Dr Xi Yang, Department of Medical Microbiology, University of Manitoba, Room 523, 730 William Avenue, Winnipeg, Manitoba, Canada R3E OW3. E-mail: email@example.com
Our previous studies, as well as those of others, have demonstrated that local or systemic Mycobacterium bovis bacille Calmette–Gue´rin (BCG) infection can inhibit de novo allergen-induced asthma-like reactions, but the effect of this infection on established allergic responses is unknown. The aim of this study was therefore to examine the effect of mycobacterial infection on established allergy in a murine model of asthma-like reaction. Mice were sensitized with ovalbumin (OVA) in alum followed by infection with BCG and subsequent intranasal challenge with the same allergen. In some experiments, mice were sensitized with OVA followed by intranasal challenge with OVA and then given BCG infection with subsequent rechallenge with OVA. Mice without BCG infection but treated with OVA in the same manner, were used as a control. The mice were examined for immunoglobulin E (IgE) response and eosinophilic inflammation, mucus production, cytokine/chemokine patterns and adhesion molecule expression in the lung. The results showed that postallergen BCG infection suppressed the established airway eosinophilia and mucus overproduction, but not IgE responses. The inhibition of asthma-like reactions by BCG infection was correlated with a shift of allergen-driven cytokine production pattern and, more interestingly, with a dramatic decrease of vascular cell adhesion molecule-1 (VCAM-1) expression in the lung. These findings suggest that intracellular bacterial infection can inhibit established allergic responses via alteration of local cytokine production and the expression of adhesion molecules.
An inverse relationship between reduced incidence of infection and increased allergy has been observed in many developed countries over the past two to three decades, which has led to the ‘hygiene hypothesis’, i.e. that the existence of microbial infections may prevent or inhibit the development of allergic diseases.1–3 Recent experimental studies have demonstrated a manipulating effect of mycobacterial infection and bacterial products on allergic inflammation and cytokine production induced by allergen, suggesting that pre-existing mycobacterial infection can inhibit the development of de novo allergic responses.4–10
The effect of live intracellular bacterial infection on established allergic reactions has yet to be reported. Although studies examining the effect of infection on de novo allergy are informative, the influence of infection on established allergy is a much more relevant question in the real world. Although some studies showed inhibitory effects of killed bacteria on established immunoglobulin E (IgE) responses and eosinophilic inflammation in established allergy,11–13 it remains unclear whether natural bacterial infection can manipulate established allergic responses. This point is important because inhibition of allergy by large doses of dead micro-organisms or bacterial components does not necessarily mean a natural infection of this organism having the same effects. A conclusive elucidation of the mechanism underlying the documented inverse correlation between allergy and intracellular bacterial infection can only be derived from studies involving live infections.
To directly examine the effect of intracellular bacterial infection on an established allergic reaction, we studied the asthma-like reaction in bacille Calmette–Guérin (BCG)-infected mice that had been sensitized with ovalbumin (OVA) (or sensitized plus intranasally challenged with OVA) before the infection, following final intranasal challenge (or rechallenge) with the same allergen. The results showed that postallergen infection with BCG suppressed established eosinophilia and mucus oversecretion induced by subsequent intranasal challenge with the allergen, but not IgE responses. The inhibitory effect is highly associated with alteration in vascular cell adhesion molecule-1 (VCAM-1) expression and cytokine production.
Materials and methods
Animals and immunization
Female C57BL/6 mice were purchased from Charles River Canada (St. Constant, PQ, Canada). Animals were used in accordance with the guidelines issued by the Canadian Council on Animal Care. Mice were treated using two protocols. For most experiments, protocol 1 was used. Briefly, mice were initially sensitized intraperitoneally (i.p.) with 2 µg of OVA (ICN Biomedicals, Montreal, Canada) in 2 mg of Al(OH)3 adjuvant (alum). Two weeks after sensitization, mice were infected intravenously with Mycobacterium bovis BCG [1 × 106 colony-forming units (CFU)] and then challenged intranasally with 50 µg of OVA (40 µl) at 20–45 days post-BCG infection. Mice were killed and analysed for allergic and immune responses at various time-points (2–10 days) following allergen challenge. For protocol 2, mice were sensitized with OVA (2 µg in alum) i.p. and then challenged intranasally with OVA (50 µg) on day 14 postsensitization. Intravenous infection with live BCG was performed 20 days following OVA challenge. On day 40 post-BCG infection, mice were rechallenged with OVA (50 µg) and killed 7 days later for analysis.
