|Home | About | Journals | Submit | Contact Us | Français|
The role of CD8+ T cells in oral tolerance remains unclear. To address this, we developed a model to induce CD8+ Tregs by feeding the MHC Class I immunodominant epitope of OVA, OVA(257–264). OVA(257–264)-feeding induced tolerance similar to that observed in OVA protein-fed mice, capable of suppressing the production of Th1 and Th17 cytokines and inhibiting a Th1-driven DTH response following immunization with whole OVA protein. OVA(257–264)-peptide induced suppression could be transferred to naïve mice with CD8+ cells, but not CD8-depleted cells, isolated from MLNs of peptide-fed mice. Interestingly, while capable of inhibiting Th1 and Th17 responses, OVA(257–264)-feeding could not suppress any feature of a Th2 inflammatory response, though OVA protein-feeding could, suggesting that these cells function through a different mechanism than their CD4+ counterparts generated in response to feeding with whole OVA. Thus, CD8+ T cells are functionally capable of mediating tolerance to Th1 and Th17 responses.
Tolerance to ingested antigens is an essential feature required for normal homeostasis and maintenance of barrier function in the gastrointestinal tract. Breakdown of tolerance is associated with the development of chronic inflammatory diseases, such as Crohn’s disease and ulcerative colitis, as well as food allergies and celiac disease1. Although the specific mechanisms involved in the generation and maintenance of oral tolerance (OT) are poorly understood, numerous studies have suggested that it is mediated by a combination of clonal deletion, clonal anergy, and active suppression by regulatory T cellsRev. in 1–5. High doses of antigen are believed to induce clonal deletion and/or anergy of responding T cells6–8, while multiple feedings with low doses of antigen stimulate active suppression of immune responses through the generation of regulatory T cells1, 3–5, 9, 10. While high dose OT is obviously important for normal gut homeostasis, low dose OT is of particular therapeutic interest because it offers a simple, noninvasive method for generating specific regulatory cells for the treatment of autoimmune and chronic inflammatory diseases.
Early OT studies focused on CD8+ T suppressor cells as primarily mediators of suppressionrev. in 3. Feeding with low doses of myelin basic protein resulted in the generation of CD8+ regulatory T cells that suppressed the induction of experimental autoimmune encephalitis (EAE) in both mice and rats through the production of TGF-β3, 9, 11, 12. Bland and Warren (1986) described a CD8+ suppressor T cell population that obtained regulatory properties after co-culture with intestinal epithelial cells13. CD8+ Tregs generated in low dose OT models have been shown to inhibit CTL activity, suppress cytokine production, and alter antibody responses in the gut3, 11, 14–16. Previous work from our lab identified a population of human regulatory CD8+ T cells, activated by co-culture with intestinal epithelial cells (IECs), capable of suppressing through a contact dependent mechanism16, 17. Defects in this cell population were observed in patients with inflammatory bowel disease, and correlated with a failure to induce OT in these patients, suggesting that they could play an important role in tolerance in the gut18. We have also developed a murine model where tolerance was induced by antigen uptake through IECs rather than Peyer’s patches or interdigitating DCs19, supporting a role for IECs in the induction of tolerance. As our own in vitro studies, as well as prior work by Bland and Warren13, demonstrate that IECs can activate CD8+ T cells with regulatory activity, these findings can be interpreted to further support a role for CD8+ Tregs in OT. On the other hand, studies utilizing CD8+ T cell deficient mice have demonstrated that they are not essential for the generation of low dose OT, but that there is an absolute requirement for CD4+ T cells9, 20–24. As such, recent studies have focused on the generation and function of CD4+ regulatory T cells in tolerance, without taking into account the possible significant contributions of CD8+ T cells.
Though it has been established that CD8+ T cells are not required for the induction of OT in animal models, there is sufficient evidence to suggest that they can mediate tolerance effectively, and that they could play an important role in tolerance in a normal setting. Studies using knockout and transgenic mice provide adequate information about the absolute requirement of cells in a particular immune response, but the developmental artifacts often associated with such mice precludes their use to address the relative contribution and subtle interactions of T cell subsets in the development of tolerance with respect to an intact immune system. To this end, we have developed a novel model system in which CD8+ Tregs can be generated independently of CD4+ Tregs by antigen feeding in a normal, unaltered background. Using this model, we have now demonstrated that CD8+ T cells are capable of mediating OT independently of their CD4+ counterparts.
