B7H1-Ig Fusion Protein Activates the CD4⁺ IFN-γ Receptor⁺ Type 1 T Regulatory Subset through IFN-γ-Secreting Th1 Cells¹

Regulatory T (Treg)⁴ cells have become a central topic in basic and clinic immunology owing to their role in maintenance of homeostasis and in pathogenesis of clinical disease (1, 2, 3). Two main categories of Treg cells, natural and adaptive, have been proposed (3, 4, 5). As representatives of these, CD4⁺CD25⁺ T and the type 1 Treg (Tr1) cells have received great attention in recent years. Treg cells are believed to be active in autoimmune disease (6, 7), tumorigenesis (8, 9, 10, 11), transplantation tolerance (12, 13), allergic disease (14, 15, 16, 17), infection (18, 19, 20, 21, 22), oral tolerance (23), and the maternal-fetal relationship (24).

The naturally occurring CD4⁺CD25⁺ T cells differentiate in the thymus and emerge into peripheral tissues, where they suppress the activation of self-reactive T cells in a basically cell contact-dependent manner (1, 2, 3, 4, 5). Their development is characterized by the influence of transcription factor Foxp3, also leading to a constitutive expression of Foxp3 as a new surface marker (25, 26). The CD4⁺CD25⁺Foxp3⁺ Treg cells are thus distinguishable from the CD25-inducibly expressed, Ag-activated T cells that possess a CD4⁺CD25⁺Foxp3⁻ phenotype. In contrast, Tr1 cells arise in the periphery upon encountering Ags in the presence of exogenous IL-10. The unique cytokine profile of IL-2⁻IL-4⁻IL-10⁺TGF-β⁺ places Tr1 cells into a subset distinct from those of Th0, Th1, or Th2 cells (27, 28, 29, 30, 31). This is despite the fact that Tr1 cells also produce IFN-γ at levels lower than that produced by Th1 cells. There is still intensive interest in better defining the origins, development, phenotype, and potential clinical application of these cells. An active role for dendritic cells (DCs), for example, has been demonstrated in Tr1 activation (32, 33, 34, 35). G-CSF, it is also reported, can induce Tr1 cells via IL-10 and IFN-α (36, 37). Stimulation of T cells in the presence of immunosuppressant such as vitamin D3 and dexamethasone has a similar Tr1-inductive effect (38, 39). However, little is known about the role of costimulatory molecules and their possible interactions with functional T subsets in Tr1 development.

In the 1980s, thousands of severe burn patients were successfully rescued in China by treatment using intermingled skin grafting (40, 41). The grafted skin was elaborated by creating a mosaic of allogeneic skin sheets imbued with many tiny autoskin pieces. The presence of the autoskin (skin islets) in the transplant could induce strong local tolerance to protect the intermingled skin from rejection, which was helpful in preventing infection. Indeed, we were able to show that the keratinocytes, a type of nonprofessional APC within the autoskin pieces, could be activated to express the B7H1 molecule. This, in turn, activated IL-10-secreting cells to create a local environment of immune low responsiveness (42). Moreover, blocking of B7H1 with a mAb completely abrogated the ability of the autokeratinocytes to induce tolerance.

B7H1 (PD-L1), a member of the B7 costimulatory family capable of costimulating T cells to proliferate and secrete IL-10, was first reported in humans by Dong et al. (43) and Freeman et al. (44). It was further demonstrated that the down-regulatory activity exerted by B7H1-activated T cells was dependent, not only upon IL-10 production, but also upon PD-1, a specific receptor bearing the ITIM. No reports to date, however, deal with the role of B7H1 in induction of Tr1.

Based upon our work on transplantation tolerance as induced by B7H1-expressed keratinocytes in the intermingled skin graft, we undertook to examine the role of B7H1 in induction of Tr1 and associated underlying mechanisms in an APC-absent system. Purified mouse naive CD4⁺ T cells were stimulated with anti-CD3 mAb and mouse B7H1-Ig fusion protein. In this way, the costimulatory molecule B7H1, rather than anti-CD28 Ab, was able to induce Tr1 cells. IFN-γ and IFN-γ-secreting Th1 cells were seen to be critical for the differentiation of Tr1. A Tr1 population expressing IFN-γR was accordingly identified with phenotype CD4⁺Foxp3⁻CD45RB^lowIFN-γR⁺IL-10R⁺. Thus, for the first time, activation of Tr1 cells can be seen to be directly costimulated by B7H1 in the presence of Th1-secreted IFN-γ.

