ABSTRACT
The environmental saprophyte Mycobacterium vaccae induces a Th1 response and cytotoxic T cells that recognize M. tuberculosis, and by subcutaneous injection, it is therapeutic for pulmonary tuberculosis (TB) induced by high-dose challenge in BALB/c mice. However, M. vaccae also drives regulatory T cells that inhibit Th2 responses, and this is seen in allergy models, not only following subcutaneous injection but also after oral administration. An oral immunotherapeutic for TB would be clinically useful, so we investigated M. vaccae given orally by gavage at 28-day intervals in the TB model. We used two different protocols: starting the oral M. vaccae either 1 day before or 32 days after infection with M. tuberculosis. Throughout the infection (until 120 days), we monitored outcome (CFU), molecules involved in the development of immunoregulation (Foxp3, hemoxygenase 1, idoleamine 2,3-dioxygenase, and transforming growth factor β [TGF-β]), and indicators of cytokine balance (tumor necrosis factor, inducible nitric oxide synthase, interleukin-4 [IL-4], and IL-4δ2; an inhibitory splice variant of IL-4 associated with improved outcome in human TB). Oral M. vaccae had a significant effect on CFU and led to increased expression of Th1 markers and of IL-4δ2, while suppressing IL-4, Foxp3, and TGF-β. When administered 1 day before infection, oral M. vaccae induced a striking peak of expression of hemoxygenase 1. In conclusion, we show novel information about the expression in TB of murine IL-4δ2 and molecules involved in immunoregulation and show that these can be modulated by oral administration of a saprophytic mycobacterium. A clinical trial of oral M. vaccae in extensively drug-resistant TB might be justified.
The immediate reason for the renewed interest in immunotherapy for tuberculosis (TB) is the emergence of extensively drug-resistant TB (XDR-TB) (2). If drugs on their own fail to treat TB, modulation of the host immune response with immunotherapeutic agents may provide additional therapeutic support. In addition, immunotherapy would also be valuable for drug-sensitive TB. The current 6-month treatment regimens are too long and cause problems with logistics and compliance. Immunotherapy might act in synergy with the currently available drugs and so improve their efficacy. Moreover, if immunotherapy can switch the immune system away from immunopathology and the Koch phenomenon toward an optimal protective response (although this remains poorly defined), it might reduce the duration of chemotherapy.
An injectable preparation of killed Mycobacterium vaccae was previously shown to be therapeutic in a high-dose intratracheal challenge model of pulmonary TB in BALB/c mice, even if given when the disease was already well advanced (14, 24). In this model, disease is characterized by an initial phase of partial immunity dominated by T-helper 1 cytokines plus tumor necrosis factor (TNF) and interleukin-1 (IL-1), followed by a phase of progressive disease accompanied by increasing expression of IL-4, diminishing expression of IL-1 and TNF (11), and exquisite sensitivity to the toxic effects of injected recombinant TNF (13). When mice in this late progressive phase were treated with two injections (day 60 and day 90) of 0.1 or 1.0 mg of heat-killed M. vaccae, there were a fall in CFU, a fall in IL-4 levels, a return to a type 1 cytokine profile, a switch from pneumonia to granuloma, and restoration of expression of IL-1 and TNF without any apparent toxicity or any increase in the toxicity of injected recombinant TNF (14). Mechanisms likely to contribute to these protective effects of M. vaccae include enhanced Th1 activity and induction of CD8+ cytotoxic T cells that can lyse macrophages infected with Mycobacterium tuberculosis (28). However, studies with IL-4−/− mice (10) and with neutralizing antibodies to IL-4 indicate that in high-dose TB challenge models, reduction in IL-4 production is also therapeutic (14, 26), and the suppression of IL-4 expression induced by M. vaccae in the BALB/c TB model was striking, suggesting a third protective mode of action (14). This ability of M. vaccae to suppress even a preexisting IL-4/Th2 response has been investigated in experimental allergy models (35, 36) and clinical trials (21) and is mediated by CD25+ CD45RBlow regulatory T cells (Tregs) (1, 35, 36).