To investigate the role of CD8+ T cells in tolerance in an immunologically intact mouse, we fed mice the MHC Class I immunodominant epitope of OVA, OVA(257–264) (herein referred to by its amino-acid sequence, SIINFEKL), for five consecutive days (Fig. 1a). Peptide purity was demonstrated by HPLC, and the inability of SIINFEKL to induce proliferation of OT-II cells (specific for the OVA(323–339) peptide) (data not shown). Mice were fed wOVA or PBS as positive or negative controls for tolerance, respectively. To assess the establishment of OT, mice were immunized with wOVA in CFA and draining LN (dLN) cells and splenocytes were subjected to in vitro restimulation with wOVA. dLN cells from PBS-fed, OVA-immunized mice produced substantial levels of inflammatory cytokines (IL-2, IFNγ, and IL-17a) following in vitro restimulation with wOVA (Figs. 2 and and3).3). Mice fed wOVA prior to immunization, demonstrated significantly reduced IFNγ, IL-17a (p<0.01), and IL-2 (p<0.05) production at all time-points assayed (Figs. 2 and and3).3). Cells isolated from mice fed SIINFEKL produced significantly less IFNγ and IL-17a at both 48h and 72h (p<0.01), as well as, diminished IL-2 at 48h following in vitro restimulation with wOVA, compared to PBS-fed mice (Fig. 2). A similar result was obtained when mice were both fed and immunized with SIINFEKL, although the overall cytokine response was substantially reduced in response to immunization with peptide rather than whole protein (Fig. S1); IL-17a levels were not determined. The inhibition of inflammatory cytokine production in dLN cell cultures from mice fed SIINFEKL closely modeled that observed in cultures derived from mice fed whole OVA protein (Fig. 2). To confirm the ability of CD4+ T cells to function independently of CD8+ T cells in the development of tolerance in a normal immune system, mice were also fed OVA(323–339), the MHC class II immunodominant epitope of OVA. Similar to tolerance induced in response to SIINFEKL feeding, dLN cells isolated from mice fed OVA(323–339) produced lower levels of IFNγ, IL-17a, and IL-2 following in vitro restimulation with wOVA (Fig. 3). IL-4, IL-10, IL-13, and TGFβ were undetectable by ELISA at all time-points assayed (data not shown). In vitro restimulation of splenocyte cultures resulted in a greatly attenuated response compared to dLN cells, making comparisons between feeding groups impossible (data not shown).
To demonstrate that the response to SIINFEKL feeding was mediated by CD8+ T cells, we first utilized CD8−/− mice. It has been previously established that CD8+ T cells are not required for tolerance to protein antigens in animal models9, 20–24. Our results support these previous findings, as wOVA-fed CD8−/− mice produced significantly reduced levels of IL-2 at 48h, and IFNγ and IL-17a at 48h and 72h, following in vitro restimulation with wOVA, compared to PBS-fed CD8−/− mice (Fig. 4). dLN cultures from SIINFEKL-fed CD8−/− mice produced similar levels of IL-2 at 48h, and nearly identical levels of IFNγ and IL-17a at 48h and 72h, compared to PBS-fed mice (Fig. 4).
To confirm the development of a regulatory CD8+ T cell population in response to SIINFEKL feeding, CD8+ cells were isolated from the MLNs of SIINFEKL-fed mice on day 5 of peptide feeding and transferred IV into naïve mice, rendering recipient mice tolerant to subsequent immunization with wOVA protein (Fig. 5). dLN cells from naïve recipients of SIINFEKL-fed CD8+ MLN cells produced significantly lower levels of IFNγ at 48h and 72h and IL-17a at 72h compared to recipients of CD8+ cells from OVA-fed mice (Fig. 5a). IL-17a was not detected in 48h cultures. A similar trend was observed when comparing SIINFEKL-fed and PBS-fed CD8+ recipient mice, although this did not reach statistical significance. Reduced cytokine production was not observed when CD8 depleted MLN cells were transferred from SIINFEKL-fed mice; however, tolerance to wOVA could be transferred with CD8 depleted MLNs from OVA-fed mice compared to PBS-and SIINFEKL-fed cell transfer recipients (Fig. 5b). Therefore, CD8+ T cells are required for SIINFEKL-mediated, but not OVA-mediated tolerance.
These data clearly demonstrate that tolerance can be induced in wildtype mice by feeding OVA derived peptides at a level comparable to that observed in mice fed the intact protein. They suggest that tolerance occurs through the induction of peptide-specific regulatory cells because feeding with either SIINFEKL or OVA(323–339) suppresses the response to other OVA epitopes presented following systemic immunization with wOVA. As 1) SIINFEKL binds only to MHC class I and is presented only to CD8+ T cells, 2) tolerance is not observed in CD8−/− mice, and 3) tolerance can be transferred to naïve mice with CD8+ cells, but not CD8 depleted cells, isolated from SIINFEKL-fed mice, these data strongly suggest that SIINFEKL feeding induces CD8+ T cells with regulatory activity that are effective at mediating OT, independently of the induction of CD4+ Tregs.
