γδ T Cell–Secreted XCL1 Mediates Anti-CD3–Induced Oral Tolerance

This article is featured in In This Issue, p.2559

The gastrointestinal immune system has the unique capacity to discriminate between potentially dangerous and harmless material, promoting an inflammatory immune response against pathogenic microbes and toxins while inducing tolerance to food Ags and commensal microbes. Dysfunction of this balance can lead to pathologic conditions such as food allergy, autoimmune diseases, and infections. In this context, oral administration of foreign Ag induces local and systemic hyporesponsiveness to a subsequent challenge with the fed Ag, and this phenomenon has been named “oral tolerance” (1).

Multiple mechanisms have been proposed to explain the immune hyporesponsiveness to fed Ags: low doses of orally administered Ag favor active suppression with the generation of regulatory T (Treg) cells, whereas high doses favor clonal anergy/deletion (2). However, induction of Treg cells expressing the transcription factor Foxp3 and the latency-associated peptide (LAP; a membrane-bound TGF-β) stands out as the major player in oral tolerance (3, 4). Although oral tolerance has classically involved oral administration of Ags, we have previously shown that oral administration of anti-CD3 mAb induced tolerance in several animal models of autoimmune and inflammatory diseases, including experimental autoimmune encephalomyelitis (EAE) (4), streptozotocin-induced and NOD autoimmune diabetes (5–7), type 2 diabetes in the Ob/Ob mouse (8), lupus-prone SNF1 mice (9), and atherosclerosis (10). Moreover, oral anti-CD3 has also been tested in a single-blind, randomized, placebo-controlled phase 2a study in subjects with nonalcoholic steatohepatitis and altered glucose metabolism that included subjects with type 2 diabetes. Positive results, including a reduction in liver enzymes and reduced blood levels of glucose and insulin, were found (11). Importantly, oral tolerance induced by anti-CD3 involved Treg cell expansion in both animal models (4, 12) and humans (11), but the mechanism underlying this effect is not known. The fact that the Fc portion of anti-CD3 was not required for oral tolerance induction, as anti-CD3 F(ab′)₂ fragment is active orally and induces Treg cells (13, 14), suggests that the tolerogenic effects of anti-CD3 depends on T cell activation rather than an indirect effect through a putative Fc receptor activation on APCs in the gut. However, because of the indispensable role of dendritic cells (DCs) in promoting Treg cell differentiation (15, 16), tolerogenic DCs are likely to be indirectly involved in anti-CD3–induced oral tolerance.

Generation of Treg cells requires several steps with a critical participation of the innate immune system present in the gut lamina propria (LP) called gut-associated lymphoid tissue. Ag uptake by DCs underlying regular villus epithelium is critical for the development of oral tolerance (17). After sampling food or microbe Ags, tolerogenic DCs migrate to the mesenteric lymph node (mLN), where they induce Treg cells by releasing TGF-β and retinoic acid (RA) (15). Two major subtypes of tolerogenic DCs responsible for oral tolerance induction have been recently characterized. IRF4-dependent migratory DCs, also called conventional DC (cDC) type 2 (cDC2), express CD11c, CD11b, CD103, and the signal-regulatory protein α (Sirpα; also known as CD172a) and are distinguished from the IRF8/BATF3-dependent migratory DCs (named cDC type 1 [cDC1]) that are CD11c⁺, CD11b⁻, CD103⁺ and express the lymphotactin (XCL1) receptor XCR1. Importantly, cDC1 is the most potent tolerogenic subset because of the expression of high levels of TGF-β and the RA-catalyzing enzyme RALDH (18). The primary factor responsible for DC migration to the secondary lymphoid organs such as the mLN is the chemokine receptor CCR7, which binds to the chemokines CCL19 and CCL21 that are highly expressed in these sites (19). Consistent with this, mice deficient for CCR7 failed to induce oral tolerance (20). Importantly, lymphocytes from both intraepithelial lymphocyte (IEL) and LP compartments have been shown to secrete XCL1, which binds to its receptor XCR1 expressed on CD103⁺ DCs from the gut LP, likely resulting in CCR7 upregulation on these DCs and migration to the mLN (21). Once in the mLN, CD11c⁺CD103⁺ DCs present Ags to cognate CD4⁺ T cells and differentiate them into Treg cells (15).