C57BL/6 mice were originally obtained from The Jackson Laboratory and maintained at Sino-British Sippr/Bk Laboratory Animal. Our study was approved by the Scientific Investigation Board of Shanghai Jiaotong University School of Medicine. Mouse naive CD4⁺ T cells were isolated from splenocytes by using nylon wool column (Wako Chemicals) and MACS (Miltenyi Biotec). After checking for phenotype as CD4⁺CD62L⁺ by flow cytometry, the naive T cells were further fractionated using MACS with anti-CD25-coupled magnetic beads to obtain the CD4⁺CD25⁻ T and CD4⁺CD25⁺ T subsets. The purity of isolated T cells and their subsets reached >95%, as determined by flow cytometry.

The naive CD4⁺ T cells were stimulated in vitro with anti-CD3 mAb together with one of the following costimulators: mouse fusion protein B7H1-Ig (R&D Systems) (45), anti-CD28 mAb (BD Pharmingen), or unrelated Ig (including mouse IgG2a; R&D Systems) as an unrelated Ab for negative control. Before the naive T cells were seeded at a concentration of 2 × 10⁶ cells/ml on culture plates, the anti-CD3 mAb (200 ng/ml; BD Pharmingen) was precoated with one of the costimulators on 96-well flat-bottom plates (Costar). Coating proceeded overnight to produce three combinations: anti-CD3 plus B7H1-Ig (5 μg/ml), anti-CD3 plus anti-CD28 (2.5 μg/ml), and anti-CD3 plus Ig-control (5 μg/ml). The activated cells were thus designated T-B7H1, T-CD28, and T-control, respectively.

Lymphoproliferation was assayed by adding [³H]TdR (sp. act., 32 Ci/mM; Shanghai Institute of Atomic Nucleus, Chinese Academy of Sciences) at 1 μCi/well 16 h before termination of culture. Isotope incorporation was determined with a liquid scintillation counter (MicroBeta TriLux). Results are expressed either as mean ± SD of cpm for triplicates/quadruplicates or as relative response (RR). RR (percentage) = (cpm of experimental combination/cpm of control) × 100.

In some cultures, anti-IL-10, anti-TGF-β, and anti-IFN-γ (R&D Systems) or anti-PD-1 (eBioscience) mAbs were added separately at concentration of 20 μg/ml.

To enrich the Tr1 cell population, achieving higher purity and concentration, the naive CD4⁺ T cells were twice stimulated with anti-CD3 plus costimulator. The primary stimulation was conducted in 24-well culture plates (Costar) on which anti-CD3 mAb, B7H1-Ig, or Ig-control was precoated, as described above, and the T cells were added at 2 × 10⁶ cells in 1 ml of culture medium/well. Three days later, 1 ml of mouse rIL-2 (rmIL-2) (20 ng/ml; R&D Systems) was added to allow cell expansion. On day 7, the cells were collected, washed, and restimulated with the same stimuli under the same conditions. The resulting cells were washed to remove residual rmIL-2. Those twice stimulated with anti-CD3 plus B7H1-Ig are referred to as T-(B7H1) to distinguish them from the T-B7H1 that were activated only once, and those twice stimulated with anti-CD3 plus Ig-control are designated as T-(control). As mentioned above, when anti-IL-10R, anti-TGF-β, or anti-IFN-γR (R&D Systems) mAbs were added into the culture on day 0 together with anti-CD3 plus costimulators, the resultant twice-stimulated cells are referred as, for example, T-(B7H1 + anti-IL-10R) or T-(B7H1 + anti-IFN-γR).