Bronchoalveolar lavage (BAL) and cell counting
As a previous kinetics study showed that airway inflammatory cell recruitment, including eosinophils, was apparent at 2 days, peaking at 6–8 days, and then gradually declined following intranasal challenge with OVA,14 the time-point we chose for most of the experiments was that of peak cellular infiltration into the lung (day-7 postchallenge). This was also the optimal time-point for measuring secondary OVA-specific IgE responses.14 In some experiments, mice were also killed and examined on day 2 or 10 after the final allergen challenge. For collecting BAL fluids, the trachea were cannulated following killing and the lungs were washed twice with 1 ml of phosphate-buffered saline (PBS). The total cell number in BAL was determined, the BAL was centrifuged and the supernatant used to test for chemokine production and the cell pellet used to prepare slides for differential cell counting. A Fisher Leukostat Stain Kit (Fisher Scientific, Ontario, Canada) was used for leucocyte differential analysis, and the number of monocytes, lymphocytes and eosinophils in a total of 200 cells were counted in each slide.
Lung tissues were routinely fixed using 10% formalin, embedded, sectioned and stained by haematoxylin and eosin (H & E). Slides were examined for pathological changes under light microscopy, as described previously.14 Mucus and mucus-containing goblet cells within bronchial epithelium were analysed by Thionin staining using the Mallory method15 and counterstaining with orange G. The histological mucus index (HMI) is a quantitative method for measuring mucus secretion.16 Briefly, slides were examined at ×400 final magnification with a rectangular 10-mm square reticule grid inserted into one eyepiece of the microscope. Intersections of airway epithelium with the reticule grid were counted, distinguishing mucus-containing or normal epithelium. The HMI represents the ratio of total number of mucus-positive intersections and the total number of all intersections.
Measurement of VCAM-1 expression using immunohistochemical staining
For analysis of VCAM-1 expression, lung tissues were snap-frozen in liquid nitrogen when the mice were killed and then stored at −80° until required for sectioning. Sections (10 µm) were placed on slides and fixed by 99·6% acetone. Slides were incubated at room temperature for 1 hr with either purified rat anti-mouse VCAM-1 or isotype-matched control antibodies, which were purchased from PharMingen (San Diego, CA). After successive washes in wash buffer, the sections were further incubated with a secondary antibody [(rabbit anti-rat immunoglobulin G (IgG)], conjugated to horseradish peroxidase (HRP), and developed with 3-amino-9-ethylcarbazol (AEC) chromogen. A DAKO EnvisonTM system Kit (DAKO Corporation, Carpinteria, CA) was used in this tissue-staining process. The intensity of VCAM-1 expression on the pulmonary vascular endothelium was classified according to the criteria defined by Briscoe et al.17
Cytokine and chemokine analysis
For examination of cytokine production, cells from draining lymph nodes (mediastinal lymph nodes) and spleen were cultured as previously described.14 Briefly, single-cell suspensions were prepared and cultured at 7·5 × 106 cells/ml (2 ml/well) for spleen cells and at 5 × 106 cells/ml (1 ml/well) for lymph node cells, with or without OVA (1 mg/ml), in the presence or absence of anti-CD4 monoclonal antibody (mAb) (YTS 191.1) at 5 µg/ml. Culture supernatants were harvested at 72 hr for measurement of various cytokines using sandwich enzyme-linked immunosorbent assays (ELISAs). Purified (capture) and biotinylated (detection) Abs purchased from PharMingen were used for the measurement of interleukin (IL)-4, IL-5, interferon-γ (IFN-γ) and IL-12, as previously described.8,14 IL-13 was determined using paired antibodies purchased from R & D Systems (Minneapolis, MI). In addition, the presence of eotaxin and macrophage chemotactic protein-1 (MCP-1) in BAL was determined by ELISAs using paired antibodies purchased from R & D Systems.
ELISAs were used for determination of OVA-specific IgE, IgG1 and IgG2a antibodies. Sera were examined for OVA-specific IgG1 and IgG2a using biotinylated goat anti-mouse IgG1 or goat anti-mouse IgG2a antibodies purchased from Southern Biotechnology Assoc. Inc. (Birmingham, AL), as described previously.14 For determination of OVA-specific IgE, sera were incubated twice with a 50% slurry of protein G–Sepharose (Pharmacia) in order to remove most of the serum IgG1, and measured for IgE using biotinylated anti-mouse IgE antibody purchased from PharMingen. The treatment with protein G–Sepharose allowed the removal of ≈ 95% of total IgG1 without affecting the concentration of IgE.18
Cytokine and chemokine levels and differential BAL cell counts in different groups were analysed by using the unpaired Student's t-test. OVA-specific antibody titres were first transformed into log10 and then analysed using the unpaired Student's t-test.