As mentioned above, oral administration of anti-CD3 is known to induce Treg cells, but the mechanism underlying this effect remains elusive. In this study, we show that orally administered anti-CD3 induces a population of γδ T cells in the small intestine LP (SILP) to produce the XCL1 that, in turn, leads to XCR1-expressing tolerogenic cDC1 to upregulate CCR7 and migrate to the mLN to induce Treg cells and tolerance. Importantly, XCL1 was not required for oral tolerance induced by fed myelin oligodendrocyte glycoprotein (MOG) peptide MOG_35–55, indicating that the mechanisms involved in anti-CD3 and fed Ag–induced oral tolerance are different. Accordingly, oral anti-CD3 potentiated oral tolerance induced by MOG_35–55, resulting in less severe EAE, a rodent model for multiple sclerosis (MS). Thus, we provide strong evidence that the mechanism by which anti-CD3 induces oral tolerance relies on the increased migration of tolerogenic DCs from the SILP to the mLN in an XCL1/XCR1-dependent fashion and the subsequent induction of Treg cells. Moreover, anti-CD3 may serve as an adjuvant to enhance oral tolerance to fed Ags for the treatment of autoimmune diseases such as MS.

Male and female 6–10-wk-old mice on a B6 genetic background were used in this study. C57BL/6J wild-type (WT), TCRδ^−/−, zDC-DTR, and BDCA2-DTR mice were purchased from The Jackson Laboratory and housed in a conventional specific pathogen–free facility at the Hale Building for Transformative Medicine according to the animal protocol with the full knowledge and permission of the Standing Committee on Animals at Harvard Medical School and Brigham and Women’s Hospital.

In the experiments for which sorted cells were required, cells from the IEL compartment of C57BL/6 mice were sorted for CD45⁺CD3⁺TCRβ⁺ or CD45⁺CD3⁺TCRγδ⁺ cells using allophycocyanin–anti-CD45 (30-F11, 1:300; BioLegend), Alexa Fluor 700–anti-CD3ε (17A2, 1:300; BioLegend), BV605–anti-TCRβ (H57-597, 1:300; BioLegend), and BV421–anti-TCRγδ (GL3, 1:200; BioLegend). To sort αβ and γδ T cells from the SILP, cells were first enriched using CD45 microbeads (Miltenyi Biotec) and sorted using the fluorescent-labeled Abs described for IEL cell sorting. Dead cells were excluded based on 7-AAD (BD Biosciences) staining. In some experiments, the fixable viability dye Aqua Zombie (1:1000; BioLegend) was used to exclude dead cells.

For intracellular cytokine staining, cells were first stimulated for 4 h with PMA (50 ng ml⁻¹; Sigma-Aldrich), ionomycin (1 μM; Sigma-Aldrich), and a protein-transport inhibitor containing monensin (1 μg ml⁻¹ GolgiStop; BD Biosciences) before detection by staining with Abs. Surface markers were stained for 25 min at 4°C in Mg²⁺- and Ca²⁺-free HBSS with 2% FCS, 0.4% EDTA (0.5 M), and 2.5% HEPES (1 M) and then fixed in Cytoperm/Cytofix (eBioscience) and permeabilized with Perm/Wash Buffer (eBioscience). Flow cytometric acquisition was performed on a Fortessa (BD Biosciences) using DIVA software (BD Biosciences), and data were analyzed with FlowJo software version 10.1 (Tree Star). Intracellular staining Abs used included FITC–anti-Foxp3 (FJK-16s, 1:100; Thermo Fisher Scientific), BV421–anti–IFN-γ (XMG1.2, 1:300; BioLegend), PE–Cy7–anti–IL-17A (eBio17B7, 1:300; eBioscience), and PE–anti–IL-10 (JES5.16E3, 1:100; eBioscience). Other Abs included Alexa Fluor 488–anti-CD45 (30-F11, 1:200; BioLegend), BV605–anti-CD4 (RM4.5, 1:300; BD Biosciences), allophycocyanin–anti-TCRγδ (GL3, 1:100; BioLegend), PerCP–Cy5.5–anti-CD11c (N418, 1:300; BioLegend), BV421–anti-CCR7 (4B12, 1:200; BioLegend), allophycocyanin–anti-XCR1 (ZET, 1:200; BioLegend), PE–Cy7–anti-Sirpα (P84, 1:300; BioLegend), AF700–anti-CD11b (M1/70, 1:300; BioLegend), and PE–anti-CD103 (2E7, 1:200; BioLegend).