Recently it was found that in a mouse model of allergic pulmonary inflammation killed M. vaccae is as effective when given as an oral immunotherapeutic as it is when administered subcutaneously (s.c.) (16). Since the oral route would clearly have enormous advantages for multidose immunotherapeutic strategies in humans, we have now tested the efficacy of oral M. vaccae in the BALB/c model of pulmonary TB in which M. vaccae was previously shown to be effective by the s.c. route. Moreover, in view of the ability of M. vaccae to drive the development of Tregs (35, 36), we also investigated the time course of expression of several molecules associated with induction of Tregs. These molecules include indoleamine 2,3-dioxygenase (IDO) (15) and hemoxygenase-1 (HO-1) (33), which are enzymes involved in the maturation of CD4+ CD25hi Tregs, in addition to Foxp3, transforming growth factor β (TGF-β), IL-4, and IL-4δ2. The latter is a splice variant of IL-4 thought to be a natural inhibitor of IL-4 in humans and mice (30, 34) and is associated with protection and with stable latent TB in humans (5).
MATERIALS AND METHODS
Murine model of progressive pulmonary TB.The experimental model of progressive pulmonary TB has been described in detail (11, 13, 14). Briefly, the virulent strain of M. tuberculosis H37Rv was cultured in 7H9 liquid medium. After 1 month of culture, mycobacteria were harvested and adjusted to 2.5 × 105 cells in 100 μl of phosphate-buffered saline (PBS), aliquoted, and maintained at −70°C until used. Before use, bacteria were recounted and viability was checked (11, 13, 14). Pathogen-free male BALB/c mice, 6 to 8 weeks old, were anesthetized (sevoflurane; Abbott Laboratories, Abbott Park, IL), and the trachea was exposed via midline incision. The animals were infected using an insulin syringe administering 2.5 × 105 viable bacteria suspended in 100 μl of PBS. After suturing of the incision, mice were maintained in vertical position until the effect of anesthesia waned. Infected mice were kept in groups of five in cages fitted with microisolators connected to negative pressure. All procedures were performed in a biological security cabinet in a biosafety level III facility. The protocol was approved by the Ethics Committee for Experimentation in Animals of the Instituto Nacional de Ciencias Médicas y Nutrición “Salvador Zubirán.”
To evaluate the effect of administration of heat-killed M. vaccae on the course of infection, two different experimental protocols were used, and each was performed twice. In the first protocol, heat-killed M. vaccae cells at a dose of 0.1 μg suspended in 100 μl of saline or saline alone was administered using an intragastric cannula. Then 23 h later, the animals were infected with H37Rv via the trachea as described above. Further similar doses of heat-killed M. vaccae were administered intragastrically on days 28, 56, 84, and 112 after H37Rv infection. Groups of six animals were killed on days 1, 3, 7, 14, 21, 30, 60, and 120 after H37Rv infection. In the second type of experiment, animals received heat-killed M. vaccae using the same dose and route of administration, but starting after the pulmonary TB infection was well established. Thus, intragastric heat-killed M. vaccae was given 32, 60, and 88 days after infection with H37Rv, and groups of six animals were sacrificed 68, 75, 90, and 120 days after infection.
Assessment of CFU in infected lungs.Right or left lungs from three mice per group at each time of killing were homogenized with a Polytron (Kinematica, Lucerne, Switzerland) in sterile tubes containing 2 ml of PBS-Tween 80 at 0.05%. Four or five dilutions of each homogenate were spread onto duplicate plates containing Bacto Middlebrook 7H10 agar (Difco Laboratories, Detroit, MI) enriched with oleic acid-, albumin-, catalase-, and dextrose-enriched medium (OACD). The number of colonies was counted 21 days later.
Preparation of lung tissue for histological and morphometric studies.Right or left lungs from three mice per group at each time of killing were perfused via the trachea with 10% formaldehyde dissolved in PBS. Parasaggital sections were dehydrated and embedded in paraffin, sectioned, and stained with hematoxylin-eosin. The percentage of lung surface affected by pneumonia was determined using an automated image analyzer (Q Win Leica; Milton Keynes, Cambridge, United Kingdom). Measurements were taken blindly, and data are reported as means ± standard deviations from three different mice at each time point in two different experiments.
