL-23, which is of particular importance in the inflammatory response, there is clearly a need for further detailed studies on the signals that control the trafficking of T-cell subsets and their importance in the expression of vaccine-induced pulmonary immunity.³⁸

Our study demonstrates a profound influence of the Th1/Th2 balance on the developmental kinetics as well as on the structure and cellular composition of the granuloma during infection. In the Th1 skewed mice we observe accelerated granuloma formation and, as indicated by flow cytometric analysis, the most clearcut difference between protected versus non-protected animals was at day 14. Early granuloma formation as a key feature of protective immunity in TB has been suggested before^39,40. This rapid formation of an effective granuloma serves to control the bacterial infection in the initial early stage of infection and thereby limits further bacterial multiplication as assessed by lower bacterial loads in Th1 skewed mice (Table 2). In the present study, we demonstrated that these early protective granulomas were characterized by large amounts of iNOS-positive cells. This is consistent with enhanced Th1 cytokine (IFN-γ and TNF-α) activity, central to the up-regulation of the iNOS enzyme within macrophages which in turn results in the production of nitric oxide and reactive nitrogen intermediates essential in controlling mycobacterial growth.^41,42 In contrast, with the most polarized Th2 response we demonstrate a markedly different inflammatory infiltrate within granulomas at the day 14 time point with few lymphocytes and iNOS-positive macrophages abundant neutrophil granulocytes and many bacteria. Such a finding is consistent with previous studies identifying these granulocytes as particularly predominant during early granuloma formation in a range of laboratory animal models of TB.⁴³ In such a context neutrophils are thought to participate in the initial, non-specific response to infection before the emergence of the more targeted acquired immune response.⁴⁴ However, it is also possible that the increased neutrophilic influx into the granulomas of Al(OH)₃-immunized and naïve mice represents less effective granuloma formation, perhaps in response to increased intra-lesional cell necrosis. Interestingly, neutrophil-rich responses within lesions are a feature of disease progression in mice approaching the end of their natural lifespan when their CD4 T cells have become dysfunctional.⁴⁵

Our study illustrates the complexity of using pathology as a measure of protection against TB. In our model, although early granuloma formation is a hallmark of protection and correlates with subsequent control of disease, the cellular constituents within lesions requires careful analysis to distinguish protective lymphocyte and macrophage-rich lesions from less-protective granulocyte-rich granulomas. Such analysis becomes increasingly important as vaccine evaluation moves from crude measures of protection such as survival and tissue bacterial loads to multi-parameter assessment in which histopathology plays an increasingly important role.