Oral Combined Therapy with Probiotics and Alloantigen Induces B Cell–Dependent Long-Lasting Specific Tolerance

The main limitation of allogeneic hematopoietic stem cell transplantation (aHSCT) for the treatment of malignant and nonmalignant disease is the development of graft versus host disease (GVHD), a complex multiorgan inflammatory syndrome that results from recognition of genetically disparate tissues of the recipient patient by the incoming allogenic donor T cells (1). GVHD treatment is based on use of immunosupressive agents that block T cell function in a nonspecific way, leading to undesired consequences—because these alloreactive T cells also facilitate engraftment of hematopoietic stem cells, accelerate immune reconstitution (1–4), and contribute to the graft-versus-leukemia (GVL) effect (4). Because of the close association between GVH and GVL reactions (4), much effort has been put forth to develop therapies that reduce the first while maintaining the latter (1).

Oral tolerance is classically defined as the suppression of immune responses to Ags that have been previously administrated by the oral route (5). Far from being just a matter of hyporesponsiveness, oral tolerance is an active immunologic event influenced by numerous factors such as Ag dose, microbiota, and costimulatory molecules, resulting from multiple mechanisms of action (6). Induction of oral tolerance has been shown to be effective in various experimental models of immune disorders (7–9) including GVHD (10–12). In fact, Nagler et al. (12) demonstrated that it is possible to alleviate acute GVHD (aGVHD) severity by the induction of oral tolerance in the recipient, although the maintenance of the GVL effect was not addressed in that study.

Administration of probiotics by the oral route also have been shown to modulate the immune response, favoring either pro- (13, 14) or anti-inflammatory (15, 16) responses. Recent studies suggest that TLRs and NF-κB signaling pathways play a crucial role in the outcome of the response (17–21). Also, probiotics can influence the integrity of the gut epithelial cell barrier through modulation of mucus, defensins, and IgA production, reduction of bacterial adhesion, and enhancement of tight junction and cell survival (22–24).

The relationship between the endogenous microbiota and development of aGVHD was proposed ∼40 y ago in experimental models (25) and confirmed in humans later on (26). As hypothesized by Hill et al. (27), damage of the host intestinal epithelium by the conditioning regiment results in systemic exposure to microbial Ags that ultimately lead to an inflammatory milieu that is adequate for donor T cell activation. With those results in mind, intestinal decontamination became a common practice in aHSCT (26, 28). More recently, Jenq et al. (29) showed that composition of the gut flora of recipient mice that developed GVHD had a dramatic loss of bacterial diversity and a distinct composition compared with recipient mice that did not develop GVHD. In addition, the authors revealed that antibiotic-treated mice showed an emergence of species that was associated with exacerbated GVHD. Interestingly, Gerbitz et al. (30) demonstrated that oral administration of Lactobacillus rhamnosus GG before and after transplantation not only improved survival and ameliorated aGVHD clinical scores but also reduced the translocation of enteric bacteria. Both works suggest that manipulation of the gut microbiota might be as good as decontamination with antibiotics to prevent GVHD.

Although promising, the clinical application of both protocols, oral tolerance induction (12) and probiotic treatment (30), cannot be used in transplanted patients. Once the conditioning regimen alters the intestinal epithelial barrier, these patients require special attention regarding their diet. As so, no live or raw food is permitted during the peritransplant phase, precluding oral administration of Ags or probiotics.

In this report, we reasoned if tolerance induction in the donor, rather than in the recipient (12, 30), could work as a strategy to dissociate GVH and GVL effects. We propose a new protocol that combines the induction of oral tolerance in bone marrow transplant donors to Ags from the recipient, with the use of the potential probiotic Lactococcus lactis NCDO2118 as tolerogenic adjuvant (combined therapy). Using a haploidentical model of aGVHD, we found that treatment of donor mice with the combined therapy protects recipients from aGVHD while maintaining the GVL effect. This protection is long lasting, Ag specific, and dependent on donor B cells, which induce regulatory T cells (Tregs) in the host.