BCG infection of OVA-sensitized/challenged mice inhibits bronchial eosinophilia and mucus overproduction
We have reported previously that i.p. immunization with OVA (2 µg in alum) induces predominant T helper 2 (Th2) cytokine production and allergen-specific IgE responses, peaking at 10–14 days following primary immunization,8,15–21 which plays an important role in the asthma-like reaction following local challenge with allergen. Hence, we examined the effect of M. bovis BCG on asthma-like reactions induced by local challenge with OVA in animals which had established allergy (IgE and Th2 responses) before the infection. Mice were first sensitized with OVA and then infected with BCG at the time of peak primary IgE production (day 14). The establishment of IgE responses was confirmed by examining serum IgE 1 day before (day 13) BCG infection. At 20–45 days postinfection, mice were challenged intranasally with OVA and analysed for asthma-like reactions. As shown in Fig. 1(a), the number of eosinophils in the BAL of OVA-sensitized mice with subsequent BCG infection were significantly lower (20–50-fold) than in those without BCG infection (P < 0·01). The percentage of eosinophils in the BAL of BCG-infected mice was less than 10% of the total infiltrating cells, while it was 60–80% in the mice without infection (Fig. 1b). In contrast, the absolute number and percentage of monocytes/macrophages in the BAL of the mice with BCG infection were significantly higher than those in the mice without BCG infection. The percentage of lymphocytes in the BAL of mice with BCG infection was also significantly higher than that in mice without BCG infection, mainly because of the dramatic decrease in the eosinophil component of the total infiltrating cells in BCG-infected mice. Similarly, H & E staining of the lung sections showed that, in contrast to the massive eosinophilic infiltration in the peribronchial and perivascular areas of the control mice without BCG infection (Fig. 2a), the OVA-sensitized/challenged mice with postsensitization BCG infection showed significantly less cellular infiltration, which was predominated by lymphocytes and monocytes/macrophages (Fig. 2b). Similar differences in airway cellular infiltration were observed between BCG-infected and -uninfected mice in the experiments where mice were treated following protocol 2, i.e., the mice were infected with BCG after both OVA sensitization and intranasal challenge and then given a secondary challenge with OVA (data not shown). Moreover, mucus staining showed much less mucus production in the mice with BCG infection than in those without infection (Fig. 2d). HMI, a quantitative method for measuring goblet cell development and mucus secretion, showed that ≈ 40% of airway epithelium was mucinous in mice without BCG infection, while it was < 20% mucinous in BCG-infected mice following OVA sensitization/challenge (Fig. 3). In all of these experiments, BCG-infected mice showed significantly lower airway eosinophilia compared with uninfected mice. The results clearly indicate that BCG infection following allergen exposure (sensitization alone or sensitization plus local challenge) is able to inhibit established and airway eosinophilic inflammation and mucus production elicited by local challenge or rechallenge with the allergen.
M. bovis BCG infection alters established Th2-type cytokine responses but fails to inhibit established IgE responses
As we and others reported previously, OVA alum sensitization, with or without intranasal challenge, induces Th2-type cytokine responses, characterized by predominant IL-4 and IL-5 production with minimal IFN-γ and IL-12 production.2,8,9 In this study, we further examined the effect of postallergen BCG infection on established Th2 responses. As shown in Table 1, Th2 cytokine production (including IL-4, IL-5 and IL-13) by draining lymph node and spleen cells from the post-OVA BCG-infected mice was significantly lower than that from the mice without BCG infection. In contrast, the levels of OVA-driven IFN-γ production by cells from the OVA-sensitized/challenged mice with BCG infection were three to five times higher than those without BCG infection. Antigen-driven IL-4, IL-5, IL-13 and IFN-γ production in the culture supernatants was virtually blocked (> 80%) by anti-CD4 mAb, suggesting that the CD4 T cell was the major cell type responsible for production of these cytokines (data not shown). In parallel with the significant increase of IFN-γ production in BCG-infected mice, IL-12 production in the culture from BCG-infected mice was significantly higher than that in uninfected mice. Similar differences in cytokine production between BCG-infected and -uninfected mice were observed in mice killed on day 2 or 10 after the final challenge with OVA (data not shown). These results, in combination with previous findings, indicate that intracellular bacterial infection is able to switch the cytokine pattern to allergen, not only in de novo but also in established allergic responses.
