TGF-β–Induced Myelin Peptide-Specific Regulatory T Cells Mediate Antigen-Specific Suppression of Induction of Experimental Autoimmune Encephalomyelitis Free

The balance between activating and inhibitory signals is crucial for the ability of immune system to effectively eliminate foreign Ag, while maintaining tolerance to self-Ag (1). CD4⁺ T cells play an essential role in this delicate balance by virtue of their ability to differentiate into both effector T cells and regulatory T cells (Tregs). It is widely accepted that Tregs play a critical role in both the establishment and maintenance of self-tolerance. In line with this hypothesis, there are numerous reports indicating that a decline in Treg number and/or function underlies the pathogenesis of many autoimmune diseases (2–4). Manipulating the frequency or suppressive activity of Foxp3⁺ Tregs has been employed to successfully modulate both autoimmune diseases (5–10) and organ transplantation (11, 12) in animal models. Most studies have focused on the utilization of CD4⁺CD25⁺Foxp3⁺ natural Treg cells (nTregs) as a therapeutic strategy in immune-mediated diseases. Although these findings suggest the potential of utilizing nTregs for treatment of human diseases, the therapeutic potential of nTregs may be limited due to the small number of nTregs in the circulation specific for any particular Ag, difficulties in inducing expansion of human Ag-specific nTregs ex vivo, and concern over the nonspecific suppressive nature of polyclonally expanded nTregs (13). As an alternative to the use of polyclonally expanded nTregs in disease therapy, there has been considerable interest in defining conditions capable of inducing the in vitro conversion and expansion of sufficient numbers of TGF-β–induced Tregs (TGF-β–iTregs) specific for the desired target Ags (14–18).

Although concerns still remain as to whether regulation of autoimmune disease by autoantigen-specific iTregs producing regulatory cytokines may lead to bystander suppression after in vivo transfer, recent evidence indicates that iTreg regulation may be Ag-specific. Transfer of myelin basic protein-specific TCR transgenic (Tg) Tregs was shown to prevent the development of experimental autoimmune encephalomyelitis (EAE) in a spontaneous EAE model, whereas nonmyelin basic protein-nTregs confer only limited protection from disease (19). Likewise, TGF-β–iTregs derived from BDC-6.9 TCR Tg T cells prevented diabetes induced by BDC-2.5 TCR Tg T cells in NOD mice, but not in NOD.C6 mice, which do not express the islet Ag recognized by BDC-6.9 TCR Tg T cells (20). Furthermore, B6 (H-2^b) Tregs activated in vitro with semiallogeneic B6D2F1 (H-2^b/d) APCs protect B6D2F1 bone marrow grafts, but fail to protect B6CBAF1 (H-2^b/k) bone marrow grafts (21). However, none of the aforementioned models distinguishes between Ag-specific activation of Tregs and Ag-specific suppression of target responder cells. It has been shown that nTregs activated in the presence of cognate Ag are able to suppress the proliferation of responder T cells with different specificity in vitro (22). This suggests that Ag specificity may hypothetically contribute only to the activation stage, but not the effector/suppressive stage of the in vivo Treg function in the aforementioned settings. Therefore, the Ag-nonspecific effects of Tregs hold the potential risk of introducing pan-suppression in an in vivo setting in which the transferred Tregs may be activated.