The same paraffin-embedded material was used to determine the presence of HO-1 by immunohistochemistry. Lung sections from treated and control mice at each time point were mounted on silane-covered slides and deparaffinized, and the endogenous peroxidase was quenched with 0.03% H2O2 in absolute methanol. Sections were incubated overnight at room temperature with rabbit-specific polyclonal antibodies against HO-1 (BioVision, CA) diluted 1/300 in PBS. Bound antibodies were detected with biotinylated antibodies (biotin-conjugated anti-rabbit immunoglobulin G [Vectastain system; Vector Laboratories]) diluted 1/200, followed by incubation with horseradish peroxidase-conjugated avidin for 30 min. The reaction was revealed by 3,3-diaminobenzidine-hydrogen peroxide for 5 to 10 min at room temperature. Tissue sections were counterstained with hematoxylin. The negative controls consisted of performing the whole procedure using normal rabbit serum or an irrelevant antibody instead of the primary antibody.
Real-time RT-PCR analysis of cytokines and iNOS expression in lung homogenates.Three lungs, right or left, from the same number of animals in two different experiments were used to isolate RNA at each time of killing using the RNeasy minikit (Qiagen, CA). Each lung was placed in 2 ml of RPMI medium containing 0.5 mg/ml of collagenase type 2 (Worthington Biochemical, NJ); incubated for 1 h at 37°C; and then passed through a 70 μm cell sieve, crushed with a syringe plunger, and rinsed with the medium. Cells were centrifuged, the supernatant was removed and red cells were eliminated with lysis buffer. After counting, 350 μl of RLT buffer (Qiagen, CA) was added to 5 × 106 cells and RNA was extracted using the RNeasy mini kit (Qiagen, CA), according to the manufacturer's instructions. The quality and quantity of RNA were evaluated through spectrophotometry (260/280 nm) and on agarose gels. Reverse transcription (RT) of the mRNA was performed using 5 μg RNA, oligo(dT), and the Omniscript kit (Qiagen, CA). Real-time PCR was performed using the 7500 real-time PCR system (Applied Biosystems, CA) and Quantitect Sybr green master mix kit (Qiagen, CA). Standard curves of quantified and diluted PCR product, as well as negative controls, were included in each PCR run. Specific primers were designed using the program Primer Express (Applied Biosystems, CA) for the following targets: glyceraldehyde-3-phosphate dehydrogenase (G3PDH), 5′-CATTGTGGAAGGGCTCATGA-3′ and 5′-GGAAGGCCATGCCAGTGAGC-3′; inducible nitric oxide synthase (iNOS), 5′-AGCGAGGAGCAGGTGGAAG-3′ and 5′-CATTTCGCTGTCTCCCCAA-3′; TNF-α, 5′-TGTGGCTTCGACCTCTACCTC-3′ and 5′-GCCGAGAAAGGCTGCTTG-3′; IL-4, 5′-CAGGAGAAGGGAACACCAC-3′ and 5′-GCTGTTTAGGCTTTCCAGGAAG-3′; Foxp3, 5′-GGGCAGGCAACAACTCAGTC-3′ and 5′-GAAGCTCGACCGGACATTTG-3′; IDO, 5′-GACTTTGTGGACCCAGACACG-3′, 5′-ACCCCCTCATACAGCAGACCT-3′; and HO-1, 5′-GTGATGGAGCGTCCACAGC-3′ and 5′-TCTCGGCTTGGATGTGTACCT-3′. The following cycling conditions were used: initial denaturation at 95°C for 15 min, followed by 40 cycles at 95°C for 20 s, 60°C for 20 s, and 72°C for 34 s. Quantities of the specific mRNA in the sample were measured according to the corresponding gene-specific standard. The mRNA copy number of each cytokine was related to one million copies of mRNA encoding G3PDH.
