C57BL/6 mice were obtained from Universidade Federal de Minas Gerais (UFMG), Brazil. TLR4^−/−, TLR2^−/−, MAL/TIRAP^−/−, and MyD88^−/− (in C57BL/6 background) were kindly provided by Dr. Sergio Costa (UFMG, Belo Horizonte, Brazil), and IL-10^−/− (in 129 Sv/Ev background) was provided by Dr. Donna Marie McCafferty (Calgary University, Calgary, AB, Canada). All mice were maintained in a specific pathogen-free facility at UFMG, Brazil. All animal procedures were performed in accordance with the guidelines from Conselho Nacional de Controle de Experimentação Animal (Brazil, http://www.mct.gov.br/index.php/content/view/310553.html) and approved by the University Ethical Committee for Animal Research (Protocol # 114/2010, CEUA-UFMG, Brazil).

As described elsewhere (43), a recombinant L. lactis strain NCDO2118 able to secrete Mycobacterium leprae Hsp65, using a xylose-inducible expression system, was constructed. The constructed vector (pSEC:hsp65) directed the expression of Hsp65 to the extracellular medium. L. lactis NCDO2118 harboring an empty vector (pXylT:SEC without hsp65) was used as a negative control in all experiments.

The L. lactis NCDO2118 strains were grown in M17 broth (Difco, Detroit, MI, USA) supplemented with 0.5% glucose (GM17) or 1% xylose (XM17; Synth, São Paulo, São Paulo, Brazil) at 30°C, without agitation. When required, chloramphenicol (10 µg/ml; Sigma-Aldrich, St. Louis, MO, USA) was added to the media.

On the first day, a single colony of recombinant L. lactis harboring an empty vector (pXylT:SEC) or recombinant L. lactis NCDO2118 (pXylT:SEC:hsp65) was grown at 30°C, without agitation, in 5 ml GM17 containing chloramphenicol (Cm, 10 µg/ml). On the second day, the overnight culture was diluted 1:10,000 in 1% xylose fresh M17 (XM17) supplemented with Cm (10 µg/ml) to provide ideal conditions for bacterial growth and M. leprae hsp65 gene expression. On the third day, when an optical density of 2.0 at 600 nm was reached, which corresponds to 2.5 × 10⁸ CFU/ml, mice treatment was performed.

During 4 days, C57BL/6, IL-10^−/−, TLR4^−/−, TLR2^−/−, MAL/TIRAP^−/−, MyD88^−/− mice were continuously fed either medium (control group) empty-vector-bearing L. lactis (CT-LL) or M. leprae-Hsp65-LL. Fresh L. lactis total culture (bacteria plus supernatant) obtained as previously described was offered to mice daily. Since each mouse drank about 5 ml of culture per day (data not shown) containing 7 µg/ml (44) of M. leprae Hsp65, the total dose of bacteria per mouse was estimated to be 5 × 10⁹ CFU and the total daily dose of M. leprae Hsp65 was about 35 µg per mouse.

Alternatively, animals received 35 µg of zymosan (Sigma-Aldrich, St. Louis, MO, USA) dissolved in 5 ml fresh empty-vector-bearing L. lactis NCDO2118 total culture per day during 4 days as a control for Hsp65-LL.

Ten days after the last day of oral treatment with Hsp65-producing L. lactis, control L. lactis, or zymosan, sex- and age-matched animals received 1.5% (w/v) DSS (40 kDa, ICN no. 160110; MP Biomedicals, Santa Ana, CA, USA) in the drinking water, ad libitum, for 7 days to induce colonic inflammation (colitis). The amount of DSS and water consumed per animal was monitored; there was no marked difference between experimental groups.