Oral tolerance was induced by gavaging mice with 10 μg/200 μl/mouse of anti-CD3 (145-2C11; Biolegend) or isotype control (IC) for 5 d. In another experiment, 250 μg/200 μl/mouse of MOG_35–55 peptide (Genemed Synthesis) was given orally for 5 consecutive d. Control animals received only 200 μl of PBS. Three days after the last feeding, mice were immunized with 100 μg of MOG_35–55 in CFA (BD Difco) in the ventral flanks. In vitro recall responses were measured at day 10 postimmunization. For this, splenocytes were stimulated with 4, 20, and 100 μg/ml MOG_35–55 (Ag-specific stimulation) or 0.01, 0.1, and 1 μg/ml anti-CD3 (Ag-nonspecific stimulation), and proliferation was measured using [³H]thymidine incorporation. Proliferation index was calculated by dividing the cpm of stimulated cells by the average of the cpm of nonstimulated cells from the same group. To investigate the role of XCL1 in oral tolerance, 100 μg and 50 μg/200 μl/mouse of anti-XCL1 mAb (80222; R&D Systems) was injected i.p. 1 d before oral administration of anti-CD3 or MOG_35–55 and on the fourth day of the oral treatments, respectively.

For transient diphtheria toxin (DT) ablation, BDCA2-DTR mice or WT mice reconstituted with zDC-DTR mouse bone marrow were i.p. injected with 20 ng of DT (Sigma-Aldrich) per gram of bodyweight. Mice were euthanized 48 h after DT injection, and mLN and spleen were used for controlling depletion. To maintain DT ablation during oral tolerance induction, mice received three doses of DT, on day −1, day +1, and day +3 of oral tolerance induction, by feeding anti-CD3 or IC for 5 consecutive d, as described above.

EAE was induced by injecting mice with 80 μg of MOG_35–55 peptide (Genemed Synthesis) emulsified in CFA (BD Difco) per mouse s.c. in the flanks, followed by i.p. administration of 150 ng of pertussis toxin (List Biological Laboratories) per mouse on the day of immunization and 48 h later. Clinical signs of EAE were assessed according to the following criteria: 0, no signs of disease; 0.5, partial tail paralysis; 1, tail paralysis or waddling gait; 1.5, partial tail paralysis and waddling gait; 2, tail paralysis and waddling gait; 2.5, partial limb paralysis; 3, paralysis of one limb; 3.5, paralysis of one limb and partial paralysis of another; 4, complete hind-limb paralysis; 4.5, complete hind-limb paralysis and front-limb weakness; 5, moribund.

SILP and IEL αβ and γδ T cells were sorted from naive, IC, or anti-CD3–treated mice 3 d after the last dose, and RNA was extracted with an RNeasy Plus Micro Kit (QIAGEN) and then reverse-transcribed with a high-capacity cDNA reverse transcription kit (Thermo Fisher Scientific) and analyzed by quantitative RT-PCR with a Vii 7 Real-Time PCR system (Applied Biosystems) with the following primer and probe (from Thermo Fisher Scientific; identifier in parentheses): Xcl1 (Mm00434772_m1). The comparative threshold cycle method and the internal control Gapdh (Mm99999915-g1) was used for normalization of Xcl1 target gene.

GraphPad Prism 7.0 was used for statistical analysis (unpaired, two-tailed Student t test or one-way ANOVA, followed by Tukey multiple comparisons). Two-way ANOVA was used for EAE experiments. Differences were considered statistically significant with a p value <0.05.

To investigate the mechanisms involved in oral tolerance induced by anti-CD3, we first treated mice with oral anti-CD3 or IC for 5 consecutive d, and 3 d later, we immunized them with MOG_35–55 emulsified in CFA. Oral tolerance was measured 10 d later by splenocyte proliferation upon MOG_35–55 stimulation (Supplemental Fig. 1A). As expected (4), anti-CD3, but not IC, induced oral tolerance as shown by decreased splenocyte proliferation after in vitro stimulation with 4, 20, and 100 μg/ml MOG_35–55 (Supplemental Fig. 1B, 1C). Consistent with this, anti-CD3–treated mice showed increased frequencies of MOG_35–55-specific CD4⁺ Treg cells (Vβ11⁺) expressing LAP in the spleen (Th3 cells) as well as Foxp3 and LAP in the inguinal lymph node (Supplemental Fig. 1D), which drains the inguinal s.c. region where immunization was performed.