RESULTS
Effects of oral heat-killed M. vaccae on the bacterial load in the lung.When M. vaccae was administered at 28-day intervals from the day before infection, and the number of CFU/lung was analyzed by two-way analysis of variance (ANOVA) with Bonferroni posttests (Fig. 1), the overall growth curves were significantly different (P = 0.0001), and from day 30 the CFU at individual time points were significantly reduced in the M. vaccae-treated animals (P < 0.001). When M. vaccae was administered monthly from day 32 and numbers of CFU were assessed from day 68 (i.e., from 8 days after the second oral treatment), the overall growth curves differed significantly (P = 0.0096, pooled data from two experiments), although differences at individual time points did not achieve significance.
Effect of intragastric heat-killed M. vaccae on bacterial load in the lungs. (a) M. vaccae was administered at 28-day intervals from the day before infection. The curves were different (P = 0.0009), and from day 30, the CFU at individual time points were significantly reduced in the M. vaccae-treated animals (P < 0.001). (b) M. vaccae was administered monthly from day 32. There was a trend toward reduced CFU in the treated animals (P = 0.078, pooled data from two experiments, two-way ANOVA with Bonferroni posttests).
Effects of oral heat-killed M. vaccae on expression of cytokines and iNOS in the lung.Compared to controls, mice treated with M. vaccae administered at 28-day intervals from the day before infection showed increased expression of mRNAs encoding TNF and iNOS (Fig. 2). In contrast, mRNA for IL-4 was decreased. (For these three molecules, the curves differed significantly [P < 0.0001]. P values for individual time points are indicated on the graphs.) Expression of IL-4δ2 was low and was not affected by treatment with M. vaccae. Since TNF (and gamma interferon [IFN-γ], which was not measured) enhances expression of iNOS, while IL-4 inhibits it, these results are compatible with a downregulation of the Th2 response that occurs in this model of TB and a return to a Th1 phenotype.
Intragastric heat-killed M. vaccae administered at 28-day intervals from the day before infection caused increased TNF and iNOS, while inhibiting IL-4. Expression of mRNAs encoding TNF (a) and iNOS (b) was increased. In contrast, that of mRNA encoding IL-4 (c) was decreased. Expression of IL-4δ2 (d) was low and was not affected by M. vaccae. pulm, pulmonary. P values for individual time points are noted as follows: *, P < 0.05; **, P < 0.01; and ***, P < 0.001 (two-way ANOVA with Bonferroni posttests).
When given from day 32, the effect of oral M. vaccae on cytokines was similar to that seen when it was given from the day before infection. Expression of mRNAs encoding TNF and iNOS was increased (Fig. 3). Moreover, there were decreased IL-4 and increased IL-4δ2 on days 68, 75, and 90, although IL-4 itself was more abundant than the splice variant at all times. These effects on the IL-4 variants were lost by day 120.
Effect of intragastric heat-killed M. vaccae given monthly from day 32 on expression of cytokines and iNOS. M. vaccae caused increased expression of mRNAs encoding TNF (a), iNOS (b), and IL-4δ2 (d) and a decrease in IL-4 (c). P values for individual time points are noted as follows: *, P < 0.05; **, P < 0.01; and ***, P < 0.001 (two-way ANOVA with Bonferroni posttests).
Effects of oral M. vaccae on expression in the lung of Foxp3, TGF-β, HO-1, and IDO.In untreated BALB/c mice with pulmonary TB, there occurred a biphasic increase in expression of Foxp3 (Fig. 4), consistent with flow cytometry data obtained by others (18), and a progressive rise in TGF-β as reported previously (10). Interestingly, intragastric M. vaccae given monthly from the day before infection caused a sustained reduction in expression of Foxp3 and TGF-β. In addition, the treatment caused early increases in expression of mRNA encoding HO-1 and IDO. For HO-1, this was confirmed by immunohistochemistry in several lung tissue compartments, and the increases were very significant (Fig. 5 and 6). Despite the later fall in mRNA, protein levels as detected by immunohistochemistry remained elevated in the M. vaccae-treated animals at day 120 (Fig. 5). Activity of these enzymes is associated with increased development of Foxp3+ Tregs (15, 33), so it is interesting that the mRNA levels had returned almost to control levels by the time the reduction in Foxp3 and TGF-β became striking in the M. vaccae-treated animals, although the protein levels had not (Fig. 5 and 6).