To test suppressive capacity, an MLC was established with syngeneic CD4⁺ T cells of C57BL/6 origin as responders and 6000-rad irradiated BALB/c (H-2^d) (Sino-British Sippr/Bk Laboratory Animal) CD3-depleted splenocytes as stimulators. Dosage of both responders and stimulators in MLC was 5 × 10⁵ per well. T-(B7H1) or T-(control) cells were added to the MLC at the same concentration in 200 μl of complete medium in 96-well round-bottom plates (Costar). In some experimental combinations, anti-IL-10 (20 μg/ml) and/or anti-TGF-β (20 μg/ml) mAbs were added to the MLC. Lymphoproliferation was assayed with [³H]TdR by the procedure described above. Percentage of suppression is calculated as 100 − RR.

To induce EAE, female C57BL/6 mice (6–10 wk old) were immunized by s.c. injection in three sites on the flank with a total of 200 μg of antigenic peptide MOG_35-C55 (MEVGWYRSPFSRVVHLYRNGK) (46) (synthesized by GL Biochem) in 200 μl of CFA containing 800 μg of Mycobacterium tuberculosis H37Ra (Difco). On days 0 and 2, each mouse received an additional 200 ng of pertussis toxin (Sigma-Aldrich) in 200 μl of PBS. Evaluation of clinical scores was done using a scale of 0–5 in a blinded manner according to Tompkins et al. (47): 0, no abnormality; 1, limp tail; 2, limp tail and hind limb weakness; 3, hind limb paralysis; 4, hind limb paralysis and forelimb weakness; 5, moribund. Mean daily clinical scores were recorded and compared with age- and sex-matched controls.

For initiation of EAE by adoptive transfer of sensitized T cells, draining lymph nodes (LN) were collected from mice that had been challenged with peptide MOG_35-C55 for 7–11 days. LN cells were cultured (10⁷/ml) in RPMI 1640 complete medium with the peptide (40 μg/ml) and rmIL-2 (20 ng/ml) for 72 h. The MOG_35-C55-sensitized T cells were harvested, washed, and resuspended at 5 × 10⁷/ml in RPMI 1640. On day 0, the cells were injected i.v. into naive C56BL/6 mice at a dose of 5 × 10⁷ cells per injection. The mice were treated with 200 ng of pertussis toxin in 200 μl of PBS on days 0 and 2 (48).

MOG_35–55-sensitized CD4⁺ T cells were isolated using MACS from LN cells of EAE mice. The cells were cocultured for 72 h at 5 × 10⁵/well with increasing numbers of T-(B7H1) or T-(control) cells, together with 4000- rad irradiated syngeneic CD3⁻ splenocytes (5 × 10⁶/well) and the peptide MOG_35–55 (40 μg/ml). The suppressive effects of T-(B7H1) on the second response of sensitized T cells to peptide MOG_35–55 were assessed by determination of lymphoproliferation.

Two protocols were adopted to evaluate the effects of T-(B7H1) on EAE in vivo. First, 5 × 10⁷ of T-(B7H1) or T-(control) cells was transferred i.v. into mouse recipients 3 days before EAE was actively induced with peptide MOG_35–55 plus CFA. Second, EAE was adoptively induced in mice by cotransferring the MOG_35–55-sensitized lymphoblasts and T-(B7H1) or T-(control) cells on day 0. To assess the effects of T-(B7H1) in vivo, clinical scores were recorded and spinal cords were collected and fixed in 10% phosphate-buffered Formalin for histopathological analysis. The paraffin-embedded tissue sections were stained either with H&E or with Luxol fast blue to examine infiltration of mononuclear cells or demyelination, respectively.