Systemic antibody levels remained unaffected by the induction of OT using SIINFEKL-fed wild type mice. Reproducible differences were not observed when comparing total, IgG1, or IgG2c OVA-specific antibody levels in serum from PBS- or antigen-fed mice (Fig. 6a). The ability of OT to effectively modulate B cell responses has been controversial. Some studies have demonstrated reduced antibody production in response to feeding, while others have reported a limited effect on serum antibody levels; thus, these findings are not surprising and are consistent with previously published observationsrev in,5. On the other hand, local IgA production was enhanced in antigen-fed mice compared to mice fed PBS (Fig. 6b). Significantly higher levels of total IgA were observed in fecal extracts from wOVA-fed mice compared to PBS-fed mice. Increases were also observed in SIINFEKL- and OVA(323–339)-fed mice, but did not reach statistical significance. Though the relationship between the induction of OT and mucosal IgA production remains unclear25, these data support previously published reports demonstrating increased mucosal IgA production in conjunction with suppression of systemic responses following antigen feeding5, 26, 27. Further study is required to clarify the importance of SIgA with respect to tolerance induction.
To determine the ability of SIINFEKL-induced regulatory cells to suppress inflammation in vivo, we employed a mouse model of delayed type hypersensitivity (DTH) (Fig. 1b). Twelve days after immunization, antigen-fed mice had wOVA injected into the right ear, while PBS was injected into the left ear as a control. The induction of a DTH response was demonstrated by an increase in ear thickness in the right ear (OVA) as compared to the left ear (PBS). 24h after injection of wOVA, both SIINFEKL- and OVA-fed mice had significantly reduced ear swelling compared to PBS-fed mice (p<0.05 and p<0.001, respectively), though wOVA-fed mice had a more pronounced reduction (Fig. 7a). A similar trend was observed at 48h, although the difference did not reach statistical significance when comparing SIINFEKL-fed and PBS-fed mice (Fig. 7a). When ear swelling was monitored over the course of four days, at no time did swelling in SIINFEKL- or wOVA-fed mice reach the level observed in PBS-fed mice (data not shown). Reduced ear swelling was observed following the induction of a DTH response in transfer recipients of CD8+ MLN cells from SIINFEKL-fed mice, compared to recipients of CD8+ MLN cells from PBS-fed or OVA-fed mice (Fig. 7b). Ear swelling was not reduced when CD8+ depleted MLN cells were transferred from SIINFEKL-fed mice, but was observed when cells were transferred from OVA-fed mice, compared to PBS-fed recipients (Fig. 7c). Thus, SIINFEKL mediated suppression of inflammation is mediated by CD8+ cells. In most experiments, mice were sacrificed at 48h and the ears and popliteal LNs (draining the site of primary immunization) were isolated. The presence of inflammatory cells in the ears was confirmed histologically by H&E staining, with a clear reduction in cell infiltrates observed in OVA-fed mice and a modest reduction observed in SIINFEKL-fed mice compared to PBS-fed mice (data not shown). Tolerance was also confirmed by the reduced production of IFNγ and IL-17a following in vitro restimulation of popliteal LN cells as described in Fig. 2 (data not shown).
Having demonstrated that SIINFEKL feeding could suppress Th1 and Th17 cytokine production in vitro and inhibit Th1 driven inflammation in vivo, we next assessed its ability to suppress Th2 mediated inflammation using a mouse model of allergic lung inflammation. Allergic lung inflammation was induced in antigen-fed mice by sensitizing with 10 µg wOVA in alum on day eight followed 14 days later by five consecutive daily intranasal instillations of wOVA in saline (Fig. 1c). Control mice received wOVA intranasally in the absence of sensitization. Contrary to our Th1-driven DTH results, SIINFEKL-fed mice (7 out of 10) had a robust inflammatory response (inflammatory score=2.65 ± 0.95) comparable to PBS-fed mice (inflammatory score=3.38 ± 0.55), consisting of eosinophil-rich infiltrates in the peri-bronchiolar and peri-vascular regions of the lung, hypertrophy of the bronchiolar epithelium, and plugging of some airways (Figs 8a and S2a). By contrast, OVA-fed mice had much fewer and smaller infiltrates, involving less of the total lung tissue, and no plugging of the airways (inflammatory score=1.85 ± 0.66, p<0.05 versus PBS- and SIINFEKL-fed mice) (Figs 8a and S2a). Correspondingly, there were significantly fewer inflammatory cells in BAL washes derived from OVA-fed mice, which were themselves comprised of a significantly smaller percentage of eosinophils, compared to PBS- and SIINFEKL-fed mice (p<0.01) (Fig. 8b). Mucus production was assessed by PAS staining. Positive staining was generally limited to larger airways, and substantial differences in staining were not observed between feeding groups (Fig. S2b); however, staining was much more prominent in sensitized mice compared to non-allergic controls (Fig. S2b). A trend towards diminished IL-4 levels was observed in BAL fluids from OVA-fed mice as compared to PBS- and SIINFEKL-fed, while IL-10 levels were increased (Fig. 8c). IL-13 was not detected in any group of mice (data not shown), and, interestingly, IL-5 was elevated only in BAL fluids derived from SIINFEKL-fed mice (Fig. 8c). Finally, the level of OVA-specific total and IgG1 antibodies, and total IgE in serum as well as OVA-specific IgG1 antibody in BAL fluids was significantly reduced in OVA-fed mice compared to PBS- and SIINFEKL-fed mice (Fig. 8d).