Male 8- to 10-wk-old C57BL/6 (B6) (H-2^b), B6 IL-10 knockout, or F₁ (BALB/c × B6^dXb) mice were used as cell donors for GVHD induction experiments and female 10- to 12-wk-old F₁ (BALB/c × B6 − H-2^dXb) mice were used as recipients. For skin transplants, male 12- to 14-wk-old BALB/c (H-2^d), B10.A (H-2^a), or B6 mice were used as donors, and female B6 mice or F₁ (BALB/c × B6) mice were used as recipient. All animals were bred at the Brazilian National Cancer Institute animal facility (Rio de Janeiro, Brazil), housed in sterilized microisolators, and were handled according to the institutional guidelines. Animal care and all animal procedures were previously approved by Ethical Committee for Animal Experimentation of the Brazilian National Cancer Institute.

The wild-type L. lactis ssp lactis NCDO2118 strain (obtained from Laboratório de Genética Celular e Molecular from Universidade Federal de Minas Gerais, MG/Brazil - taxonomy ID 1117941 at the National Center for Biotechnology Information database). The bacteria were grown in M17 broth (Difco) supplemented with 0.5% glucose at 30°C without agitation for 24 h. In the second day, when a 2.0 OD at 600 nm (OD, nm) was reached, an inoculum with 5 μl was diluted in 25 ml supplemented M17 broth and offered to the animals.

To prepare the protein extract, F₁ (BALB/c × B6) spleen cells were resuspended on ice-cold PBS and filtered through a 40-μm cell strainer, cells were disrupted by freezing and thawing five times and boiled for 10 min. Soluble fraction was obtained by centrifugation at 14,000 rpm, 30 min, at 4°C and used as oral treatment. Total protein was quantified with Bio-Rad protein kit assay, according to the manufacturer’s specifications.

Donor B6 mice received 50 μg F₁ (BALB/c × B6) spleen protein extract diluted in 100 μl PBS, by gavage, daily for 5 d. During this period, ad libitum doses of L. lactis NCDO2118 were offered in replacement of drinking water. Four days later, the animals were immunized with 10⁶ F₁ splenocytes i.p. (31), and donor splenic cells were collected 1 wk after.

Bone marrow transplantation (BMT) was modified from described previously (32). F₁ mice were lethally irradiated with 950 cGy total body irradiation (TH780C irradiator with a cobalt 60 [⁶⁰Co] source), followed by the i.v. transfer of 5 × 10⁶ bone marrow cells from healthy donors along with 5 × 10⁶ T cells from total spleen of treated or untreated donors. When indicated, spleen cells were depleted of CD25⁺ cells or CD19⁺ cells. In GVL experiments, 10⁵ P815 GFP-expressing mastocytoma cells were injected together with the bone marrow inoculum.

Mice were monitored daily for survival and weekly for GVHD clinical score. Clinical evaluation was modified from the literature (33, 34) and was based on six parameters: weight loss, fur texture, activity, posture, fecal consistency, and survival rate. For histopathologic examination, samples of skin, liver, and colon collected at day 21 posttransplant were formalin preserved, paraffin embedded, sectioned, and stained with H&E. The pathologic score system was based on evaluation of individual parameters for each organ being the final score obtained by the sum of them. For skin sections, inflammatory infiltration, fibrosis, loss of appendices, epidermal chances, and ulceration were evaluated. For liver samples, parenchyma suffering, inflammatory infiltration, and portal space destruction were evaluated. For colons, lamina propria infiltration, deeper layer infiltration, structural changes, and damage extension were analyzed.

At day 25 posttransplant, samples of spleen, liver, bone marrow, total lymph nodes, and blood were analyzed by flow cytometry for the presence of P815-GFP⁺ cell. For determination of tumor total elimination, recipients were monitored daily for tumor-related morbidity and mortality. Tumor-related death was determined if mice had paralysis (in that case they were euthanized immediately and examined) and if autopsy identified hepatosplenomegaly and presence of macroscopic tumor nodules on the liver, spleen, or bone marrow.