In the current study, we took advantage of a recent finding that retinoic acid (RA) can enhance naive CD4⁺ T cell differentiation into TGF-β–iTreg cell phenotype (23). We began by optimizing the in vitro culture conditions in which substantial numbers of proteolipid protein (PLP)_139–151-specific iTregs (139-iTregs) could be generated from naive PLP_139–151-specific CD4⁺CD25⁻ 5B6 TCR Tg T cells. The present data show that the resulting iTregs express an intermediate level of CD62L and a high level of CD103 on the cell surface. Despite this inflammation-seeking phenotype (24), which hypothetically would lead the Tregs to inflamed sites, our data show that in vitro-generated 139-iTregs home to and proliferate in local draining lymph nodes (dLNs) following in vivo transfer into PLP_139–151/CFA-primed recipients, but did not respond to priming with a non–cross-reactive PLP_178–191 peptide, indicating Ag-specific activation and proliferation of 139-iTregs in vivo. Furthermore, our data show that 139-iTregs carry out Ag-specific suppression of in vivo induction of relapsing EAE (R-EAE) by three lines of evidence. First, 139-iTregs are able to suppress R-EAE induced by PLP_139–151, but not PLP_178–191 or a mixture of the two peptides. Second, 139-iTregs effectively inhibit delayed-type hypersensitivity (DTH) responses to PLP_139–151, but not to coprimed peptides. Lastly, in vivo proliferation of PLP_139–151- but not myelin oligodendrocyte glycoprotein (MOG)_35–55-specific CD4⁺ T cells is inhibited by 139-iTregs in SJL/B6 F1 mice primed with a combination of PLP_139–151 and MOG_35–55. Collectively, these data indicate that 139-iTregs, which express an inflammation-seeking phenotype, are able to migrate into and proliferate in dLNs in which they specifically suppress T cell-mediated immune responses to the cognate peptide without inducing pan-suppression of the immune responses to non–cross-reactive peptides in the same dLNs.

PLP_139–151-specific 5B6 TCR Tg mice were obtained from Dr. Vijay Kuchroo (Harvard Medical School, Boston, MA) and crossed to SJL CD90.1⁺ congenic mice. 5B6 CD90.1⁺CD90.2⁺ Tg mice were obtained by crossing 5B6 CD90.1⁺ Tg mice with wild-type (WT) SJL CD90.2⁺ mice. GFP-Foxp3 mice were obtained from Dr. Alexander Rudensky (Memorial Sloan-Kettering Cancer Center, New York, NY) and crossed to the SJL background. SJL and SJL/B6 F1 mice were purchased from The Jackson Laboratory (Bar Harbor, ME). The mice were housed under specific pathogen-free conditions in the Northwestern University Center for Comparative Medicine Barrier Facility (Evanston, IL). Paralyzed animals were provided easier access to food and water. All protocols were approved by Northwestern University Animal Care and Use Committee. PLP_139–151 (HSLGKWLGHPDKF), PLP_178–191 (NTWTTCQSIAFPSK), MOG_35–55 (MEVGWYRSPFSRVVHLYRNGK), and OVA_323–339 (ISQAVHAAHAEINEAGR) were purchased from Genemed Synthesis (San Francisco, CA).

CD4⁺CD25⁻ T cells were negatively isolated from pooled spleen and lymph nodes using the CD4⁺ T cell isolation kit (Miltenyi Biotec, Auburn, CA) and AutoMacs (Miltenyi Biotec), except the addition of anti-CD25–biotin in the Ab mixture. The purity of CD4⁺CD25⁻Foxp3⁻ ranged from 95–99%. Whole splenocytes (irradiated, 3000 rad) from naive SJL mice were used as APCs. A total of 4 × 10⁶ of CD4⁺CD25⁻ T cells and APCs at the T/APC ratio of 1:1 were cultured for 5–7 d in 24-well plates with 5 μM PLP_139–151 (for 139-iTregs) or 1 μg/ml anti-CD3 (for polyclonal-iTregs [poly-iTregs]), 100 nM RA (Sigma-Aldrich, St. Louis, MO), and 20 U/ml IL-2 in complete RPMI 1640. For function test, APCs were removed using Lympholyte M, and iTregs were positively selected by using anti-CD25–biotin plus antibiotin beads (Miltenyi Biotec) and AutoMacs (Miltenyi Biotec). Purity of CD4⁺CD25⁺Foxp3⁺ cells ranged from 70–80%. Alternatively, GFP-iTregs were isolated on a MoFlo (Beckman Coulter, FL), gating on CD4⁺ and GFP⁺ (Foxp3) cells. Sorted cell purity ranged from 95–99%.