Importantly, flow cytometric analysis of mLN T cells 3 d after the last dose of anti-CD3 or IC showed increased frequencies of Foxp3⁺ Treg cells in mice treated with anti-CD3 but not IC (Fig. 1A). Moreover, these Foxp3⁺ Treg cells expressed more LAP (Fig. 1A), indicating a higher activation status. Furthermore, Foxp3⁺ Treg cells were increased in the SILP 5 d after the last dose of oral anti-CD3 but not IC (Fig. 1B), suggesting that expanded Treg cells may have migrated from the mLN to the SILP, as previously demonstrated during oral tolerance induced by fed Ags (22). However, no difference was observed in LAP-expressing Treg cells (Fig. 1B). Thus, oral administration of anti-CD3 induces Foxp3⁺LAP⁺ Treg cells, which play a pivotal role in the induction of oral tolerance by anti-CD3 (4).

Because DCs are critical for Treg cell induction (15, 16, 18), we investigated whether DCs migrated to the mLN after oral anti-CD3 administration to induce Treg cell expansion (Fig. 1A, 1B), as previously shown for oral tolerance induced by fed Ags (18). We analyzed the two major intestinal migratory DC populations, cDC1 (CD11c⁺CD11b⁻CD103⁺XCR1⁺) and cDC2 (CD11c⁺CD11b⁺CD103⁺Sirpα⁺), in the mLN from mice 3 d after the last dose of anti-CD3 or IC. We found a significant increase in the frequency of cDC1, but not cDC2 in the mLN from mice treated with anti-CD3 but not IC (Fig. 1C), suggesting that oral administration of anti-CD3 may induce intestinal XCR1-expressing DC migration to the mLN and that these cells may play a critical role in the oral tolerance induced by anti-CD3.

To confirm the involvement of DCs in the oral tolerance induced by anti-CD3, we depleted the two major subsets of DCs, the plasmacytoid DCs (pDCs) and cDCs, and tested the ability of anti-CD3 to induce oral tolerance in the EAE mouse model of MS. To specifically deplete pDCs, we used the BDCA2-DTR mice (23) followed by DT injection, and to deplete cDCs, we used lethally irradiated C57BL/6J WT mice repopulated with bone marrow from the zDC-DTR (Zbtb46-DTR) mice (24) followed by DT injection (Fig. 2A). The use of chimera mice in this procedure is critical because DT administration in zDC-DTR mice is fatal within 24–48 h because of a yet-unknown essential group of radioresistant cells (25). We used the zDC-DTR animals instead of other transgenic mice for DC depletion because zDC (Zbtb46, Btbd4) is specifically expressed by cDCs and committed cDC precursors but not by monocytes, pDCs, or other immune cell populations. Accordingly, in contrast to the previously characterized CD11c-DTR mice, for example, non-cDCs, including pDCs, monocytes, macrophages, and NK cells, are spared after DT injection in zDC-DTR mice, indicating a selective ablation of cDCs (24). Of note, DT was injected three times during oral anti-CD3 or IC administration to assure depletion of DCs in the critical period of oral tolerance induction, and a significant depletion of pDCs and cDCs was observed in secondary lymphoid organs (mLN and spleen) of BDCA2-DTR and zDC-DTR mice, respectively (Fig. 2B, 2C). We found that depletion of cDCs but not pDCs completely abrogated the tolerogenic effect of orally administered anti-CD3, as shown by the reversion of the decreased severity of EAE induced by anti-CD3 (Fig. 2D, 2E). This is consistent with our findings that cDC1, which are cDCs, are increased in the mLN from mice treated with anti-CD3 (Fig. 1C). Thus, cDCs play a crucial role in the anti-CD3–induced oral tolerance.

It has been suggested that the chemokine XCL1, upon binding to its receptor XCR1, expressed on intestinal cDC1, induces the upregulation of CCR7, a chemokine receptor critical for DC migration to the mLN (21). Because anti-CD3 must bind to the CD3 molecule, which is solely expressed on T cells, we investigated whether oral administration of anti-CD3 induced XCL1 production by T cells from the small intestine IEL compartment and SILP. To address this question, we fed mice with anti-CD3 for 5 consecutive d and analyzed αβ and γδ T cells from IEL and SILP 3 d after the last dose of anti-CD3 (Fig. 3A). We found that frequency of αβ and γδ T cells did not change after anti-CD3 administration in both IEL and SILP compartments (data not shown). Compared with αβ T cells, IEL and SILP γδ T cells naturally expressed more Xcl1 mRNA, regardless of anti-CD3 administration (Fig. 3A, 3B). However, only SILP γδ T cells upregulated Xcl1 mRNA after oral administration of anti-CD3 (Fig. 3C–F). Thus, these data suggest that XCL1 produced by γδ T cells may be the link between T cells and DCs involved in anti-CD3–induced oral tolerance.