Effects of intragastric heat-killed M. vaccae given monthly from the day before infection on expression of Foxp3 (a), TGF-β (b), HO-1 (c), and IDO (d). mRNAs encoding Foxp3 and TGF-β were inhibited at later time points, while HO-1 and IDO mRNAs were increased at early time points. pulm, pulmonary. P values for individual time points are noted as follows: *, P < 0.05; **, P < 0.01; and ***, P <0.001 (two-way ANOVA with Bonferroni posttests).
Percentage of cells staining positive for HO-1 in different compartments of lungs from control infected mice (Con) and from infected mice that received intragastric heat-killed M. vaccae (Mv) administered at 28-day intervals from the day before infection.
Representative immunohistochemical detection of HO-1 in the lungs of mice 120 days after infection with H37Rv. (A) Pneumonic patches in the lung of a mouse treated with M. vaccae show numerous positive cells. (B) In contrast, only occasional cells show faint immunoreactivity in the perivascular areas of control animals. (C) In the perivascular areas, numerous positive cells with the morphology of activated lymphocytes and macrophages are seen in the lung of a mouse treated with M. vaccae. (D) Macrophages located in perivascular granulomas also showed strong immunostaining in a mouse treated with M. vaccae.
When given from day 32, intragastric M. vaccae caused an inhibition of expression of Foxp3 mRNA (Fig. 7), just as observed when given from day 0. There was also a minimal, though statistically significant, increase in IDO mRNA, similar to that seen after day 30 when treatment was given from day 0. Expression of HO-1 was consistently suppressed.
Effect of intragastric heat-killed M. vaccae given monthly from day 32 on expression of Foxp3 (a), IDO (b), and HO-1 (c). mRNAs encoding Foxp3 and HO-1 were downregulated, while there was also a minimal, although statistically signficant, increase in IDO mRNA. P values for individual time points are noted as follows: *, P <0.05; **, P <0.01; and ***, P <0.001 (two-way ANOVA with Bonferroni posttests).
DISCUSSION
Although the reductions in CFU that we have observed in this model using intragastric M. vaccae are modest, the effects in humans, over longer treatment durations, cannot be predicted. These effects were seen in the absence of chemotherapy, whereas in humans, immunotherapy would be used as an adjunct to available chemotherapy, even if only second- and third-line drugs were appropriate, as may be the case for patients with XDR-TB. Synergy between immunotherapy and chemotherapy is likely. Such synergy between multidose M. vaccae and chemotherapy was observed in one preliminary study in Argentina, despite the use of patients infected with fully drug-sensitive organisms who were therefore responding rapidly to the chemotherapy itself (6).
The induced changes in cytokine balance and in regulatory molecules were striking and suggest diminution of the IL-4 response that develops in this model during established disease and a switch back to protective Th1 responses with consequent induction of iNOS. The splice variant of IL-4, IL-4δ2, has received little attention in mice, although its presence was reported in 2004 (34). However, a series of studies in humans suggest that IL-4δ2 behaves like a Th1 cytokine. In TB, a low ratio of expression of IL-4δ2 relative to the agonist, IL-4, correlates with severity of disease (32) and with a risk of progression in recent contacts (31). In contrast, high expression of IL-4δ2 is characteristic of individuals with stable, well-established latent TB that does not progress (5, 7). A rise in IL-4δ2, accompanied by a fall in IL-4, such as that seen in the mice receiving M. vaccae from day 28, was characteristic of clinical improvement during treatment in a recent South African study (27).
The concomitant fall in expression in TGF-β (Fig. 4b) might be secondary to the fall in IL-4, because the rising TGF-β levels are largely secondary to IL-4 release in this model (10).