Mice were immunized with peptides in adjuvant as previously described (25, 26). Each mouse received 100 μl emulsion in CFA containing 200 μg Mycobacterium tuberculosis H37Ra (Difco, Detroit, MI) and 50 μg PLP_139–151, 100 μg PLP_178–191, or a mixture of 50 μg PLP_139–151 and 100 μg PLP_178–191 s.c. distributed into three spots on the flank. In some experiments, a mixture of 50 μg PLP_139–151 and 200 μg MOG_35–55 or OVA_323–339 was used to prime the mice. Initial disease symptoms were usually observed between 10 and 15 d postimmunization. Mice were monitored for clinical symptoms of EAE daily after disease onset. Mice were scored on a scale of 0–5 as follows: 0, no abnormality; 1, limp tail or hind limb weakness; 2, limp tail and hind limb weakness; 3, partial hind limb paralysis; 4, complete hind limb paralysis; and 5, moribund. The data are plotted as the mean daily clinical score for all animals in a particular experimental group.

DTH responses were measured using a 24-h ear-swelling system as previously described (2). The increase in ear thickness was determined 24 h after ear challenge by injecting 10 μg respective peptide (in 10 μl saline) into the dorsal surface of the ear. Results are expressed in units of 10⁻⁴ inches ± SEM.

Single-cell suspensions were blocked for at least 10–15 min with anti-CD16/32, 10% rat serum, and 10% mouse serum prestaining with a fluorescently tagged Ab-mixture directed against surface markers CD4 (RM4-5), CD25 (PC61), CD62L (MEL-14), CD90.1 (HIS5.1), CD90.2 (53-2.1), GITR (DTA-1), and CD103 (2E7) (BD Pharmingen, Franklin Lakes, NJ or eBioscience, San Diego, CA). Intracellular Foxp3 (FJK-1ba) and/or CTLA-4 (UC10-4B9) were stained using an eBioscience Foxp3 Staining kit and Abs according to the manufacturer’s instruction. Data were acquired on a Canto II cytometer (BD Biosciences, San Jose, CA) and analyzed with FlowJo software (Tree Star, Ashland, OR).

Sorted CD4⁺ responders or 139-iTregs were labeled with 4 μM CFSE (Invitrogen, Carlsbad, CA). A total of 1 × 10⁶ CFSE-labeled cells were injected i.v. into naive WT mice followed by immunization. After the indicated number of days, spleen and/or LNs were isolated and stained for surface markers. CFSE dilution was tested on responders or 139-iTregs identified by surface markers using a Canto II flow cytometer (BD Bioscience). Data were analyzed with FlowJo software (Tree Star).

Sorted CD4⁺ responders were labeled with CFSE. A fixed number of responder T cells was cultured with titrated numbers of CD4⁺CD25⁺ T cells in the presence of PLP_139–151 and irradiated splenocytes as APCs. Cells were cultured for 72 h at 37°C in HL-1 medium (BioWhittaker, Walkersville, MD) supplemented with 50 μM 2-ME, 2 mM l-glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin. Proliferation was determined by the dilution of CFSE. Discrimination of responders and Tregs was based on staining with CD90.1 and CD90.2.

Comparisons of clinical scores, DTH responses, and in vivo proliferation of TCR Tg T cells among the various groups were analyzed by unpaired Student t test.

TGF-β, in combination with TCR stimulation, has been reported to convert Foxp3⁻CD4⁺ T cells into Foxp3⁺CD25⁺ iTregs (14, 17, 27). To examine the ability and efficiency of TGF-β to induce PLP_139–151-specific Tregs, CD4⁺CD25⁻ T cells (>98% purity) were isolated from 5B6 PLP_139–151-specific TCR Tg mice and stimulated with irradiated APCs pulsed with 5–100 μM peptide in the presence of 4 ng/ml TGF-β (data not shown). As expected, this condition induced the expression of Foxp3 and CD25 in these otherwise CD25⁻Foxp3⁻ T cells with a relatively low efficiency (14, 17, 27). The conversion efficiency was slightly increased by adjusting the T/APC ratio and peptide concentration, indicating that TGF-β is able to induce the differentiation of iTregs from naive 5B6 TCR Tg T cells, but that other factors may be needed to increase the efficacy of iTreg activation.