For Treg depletion, spleen cells from donor mice were incubated with anti-CD25 (PC61.5) rat anti-mouse IgG Ab and positively selected using goat anti-rat IgG Dynabeads (Invitrogen, Carlsbad, CA). B cell depletion was done by positive selection using goat anti-mouse IgG Dynabeads (Invitrogen). After depletion, cell suspensions presented <1% of CD25⁺ and 3% of CD19⁺ cells. Both procedures were done accordingly to the manufacturer’s recommendations. For in vivo Treg depletion, recipient chimeras were injected i.p. with PC61 mAb (250 μg/mouse) at day 3 posttransplant. On day 14 posttransplant another i.p. dose of PC61 mAb were injected (100 μg/mice).

CD19⁺ B cells were isolated from donor spleen by positive selection using anti-CD19 MACS microbeads (Miltenyi Biotec, Auburn, CA), according to the manufacturer’s recommendations. For adoptive transfer experiments, 1 × 10⁷ purified CD19⁺ B cells (purity > 95%) from treated or untreated donor were injected along with the graft. When indicated, the spleen used as T cell source was previously depleted of total B cells and reconstituted with purified CD19⁺ B cells.

For surface staining, cell suspensions were preincubated with 2% normal mouse serum in PBS (Sigma-Aldrich, St. Louis, MO) on ice, followed by incubation with appropriate Ab mixture for 15 min. For Foxp3 staining, cells were labeled as specified in the eBioscience kit standard protocol. All samples were acquired in a FACSCalibur (BD Biosciences, San Jose, CA) or in a FACSCanto (BD Biosciences). Acquisitions were performed using CellQuest (BD Biosciences) software and BD FACSDiva software, respectively. Analyses were performed using FlowJo software (Tree Star, Ashland, OR). For cell staining the following Abs were used: FITC or allophycocyanin anti-mouse CD4 (GK1.5); PECy7 CD8a (53-6.7); PE anti-mouse CD25 (PC61.5, 7D4); allophycocyanin anti-rat/mouse Foxp3 staining kit (FJK-16); FITC or PECy5 anti-mouse CD3e (145-2C11); biotinylated anti-mouse CD11c (HL3); FITC anti-mouse CD11b (M1/70); PE, PercP-Cy5.5, allophycocyanin, or FITC anti-mouse CD19 (1D3); PE anti-mouse B220 (RA3-6B2); FITC anti-mouse IgM (eB121-15F9); allophycocyanin anti-mouse IgD (11-26c [11-26]); FITC anti-mouse CD5 (53-7.3); PE anti-mouse CD1d (1B1); allophycocyanin-eFluor 780 anti-mouse CD21 (eBio8D9); PE anti-mouse CD43 (eBioR2/60); allophycocyanin anti-mouse CD40 (1C10); FITC anti-mouse CD86 (GL1); biotinylated anti-mouse B7-H2 (HK5.3); allophycocyanin anti-mouse I-A/I-E (M5/114.15.2); biotinylated or allophycocyanin anti-mouse H-2Kb (AF6-88.5.5.3); PE anti-mouse H-2Kd (SF1-1.1.1) and FITC, PE, PerCP, or allophycocyanin streptavidins were all purchased from eBioscience (San Diego, CA). PE anti-mouse latency-associated polypeptide (LAP) (TW7-16B4) was purchased from BioLegend (San Diego, CA).

For grouped analysis, one-way or two-way ANOVA with Bonferroni posttest were used. Unpaired Student t test was applied for statistical differences between two groups. Survival data were analyzed with log-rank (Mantel–Cox) test. Error bars represent SD. Values of p < 0.05, < 0.01, and < 0.001 are referred to in the figure legends with symbols *, **, and ***, respectively.