Though effective at suppressing Th1 and Th17 cytokine production and inhibiting Th1-induced inflammation in vivo, SIINFEKL-feeding was incapable of modulating Th2-driven inflammation. On the other hand, a marked decrease in inflammatory cells was observed in the lungs of OVA-fed mice, though the response was not completely ablated. As wOVA feeding likely induces the generation of CD4+ Tregs, this could suggest that SIINFEKL-induced CD8+ cells function through a different regulatory mechanism.
Though the specific mechanisms involved in the induction and maintenance of OT are poorly understood, it is generally accepted that high doses of antigen induce tolerance through anergy and clonal deletion, while low doses induce active suppression through the development of regulatory cells. Nevertheless, it is unlikely that these mechanisms are exclusive of one another, and it remains unclear as to what constitutes a high dose versus a low dose for any given antigen. 1 mg of wOVA is a low dose for OT induction specifically because of its demonstrated ability to induce OVA specific regulatory cells, rather than deletion/anergy. Our current studies were conducted using molar equivalents of SIINFEKL to induce tolerance; a dose response using a wide range of SIINFEKL concentrations (0.0217 µg-217.0 µg) demonstrated that both 2.17 µg and 21.7 µg induced a tolerogenic response comparable to feeding with 1 mg of OVA (data not shown). To clarify the general mechanism of tolerance induced by SIINFEKL feeding, we determined the presence of SIINFEKL-specific cells in dLNs and spleens of mice after feeding and immunization, via flow cytometry using fluorescently-labeled SIINFEKL-loaded H2-Kb pentamers. Splenocytes from OT-I Rag-1−/− mice, which contain CD8+ T cells of singular specificity for SIINFEKL, were used as a positive control for staining, while cells from CD8−/− mice were used as a negative control. Evaluation of dLN cells (Table I, Fig. S3) and splenocytes (data not shown) from PBS-, SIINFEKL-, or OVA-fed, OVA immunized and boosted mice revealed no significant differences in the numbers of total CD4+ or CD8+ T cells, or SIINFEKL-specific CD8+ T cells in the different feeding groups. Limited numbers of antigen specific cells were detected in both dLNs and spleens; however, this range was well within the expected value of antigen specific cells for a normal, nontransgenic, antigen immunized mouse. Positive staining was not observed in CD8−/− mice, nor was it readily detectable in spleens from unimmunized C57BL/6 mice (data not shown). Similar numbers of SIINFEKL-specific T cells in SIINFEKL-, PBS-, and OVA-fed mice indicates that SIINFEKL feeding does not result in T cell depletion at the concentrations used in this study, and further supports the premise that CD8+ T cells are actively regulating the tolerogenic response to OVA in our system. As a whole, these data provide clear evidence for the ability of CD8+ T cells to regulate tolerance independently of CD4+ T cell help in a normal immune setting.
The majority of studies investigating the regulation of OT have relied on the use of manipulated immune systems to study T cell function. These systems include the use of knockout mice, where the absence of an entire lineage of cells during development often results in additional unintended impairments in immune function or the development of compensatory mechanisms that are not present in a normal setting. Other studies have employed antibody to deplete entire cell lineages, or more commonly, transgenic mice with clonal populations of T cells that have singular specificity for the antigen (peptide) of interest. While these systems are valuable for the gross assessment of immune function of specific cell types, artifacts inherent to them do not allow for an intricate understanding of the immune response as it relates to a normal immune system.