To attempt to enhance the induction of PLP_139–151-specific TGF-β–iTregs (139-iTregs), we tested if the addition of other factors would enhance the efficiency and yield of iTreg conversion from naive T cell precursors. Because RA (23) and IL-2 (28) reportedly increase Treg differentiation, we tested the effect of these factors on the induction of 139-iTregs. As expected, addition of 20 U/ml IL-2 (Fig. 1A) and/or 100 nM RA (Fig. 1B, 1D) significantly enhanced the efficiency of conversion. Notably, a significant difference was observed between the cultures of serum-free media HL-1 and RPMI 1640, which contains 10% FCS by volume (Fig. 1A). The difference could potentially be due to the undefined serum components such as TGF-β. In the presence of irradiated spleen APCs pulsed with 5 μM PLP_139–151, 4 ng/ml TGF-β, 20 U/ml IL-2, and 100 nM RA in RPMI 1640 with 10% FCS, the converted Foxp3⁺ T cells underwent extensive expansion with over seven divisions in Foxp3⁺ T cells compared with less than three to four divisions in a majority of the Foxp3⁻ T cells (Fig. 1C). It is not clear whether the cells are first converted to Foxp3⁺ cells that in turn divide extensively, if Foxp3⁺ T cells are converted from activated T cells with minimum division, or if both mechanisms are involved. Nevertheless, these data show that a net effect of the culture conditions is the expression of Foxp3 by the activated PLP_139–151-specific T cells or the selective growth of Foxp3⁺ T cells over the non-Foxp3⁺ T cells (Fig. 1B, 1D). A total of 5 μM PLP_139–151-pulsed irradiated spleen APC, 4 ng/ml TGF-β, 20 U/ml IL-2, 100 nM RA in RPMI 1640 with 10% FCS, and 5–7 d of incubation were thus routinely used to culture the cells. These conditions routinely led to an iTreg cell conversion efficiency of 50–90% of both naive 5B6 TCR Tg T cells and naive T cells derived from 5B6 TCR Tg mice crossed with GFP-Foxp3 knockin mice (GFP-Foxp3 × 5B6 TCR Tg mice) (Fig. 1D).

The expression of CD25, CD73, GITR, and CTLA-4 are linked to the suppressive capacity of Tregs, and the expression patterns of homing molecules such as CD62L, CCR7, and CD103 (αEβ7 integrin) are linked to different subsets of Tregs (29–33). In particular, CD62L^high Tregs have been shown to migrate to lymphoid tissues and suppress the sensitization of naive T cells, whereas CD62L^lowCD103⁺ cells can efficiently enter inflamed tissues and appear to be superior to their CD103⁻ counterparts in suppressing established immune responses (30, 32, 33). Therefore, CD103 is referred to as a marker of effector/memory Tregs (24, 33). We thus determined the expression of CD25, CD73, GITR, CTLA-4, CD62L, and CD103 on Foxp3⁺ T cells from in vitro-differentiated 139-iTregs and compared the expression pattern with natural Tregs from freshly isolated spleen cells from either 5B6 TCR Tg mice (139-nTreg) or wild-type SJL mice (WT nTreg). As shown in Fig. 2, the Foxp3⁺ 139-iTregs are comparable to the nTregs in the expression of CD25, CD73, GITR, and CTLA-4, but they downregulated CD62L and upregulated CD103. Specifically, the nTregs have either CD62L^high or CD62L⁻ populations, whereas 139-iTregs exhibit one CD62L^int population. Furthermore, ∼30% of nTregs in freshly isolated spleen cells from either Tg mice or WT mice expressed CD103, whereas >90% of Foxp3⁺ 139-iTregs expressed CD103. These findings suggest that these 139-iTregs express an effector/memory Treg cell phenotype and/or may represent populations of cells generated in vivo.