Macroscopic score of DSS-induced colitis was derived by separately scoring three major clinical signs, which were weight loss, diarrhea, and rectal bleeding, 7 days after DSS administration as described by Murthy et al. (45). Loss of body weight was calculated as the difference between the initial and actual weight. Diarrhea was showed as mucus/fecal material adherent to anal fur. The presence or absence of diarrhea was confirmed by examination of the colon following completion of the experiment. Mice were sacrificed and the colons were excised. Diarrhea was defined by the absence of fecal pellet formation in the colon and the presence of continuous fluid fecal material in the colon. Rectal bleeding was defined as diarrhea containing visible blood and gross rectal bleeding and scored as described for diarrhea. The macroscopic score was calculated from the score of the clinical signs using the following formula: (weight loss score) + (diarrhea score) + (rectal bleeding score). Samples of colon were fixed in formalin and processed for microscopic analysis. Hematoxylin-eosin (Doles, Panamá, Goias, Brazil)-stained sections were blindly scored based on a semi-quantitative scoring system described previously (46) where the following features were graded: extent of destruction of normal mucosal architecture (0 represents normal; 1, 2, and 3 represent mild, moderate, and extensive damage, respectively); presence and degree of cellular infiltration (0 represents normal; 1, 2, and 3 represent mild, moderate, and transmural infiltration, respectively); extent of muscle thickening (0 represents normal; 1, 2, and 3 represent mild, moderate, and extensive thickening, respectively); presence or absence of crypt abscesses (0 represents absent; 1 represents present); and the presence or absence of goblet cell depletion (0 represents absent; 1 represents present). Scores for each feature were summed up to a maximum possible score of 11.

The colon was washed with PBS and placed in a buffer solution (1 ml/g) containing Tween-20 0.05% (Sigma-Aldrich, St. Louis, MO, USA), phenylmethylsulfonyl fluoride 0.1 mM (Sigma-Aldrich, St. Louis, MO, USA), benzethonium chloride 0.1 mM (Sigma-Aldrich, St. Louis, MO, USA), EDTA 10 mM (Synth, São Paulo, São Paulo, Brazil), and aprotinin A 20 KIU (Sigma-Aldrich, St. Louis, MO, USA). Then, it was homogenized, centrifuged at 3,000 g for 10 min and the supernatants collected for cytokine assay. Plates were coated with purified monoclonal antibodies reactive with cytokines IL-6, TNF-α, TGF-β (active form), IFN-γ, IL-4, IL-17, and IL-10 (BD Bioscience, San Jose, CA, USA), overnight at 4°C. In the following day, wells were washed, supernatants were added, and plates were incubated overnight at 4°C. On the third day, biotinylated monoclonal antibodies against cytokines (BD Bioscience, San Jose, CA, USA) were added and plates were incubated for 2 h at room temperature. Color reaction was developed at room temperature with 100 μl/well of orthophenylenediamine 1 mg/ml (Sigma-Aldrich, St. Louis, MO, USA), 0.04% H₂O₂ substrate in sodium citrate buffer. Reaction was interrupted by the addition of 20 μl/well of 2 N H₂SO₄. Absorbance was measured at 492 nm by ELISA reader (Bio-Rad, Hercules, CA, USA).

Levels of IgA were determined in the intestinal lavage. Small intestine was rinsed with 10 ml of cold PBS. Intestinal lavages were centrifuged at 12,000 g for 20 min at 4°C, and levels of IgA in the supernatants were determined by ELISA. Briefly, 96-well plates were coated with Ig goat anti-mouse unlabeled antibody in coating buffer (pH 9.6) overnight at 4°C. Wells were washed and blocked with 200 µl PBS containing 0.25% casein for 1 h at room temperature. Samples were added to the plates and incubated for 1 h at 37°C. The plates were then washed, peroxidase-streptavidin goat anti-mouse IgA-HRP (Southern Biotechnology, Birmingham, AL, USA) diluted 1:10,000 was added, and the plates were incubated for 1 h at 37°C. Color was developed at room temperature with 100 μl/well of orthophenylenediamine (1 mg/ml; Sigma-Aldrich, St. Louis, MO, USA) and 0.04% H₂O₂ substrate in sodium citrate buffer. The reaction was interrupted by the addition of 20 μl/well of 2 N H₂SO₄. Absorbance was measured at 492 nm using an ELISA microplate reader (Bio-Rad, Hercules, CA, USA).