Foxp3 and Neuropilin-1 (Nrp1) have been recently identified as specific markers for CD4⁺CD25⁺ Treg (25, 26, 49). It has also been reported that chemokine receptor CCR5 and transcription factor T-bet are predominantly expressed in Th1 cells (50, 51), and that CCR3 and GATA3 are restricted to expression in Th2 cells (51, 52). To quantitatively determine levels of Foxp3, Nrp1, IL-10, IFN-γ, CCR5, CCR3, T-bet, GATA3, IFN-γR, and IL-10R at mRNA level, real-time PCR was performed using LightCycler RNA Master SYBR Green I Kit (Roche Applied Science) with the following primers: Foxp3, sense ATTGGTTTACTCGCATGTTCG and antisense GTCAAGGGCAGGGATTGG; Nrp1, sense TCACATTGGGCGTTATTG and antisense CACTGTAGTTGGCTGAGAAA; IL-10, sense ACCTGGTAGAAGTGATGCC and antisense CACCTTGGTCTTGGAGCT; IFN-γ, sense TGAGACAATGAACGCTACA and antisense TTCCACATCTATGCCACT; CCR5, sense CTGAAGAGCGTGACTGAT and antisense-ACATTATGTTCCCAAAGAC; CCR3, sense TCTCCTGAGATGTCCCAATA and antisense TCACCAACAAAGGCGTAG; IL-10R, sense ACATTCGGAGTGGGTCAA and antisense GAGAAACGCAGGTGTAAAG; and IFN-γR, sense TCCTACATACGAAACATACGG and antisense TCTAACTTGCCAGAAAGATGA.

Samples were run in duplicate, and expression levels were normalized to expression of hypoxanthine phosphoribosyltransferase in each set of samples to calculate the degree of change.

For cytokine analysis, cell cultures were set up as described above and supernatants were collected at indicated times and stored at −20°C. Supernatant concentrations of IL-2, IL-4, IL-10, IL-12, IFN-γ, and TGF-β were determined using cytokine ELISA kits (Bender MedSystems). Results were determined on a MR7000 plate reader (Dynatech Laboratories) at 450 nm and analyzed using BIOLINX software.

Percentages of cells with positive surface markers labeled with fluorescent dye-conjugated mAbs were assayed using flow cytometry. For example, identification of two CD4⁺ populations with or without IL-10R and IFN-γR was accomplished using FITC-conjugated anti-biotin (Sigma-Aldrich), biotin-conjugated anti-IFN-γR (BD Pharmingen), and PE-conjugated anti-IL-10R (BD Pharmingen) mAbs. Analysis was performed on FACSCalibur (BD Biosciences) using CellQuest software.

A Student’s paired t test was used to determine significant differences (p < 0.05). Error bars in figures represent the SD of triplicates or quadruplicates for cell cultures.

In our previous work, IFN-γ-induced B7H1⁺ keratinocytes and B7H1-transfected mouse endothelial cells were successfully used to induce IL-10-secreting T cells (42). To exclude the possibility of interference from accessory cells, highly purified mouse naive CD4⁺CD62L⁺ T cells were used in this study and were stimulated with anti-CD3 mAb plus fusion protein B7H1-Ig. Anti-CD28 mAb and unrelated Ig-control were used as positive and negative controls in the presence of anti-CD3 mAb. Fig. 1 A shows two distinguishable reaction patterns of lymphoproliferation during a 6-day period. The early phase (day 1–3) was characterized by a rapid synchronous proliferation of the CD4⁺ T cells for all three kinds of costimulatory molecule, especially for B7H1-Ig and anti-CD28. During the later phase (day 4–6), however, reaction patterns diversified. After costimulation with anti-CD28, for example, the intensity of lymphoproliferation was maintained. In contrast, when costimulated with B7H1-Ig, the [³H]TdR uptake for T cell proliferation dropped sharply on day 4 and continued to decrease. This indicated that the B7H1-Ig-costimulated CD4⁺ T cells lost their proliferative capacity during this phase.

To confirm the differential effects of the costimulatory B7H1-Ig in the two phases, we examined the dose-effect curves on days 3 and 6, respectively. As indicated in Fig. 1, B and C, the enhancement and depression patterns of lymphoproliferation reversed, while the response levels in the control (treated with Ig-control) remained basically unchanged. The cpm value for lymphoproliferation on day 6, for example, dropped to 75% suppression when the dose of B7H1-Ig was increased to 10 μg/ml. These results suggest that some CD4⁺ T cell-derived components (cells and/or soluble factors) with inhibitory activity began to emerge by day 4 of lymphoproliferation and reached their maximum levels by the sixth day.