Because we found that γδ but not αβ T cells upregulated Xcl1 mRNA after oral administration of anti-CD3, we investigated whether γδ T cell–deficient (TCRδ^−/−) mice had impaired oral tolerance to anti-CD3. To address this, WT and TCRδ^−/− mice were fed anti-CD3 or IC for 5 consecutive d, immunized with MOG_35–55/CFA 3 d after the last dose of anti-CD3 or IC, and sacrificed 10 d postimmunization for spleen removal and recall assay. A group of WT and TCRδ^−/− mice treated with anti-CD3 or IC were sacrificed before immunization for flow cytometric analyses of DCs from the SILP and mLN (Fig. 4A). We found that anti-CD3 failed to induce oral tolerance in TCRδ^−/− mice, as shown by increased splenocyte proliferation upon MOG_35–55 stimulation (Fig. 4B). Consistent with the fact that TCRδ^−/− mice may have reduced production of XCL1, we observed a trend in cDC1 but not cDC2 accumulation in the SILP of these mice (Fig. 4C). Accordingly, the increased frequency of CCR7-expressing cDC1 in the SILP observed after oral administration of anti-CD3 in WT mice was completely abolished in TCRδ^−/− mice (Fig. 4D), suggesting an impaired migration of these cells to the mLN. In fact, anti-CD3 treatment induced increased frequencies of cDC1 in the mLN from WT mice (Fig. 1C), and these cells expressed higher levels of CCR7, which we did not observe in the mLN from TCRδ^−/− mice (Fig. 4D, 4E), confirming that the migration of tolerogenic SILP cDC1 is impaired in TCRδ^−/− mice. Thus, upon oral administration of anti-CD3 in WT mice, cDC1 migrate from the SILP to the mLN. However, in TCRδ^−/− mice, production of XCL1 is impaired because of the lack of γδ T cells, and XCR1-expressing DCs (cDC1) tend to accumulate in the LP (and consequently not to migrate to the mLN, where Treg cells are induced and oral tolerance is established) because of the lower expression of CCR7, which is likely induced by the interaction between XCL1 and its receptor XCR1.

Thus, the impaired oral tolerance induced by anti-CD3 in TCRδ^−/− mice may be associated with a deficient migration of intestinal cDC1 to the mLN because of an impaired XCL1/XCR1 signaling pathway in DCs from these mice.

To investigate the role of XCL1 in the oral tolerance induced by anti-CD3, WT mice were fed either anti-CD3 or IC for 5 consecutive d with or without anti-XCL1 mAb treatment, immunized with MOG_35–55/CFA 3 d after the last dose of anti-CD3 or IC, and sacrificed 10 d postimmunization for spleen removal and recall assay. A group of WT mice treated with anti-CD3 or IC that received or not anti-XCL1 mAb were sacrificed before immunization for flow cytometric analyses of DCs from the mLN (Fig. 5A). We found that anti-CD3 failed to induce oral tolerance in mice treated with anti-XCL1, as shown by increased splenocyte proliferation upon MOG_35–55 stimulation (Fig. 5B), confirming that XCL1 plays a critical role in the oral tolerance induced by anti-CD3. Importantly, this effect was associated with decreased frequencies of cDC1 but not cDC2 in the mLN of oral anti-CD3–fed mice treated with anti-XCL1 mAb (Fig. 5C). Moreover, mLN cDC1 from these mice showed a reduced expression of CCR7 (Fig. 5D), indicating that XCL1 is indeed crucial for cDC1 migration to the mLN upon oral administration of anti-CD3.

To investigate whether XCL1 also played a role in oral tolerance induced by fed Ags, WT mice were fed MOG_35–55 or PBS for 5 consecutive d with or without anti-XCL1 mAb, immunized with MOG_35–55/CFA 3 d after the last dose of MOG_35–55 or PBS, and sacrificed 10 d postimmunization for spleen removal and proliferation assay (Supplemental Fig. 2A). We found that oral administration of MOG_35–55 induced tolerance as expected, and this effect was not affected by XCL1 blockage (Supplemental Fig. 2B), suggesting that XCL1 is only required for anti-CD3–induced oral tolerance and that a different immune pathway is involved in the oral tolerance induced by Ag feeding.