To circumvent the need for a manipulated immune system we developed an OT model designed to induce the development of a CD8+ Treg population, independently of CD4+ Tregs, in a normal immune system through the feeding of the MHC class I immunodominant epitope of OVA, SIINFEKL. Feeding with SIINFEKL peptide suppressed the production of pro-inflammatory cytokines (IFNγ, IL-17a, IL-2) in vitro and inhibited Th1-driven delayed type hypersensitivity in vivo (Figs. 2 and and6).6). These data suggest that this occurs through the induction of a CD8+ regulatory cell population. The OVA protein is comprised of several highly immunogenic epitopes in addition to SIINFEKL, most notably OVA(323–339), against which the majority of the CD4+ T cell response to OVA is generated. If SIINFEKL-feeding induced anergy or deletion of peptide specific cells, the T cell response to other OVA epitopes would continue unabated, resulting in similar inflammation/cytokine production compared to PBS-fed mice. Clearly, neither anergy nor deletion were the primary mechanism of tolerance in this case, as the response to immunization with wOVA was significantly diminished, and staining with SIINFEKL-loaded H2-Kb pentamers clearly demonstrated that SIINFEKL-specific cells were not deleted following peptide feeding (Table I). Furthermore, SIINFEKL-induced tolerance required the presence of CD8+ T cells as peptide-induced tolerance could not be achieved in CD8−/− mice. Finally, and most importantly, tolerance to SIINFEKL could be transferred to naïve mice with CD8+ cells isolated from MLNs of SIINFEKL-fed mice, but not with CD8 depleted cells (Figs. 5 and and7).7). In total, these data strongly suggest that peptide feeding establishes a CD8+ Treg population that actively suppresses the response to other immunogenic OVA epitopes (bystander suppression).
CD8+ cells from SIINFEKL-fed mice were capable of suppressing inflammation in vivo in a Th1-driven DTH response; however, the level of suppression observed in SIINFEKL-fed mice was reduced compared to that observed in OVA-fed mice. This was anticipated as feeding with wOVA likely induced the generation of a mixture of regulatory cells of different specificities in conjunction with anergy and/or deletion of other OVA epitope-specific T cell populations. SIINFEKL-feeding would generate a regulatory cell population of singular specificity, likely with a single mechanism of action, which, comparatively, would have a more localized effect. Interestingly, our transfer of tolerance data suggest that feeding with wOVA does not induce the development of CD8+ regulatory cells, as transfer of CD8+ MLN cells from OVA-fed mice did not transfer tolerance, while transfer of CD8+ depleted cells did (most likely due to the presence of CD4+ Tregs). This was not unexpected; despite the fact that mice were administered molar equivalents of SIINFEKL and wOVA, the bioavailability of peptide derived through the digestion of whole protein versus the bioavailability of freely administered peptide is likely to be different. The induction of low dose oral tolerance is tightly controlled by antigen dose; thus, while a molar equivalent was used (compared to SINFEKL peptide), the amount of SIINFEKL (or other CD8 peptides) derived from the wOVA protein for presentation could be outside of the range necessary to induce tolerance in CD8+ cells.
Of significant interest was the finding that SIINFEKL-feeding did not suppress any aspect of a Th2-driven model of inflammation, while OVA-feeding did substantial reduce (though it did not completely prevent the response) the recruitment of inflammatory cells to the lungs. One explanation is that the CD8+ Treg cell population generated in response to SIINFEKL-feeding was not recruited to the site of inflammation, and therefore was not capable of mediating suppression in this instance. However, it has been shown that previously activated CD8+ T cells can be switched to produce IL-5 in the presence of a Th2 micro-environment28. Therefore, it is possible that the SIINFEKL specific cells are being recruited to the site inflammation in our model, but are being switched to an IL-5 producing profile in response to the Th2 microenvironment. As a result, they would be unable to suppress the Th2 response. This hypothesis is supported by the increased level of IL-5 observed in BAL fluids from SIINFEKL-fed mice, as compared to all other groups. Nevertheless, if these cells were switched to an IL-5 producing phenotype, we would expect an increased level of eosinophilia in the lungs of SIINFEKL-fed mice compared to the other groups, which was not observed. We are currently designing experiments to investigate the specific method of suppression used by SIINFEKL-induced CD8+ Tregs, as well as, attempting to create a phenotypic profile of these cells (including chemokine receptors, and homing markers) to address their ability to be recruited to different sites of inflammation. It is important to note, however, that we do not believe that the suppression of IFNγ or IL-17a in our dLN cell cultures is mediated through the production of Th2 cytokines. We have never detected the production of Th2 cytokines (IL-4 and IL-13) at any time point in any of our dLN cell cultures (data not shown).