To examine the cytokine profiles, three kinds of costimulators were tested separately. When CD4⁺ T cells were treated with anti-CD3 plus anti-CD28, there ensued an enhanced secretion of typical Th1-related cytokines IFN-γ and IL-2 (Fig. 2, A and B) rather than IL-10, TGF-β, and IL-4 (Fig. 2, C–E). In contrast, when anti-CD28 was replaced by B7H1-Ig, the secretion levels of IL-10 and TGF-β were dramatically increased during the later phase (day 4–6), rapidly reaching high levels (Fig. 2, C and D). It is of interest to note that B7H1-Ig was also capable of stimulating some production of IFN-γ on days 2 and 3 (Fig. 2,B). On day 4, its concentration decreased to a moderate level, although the secretion of IL-2 never reached a significant amount (Fig. 2,A). Fig. 2 E indicates that T cells costimulated either with anti-CD28 or with B7H1-Ig produced little detectable IL-4, suggesting no involvement of a Th2 subset in the response. The high levels of IL-10 and TGF-β that were detectable only for the B7H1-Ig-costimulated CD4⁺ T cells strongly implicate another T cell subset, Tr1, as being activated in the later phase.

The mAb-blocking experiment further indicated that the suppression induced by B7H1-Ig-costimulated CD4⁺ T cells could be partially diminished by adding mAbs against IL-10, TGF-β, or PD-1, respectively (Fig. 3,A). Among them, anti-IL-10 had the strongest blocking activity. PD-1 is an ITIM-bearing inhibitory receptor capable of inducing suppression or apoptosis when ligated with the B7H1 molecule (44, 53). Some reverse effects of suppression were also observed when anti-PD-1 mAb was used (p < 0.05, Student’s t test). The combination of these three mAbs yielded the strongest recovery from B7H1-Ig-induced suppression, suggesting that, besides cytokines IL-10 and TGF-β, the effective function of PD-1 might also contribute to the hyporeaction during the later phase. This could be further verified by the recovery pattern that did not reach its maximum when only anti-IL-10 and anti-TGF-β were combined together (Fig. 3 A).

Quite interestingly, anti-IFN-γ mAb could not diminish the B7H1-Ig-induced suppression. Instead, the intensity of lymphoproliferation was weakened somehow (Fig. 3,A), suggesting that the IFN-γ might exert its role on Tr1 differentiation instead as an effective factor for B7H1-Ig-induced suppression. This stimulated our interests for further exploration. When IL-10 was taken into account as effective molecule, for example, it was revealed that the production of IL-10 by the B7H1-Ig-costimulated T cells was increased and much earlier when exogenous IFN-γ was added into culture at day 0 (Fig. 3,B). Both anti-IFN-γ and anti-IFN-γR mAbs were able to suppress IL-10 production dramatically. There was no effect detected, however, in the presence of exogenous IFN-γ in the Ig-control-costimulated CD4⁺ T cells (Fig. 3 B).

When CD4⁺ T cells were isolated from IFN-γ gene knockout mice (IFN-γ^−/−) (originally from The Jackson Laboratory) and assayed by intracellular staining for IL-10-producing cells (IL-10⁺ T) after treatment with anti-CD3 plus B7H1-Ig, the percentage of IL-10⁺ T cells was reduced by 84.7% (i.e., from 17.0% for wild-type mice to 2.6% for IFN-γ^−/− mice). Adding back IFN-γ could help restore the percentage of IL-10⁺ T cells to 20% (Fig. 3 C).

Several important results were observed when cell surface markers and the Th1/Th2-specific transcription factors T-bet/Gata3 were examined using real-time PCR. 1) As expected, for anti-CD28 costimulation, the expression of Th1-related CCR5 molecule and T-bet was greatly increased in CD4⁺ T cells from day 2 to day 6 (Fig. 4, A and B). In contrast, costimulation with B7H1-Ig induced the expression of CCR5 and T-bet at only a moderate level during the early phase. It is notable that expression was rapidly depressed during the later phase. 2) Compared with the negative control (costimulated with Ig-control), expression of IL-10R at very high levels could be seen during the later phase when CD4⁺ T cells were costimulated with B7H1-Ig instead of with anti-CD28 (Fig. 4,C). 3) Both CD28- and B7H1-Ig-costimulated T cells were able to express more IFN-γR molecule as compared with baseline levels (Fig. 4,D). 4) No CCR3⁺Gata3⁺ T cells were detectable (Fig. 4, C and D) for the three costimulators evaluated, confirming again that a Th2 subset was not involved in the Tr1-related suppression.