Because we found that oral tolerance induced by Ag feeding and anti-CD3 involves different pathways, we hypothesized that anti-CD3 may enhance MOG_35–55-induced oral tolerance and thus function as an adjuvant. To address this, WT mice were fed both anti-CD3 and MOG_35–55 for 5 consecutive d, immunized with MOG_35–55/CFA 3 d after the last dose of anti-CD3/MOG, and sacrificed 10 d later for spleen removal and recall assay as well as flow cytometric analyses of Treg cells (Fig. 6A). We found that anti-CD3 enhanced oral tolerance induced by MOG_35–55, as shown by reduced splenocyte proliferation upon MOG_35–55 (specific) and anti-CD3 (unspecific) stimulation (Fig. 6B). Thus, anti-CD3 works as an adjuvant that enhances oral tolerance induced by fed Ags.

To further confirm the role of anti-CD3 as an adjuvant for oral tolerance induced by fed Ags, WT mice were fed both anti-CD3 and MOG-_35–55 for 5 consecutive d and EAE was induced (Fig. 6A). We found that mice treated with both oral anti-CD3 and MOG_35–55 showed less severe EAE (Fig. 6C), which was associated with decreased total CD3⁺ T cell and CD4⁺Vβ11⁺ MOG_35–55-specific T cell infiltration into the spinal cord as compared with the other groups (oral vehicle, anti-CD3, or MOG_35–55 separately) (Fig. 6D). Among the infiltrating cells, frequency of IFN-γ– and IL-17–producing cells was significantly reduced in mice treated with both anti-CD3 and MOG_35–55 (Supplemental Fig. 3). This suggests that the combination of anti-CD3 and a fed Ag prevents effector T cell infiltration into the spinal cord during an inflammatory process. This effect likely occurred because of an expansion of Treg cells in the periphery, as we found that oral administration of both anti-CD3 and MOG_35–55 induced increased Treg cell frequencies in the spleen 10 d after MOG_35–55/CFA immunization and, thus, before EAE onset (Supplemental Fig. 4). Thus, anti-CD3 functions as an adjuvant for the classical fed-Ag oral tolerance induction by decreasing T cell infiltration and the consequent inflammation in the spinal cord.

A major goal of immunotherapy of autoimmune diseases is the induction of Treg cells that mediate immunologic tolerance. The intestinal mucosa is a perfect site for this purpose because of its tolerogenic microenvironment. However, for several autoimmune diseases, the self-antigens recognized and attacked by the immune system are not known, precluding the use of a specific oral therapy to promote Treg cell expansion and tolerance in these cases. An ideal approach would be the oral administration of a compound that induces tolerance independent of the Ag involved in the autoimmune disorder but that does not cause immunosuppression. Based on this idea, we have found that oral administration of anti-CD3 mAb induced oral tolerance through a mechanism dependent on both CD4⁺CD25⁺LAP⁺ and CD4⁺CD25⁻LAP⁺ Treg cells in mLN and spleen (26). Importantly, oral anti-CD3 alleviated disease in several mouse models of autoimmunity and inflammation (4–10, 12), indicating that anti-CD3 is a good candidate for oral tolerance induction to treat autoimmune and inflammatory diseases. In fact, human trials using humanized anti-CD3 have shown promising results, and the mechanisms underlying the tolerogenic effects in people also appear to involve the generation of Treg cells (11). However, how oral administration of anti-CD3 increases Treg cell population is unknown, and the current study aimed to investigate this mechanism.

A critical question that needed to be answered was whether anti-CD3 directly expanded Treg cells by binding to the CD3 complex expressed on these cells or whether it indirectly modulated DCs that, in turn, induced Treg cell differentiation and expansion, as previously shown during oral tolerance induced by oral administration of Ags (18). We showed that mice lacking cDCs but not pDCs did not develop tolerance upon oral administration of anti-CD3, providing strong evidence that the mechanism underlying Treg cell induction by oral administration of anti-CD3 likely relies on its ability to modulate migration of tolerogenic cDCs to the mLN. Consistent with this, we found that both frequencies and numbers of cDC1 but not cDC2 increased in the mLN from mice orally treated with anti-CD3 as soon as 3 d after the last dose of the Ab. This correlated with Treg cell expansion at the same time point in the mLN, with the subsequent increase in the frequency of these cells in the SILP 2 d later. cDC1 migration appeared to be dependent on CCR7 expression because CCR7-expressing cDC1 frequencies were increased in the mLN from anti-CD3–treated mice. This suggested that T cells in the gut, upon activation by anti-CD3, modulated DC migration. Consistent with this, it has been recently shown that DCs from mice deficient in either the chemokine XCL1 or its receptor XCR1 have impaired migration to the mLN because of reduced expression of CCR7 on these cells, suggesting that the XCL1/XCR1 axis controls CCR7 expression on DCs (21).