Lamina propria cells were isolated by a modified version of the method described by Davies and Parrot (47). Briefly, the entire length of large intestine was dissected, opened longitudinally, washed with PBS, and cut into small pieces. Tissue fragments were placed in Petri dishes and washed three times in calcium- and magnesium-free HBSS containing 1 mM dl-dithiothreitol (DTT; Sigma-Aldrich, St. Louis, MO, USA) for 30 min. Supernatants were discarded. After that, tissue fragments were incubated with 100 µl/mL of collagenase II (Sigma-Aldrich, St. Louis, MO, USA) for 60 min at 37°C on a shaker. Supernatants were passed through a 70 µm cell strainer (Falcon, Corning, NY, USA) and then resuspended in medium. Cells isolated from LP were then submitted to flow cytometry labeling.

Fluorescein isothiocyanate-conjugated mAbs to CD4 (clone RM4-5, IgG2a); phycoerithrin (PE)-conjugated mAbs to CD4 (clone RM4-5, IgG2a); CD45Rb (clone 16A, IgG2a); Foxp3 (clone R16-715, IgG2a); PE-Cy5-conjugated mAbs to CD4 (clone RM4-5, IgG2a); PerCP-Cy5.5 mAbs to CD25 (clone PC61, IgG1); biotin-conjugated mAbs to CD25 (clone 7D4, IgM); streptavidin-Cy5-chrome and streptavidin-allophycocyanin (APC), rat IgG1, IgG2a, IgG2b, and IgM isotype controls were purchased from BD Biosciences (San Jose, CA, USA). The biotin anti-LAP antibody (clone BAF246, IgG) was purchased from R&D Systems (Minneapolis, MN, USA). Surface staining was performed according to standard procedures at a density of 0.5–1 × 10⁶ cells per 25 µl, and volumes were scaled up accordingly. For Foxp3 staining, cells were fixed in Fix/Perm buffer (eBioscience, San Diego, CA, USA) after the surface staining, followed by permeabilization in Perm buffer (eBioscience, San Diego, CA, USA) and staining for PE-anti-Foxp3 according to the manufacturer’s instructions. Samples were acquired in a FACSCan or in a FACSCalibur cytometer (BD Biosciences, San Jose, CA, USA) and analyzed using the FlowJo software (TriStar, San Carlos, CA, USA). At least 30,000 events were acquired for each sample.

Results were expressed as the mean ± SEM. Normal distribution of samples was confirmed by the Kolmogorov–Smirnov test. Significance of differences among groups was determined by Student’s t-test or analysis of variance (Tukey’s post hoc test). Means were considered statistically different when p < 0.05.

We previously constructed a recombinant L. lactis strain able to produce and secrete Hsp65 from M. leprae. This L. lactis-expressing Hsp65 preparations contained less endotoxin (lipopolysaccharide, LPS) than the limit set by the Food and Drug Administration (34). To verify the efficacy of this new therapeutic concept, C57BL/6 wild-type (WT) mice were continuously fed Hsp65-LL for 4 consecutive days. The total dose of bacteria per mouse was estimated to be 5 × 10⁹ CFU and the total daily dose of M. leprae Hsp65 was about 35 µg per mouse. The control groups received either medium (CT) or empty vector-bearing L. lactis (CT-LL). After 10 days, mice received 1.5% DSS dissolved in drinking water during 7 days to induce colitis (Figure 1A). We showed that DSS-induced colitis led to colon shortening in mice, which was prevented by the pretreatment with Hsp65-L. lactis (Figure 1B). Macroscopically, mice fed Hsp65-L. lactis showed similar score (weight loss, diarrhea, and rectal bleeding) to naïve mice (Figure 1C). DSS-induced colitis caused epithelial damage, loss of goblet cells, and crypts, and cellular infiltration starting at day 3. Accordingly, the histological score, devised to allow quantification of histological changes, revealed that pretreatment with Hsp65-L. lactis completely prevented the loss of mucosal architecture and inflammatory cell infiltration in mucosa and submucosa layers of the mice (Figures 1D,E).