Table 1. Bacille Calmette–Guérin (BCG) infection: altered allergen-driven cytokine production
Alteration of allergen-driven cytokine-production patterns caused by postsensitization bacille Calmette–Guérin (BCG) infection. Mice were sensitized and challenged with ovalbumin (OVA) with (OVA/BCG/OVA) or without (OVA/OVA) BCG infection, as described in the legend to Fig. 1, and were killed on day 7 post-intranasal challenge with OVA. The splenocytes and local draining lymph node cells from individual mice were cultured with allergen-specific stimulation. Culture supernatants were harvested at 72 hr and tested for interleukin (IL)-4, IL-5, IL-13, interferon-γ (IFN-γ) and IL-12 using sandwich enzyme-linked immunosorbent assays (ELISAs). Data are presented as mean±SD of each group. Statistical analysis to compare OVA-sensitized/challenged mice with or without BCG infection were performed using the Students' t-test. One representative experiment of five independent experiments is shown.
3358 ± 424
914 ± 680
5984 ± 221
894 ± 383
149 ± 8
21 ± 10
129 ± 48
33 ± 25
5278 ± 1017
2746 ± 993
2502 ± 544
575 ± 568
55 ± 11
205 ± 78
13 ± 13
42 ± 19
118 ± 49
648 ± 262
300 ± 64
1232 ± 111
To elucidate the effect of BCG infection on established IgE responses, we examined OVA-specific IgE in OVA-sensitized/challenged mice, with or without BCG infection, following intranasal challenge with OVA. Interestingly, and in contrast to their switched cytokine patterns from Th2 to Th1 (Table 1), BCG-infected mice showed slightly, but significantly (3- to 4-fold) higher, rather than lower, OVA-specific IgE production compared to those without BCG infection (Fig. 4). A similar trend was observed in OVA-specific IgG1 and IgG2a production. The results, together with the data from cytokine analysis, imply that the antibody response in established allergy may not be closely associated with the allergen-specific T-cell response and local eosinophilic inflammation under certain circumstances.
BCG infection inhibits VCAM-1 expression on airway vascular endothelium induced by OVA challenge
Besides the potent efficacy of Th2 cytokines in the development and recruitment of eosinophils into airways, chemokines and adhesion molecules play important roles in this process.22–26 To test the effect of postsensitization BCG infection on chemokine production and adhesion molecule expression induced by intranasal challenge with OVA, we analysed MCP-1 and eotaxin levels in the BAL, and VCAM-1 expression on airway vascular endothelium, in mice with or without BCG infection. As shown in Fig. 5, the levels of the tested inflammatory cell-attracting chemokines in BAL were comparable (MCP-1, P = 0·65; eotaxin, P = 0·40) between mice with or without BCG infection. In sharp contrast, VCAM-1 expression on vascular endothelial cells was dramatically different between BCG-infected and -uninfected mice following OVA challenge (Fig. 2e,f). Kinetic study in our model showed that VCAM-1 expression on the endothelium of airway blood vessels of sensitized mice was readily detectable at day 2 following intranasal challenge with OVA, maintaining an increase until 6–8 days after challenge and remaining at high levels for at least 2 weeks, which was paralleled with the process of eosinophil recruitment into airways (data not shown). As shown in Fig. 2(e,f), high density of VCAM-1 expression on airway endothelium was observed in mice without BCG infection following OVA challenge, while very faint or undetectable VCAM-1 expression was observed in the OVA-exposed mice with BCG infection. Scores of VCAM-1 expression on the vascular endothelium of the BCG-infected OVA-sensitized/challenged mice was significantly lower than that in the mice identically treated with OVA but without BCG infection (Fig. 6). The results suggest that suppression of VCAM-1 expression on airway vascular endothelium may be an important mechanism by which BCG infection inhibits local eosinophilic inflammation.