Interestingly, XCR1-expressing DCs are potent tolerogenic cells likely because of their high expression of TGF-β and RA, which are critical factors for Treg cell induction and migration from the mLN to the gut to promote tolerance (18). Because T cells from the intestinal LP and IEL compartments can secrete XCL1 (21), we hypothesized that oral anti-CD3 induced XCL1 production by intestinal T cells. In fact, γδ T cells from both LP and IEL compartments expressed higher Xcl1 mRNA levels than αβ T cells, and only γδ T cells from the SILP upregulated Xcl1 mRNA after oral administration of anti-CD3. This suggests that the CD3 complex from γδ T cells may be more sensitive to stimulation by low doses of anti-CD3 or express more CD3 ε-chains than αβ T cells (27). However, the reason why anti-CD3 preferentially activates γδ T cells from the SILP is less clear, but it could be related to the fact that anti-CD3 accumulates in the LP upon oral administration (4). Thus, it is possible that anti-CD3 reaches the LP in a way that it does not encounter γδ T cells from the IEL compartment, such as through microfold cells. Furthermore, the mechanisms by which anti-CD3 induces the expression of XCL1 in γδ T cells are not clear. We believe that the CD3 signaling pathway may explain the link between XCL1 production and CD3 engagement. This is because chemokine expression is usually induced by transcription factors such as NF-κB and AP-1 (28), which are both activated upon CD3/TCR engagement and the subsequent signaling pathway. Thus, we hypothesize that upon anti-CD3 binding to the CD3/TCR complex, NF-κB and AP-1 may induce XCL1 gene expression.

The role of γδ T cells in the mechanism of DC migration induced by oral anti-CD3 became clearer when we attempted to induce oral tolerance in TCRδ^−/− mice. These animals did not develop tolerance upon oral anti-CD3 administration, and this effect correlated with decreased CCR7 expression on cDC1 from the SILP and the consequent impaired migration of these cells to the mLN. Furthermore, by blocking XCL1 using an mAb, we recapitulated the findings observed in TCRδ^−/− mice, indicating that XCL1 secreted by γδ T cells is the link connecting DC migration and Treg cell induction by oral administration of anti-CD3.

Strikingly, the XCL1/XCR1 axis was not required for the oral tolerance induced by fed MOG_35–55. Because both anti-CD3– and fed Ag–induced oral tolerance depends on Treg cell induction in the mLN (4, 18), this finding suggests that the mechanism involved in DC migration to the mLN may differ in anti-CD3 versus Ag feeding. In this context, orally administered Ags are sampled by DCs from the SILP, processed, and presented to T cells in an MHC class II–dependent manner after migration to the mLN. In contrast, anti-CD3 may potentiate SILP Ag-loaded DC migration to the mLN by inducing XCL1 production through γδ T cells. Based on this concept, we hypothesized that anti-CD3 could serve as an adjuvant for fed Ag–induced oral tolerance. Consistent with this, we found that oral administration of anti-CD3 enhanced oral tolerance to MOG_35–55, as observed by decreased splenocyte proliferation and reduced EAE severity, an effect associated with decreased effector T cell infiltration into the spinal cord and increased Treg frequencies in the spleen of EAE mice. Thus, anti-CD3 may be an important adjuvant that can potentiate oral tolerance induced by fed Ags to treat autoimmune conditions, particularly when the Ag involved in the autoimmune disease is known.

In summary, we have identified the mechanism by which oral tolerance induced by anti-CD3 promotes regulatory T cell expansion and propose that anti-CD3 may be used in combination with fed Ags to potentiate oral tolerance and the consequent immunoregulation for the treatment of autoimmune diseases.

We thank Deneen Kozoriz for excellent technical support in cell sorting.

The authors have no financial conflicts of interest.