Effective suppression of responder T cells by Tregs in vivo requires the presence of these two cell populations simultaneously in the same microenvironment. To determine the potential of 139-iTregs to suppress induction of R-EAE initiated by CD4⁺ T cell sensitization in the local dLNs, we first tested the ability of 139-iTregs to migrate to and proliferate in the dLNs in response to immunization. CD90.1⁺–139-iTregs were labeled with CFSE and transferred into WT CD90.2⁺ SJL recipients followed by s.c. priming with PLP_139–151, PLP_178–191, or a combination of PLP_139–151 and PLP_178–191 in CFA. On day 3 or 4 and day 7 or 8 postimmunization, spleen, cervical LNs, and dLNs were tested for the proliferation of 139-iTregs (as identified by CD4⁺CD90.1⁺ expression) via CFSE dilution, and a similar pattern of results was seen at both time points. As shown in Fig. 3, 139-iTregs underwent a significant amount of proliferation in the dLNs of WT SJL mice primed with either the cognate peptide PLP_139–151 alone or in combination with PLP_178–191, but not upon priming with PLP_178–191 alone. These results suggest that the CD62L^intCD103⁺ 139-iTregs are able to migrate to dLNs and expand specifically to their cognate peptide.

R-EAE in the SJL mouse is characterized by a relapsing-remitting clinical disease course in which the disease relapses are driven by T cell responses to endogenous myelin epitopes released as a result of the Th1/17-mediated inflammatory response to the initiating myelin peptides (i.e., epitope spreading) (25, 34, 35). Responses to the initiating and spread epitopes can be detected in vivo via DTH responses upon challenge with the cognate peptides (36). The correlation of disease severity and detectable T cell responses indicate that R-EAE is a valuable model to study the pathogenesis of CD4⁺ T cell-mediated autoimmune responses and to test the effect of immunotherapy on autoreactive T cells. Previous studies in our laboratory showed that an adoptive transfer of 2–5 × 10⁶ polyclonal nTregs suppressed active EAE induction (6). The fact that the 139-iTregs are able to proliferate in the dLNs of WT SJL mice primed with cognate peptide PLP_139–151/CFA (Fig. 3) suggests that the 139-iTregs may be more efficient than poly-iTregs in preventing R-EAE induced by cognate peptide PLP_139–151/CFA. To determine whether the 139-iTregs are similarly or more suppressive than poly-iTregs, we generated GFP–139-iTregs and GFP–poly-iTregs from GFP-Foxp3 × 5B6 TCR Tg mice and GFP-Foxp3 knockin mice, respectively. A total of 2 × 10⁵ GFP–139-iTregs or GFP–poly-iTregs were transferred into WT SJL, which were subsequently primed with PLP_139–151/CFA. As shown in Fig. 4A, mice receiving GFP–139-iTregs exhibited significantly decreased clinical scores when compared with saline-treated control mice. In fact, transfer of 10-fold less GFP–139-iTregs also significantly reduced the disease severity (data not shown). In contrast, no significant differences in disease score were observed between the mice receiving the poly-iTregs and saline-treated controls (Fig. 4A). Thus, these data demonstrate that the 139-iTregs can suppress PLP_139–151/CFA-induced R-EAE more efficiently than poly-iTregs.

Tregs activated via their Ag-specific TCR can suppress the responders with different specificities in vitro (22). In the current study, in vitro-generated 139-iTregs have an activated phenotype (24) (i.e., downregulation of CD62L and an upregulation of CD103) (Fig. 2). Therefore, we wished to determine whether the 139-iTregs could suppress immune responses induced by noncognate peptides without further activation in vivo. GFP–139-iTregs were generated from GFP-Foxp3 × 5B6 TCR Tg mice, purified by FACS sorting, and transferred into WT SJL mice that were primed with either PLP_139–151 or PLP_178–191 in CFA. Peptide-specific T cell responses were then assessed by monitoring development of clinical EAE (Fig. 4B) and testing DTH responses following challenge with PLP_139–151 or PLP_178–191 (Fig. 5A). Transfer of GFP–139-iTregs successfully inhibited EAE induction and disease and DTH responses to PLP_139–151 in the mice primed with PLP_139–151, but the GFP–139-iTregs failed to inhibit the disease (Fig. 4B) or DTH responses to PLP_178–191 (Fig. 5A) in the mice primed with PLP_178–191. This finding indicates that in vitro-activated 139-iTregs are not able to suppress immune responses to the irrelevant peptide PLP_178–191 in vivo.