These results, together with the proliferation patterns and the cytokine profiles depicted in Figs. 2 and 3, strongly suggest that, for costimulation with B7H1-Ig, a Th1 (or a Th1-like) subset was activated during the early phase, followed by differentiation into an IL-10-secreting T subset with suppressive properties.

To confirm this, IL-10R and IFN-γR were invoked as two parameters with which to define the two populations by flow cytometry in the 6-day course of B7H1-Ig-costimulated lymphoproliferation. As indicated in Fig. 5, the first population emerging on day 2–4 was phenotypically IFN-γ⁺IL-10R⁻. The other population, characterized as IFN-γ⁺IL-10R⁺, appeared later and began to be emerged on day 3 (Fig. 5,E). Its proportion could gradually reach 93.4% on day 6 from 11.6% on day 3 and 76.1% on day 5 (Fig. 5,G), accompanied by a decrease in the first population from 72.8% on day 3 to 2.7% on day 6. In contrast, when anti-CD28 was used as costimulator, the IFN-γ⁺IL-10R⁻ T population kept expanding on a very high level (>90%) from day 1 to day 5 with no activation and expansion of IFN-γ⁺IL-10R⁺ T cells (Fig. 5, J–N).

As indicated above, most of the CD4⁺ T cells stimulated with anti-CD3 plus B7H1-Ig developed into a subset with down-regulatory ability by day 6. To further purify and enrich the subset, the CD4⁺ T cells primarily activated for 3 days with anti-CD3 plus B7H1-Ig were expanded with rmIL-2 for another 4 days and then restimulated with anti-CD3 plus B7H1-Ig for an additional 6 days. The resultant subset (cultured for 14 days total) is designated as T-(B7H1) to distinguish it from T-B7H1 costimulated once with B7H1-Ig for 6 days. As indicated in Fig. 5,H, the IL-10R⁺IFN-γR⁺ T-(B7H1) could reach a purity as high as 98.5%. In contrast, the histogram patterns of the T-(control) remained unchanged either for single (6 days) or double (14 days) treatment (Fig. 5, O and P) in comparison with the T cells receiving no treatment (Fig. 5, A and I). Proliferation assays (Fig. 6,A) verified that IL-10 secretion by T-(B7H1) was much greater and peaked earlier than that of T-B7H1. Functionally, T-(B7H1) was able to exert strong suppression on MLC in which syngeneic CD4⁺ T cells were set to respond to allogeneic splenocytes, although the T-(B7H1) themselves were not able to proliferate to the allogeneic stimulation (Fig. 6 A).

In contrast to the Th1-related cytokine profile seen in the control group (Fig. 6,C), the suppression of T-(B7H1) cells on the MLC was correlated to their high levels of IL-10 and TGF-β secretion (Fig. 6,B). It is also notable that production of IL-4 remained at very low levels in experiments using either T-(B7H1) or T-(control) (Fig. 6, B and C).

To explore the possible connection with CD4⁺CD25⁺ T cells, CD4⁺ T cells were fractionated into CD25-positive and CD25-negative subsets and two CD4⁺CD25⁺ Treg-specific markers, Foxp3 and Nrp1, were assayed using real-time PCR. As indicated in Fig. 7,A, the markers were not detected in CD4⁺CD25⁻ T cells either before or after treatment with anti-CD3 plus qB7H1-Ig. For total T cells, the relative expressions of the two markers remained unchanged during costimulation with B7H1-Ig. This implies that the B7H1-Ig-costimulated T-(B7H1) cells might be irrelevant to the lineage of CD4⁺CD25⁺Foxp3⁺ Treg cells. Functional analysis further supported this point because both T-(B7H1) and CD4⁺CD25⁻ T-(B7H1) had similar capabilities for inducing strong suppression on MLR (Fig. 7,C). However, when CD45RB on T cells was examined by flow cytometry before and after treatment with anti-CD3 plus B7H1-Ig, its mean expression intensity (indicated as Y-mean) shifted from 1236 to 462, or from CD45RB^high to CD45RB^low (Fig. 7 B). CD45RB^low has been reported as a marker for Treg cells in mice (54, 55).