We started investigating the mechanisms involved in colitis prevention by Hsp65-LL by assessing whether oral administration of this recombinant bacterium could alter cytokine production in colon tissue. The chosen cytokines are either implicated in the pathogenesis of IBD (IL-6, TNF-α, IL-4, IFN-γ, or IL-17) or have anti-inflammatory activity (IL-10 and TGF-β1). Cytokines were measured in the colonic tissue homogenates after 7 days of DSS administration. We found that levels of IL-6, TNF-α, and IL-4 were increased in the CT group, but they resembled the naïve group in Hsp65-L. lactis-treated mice (Figures 2A,B,D). Mice pretreated with empty vector-bearing L. lactis showed intermediate levels of these cytokines (Figures 2A,D). IL-17 production did not change in any group when compared with naïve mice, whereas IFN-γ levels were increased by DSS administration, but not reduced by the oral pretreatments (Figures 2C,F). The increase of TGF-β1 levels observed in the CT group was prevented by the pretreatment with Hsp65-L. lactis (Figure 2E). However, IL-10, an important cytokine involved in the homeostasis of gut mucosa (48), was kept close to physiological levels (naïve group) after pretreatment with Hsp65-L. lactis. Conversely, IL-10 levels found in the CT and CT-LL groups were reduced (Figure 2G). Thus, Hsp65-L. lactis oral treatment maintained most of the cytokines measured at physiological levels.

Secretory IgA exerts protective functions in the intestinal mucosa such as improving gut immunological barrier and controlling commensal microbiota (49). Upregulation of its production is a possible mechanism of action for probiotic bacteria. Thus, our next step was to measure IgA in the intestinal lavage of mice. Neither acute colitis nor the oral pretreatment with either empty vector-bearing L. lactis or Hsp65-L. lactis affected intestinal IgA production (data not shown), suggesting that cytokine maintenance, particularly IL-10, at homeostatic levels rather than secretory IgA underlied the anti-colitogenic functions of oral treatment with Hsp65-L. lactis.

Since Hsp65-LL maintained the physiological levels of IL-10, we investigated whether the prevention of colitis by Hsp65-L. lactis was dependent on this anti-inflammatory cytokine. To address this question, 6-week-old IL-10-deficient (IL-10^−/−) mice and WT mice were orally fed with either medium or Hsp65-LL for four consecutive days. Ten days later, colitis was induced by DSS administration. Different from what we observed in WT mice, macroscopic and histological scores did not change with Hsp65-L. lactis oral treatment in IL-10^−/− mice (Figures 3A,B), revealing the importance of IL-10 in the protection from DSS-induced colitis.

To confirm the involvement of IL-10 in the anti-colitogenic activity of Hsp65-LL, we tested whether the spontaneous colitis developed in IL-10^−/− mice could be ameliorated by the Hsp65-L. lactis treatment. In this case, 4-week-old IL-10^−/− mice were orally fed Hsp65-L. lactis for 4 consecutive days. Macroscopic and histological scores were evaluated after 6 weeks, when IL-10^−/− mice show the signals of spontaneous colitis (48, 50). Despite the reduction in the macroscopic score when compared with medium or empty vector-harboring L. lactis groups, Hsp65-LL was ineffective in improving the histological score of IL-10^−/− mice colitis (Figures 3C,D). Thus, the IL-10 pathway is partially necessary for the Hsp65-LL immunomodulatory effect in the spontaneous colitis suggesting that other mechanisms, such as the production of TGF-β1, may also be involved. Indeed, pretreatment of IL-10^−/− mice with Hsp65-L. lactis led to an increase in TGF-β1 production in the colon tissue with no changes in IFN-γ or IL-17 (Figure 3E). Consistent with this, TGF-β levels were increased in extracts of colonic LP, and the frequency and absolute number of TGF-β1-producing CD4⁺CD25⁺LAP⁺ Tregs were found augmented in the colonic LP (Figures 3E,F; Table S1 in Supplementary Material) of IL-10^−/− treated with Hsp65-L. lactis. Taken together, these data suggest that IL-10 is crucial for the anti-colitogenic function of Hsp65-LL in DSS-induced colitis and at least partially involved in the prevention of spontaneous colitis in IL-10-deficient mice. In the complete absence of IL-10, TGF-β production increased probably as a compensatory mechanism to restore gut homeostasis.