Although it has been reported that induction of oral tolerance in the recipient can ameliorate aGVHD (12), there are no reports on whether it is possible to induce tolerance to host Ags in the donor before the transplant is performed. To determine whether donor T cells can be rendered tolerant to host tissues through the oral route, donor B6 mice received daily doses of host protein Ags by gavage, whereas L. lactis was offered ad libitum during 5 d (combined therapy). These animals were immunized i.p. with recipient spleen cells after 4 d and 1 wk later used as spleen cell donors along with normal B6 bone marrow to reconstitute lethally irradiated F₁ (BALB/c × B6) mice (Fig. 1A). Fig. 1B shows that when combined therapy is used, an elevated survival rate was observed. Also, evaluation of clinical parameters reveals an important decrease in disease severity when compared with all others groups tested (Fig. 1C, Supplemental Fig. 1). Donor treatment based on gavage of recipient protein Ags (AgF1) or L. lactis administration ad libitum (Llactis), used separately, shows only a discrete protection (Fig. 1, Supplemental Fig. 1).

To confirm the reduction in aGVHD, we examined histopathologically the skin, colon, and liver of recipient mice 21 d after BMT. The pathologic scores were significantly lower in recipients transplanted with spleen cells from treated donors (AgF1/Ll group) when compared with those who received spleen cells from untreated donors (Ctr⁺) (Fig. 2), AgF1, or L1actis groups (Supplemental Fig. 2). This reduction might be related to a higher IL-10 production found on affected tissues at that same period (Supplemental Fig. 3).

The close association between GVHD and the GVL effect reflects the need for therapies that target the undesired reaction only (1, 4). To determine whether the combined therapy is able to decrease GVHD while maintaining the benefic GVL effect, GFP-expressing P815 mastocytoma cells were injected along with the graft. Twenty-five days post-BMT, bone marrow, blood, lymph nodes, spleens, and livers of the recipients were examined for the presence of GFP⁺ cells. As expected by the absence of GVHD, GFP⁺ cells were found in all tissues analyzed of mice in the negative control disease group. In contrast, in the positive control disease group, where GVHD developed, no GFP⁺ cells were observed (Fig. 3A). However, despite being protected from aGVHD, mice in the experimental AgF1/Ll group were capable of eliminating the tumors cells from all tissues. The protected chimeras remained free of tumor for >15 wk post-BMT (Fig. 3B), whereas chimeras from negative and positive control disease groups died with clear signals of tumor growth and GVHD development, respectively.

Because the therapy proposed in this study combines two treatments with immunomodulatory effects, the protection observed could be a result of a general state of immunosuppression that would interfere with recipient response to other Ags. To test whether the immunosuppression induced by combined therapy was specific for host Ags, recipients protected for >1 y post-BMT received skin grafts from third-party and from BALB/c donors. The protected chimeras specifically rejected the unrelated skin graft but accepted the BALB/c graft (Supplemental Fig. 4A).

However, after 1 y, we cannot exclude the participation of newly generated T cells because the thymic function of the chimeras, at that posttransplant period, already had been restored and the newly emergent T cells are tolerant to the BALB/c Ags. To test the specificity of the tolerance induced for host Ags, B6 mice treated or not with the combined therapy received skin grafts from third-party and from F₁ donors. As shown in Supplemental Fig. 4B, both treated or untreated B6 recipients rejected the skin grafts. However, when female B6 mice were treated with the combined therapy using male B6 splenic protein extract on gavage, they did not reject male skin graft unlike the untreated females (Supplemental Fig. 4C). It suggests that combined therapy induces a more effective tolerance to minor rather than major histocompatibility Ags, which may explain how it is able to protect from GVHD preserving the GVL effect.

Altogether, the data show that treatment of donor mice with combined therapy efficiently reduces clinical and pathological aGVHD development on a specific and long-lasting fashion, without immunosuppressing the recipient and preserving the GVL effect.