These results, together with the functional analysis and cytokine profile examination, indicate that the T-(B7H1) belongs to CD4⁺Foxp3⁻CD25⁻CD45RB^low Tr1, which induces immune suppression via secretion of IL-10 and TGF-β.

To reveal the role of IFN-γR in Tr1-related suppression, anti-IFN-γR and anti-IL-10R mAbs were added during the induction of T-(B7H1). Two distinct subsets emerged, designated T-(B7H1 + αIFN-γR) and T-(B7H1 + αIL-10R). In addition, the control subset T-(B7H1 + Ig-control) is noted. The results presented in Fig. 7,D indicate that either T-(B7H1) or T-(B7H1 + Ig-control) could induce suppression of MLR, as expected. However, the suppression was dramatically diminished when T-(B7H1 + Ig-control) was replaced with T-(B7H1 + αIFN-γR) or with T-(B7H1 + αIL-10R), strongly suggesting that these two subsets were unable to induce suppression, or that they no longer belonged to Tr1. The effects of perturbing Tr1 differentiation using anti-IFN-γR and anti-IL-10R mAbs are similar to those seen with the T-(B7H1) that were originally isolated from IFN-γ^−/− mice in our MLR testing system (Fig. 7 C).

These results demonstrate that the ligation of IFN-γR with IFN-γ on the surface of the T-(B7H1) or T-(B7H1) precursors is essential for differentiation and activation of Tr1. The Th1 subset, from which the original IFN-γ was produced, could function as a kind of inducer for Tr1 differentiation.

To further confirm that the T-(B7H1) is functionally equal to Tr1, a mouse model with EAE was used in immunization of C57BL/6 mice with autologous Ag-derived peptide MOG_35-C55 in adjuvant. As seen in Fig. 8,A, coculture of the MOG_35–55-specific T cells in fixed number with increasing numbers of T-(B7H1) cells inhibited the T cell proliferation in a dose-dependent manner. In vivo experiments further demonstrated that when the T-(B7H1) cells were adoptively transferred into naive recipients 3 days before EAE was actively induced, T-(B7H1) conferred significant protection from the development of clinical EAE (Fig. 8,B) in comparison with mice receiving either no cells or T-(control) cells. In another adoptive transfer experiment, T-(B7H1) cells were coinjected with previously activated MOG_35–55-specific T lymphoblasts, resulting in 53% reduction of clinical symptoms in comparison with mice receiving T-(control) cells plus T lymphoblasts (Fig. 8 C). Thus, supplementation with T-(B7H1) conferred protection against progression of both actively and passively induced EAE.

When spinal cord tissue was fixed and stained with H&E to examine mononuclear cell infiltration characteristic of EAE disease progression in CNS inflammation, infiltration was substantially reduced in the T-(B7H1)-treated recipients (Fig. 8,D). This correlated well with observed changes in disease severity. When fixed spinal cord tissue was stained with Luxol fast blue, demyelination was seen to be substantially reduced in the T-(B7H1)-treated group (Fig. 8 E).

DCs, especially immature or semimature DCs, have been reported to be capable of inducing Tr1 rather than CD4⁺CD25⁺ Treg in allergic diseases (32, 33, 34, 35). It is thus believed that the activation status of the DCs and IL-10 secretion by innate cells may be the major influences on the induction of Tr1 cells (4). In addition to the roles of DCs and exogenous IL-10, treatment with IFN-α or G-CSF appeared to be effective in inducing Tr1 cells (36, 37). Stimulation of T cells in the presence of vitamin D3 and dexamethasone had a similar Tr1-inducing effect, which depended on induction of autocrine IL-10 (38). Interestingly, costimulation via CD2 or with Abs against CD46 also resulted in the generation of Tr1 cells. From these studies, conducted in the absence of APCs, it is suggested that certain soluble factors might be involved in, or even sufficient for, the differentiation of Tr1 cells. In allergen-induced airway hyperreactivity, for example, pulmonary DCs could stimulate the development of Tr1 in a manner similar to costimulation with ICOS ligand (17), a member of the B7 family. Since we reported that B7H1-positive keratinocytes able to induce local tolerance in the intermingled skin-grafting procedure via IL-10-secreting T cells (42), there have been no investigations dealing with the relationship of B7H1 and Tr1 activation. This study demonstrates that B7H1-Ig fusion protein, along with anti-CD3 mAb, can stimulate the generation of Tr1 from CD4⁺ T cells.