Given that 139-iTregs were activated and proliferated in mice primed with cognate peptide but not PLP_178–191 alone (Fig. 3), the failed suppression of PLP_178–191-induced immune responses in the mice primed with PLP_178–191 alone may simply be ascribed to a lack of specific reactivation/amplification of 139-iTregs in vivo. We thus next assessed whether nonspecific suppression of PLP_178–191-specific responses by 139-iTregs would be observed in SJL mice primed with a mixture of PLP_139–151 and PLP_178–191, in which 139-iTregs have been shown to proliferate to the same extent as in mice primed with PLP_139–151 alone (Fig. 3). The data show that GFP–139-iTregs do not suppress the R-EAE induced by priming with the peptide-mixture (Fig. 4B) and also fail to suppress PLP_178–191-specific DTH responses (Fig. 5B). To verify that this was not a phenomenon unique to the PLP_178–191-specific responses, we examined the ability of GFP–139-iTregs to suppress the DTH responses to MOG_35–55 or OVA_323–339 in the mice primed with a combination of PLP_139–151 and the second respective peptide. Similar to the results obtained in the mice primed with a combination of PLP_139–151 and PLP_178–191, GFP–139-iTregs inhibited PLP_139–151-, but not MOG_35–55- or OVA_323–339-specific DTH responses (Fig. 5C, 5D). These data demonstrate the exquisite specificity of effector phase suppression by 139-iTregs in vivo, as particularly illustrated by their failure to suppress R-EAE induced by the linked PLP_178–191 peptide.

We first demonstrated that 139-iTregs inhibited the activation and proliferation of PLP_139–151-specific effector T cells in the draining LNs of PLP_139–151-primed mice (Supplemental Fig. 1) as determined by inhibition of CFSE dilution (Supplemental Fig. 1A), lesser expression of CD25 (Supplemental Fig. 1B), and suppression of effector cell expansion (Supplemental Fig. 1C). Next, to confirm that the 139-iTregs specifically target PLP_139–151-specific responder T cells without altering the responses of T cells specific for other peptides, we transferred an equal number of naive CFSE-labeled SJL CD90.1⁺5B6 (PLP_139–151-specific) and C57BL/6 CD90.2⁺2D2 (MOG_35–55-specific) responder CD4⁺ T cells either alone or combined with 139-iTregs into (SJLxB6) F₁ recipient mice followed by immunization with a combination of PLP_139–151 and MOG_35–55. Proliferation of the T cells from dLNs was determined on PLP_139–151-specific (CD90.1⁺CFSE⁺) or MOG_35–55-specific (CD90.2⁺CFSE⁺) responders via dilution of CFSE on days 1, 3, and 5 or 6 postimmunization. To detect the PLP_139–151-specific (CD90.1⁺CFSE⁺) and MOG_35–55-specific (CD90.2⁺CFSE⁺) responder T cells in vivo, a mixture containing 10⁶ of each responder cells was injected. As this greatly increases the precursor frequency of the specific T cells, increased numbers (4 × 10⁶) of 139-iTregs were used in these experiments compared with the experiments presented in Figs. 4 and 5. Because the populations of both PLP_139–151-specific and MOG_35–55-specific responders were not distinct on days 5 or 6 (Fig. 6A) due to the extensive division in immunized mice, only the data from day 3 were analyzed for the statistical difference between the immunized mice receiving 139-iTreg versus saline-treated controls (Fig. 6C). On days 3 and 5 postimmunization, proliferation of a small, but significant, number of PLP_139–151-specific responder cells was observed in naive control recipient mice due to the responses of the 5B6 cells to C57B/6 alloantigen (Fig. 6A, 6B, right panels). However, the response of these cells in PLP_139–151-immunized mice was significantly enhanced (Fig. 6A, 6B, left panels). The administration of 139-iTregs in the immunized mice significantly inhibited the proliferation of PLP_139–151-specific responder T cells to the level observed in the naive mice (Fig. 6A, 6B, middle panels; Fig. 6C). In contrast, the immunized mice receiving 139-iTregs exhibited comparable proliferation of MOG_35–55-specific responder T cells to saline-treated controls (Fig. 6), indicating that the 139-iTregs do not affect the proliferation of responder T cells specific for the non–cross-reactive MOG_35–55 peptide.