Also of interest is the finding that OVA(323–339)-feeding required a 10-fold lower molar concentration of peptide to induce OT compared to feeding with SIINFEKL. When mice were fed a molar equivalent of OVA(323–339) (40.1 µg rather than 4.01 µg), tolerance could only be generated to subsequent immunization with OVA(323–339) peptide, but not to wOVA (data not shown). This suggests the induction of high dose tolerance, i.e., the functional inactivation or deletion of OVA(323–339) responsive cells, rather than the generation of OVA(323–339)-specific Tregs. High dose OT was not observed with up to a 10-fold greater molar concentration of SIINFEKL-peptide (217.0 µg). These data may suggest that CD4+ T cells are more sensitive to the induction of tolerance than their CD8+ T cell counterparts are. Alternatively, and perhaps more likely, different sensitivities in the induction of CD4+ vs. CD8+ T cell-mediated tolerance could be attributed to differences in the APC population involved in the activation of each cell type following antigen feeding. Previous work from our laboratory has demonstrated that CD8+ Tregs can be activated by co-culture with IECs. It is possible that IECs are playing a primary role in the activation of CD8+ Tregs in our model whilst CD4+ T cell activation is reliant on conventional APCs, such as DCs. In addition, it is possible that different subtypes of DCs might also be involved, as recent studies have provided intriguing evidence into a role for plasmacytoid DCs in the induction of CD8+ Tregs29, 30. We are currently designing experiments to assess the trafficking and uptake of our peptide antigens following feeding.
Historically, the goal of OT has been the inhibition of Th1-mediated immune responses, as these were believed to be the primary mediators of most autoimmune and chronic inflammatory diseases. More recently, however, a new inflammatory T cell ‘lineage’, Th17, has emerged as a key protagonist in a number of inflammatory diseases, including multiple sclerosis, Crohn’s Disease, and rheumatoid arthritis31, 32. In fact, in EAE, an animal model of multiple sclerosis, members of the IL-17 family, IL-17a and IL17f, have been recognized as the primary cytokines responsible for disease progression31. The ability of myelin basic protein feeding to suppress the development of EAE in mice and rats3, 11, suggests that OT is capable of suppressing the production of IL-17. However, to our knowledge, this has never been directly demonstrated in a model of OT. Here, for the first time, we have documented that feeding soluble antigen (peptide or protein) suppresses the production of the pro-inflammatory cytokine IL-17a. IL-17a production is equally suppressed by both CD4+ and CD8+ T cells, as feeding with either OVA(323–339) or SIINFEKL, respectively, resulted in significantly lower levels of IL-17a following in vitro restimulation. Thus, OT is capable of suppressing both Th1 and Th17 immune responses.
In summary, the data presented here suggest that we can induce the development of CD8+ Tregs by feeding the MHC Class I immunodominant epitope of OVA, SIINFEKL. We have demonstrated that CD8+ T cells are fully capable of mediating tolerance independently of their CD4+ counterparts in a normal immune system. Although low dose OT has been extensively studied, the specific mechanisms involved in its induction remain unclear. A complete understanding of these mechanisms is required for OT to be used effectively as therapy for chronic inflammatory/autoimmune diseases. By first establishing a functional understanding of CD8+ Tregs in OT, we can then identify how the function of these cells intersects with that of CD4+ Tregs, in turn, giving us a more complete picture of the function of regulatory T cells in mucosal tolerance.
C57BL/6 mice, originally obtained through the NCI-Frederick Animal Production Program (NCI, Frederick, MD), and CD8−/− mice, originally obtained from Jackson Laboratories (Bar Harbor, ME), were bred and housed in the SPF barrier facility at The Mount Sinai School of Medicine. The institutional animal care and use committee approved all procedures concerning the use of these mice.
OT was induced by feeding 1 mg of OVA (Sigma, St. Louis, MO), 21.7 µg of OVA(257–264) peptide (SIINFEKL), or 4.01 µg of OVA(323–339) (Peptides International, Louisville, KY or Anaspec, Fremont, CA) in 250 µl of PBS for 5 consecutive days by intragastric intubation with a 20 gauge feeding needle (Fig. 1a). Control mice received PBS alone. SIINFEKL is the MHC class I immunodominant epitope of the OVA protein presented in H2-Kb 33, stimulating CD8+ T cells. Conversely, OVA(323–339) is the MHC class II immunodominant epitope of OVA presented in both I-Ad 34 and I-Ab, stimulating CD4+ T cells. A dose response was performed to obtain the tolerogenic doses for each peptide by feeding mice 0.01–10× the molar equivalent of 1 mg of wOVA, where 21.7µg and 40.1µg represent the molar equivalent dose of SIINFEKL and OVA(323–339), respectively. Purity of the peptides was confirmed by HPLC and mass spec analysis conducted by the manufacturer (Peptides International) (Anaspec). Mice were immunized three days after the final feeding (day 8) in both hindleg footpads with 100 µg of OVA or SIINFEKL emulsified in CFA (50 µg in 50 µl per footpad) (Pierce, Rockford, IL). 12 days after primary immunization, the mice were bled and sacrificed, and dLNs (popliteal and inguinal) and spleens were collected and teased into single cell suspensions. RBCs were lysed with 0.8% NH4Cl, and 2 × 105 cells/well were cultured in 96-well U-bottom plates with 5 µg/ml Concanavalin A (positive control) (Sigma), media alone (negative control), 10 µg/ml OVA, or 100 µg/ml OVA at 37°C. The response of individual mice was assayed in triplicate. Supernatants were collected at 24, 48, and 72h and stored at −20°C.