The phenomenon of oral tolerance can occur by different mechanisms being the expansion of regulatory-suppressive T cells generally related to repeated lower doses of Ag (5, 6). To investigate the mechanism by which the combined therapy exerts the protection against aGVHD, spleen cells from treated and untreated donors were analyzed. No significant differences were observed in absolute number or frequency of splenic cell subpopulations, including Foxp3⁺ T cells (Fig. 4A, 4B). To further exclude the participation of donor Tregs, CD25⁺ T cells were depleted from spleen cells of treated donors before transplantation. Depletion of donor Tregs did not abrogate the protection, and the elevated survival rate was maintained (Fig. 4C, 4D), confirming that these cells are not necessary for the protective effect.

Several recent studies in both mice and humans have showed the participation of B cell on immunoregulation and protection against acquired inflammatory syndromes (35). In models of oral tolerance induction, B cells seem to play an important role in tolerance induced with low doses of Ag (36). Because the combined therapy is based on induction of oral tolerance with small amounts of recipient Ag, we evaluated the participation of donor B cells in the combined therapy protocol. Spleen cells from combined therapy–treated donors were depleted of B lymphocytes before transplantation. As indicated in Fig. 5A, elimination of B cells completely abrogated the protection from aGVHD, leading to a higher total clinical score and an elevated mortality rate in this group (AgF1/Ll [-L_ØB]) when compared with non-B cell–depleted one (AgF1/Ll). Because the role of B cells in GVHD pathophysiology is still controversial (37), one may argue that loss of protection could be related not only to the absence of B cells tolerized by treatment but also being exacerbated by the absence of B lymphocytes at all. To overcome this issue, spleen cells from combined therapy–treated donors depleted of B cells were reconstituted with purified splenic CD19⁺ B cells from untreated donors. The reconstitution had no effect, and the recipients still developed aGVHD. However, when CD19⁺ B cells from a treated donor were used to reconstitute the combined therapy B-depleted spleen, protection was completely restored (Fig. 5A).

We next tested whether tolerized B cells were sufficient to inhibit aGVHD by transferring CD19⁺ B cells from a treated donor along with spleen cells (and bone marrow) from a control untreated B6 donor (Fig. 5B). Of interest, depletion of B lymphocytes from untreated donor spleen cells ameliorated the GVHD, suggesting that B cells from untreated mice, at least in this model, have a pathogenic role in GVHD development. Taken together, the data show that protection of recipient mice against aGVHD is mediated by B lymphocytes with regulatory functions generated by the combined therapy treatment of donor.

To better characterize the B cells involved in the regulatory activity over aGVHD development, donor mice were treated or not with the combined therapy and their splenic B cells analyzed by flow cytometry.

When the B lymphocytes were grouped according to the expression of IgM and IgD, we found a higher frequency of transitional B cells (CD19⁺IgM⁺IgD^lo) in treated animals. No differences were observed among the immature (CD19⁺IgM^hiIgD⁻) and mature (CD19⁺IgM^loIgD⁺) B cells subpopulations (Fig. 6A), suggesting that combined therapy induced an increase in transitional Bregs. However, several different phenotypes have been attributed to B cells with regulatory functions and at least two subpopulations were described: transitional Breg, mentioned above, and the B10 cells (37). In common, both express CD21 and CD1d molecules, being that the B10 is also positive for CD5 (38). Despite the increase in overall CD21 expression in splenic B cells from treated donors (Fig. 6B), we observed that CD19⁺CD21⁺CD1d⁺ and CD19⁺CD21⁺CD1d⁺CD5⁺ subpopulations were reduced on these animals (Fig. 6E, 6F).

Although there is not a unique marker, or set of markers, that exclusively identifies regulatory B cells (Bregs), expression of molecules such as CD86 (39) and class II MHC (40) are necessary for most of their regulatory functions, regardless of the specific B cell immunophenotype. In fact, splenic B lymphocytes from combined therapy–treated donor express high amounts of both molecules (Fig. 6C, 6D).