It has been shown that activated nTregs suppress the proliferation of T cells with different specificities in vitro (22). We next wished to assess whether a nonspecific suppression of MOG_35–55-specific CD4⁺ responder T cells by 139-iTregs could be achieved in in vitro coculture. As in the above in vivo experiments, the proliferation was determined by the CFSE dilution of CFSE-labeled PLP_139–151-specific and MOG_35–55-specific responder CD4⁺ T cells, which were premixed together and stimulated with (SJLxB6) F₁ splenocytes in the presence or absence of both PLP_139–151 and MOG_35–55. PLP_139–151-specific and MOG_35–55-specific responder cells were gated as live CD4⁺CFSE⁺CD90.1⁺ and CD4⁺CFSE⁺CD90.1⁻, respectively (Fig. 7A). As shown in Fig. 7B, addition of 139-iTregs inhibited the proliferation of both PLP_139–151-specific and MOG_35–55-specific responder T cells to the extent of no stimulation control (Fig. 7B). Collectively, these results indicate that 139-iTregs, similar to nTregs reported by others (22), have the capacity to suppress the responder T cells with different specificity in in vitro microculture conditions, but inhibit EAE induction in vivo in an Ag-specific fashion.

Tregs are recognized to play a central role in the maintenance of peripheral self-tolerance. Experimental evidence from preclinical animal models has shown that a transfer of Tregs can restore tolerance to self-Ags (5–7) or alloantigens (11, 21, 37, 38). However, despite great advances in the understanding of the biology of Tregs, translation of the results from animal models to human diseases is hampered by many issues, such as development of techniques for generating sufficient numbers of Ag-specific Tregs for treatment and the risk of pan-suppression induced by Treg cell transfer. The current results demonstrate that in vitro differentiation of TGF-β-iTregs from naive T cell progenitors in the presence of RA provides an efficient way to generate substantial numbers of Tregs specific for the desired Ag and that these TGF-β–iTregs suppress R-EAE in an Ag-specific manner without inducing pan-suppression to coprimed Ags in the same dLNs.

The key observation made in this study is that 139-iTregs generated from naive 5B6 TCR Tg T cells preferentially suppress the immune response to the cognate PLP_139–151 peptide in vivo. In agreement with the findings of other laboratories (8, 10, 16), the 139-iTregs were more efficient in suppressing the induction of R-EAE than polyclonally driven iTregs converted from naive T cells derived from WT SJL mice (Fig. 4A) or polyclonally driven nTregs. For example, 10-fold fewer 139-iTregs (2 × 10⁵) were sufficient to suppress PLP_139–151-induced R-EAE in SJL mice (Fig. 4) as compared with polyclonal nTregs in which a minimum of 2 × 10⁶ cells were required (6). Additionally, the 139-iTregs suppressed R-EAE induced by PLP_139–151, but not PLP_178–191 or a combination of the two peptides (Fig. 4B). Furthermore, the 139-iTregs suppressed DTH responses to PLP_139–151 while permitting development of responses to PLP_178–191 (Fig. 5A, 5B), MOG_35–55 (Fig. 5C), and OVA_323–339 (Fig. 5D) in recipient mice primed with a mixture of PLP_139–151 and the other respective peptide. This indicates that 139-iTregs specifically suppress the immune response to the cognate peptide without altering the response to a coprimed peptide. The failure of 139-iTregs to suppress EAE induced by PLP_178–191 is especially significant, as it indicates that the 139-iTregs do not mediate linked suppression to a non–cross-reactive peptide on the same myelin protein. Finally, the current study demonstrates that the 139-iTregs specifically suppressed the proliferation of PLP_139–151-specific CD4⁺ T cells (Fig. 6, Supplemental Fig. 1), but not MOG_35–55-specific CD4⁺ T cells in SJL/B6 F1 mice that received a mixture of PLP_139–151-specific and MOG_35–55-specific CD4⁺ T cells and were primed with a combination of PLP_139–151 and MOG_35–55 (Fig. 6).