Heat shock proteins have the capacity to interact with a variety of immune molecules at the cell surface such as toll-like receptors (TLRs), particularly TLR2 and TLR4 (18, 51). Accordingly, it has been shown that the signaling pathway induced by TLRs is involved in the immunomodulation by immune cells (18, 25, 51). To verify whether the TLR signaling cascade was indeed involved in the immunomodulation of colitis induced by Hsp65-L. lactis in vivo, we pretreated TLR2^−/−, TLR4^−/−, MAL/TIRAP^−/−, and MyD88^−/− mice with medium, empty-vector-bearing L. lactis (data not shown) or Hsp65-L. lactis. As shown in Figure 4A, oral administration of Hsp65-L. lactis significantly reduced the histological score in WT and TLR4^−/− mice, but not in TLR2^−/−, MAL/TIRAP^−/−, or MyD88^−/− mice, revealing the dependence of these molecules on Hsp65-L. lactis anti-colitogenic effects. Consistent with this, IL-10 production was maintained at physiological levels after the pretreatment of WT mice with Hsp65-L. lactis, which did not occur in TLR2^−/− mice (Figure 4B). To address whether a naturally occurring exogenous TLR2 ligand such as zymozan, a glucan prepared from yeast cell wall, would improve the DSS-induced colitis in a similar fashion than Hsp65, we compared the effect of Hsp65-L. lactis with zymozan at the same concentration (35 µg per mouse) dissolved in the CT-LL supernatant. As shown in Figure S1 in Supplementary Material, oral pretreatment with zymozan did not prevent colitis in mice, suggesting TLR2 ligand specificity in the immunomodulation induced by Hsp65-LL.

Since Hsp65-LL induced an anti-inflammatory cytokine profile in the colon, we next investigated whether Hsp65-LL would affect the frequency of Tregs. Recently, we showed that the pretreatment with Hsp65-LL modulates EAE in mice by inducing CD4⁺Foxp3⁺ and CD4⁺LAP⁺ Tregs (30). In DSS-induced colitis model, the pretreatment of mice with Hsp65-LL increased the CD4⁺Foxp3⁺ cell frequency in the spleen when compared to naïve mice after 3 days of colitis induction (Figure 5A). Moreover, the reduced frequency of CD4⁺LAP⁺ cells observed in both CT and CT-LL groups was restrained by the Hsp65-L. lactis pretreatment (Figure 5B). Finally, we verified whether oral pretreatment of mice with Hsp65-L. lactis would be sufficient to increase Tregs before the colitis induction and whether TLR2 would be involved in such effect. WT and TLR2^−/− mice received the Hsp65-L. lactis by the oral route during 4 days, and 4 or 10 days after the last day of oral treatment we analyzed the frequencies and absolute numbers of Treg populations. We observed an increase in CD4⁺Foxp3⁺ (Figures 5C,D) and CD4⁺LAP⁺ (Figure 5E) cell frequencies and numbers (Tables S2 and S3 in Supplementary Material) in spleen and mesenteric lymph nodes (mLNs) of Hsp65-L. lactis fed mice 10 days after the oral treatment. However, such Treg increase in WT mice after Hsp65-L. lactis pretreatment was not observed in TLR2^−/− mice, suggesting the involvement of TLR2 in the Treg expansion after Hsp65-L. lactis oral administration.