In our testing system, naive CD4⁺CD62L⁺ T cells (purity >97%) were stimulated with anti-CD3 mAb and fusion protein B7H1-Ig. This excludes possible contamination with accessory cells or APCs, especially those capable of secreting IL-12 for Th1 differentiation. Additional examination using ELISA indicated that our naive T cells were negative for IL-12 production (data not shown). This strongly implies that some DC-independent pathways for Tr1 differentiation exist.

Because IL-10 is essential for Tr1 activation and development, autocrine mechanism might be used by the cells via their IL-10R. This study not only confirms the role of IL-10 and its receptor, but also reports for the first time a dependence upon IFN-γ for Tr1 differentiation. In our APC-free system, IFN-γ can be produced only from CD4⁺ T cells. It is apparent that the IFN-γ-secreting T cells are typical Th1 cells with a CD4⁺CCR5⁺CxCR3⁻T-bet⁺Gata3⁻ phenotype and a cytokine-secreting profile of IFN-γ⁺IL-2⁺IL-4⁻IL-10⁻ as identified in our experiments (Figs. 2, 4, and 6). As reported, Tr1 cells often develop alongside Th1 cells in several kinds of diseases, including autoimmune diseases, cancer, infectious disease, and allergies (21). It is possible that when Th1 cells exert their effects by production of IFN-γ, the generation of Tr1 cells follows. This creates a feedback mechanism down-regulating the elevated immune responses initiated by Th1 cells.

In contrast, when CD4⁺ T cells were stimulated with anti-CD3 plus anti-CD28, there appeared a typical Th1-related reaction pattern with no evidence for generation of Tr1 (Fig. 1). The critical role of costimulatory molecule B7H1 is therefore more impressive for Tr1 differentiation, confirming our previous observation that B7H1-expressing keratinocytes were active in inducing Tr1 differentiation (42). In this sense, two kinds of receptors binding the B7H1 molecule draw our attention. One is the ITIM-bearing PD-1 receptor (44, 45), which usually delivers an inhibitory signal to stop the activation pathway initiated by the ITAM-bearing receptors on the same cells. In our experiment, PD-1 was detected on T cells, and its suppressive role in the late phase was confirmed by blocking with mAb (Fig. 3,A). The other receptor, however, is unknown, although its importance has been suggested for induction of stimulation or up-regulation instead of suppression (45). It is possible that, in our experiments, B7H1 acted via such a receptor to induce the IFN-γ-secreting Th1 cells in the early phase as elucidated by a hypothetic diagram we proposed (Fig. 9). The anti-CD3/B7H1-Ig-activated Th1, in addition to producing large amounts of IFN-γ, might deliver some special signal(s) for Tr1 differentiation or via B7H1-engaged PD-1. It is evident that this signal could not be produced by the anti-CD28-costimulated Th1 cells (Figs. 1 A and 4). In this manner, the function of PD-1 would not be restricted to delivery inhibitory signals on the late phase, but also to induce the differentiation of Tr1 cells. The underlying mechanisms need further investigation.

Transcription factor Foxp3 has been regarded critical for differentiation of CD4⁺CD25⁺ Treg and used as Treg unique marker to distinguish from effector CD4⁺CD25⁺ T cells (25, 26). In contrast, there are reports that Foxp3 was also detectable for CD4⁺CD25⁻ T cells stimulated with anti-CD3/anti-CD28 (56) or TGF-β (57). However, there are no reports dealing with Foxp3-positive Tr1 cells generated by stimulation with IL-10. This suggests that whether Foxp3 is involved in activation of CD4⁺CD25⁻ Treg might depend on different stimuli and/or inducing systems. Fig. 7 A indicates that, however, the CD4⁺CD25⁻ IFN-γR⁺IL-10R⁺Tr1 cells we identified are Foxp3 negative.

The authors have no financial conflict of interest.