In this study, we have demonstrated that BCG given to mice with previous exposure of allergen inhibits an established allergic reaction, characterized by significant reduction of local eosinophilic inflammation and mucus production. The reduction of airway pathogeneses to allergen was associated with a switch of allergen-driven cytokine-production patterns of local and circulating CD4 T cells and paralleled a dramatic decrease of VCAM-1 expression on airway vascular endothelium. These findings are encouraging because a major challenge for allergists is to inhibit an established allergic response to an allergen. Modulation of established T-cell responses is more difficult than modulating primary responses.27–31 Although mycobacterial infection has been demonstrated to be able to inhibit de novo allergic responses (as shown in previous studies), it was speculated that the infection may exacerbate established allergic responses because prior allergic reaction may set up an environment for Th2 development; thus the subsequent infection with intracellular bacteria, which normally induce Th1 responses, will be directed to Th2 responses.2 This view was supported by several articles which showed that the pre-existence of IL-4 favours Th2 responses under conditions that normally induce Th1 responses.32–34 In contrast to these speculations, our present data showed that intracellular bacterial infection, which normally induces Th1 responses, can inhibit established predominant Th2 responses to allergen (except IgE responses). The results suggest that the phenotype of an established Th2 pattern of immune responses, at least at a population level, can be modulated by subsequent exposure to strong Th1 inducers. Although the approach (using live infection) in this study is unique, the finding of the inhibition of established Th2 responses by a Th1 inducer is in agreement with several recent reports which show that administration of recombinant (r)IL-12 or killed intracellular bacterial organisms inhibited established Th2 cytokine responses.12,13,35,36
A novel finding in this study is that BCG infection inhibits VCAM-1 expression on pulmonary vascular endothelium induced by exposure to OVA. VCAM-1 has been shown to be critical in the interaction between eosinophil and vascular endothelial cells during the migration of eosinophils into airways. To our knowledge, this is the first report which demonstrates that infection can alter airway eosinophilic inflammation induced by allergen via modulation of adhesion molecule expression. The mechanism by which mycobacterial infection inhibits VCAM-1 expression remains unclear. As IL-4 plays an important role in the expression of VCAM-1, and because IL-4 production in BCG-infected mice was significantly lower than that in uninfected mice, the infection may inhibit OVA-induced VCAM-1 expression via inhibition of IL-4 production. On the other hand, it has been reported that BCG infection alone induces VCAM-1 expression in the lung.37 Moreover, IL-12, a strong inducer of Th1, has also been found to induce VCAM-1 expression.18,38–40 Our own data also showed that, 3–6 weeks following BCG infection (without OVA exposure), the expression of VCAM-1 on pulmonary vascular endothelium was remarkably enhanced (X. Yang et al., unpublished). Therefore, it seems paradoxical that a VCAM-1 inducer (BCG infection) can inhibit the expression of VCAM-1 induced by another inducer (allergen). We consider that the cross-regulation between OVA-induced Th2 cells and BCG-induced Th1 cells may lead to net reduction of VCAM-1 expression on the vascular endothelium. The exact mechanism for this cross-regulation remains to be studied. Elucidation of these mechanisms may significantly enhance our understanding on the regulation of adhesion molecule expression during allergic and infectious diseases.
It is interesting that OVA-specific antibody responses in BCG-infected mice is disassociated with altered eosinophilic inflammation and cytokine production in these mice. Our results showed that, although BCG infection in OVA pre-exposed mice shifted the T-cell cytokine pattern elicited by OVA challenge from Th2 predominant to Th1 predominant, OVA-specific IgE production, as well as production of IgG1 and IgG2a, in these mice demonstrated a trend of increase compared to those without infection. Similarly, total IgE levels in OVA-sensitized/challenged mice with postsensitization BCG infection was also significantly higher (2–10-fold) than those without BCG infection (X. Yang et al., unpublished). This finding was in agreement with a previous report which showed that administration of certain levels of IL-12 enhance, rather than suppress, established IgE responses.41 As BCG infection increases IL-12 production, the IL-12 induced by the infection, in combination with other unknown factors, may enhance IgE responses in this model. Another explanation is that some OVA-specific B cells, primed in the stage of sensitization, survive as memory cells that may not need T-cell cytokines for reactivation and/or expansion following challenge exposure to allergen. It has been documented for decades that these long-term-survival allergy-specific memory B cells play an important role in the maintenance and mediation of allergic responses.42 On the other hand, the relevance of the failure to inhibit IgE responses by BCG infection is unclear. Although there is no doubt that the IgE response is essential in atopic allergy (including asthma), the level of IgE responses does not always parallel the prevalence or severity of the asthmatic reaction, possibly because IgE normally only functions as a trigger for allergic reactions.3,43 In particular, many studies in allergen immunotherapy show that, although the cytokine pattern and the clinical manifestation have been altered as a result of the therapy, the serum IgE levels in some individuals fail to decrease and even significantly increase.44,45
This work was supported by a grant from the Canadian Institutes of Health Research (CIHR) to X. Yang (MT-14680). X. Yang holds a Scholarship (salary) award from the CIHR.