Fecal pellets were collected, weighed, and stored in 1 ml of PBS containing 0.02% NaN3. Fecal extracts were generated by vortexing for 5 min and centrifuging to remove solid waste. Serum was isolated by allowing blood to clot overnight at 4°C, followed by centrifugation, and stored at −20°C for antibody analysis.
MLN were collected and pooled from antigen fed mice 3h after the final feed, and teased into single cell suspensions. CD8+ T cells were purified using a CD8+ cell magnetic bead isolation kit (Stem Cell Technologies, Vancouver, CA) according to the manufacturer’s instruction. Cells were confirmed to be >95% pure by flow cytometry. 2 × 105 CD8+ cells were transferred IV via retroorbital injection. Separate groups of mice received 1 × 106 CD8+ depleted cells. 24h after transfer, the mice were immunized and tolerance was assessed as described above.
Antigen-fed mice were injected ID with 20 µg of OVA in 20 µl of PBS in the left ear and 20 µl of PBS in the right ear 12 days after feeding (Fig. 1b). Ear thickness was measured under anesthesia at both 24h and 48h, by carefully placing a digital caliper (World Precision Instruments, Sarasota, FL) along the midline of the ear, over the site of injection. The amount of swelling was ascertained by subtracting the thickness of the right ear (vehicle) from the thickness of the left ear (antigen) (left ear thickness-right ear thickness). At 48h, the mice were sacrificed; the ears were removed and fixed in 10% formalin for histological analysis, and dLNs were isolated and restimulated in vitro as described above.
Allergic lung inflammation was induced as previously described35 (Fig. 1c). Briefly, antigen-fed mice, IP sensitized with 10 µg OVA in alum (Reheis, Berkely Heights, NJ) on day 8, were anesthetized and challenged intranasally for five consecutive days (days 22–25) with 100 µg OVA in 50 µl normal saline. 24h after the final IN dose, the mice were bled and sacrificed. Bronchoalveolar lavage (BAL) was performed by washing the lungs twice with 500 µl of 5 mM EDTA in saline (Sigma). The BAL fluids were centrifuged and the numbers of viable cells were determined by trypan blue exclusion. BAL fluids were stored at −20°C, while the cells were mounted on slides for analysis. Lung tissue was fixed in formalin (Fisher Scientific, Pittsburgh, PA) for histological analysis. Formalin-fixed tissues were processed, sectioned, and H&E stained by the Biorepository Cooperative and Histology Service Shared Resource Facility of the Mount Sinai School of Medicine. The presence of inflammatory cells was assessed using standard light microscopy at the MSSM-Microscopy Shared Research Facility, and inflammation manifested as infiltrates in the perivascular and peribronchiolar regions of the lung. Stained slides were coded and blindly examined, and the relative degree of inflammation was graded in a semiquantitative manner by assigning a score of 0–4 based upon the number, location, and size of the inflammatory infiltrates (0, no visible lesion; 1, multifocal inflammation; 2, locally extensive inflammation; 3, diffuse inflammation; and 4, severe diffuse inflammation).
Cytokine and antibody levels were determined by ELISA. For OVA-specific antibody levels, 96-well plates were coated with 10 µg/ml OVA overnight at 4°C. For fecal IgA and total IgE levels, plates were coated with 5 µg/ml goat anti-mouse IgA or goat anti-mouse IgE overnight (Southern Biotech, Birmingham, AL) at 4°C. Plates were washed 3 times and blocked with 10% FBS in PBS for 1h at room temperature. Following another wash step, samples were serially diluted across the plate and incubated for 2h at room temperature. After another wash step, the plates were incubated with optimized dilutions of HRP-linked goat anti-mouse total Ig or goat anti-mouse IgG1, or AP-conjugated goat anti-mouse IgG2a, goat anti-mouse IgA, or goat anti-mouse IgE (Southern Biotech) for 1h at room temperature. The plates were washed again and developed with TMB microwell peroxidase substrate (KPL, Gaithersburg, MD) or p-Nitrophenyl Phosphate (Sigma). Absorbance was measured using a µQuant Microplate Reader and KC4 software (Bio-tek Instruments, Winooski, VT). Absolute levels of fecal IgA were determined by inclusion of a mouse IgA myeloma protein (S107) (Southern Biotech) as a standard on each plate.