Recently, B cells with strong suppressive functions in vitro and in vivo were generated by the incubation of naive B cells with a relevant Ag conjugated to cholera toxin B subunit. In this scenario, tolerance was shown to be at least, partially dependent on B cell expressing the LAP/TGF-β (41). On the basis of these data, we looked at LAP expression in CD19⁺ cells and found a higher expression of this molecule in B cells from treated when compared with untreated mice (Fig. 6G, 6H).

Convincing data have attributed the suppressive capacity of Bregs to their IL-10 production (41–43). To check whether the protection observed in this study is due to B cell–derived IL-10, CD19⁺ B cells from B6 IL-10 knockout mice treated with the combined therapy were transferred to GVHD developing mice (irradiated F₁ receiving spleen and bone marrow cells from control B6 mice). The results showed that B cells from IL-10–deficient mice were not able to protect the recipient from GVHD development, unlike B cells from IL-10–sufficient donors pretreated with the combined therapy (Fig. 7A, 7B).

Altogether, the above results indicate that the protection observed after oral combined therapy administration depends on IL-10–producing Bregs.

A growing body of evidence has demonstrated the involvement of Bregs in the generation or recruitment of regulatory T lymphocytes (41, 44). To test whether B cells from combined therapy–treated donors were inducing Tregs, chimeras transplanted with control spleen cells or spleens depleted of B lymphocytes from both, treated or untreated donors, were evaluated for the presence of donor regulatory Foxp3⁺ cells in spleens and mesenteric and peripheral lymph nodes 4 d post-BMT. As shown in Fig. 8A and 8B, recipients reconstituted with spleen cells from treated donors had higher frequencies and absolute numbers of Foxp3⁺ cells in the spleen and lymph nodes. In addition, depletion of B lymphocytes from the graft completely abrogated this increase. Of note, the increase of donor Foxp3⁺ cells happens in the recipient mice, indicating that the regulatory phenotype is induced during the alloresponse in the host and not before because spleen cells from donor mice do not have higher numbers of Tregs nor do the spleen cells depend on the Tregs to exert their inhibitory effect (Fig. 4). Reinforcing these data, in vivo depletion of CD25⁺cells from recipient reconstituted with combined therapy–treated donors totally abrogated the protection (Fig. 9A, 9B).

Taken together, the results show that tolerized B cells from the donors mediate the protection against GVHD by inducing Foxp3⁺ cells in the transplant recipients.

HSCT is an immunotherapy for treatment of malignant and nonmalignant diseases that is seriously limited by GVHD, an immune syndrome with high morbidity and mortality rates (1–3). Prophylaxis and treatment of GVHD is based on nonspecific immunosupressive agents that, among other complications, eliminate GVLr, which elevates the risk of tumor relapse (4). Education of the immune system by induction of oral tolerance (12) and manipulation of gut microbiota by ingestion of probiotic bacteria (30) represent rationally viable and noninvasive therapies to reduce the development of GVHD. Nonetheless, their clinical application has been limited because the HSCT recipients are subjected to conditioning regimens that require a strict diet in which live or raw food is precluded.

In this study, we demonstrate that treatment of donor mice with combined therapy reduces clinical and pathological manifestations of aGVHD, improving the survival rates to 100%. This phenomenon is specific and long lasting, and it is not a result of a nonspecific immunosuppressive milieu. Maintenance of beneficial GVL effect is as important as the reduction of GVHD when HSCT is used to treat a malignant disease (45–47). In humans, although it has been suggested that GVL and GVHD are temporally separate (48) and that the severity of GVHD does not implicate in improved GVL effect (49), both reactions are clinically associated (4), and there is no specific therapeutic approach to target the undesired reaction only. In our approach, recipients reconstituted with spleen cells from combined therapy–treated donors develop less severe aGVHD and are still capable of eliminating tumor cells injected along with the graft, showing that the GVLr is maintained. This dual capacity may be related with the fact that combined therapy induces a more effective tolerance to minor histocompatibility Ags. Actually, it is expected because the Ags are administered in the form of soluble proteins that will be captured and processed, losing the entire MHC peptide complexes.