We have pointed out previously that high IL-4 levels occur in TB most often in developing countries close to the equator, particularly in low-lying areas (22). We have suggested that this is due to the combined influence of high exposure to environmental mycobacteria and to helminths, followed by high-dose challenge with M. tuberculosis (23, 25). A study in Cameroon showed that even in the normal population, the magnitude of the IL-4 response to purified protein derivative increases with age, at least up to the age of 16 years, at which time recruitment in this study ceased (29). IL-4, together with TGF-β, is able to oppose apoptosis, autophagy, and the induction and function of cytotoxic lymphocytes (3, 9, 17, 23). These are crucial mechanisms providing the immune system with mycobactericidal pathways that circumvent the ability of M. tuberculosis to damage phagosome function. This might explain why a neutralizing antibody to IL-4 is therapeutic in another high-dose challenge mouse model of TB (26) and suggests that reducing IL-4 is likely to be therapeutically useful. Similarly inhibition of TGF-β by administering soluble type III TGF-β receptors is also therapeutic in this model (12).
We have not monitored IFN-γ expression in this study because protection is not quantitatively related to levels of this cytokine, which is always expressed at high levels in our model (11, 13, 14). This lack of correlation may be because the classical IFN-γ-mediated macrophage activation pathway is inhibited by M. tuberculosis, so that protection depends to an unknown extent on “rescue” mechanisms such as apoptosis, autophagy, and cytotoxic lymphocytes that can kill via bactericidal peptides or recycle the damaged phagosome into fresh macrophages (discussed and referenced in reference 25).
The role of IDO in TB has not been fully evaluated. We were intrigued by this molecule because IDO metabolizes tryptophan, releasing kynurenine and metabolites such as 3-hydroxyanthranilic acid and quinolinic acid that modulate T-cell differentiation, function, and survival (8). Moreover, IDO-positive dendritic cells drive maturation of Treg (15). In view of the small but significant increases in expression of IDO induced by M. vaccae, particularly during the first 3 weeks, it will be interesting to test the effects of an inhibitor of the enzyme in this model.
HO-1 degrades heme to biliverdin, free divalent iron, and carbon monoxide (CO), resulting in immunomodulatory effects, perhaps mediated by CO (20). We show that expression of HO-1 increases steadily to very high levels in this TB model and that oral M. vaccae causes a striking (and precisely reproducible [data not shown]) early increase in HO-1 expression between days 7 and 21. Further work will be required to determine whether this peak of HO-1 activity results in the expansion of a beneficial subset of regulatory cells. Interestingly, activation of HO-1 had a protective effect in an allergy model that was mediated by Foxp3+ CD4+ CD25+ Tregs, IL-10, and membrane-bound TGF-β (33). However, other data suggest that enhanced HO-1 activity can drive Foxp3+ Tregs that facilitate graft tolerance across a powerful histocompatibility barrier, so there is no evidence of preferential downregulation of Th2 (19). Since the activity of this enzyme can be modulated in vivo, the relevance of the early peak of HO-1 induced by oral M. vaccae can be investigated experimentally.
In conclusion, we have used a well-established model of progressive pulmonary TB to provide new information on the time course of expression of several molecules that are involved in immunoregulation, and we have shown that a potential immunotherapeutic agent, already shown to be active in this model when given s.c., not only reduces CFU when given orally, but also modulates expression of these regulatory molecules. There is an urgent need for safe immunotherapeutics, and in a recent report, the Special Programme for Research & Training in Tropical Diseases of WHO endorsed this approach (4). Protocols for the production of M. vaccae in accordance with good manufacturing practice guidelines have been established and used successfully in the past. Moreover, M. vaccae has a proven safety record, and these findings should encourage further investigation and clinical trials. The initial data from a multidose study of M. vaccae given by the intradermal route were encouraging (6), and we feel that trials of the oral route in XDR-TB are now justified. This route would have major advantages under conditions in developing countries.
ACKNOWLEDGMENTS
This work was supported by Stanford Rook Ltd., London, United Kingdom, and by CONACyT, Mexico.
FOOTNOTES
- Received 24 July 2008.
- Returned for modification 2 September 2008.
- Accepted 19 September 2008.
- Copyright © 2008 American Society for Microbiology