Given the observation that the 139-iTregs, similar to nTregs as reported by other laboratories (22), have the ability to nonspecifically suppress CD4⁺ effector T cells in vitro (Fig. 7), our findings suggest that specific Ag recognition plays a critical role in effector suppression of target T cells by 139-iTregs in vivo. Further, it is apparent that the nonspecific suppressive ability of 139-iTregs in vitro does not reflect the specific regulation carried out by the identical cells in vivo. The nonspecific in vitro suppressive capacity of the 139-iTregs may simply reflect the production and release of regulatory cytokines, such as IL-10 and/or TGF-β (39), which can attain sufficient concentrations in a microculture well to suppress the activation of bystander cells. We continue to examine the mechanisms of suppression operative in vivo.

Interestingly, we found that in vivo-administered 139-iTregs were capable of undergoing significant expansion, particularly in the LNs draining the site of PLP_139–151/CFA priming (Fig. 3). This suggests the possibility that an important component of the ability of Ag-specific iTregs to carry out efficient immune regulation in vivo may lie in their ability to self-amplify in lymphoid sites draining the target organ of an autoimmune disease or a transplant. This would also concentrate their suppressive ability at the site where auto- or alloreactive immune effector cells would be expected to be activated in inflammatory diseases and suggest their main role is to prevent activation of effector T cells. The finding that 139-iTregs were also activated and expanded in vivo in mice primed with a combination of PLP_139–151 and a second peptide PLP_178–191 (Fig. 3), yet failed to suppress immune responses and R-EAE to the coprimed PLP_178–191 peptide (Figs. 4–6), indicates that in vivo-reactivated 139-iTregs do not induce bystander suppression even in the same dLNs. This again suggests, in the context of the current experimental model, that the major in vivo regulatory effect of these cells is in inhibiting priming of autoreactive T cells. Our ongoing work is studying the ability of these cells to suppress effector inflammatory functions of the autoreactive T cells in the CNS.

It is well established that TGF-β synergizes with TCR stimulation to induce Foxp3 expression in CD4⁺Foxp3⁻ T cells (14, 17, 27). Our data are consistent with and add to these previous results. Most studies on TGF-β–iTregs have used complex techniques that require purification of dendritic cells to be used as APCs in the inducing cultures. We find that unfractionated irradiated splenic APCs suffice for the induction of 139-iTregs from naive PLP_139–151-specific CD4⁺CD25⁻Foxp3⁻ T cells under our optimized conditions in which RA, IL-2, TGF-β, and a low dose of peptide are used.

The current work used Ag-specific iTregs converted from naive PLP_139–151-specific Tg T cells, a precursor source not relevant to the potential use of iTregs for therapy of human disease. In this regard, our ongoing work has demonstrated that our in vitro iTreg conversion conditions can induce Foxp3 expression in Ag-sensitized CD4⁺ T cells isolated from myelin peptide/CFA-primed mice. More importantly, the resulting iTregs can suppress DTH responses in vivo and proliferation of CD4⁺ T cells in vitro in an Ag-specific fashion (manuscript in preparation). These findings suggest that it may be possible to generate peptide-specific iTregs from CD4⁺ T cells isolated from peripheral blood or target tissues of patients with autoimmune disease for use in disease therapy.

In summary, this study suggests that it is possible to generate peptide-specific iTregs from naive CD4⁺ T cells. Furthermore, we provide evidence indicating that these in vitro-generated Ag-specific TGF-β/RA-iTregs specifically suppress the T cell response to the cognate peptide without inducing pan-suppression in vivo. Given the complexity of human autoimmune diseases, continued research designed to better understand the underlying mechanisms of the contribution of Ag-specific recognition to in vivo suppression is necessary for efficiently treating different disease entities. To promote the development of safe and efficacious therapeutic protocols employing Tregs, it is necessary to further characterize in vitro generated Ag-specific TGF-β/RA-iTregs with regards to their homing characteristics, stability, long-term specificity, mechanisms of suppression, and maintenance of suppressive capacity in vivo for regulating target effector T cells at various differentiation stages.

We thank Dr. Vijay K. Kuchroo for the gift of the 5B6 TCR Tg mice, Dr. Alexander Rudensky for the gift of GFP-Foxp3 mice, the Northwestern University Flow Cytometry Core Facility for cell sorting, and all of the Miller laboratory members for helpful discussion.

Disclosures The authors have no financial conflicts of interest.