Recently, B lymphocytes subsets able to modulate inflammatory responses were described previously (35). These Bregs seem to play an essential role in oral tolerance induced with low amounts of Ags (36). In accordance to that, protection obtained with combined therapy is totally abrogated in the absence of B lymphocytes. Moreover, CD19⁺ cells from treated donors, but not from untreated ones, are sufficient to inhibit aGVHD in recipients reconstituted with control spleen cells, showing that combined therapy induces the generation of a B cell subpopulation with regulatory functions.

Generation of specific Foxp3⁺ Tregs by allogeneic primary B cells (50) and by LAP/TGF-β⁺ B cells (41) have been reported previously. Indeed, our data show that combined therapy not only increases the CD19⁺LAP⁺ population but also induces donor Foxp3⁺ Tregs. Interestingly, in our model, the expansion of Tregs did not occur in treated donors because we did not observe differences in frequency or absolute number of donor splenic Foxp3⁺ cells. Moreover, elimination of CD25⁺ population in the graft before the transplant does not abrogate the protective effect from combined therapy. The induction of donor Foxp3⁺ cells, possibly from effector alloreactive T cells, happens on recipient after the transplant and is dependent on those B cells generated by the treatment, indicating that Ag-specific presentation and recognition are important steps for acquisition of a regulatory phenotype by allo-T cells and for their following activation. The abrogation of protection after in vivo depletion of CD25⁺ cells on recipient supports this hypothesis. On the basis of these data, we propose a model in which B cells from orally treated mice present cognate Ags to allospecific responder T cells responsible for the GVH reaction, explaining the Ag-specific suppression toward aGVHD only. These results are in accordance with previous reports showing that the generation of specific Tregs after allostimulation in vitro inhibits aGVHD and spare GVL/GVT responses in some experimental models (51, 52).

The heterogeneity among Bregs and the absence of a master transcriptional factor associated with this subpopulation strongly suggest that, like their T cell counterpart, they can be divided into different regulatory subsets originated from ontogenically distinct subsets (53). However, some features such as the production of IL-10 (35) and expression of molecules like CD86 (39) and class II MHC (40) along with an immature phenotype are common among the different subsets. In fact, we observed increased IL-10 in protected recipients. Moreover, combined therapy induces a high frequency of transitional B lymphocytes, with high expression of CD21, CD86, and class II MHC, a phenotype compatible with Bregs. In addition, we demonstrated that IL-10 production by splenic B cells from treated donors is necessary for the protection observed after combined therapy, indicating the B10 cells, the IL-10–producing Breg (42), is in fact the key player in this setting. The contribution of a distinct cell type in the immediate induction of B cell with regulatory function cannot be excluded, although to our knowledge there is no report showing the dependence of other cells, such as dendritic cells, on Breg activation.

Finally, B cell differentiation into a regulatory subtype was described by Lampropoulou et al. (43) as being dependent on two signals. A first signal delivered by TLR initiates IL-10 production by B cells. In a second phase, engagement of receptors classically involved with B cell survival and expansion, such as the BCR, amplifies the initial population of IL-10–producing B cells, allowing the effective suppression. Altogether, these data might support the idea that the ingestion of L. lactis gives the innate stimulus necessary to the acquisition of a regulatory phenotype by the B cells, which is dependent on IL-10 in our model. The recipient Ags in the protein extract would then provide BCR stimuli that would enable specific B cells to present the cognate Ags to T cells, generating Tregs. Activity of these alloreactive Tregs would dampen the response to recipient Ags but preserve the response to Ags derived from the tumor cells (51, 52).

Our results suggest that oral tolerance induction by the use of a probiotic bacteria and the recipient protein extract, on HSCT donors before transplantation, might prove safe, simple, and effective for preventing GVHD in human patients.

We thank Ligia Peçanha for critical reading of the manuscript and suggestions, João P. Monteiro for discussions and critical reading and Jane Littrell for English review.

The authors have no financial conflicts of interest.