IL-2, IL-4, IL-10, IL-13, IFNγ (BD Biosciences, San Jose, CA), IL-17a (eBisociences, San Diego, CA), and TGF-β levels (R&D Systems, Minneapolis, MN) were determined using commercially available ELISA kits according to the manufacturer’s instructions.
1 × 106 cells were incubated with unlabelled SIINFEKL-loaded H2Kb pentamers (Proimmune, Oxford, UK) for 10 min at room temperature. The cells were washed once with PBS containing 0.1% BSA and 0.1% NaN3, and incubated with 1 test of APC-labeled pentamer conjugate, 0.25 µg Pacific Blue-labeled anti-mouse CD3, 0.25 µg FITC-labeled anti-mouse CD8 (eBiosciences), 0.25 µg PE-labeled anti-mouse CD19, 0.25 µg PE-labeled anti-mouse CD49b, and 0.25 µg PerCP-labeled anti-mouse CD4 (BD Biosciences) for 20 min. at 4°C. The cells were washed twice, resuspended in 1.5% paraformaldehyde containing 1% FBS, and analyzed using an LSRII flow cytometer (BD Biosystems) and FlowJo analysis software (Treestar inc., Ashland, OR). 100,000 events were collected in the lymphoid gate for analysis.
One-way ANOVA with a Bonferroni t-test post test was performed using GraphPad Prism version 4.00 for Windows (GraphPad Software, San Diego, CA).
SIINFEKL-feeding induces tolerance to subsequent immunization with SIINFEKL peptide. C57BL/6 mice were fed PBS, SIINFEKL, or wOVA followed by immunization with SIINFEKL in CFA. dLN cells were isolated and restimulated with 0, 10, or 100 µg/ml wOVA for 48, or 72h. IFNγ protein levels in culture supernatants were determined by ELISA at the indicated timepoints; data are reported in pg/ml ± SE. * p<0.05, ** p<0.01, † p<0.001.
SIINFEKL feeding does not inhibit Th2-driven inflammation. C57BL/6 mice were fed PBS (n=9), SIINFEKL (n=10), or wOVA (n=10) followed by IP sensitization with wOVA in alum. Beginning on day 15, mice received 100 µg wOVA in 50 µl saline intranasally for five consecutive days. Non-allergic controls received OVA intranasally in the absence of sensitization (n=9). a, representative H&E-stained lung section from mice fed PBS, SIINFEKL, wOVA, or non-allergic controls (400× magnification). Insets are non size-reduced sections of the same image identifying the presence of eosinophils in the infiltrates; black boxes indicate the source of the insets. b, PAS staining for mucus production in lung sections from mice fed PBS, SIINFEKL, wOVA, or non-allergic controls (200× magnification). Insets (400× magnification) are of the same image indicating positive PAS staining; black boxes indicate the source of the insets.
SIINFEKL-feeding does not induce deletion of SIINFEKL specific cells. Histograms depicting lymphocyte gating (left column), CD3+CD8+ staining (center column), and CD8+Pentamer+ staining in PBS-fed, SIINFEKL-fed, or wOVA-fed mice are representative of those used to generate Table I. Lymphocyte populations were first gated according to expected forward and side scatter. CD8+ T cells were quantified by gating double-positive cells on the CD3 and CD8 staining profiles of total lymphocytes. Pentamer+ cells were quantified by drawing quadrants on CD8 and pentamer+ staining profiles of total lymphocytes. Gating first on CD3+CD8+ cells, followed by gating on CD8+Pentamer+ did not alter the number of pentamer positive cells. Cells were also stained with anti-mouse CD19 and anti-mouse CD49b to eliminate background staining caused by non-specific binding of the pentamer to B cells and NK cells. Inclusion or exclusion of these markers, while greatly influencing background staining did not alter the number of pentamer positive cells or comparisons between feeding groups.
The authors wish to thank M. Cecilia Berin, Ph.D.; Stephanie Dahan, Ph.D.; and Keren Rabinowitz for technical assistance and helpful discussions. This work was supported by National Institutes of Health Grants: AI 044236, AI 066738, DK 072201, and Training Grant DK 07792.
The authors have no conflict of interest to declare.