Crosspresentation by nonhematopoietic and direct presentation by hematopoietic cells induce central tolerance to myelin basic protein

Classic MBP Is Tolerogenic Before Myelination.

Thymocytes in MHC class I-restricted MBP79–87-specific 8.6 TCR transgenic mice on a WT (MBP^+/+) background are deleted at a late stage of maturation, manifested by a loss of mature CD8 single-positive (SP) HSA (CD24)^lo CD3⁺ thymocytes (11). To define the role of classic MBP in this central tolerance, we bred 8.6 mice onto the golli^−/− background. Positive selection is observed in 8.6 MBP^−/− mice, whereas extensive negative selection is seen in both MBP^+/+ and golli^−/− 8.6 mice, confirming that classic MBP is sufficient to mediate tolerance (Fig. 1 A–C). Classic MBP transcripts are detected only at very low levels in the neonatal nervous system, and classic MBP proteins are not detected until day 6 or later, coinciding with the beginning of active myelination (14). Therefore, to determine whether 8.6 T cell tolerance depends on classic MBP derived from myelin, we analyzed 3-day-old mice. Surprisingly, both 3-day-old MBP^+/+ and golli^−/− 8.6 mice exhibited strong negative selection, whereas 3-day-old MBP^−/− 8.6 mice exhibited positive selection (Fig. 1 D–F). Deletion was very efficient in 3-day-old mice as the decrease in the average percentage of CD8 SP HSA^lo thymocytes in 8.6 MBP^+/+ and 8.6 golli^−/− mice compared to 8.6 MBP^−/− mice was 20-fold and 12-fold, respectively, whereas the same comparisons in adult mice showed 6-fold and 9-fold reductions (Fig. 1G). These data indicate that there is sufficient classic MBP to mediate central tolerance in neonates despite the lack of myelin and classic MBP protein expression in the nervous system. In contrast to CD8⁺ T cells, negative selection of CD4⁺ classic MBP121–140-specific T cells proceeds gradually after birth and is not complete until 10 weeks of age, reflecting the dependence of negative selection of these T cells on generation of myelin (5). Thus, classic MBP-mediated deletion of 8.6 thymocytes in 3-day-old mice suggests that the source of classic MBP is not myelin.

BM-Derived Cells Synthesize and Directly Present Classic MBP.

To identify the cell types and sources of MBP that mediate negative selection of MBP79–87-specific thymocytes, we generated a series of BM chimeras whose phenotypes and interpretations are summarized in Table 1. To limit golli and classic-MBP expression to BM-derived cells, BM chimeras were constructed by using MBP^−/− hosts. Control chimeras generated by transplanting 8.6 MBP^−/− BM into MBP^−/− mice and 8.6 MBP^+/+ BM into MBP^+/+ mice recapitulated the positive and negative selection observed in intact 8.6 MBP^−/− and 8.6 MBP^+/+ mice, respectively (Fig. 2 A and B). Deletion of CD8 SP HSA^lo thymocytes when 8.6 MBP^+/+ BM was transplanted into MBP^−/− mice indicates that BM-derived cells can serve as a source of golli and/or classic MBP to mediate central tolerance (Fig. 2C). To determine whether the tolerogenic protein was presented directly by the BM-derived cells that synthesized it, we transplanted 8.6 β₂-microglobulin (β2m)^−/−MBP^+/+ BM into MBP^−/− recipients so that BM-derived cells could provide MBP but not present it because of the lack of MHC class I molecules. No deletion was observed in these chimeras (Fig. 2D), indicating that negative selection does not occur via crosspresentation of BM-derived MBP by TECs. The absence of negative selection in these chimeras also indicates that β2m^−/− BM-derived cells do not acquire β2m from host cells. We could not determine whether golli MBP synthesis by BM-derived cells is sufficient to induce tolerance because there are no mouse models that express golli but not classic MBP. However, negative selection occurred when 8.6 golli^−/− BM was transplanted into MBP^−/− mice, indicating that classic MBP synthesized and presented by BM-derived cells is sufficient to induce efficient central tolerance (Fig. 2E). Indeed, there is no significant difference between the average percentage of CD8 SP HSA^lo thymocytes when 8.6 MBP^+/+ BM is transplanted into MBP^+/+ mice vs. when 8.6 golli^−/− BM is transplanted into MBP^−/− mice (P = 0.9, Fig. 2F).

We attempted to investigate whether BM-derived cells crosspresent MBP in the MHC class I pathway because we previously found that BM-derived cells present exogenous classic MBP in the MHC class II pathway (5). We constructed BM chimeras that restricted expression of MBP to nonBM-derived cells and MHC class I to BM-derived cells. However, 8.6 thymocytes were not positively selected in control BM chimeras in which 8.6 MBP^−/− BM was transplanted into β2m^−/−MBP^−/− mice (data not shown), indicating that, unlike some CD8⁺ T cells (15, 16), MHC class I⁺ hematopoietic cells do not positively select CD8⁺ MBP79–87-specific T cells. Thus, 8.6 thymocytes could be negatively selected via crosspresentation of MBP by BM-derived cells; however, this mechanism is not required for central tolerance because of the highly efficient negative selection mediated solely by BM-derived cell synthesis and direct presentation of classic MBP.

TECs Mediate Deletion via Crosspresentation of Classic MBP and Synthesis of Golli MBP.

To determine whether MBP produced by nonBM-derived cells also induces tolerance, we constructed BM chimeras in which MBP synthesis was restricted to nonBM-derived cells. When 8.6 MBP^−/− BM was transplanted into both MBP^+/+ and golli^−/− recipients, 8.6 thymocytes were negatively selected, demonstrating that classic MBP alone obtained from nonBM-derived cells is sufficient to induce tolerance (Fig. 3 A and B). We then asked whether TEC presentation of MBP produced by nonBM-derived cells was sufficient to induce tolerance by transplanting 8.6 β2m^−/−MBP^−/− BM, which lacks both MBP and MHC class I expression, into MBP^−/−, MBP^+/+, and golli^−/− recipients. Positive selection was observed in the control MBP^−/− recipients, but negative selection occurred in both MBP^+/+ and golli^−/− recipients (Fig. 3 C–E), demonstrating that TECs induce tolerance by using a nonBM-derived source of classic MBP. The extent of deletion was similar in all cases with no significant difference in the average percentage of CD8 SP HSA^lo thymocytes (P > 0.2, Fig. 3F).

The classic MBP presented by TECs could be synthesized by TECs themselves as part of their program of promiscuous gene expression, or it could come from an exogenous source such as myelin. Golli MBP may also contribute to TEC-mediated tolerance as golli transcripts have been detected in TECs (17). To address these questions, we generated BM chimeras in thymectomized recipients that were given thymus grafts to restrict MBP synthesis to nonBM-derived cells in either the thymus or periphery. Development of 8.6 thymocytes was normal in the thymus grafts, as shown by positive selection when 8.6 β2m^−/−MBP^−/− BM was transplanted into thymectomized MBP^−/− mice containing a MBP^−/− thymus graft and by negative selection when the same BM was transplanted into thymectomized MBP^+/+ mice containing a MBP^+/+ thymus graft (Fig. 4 A and B). In BM chimeras in which synthesis of golli and classic MBP was restricted to nonBM-derived cells within the thymus graft, 8.6 thymocytes were deleted, indicating that synthesis and presentation of golli and/or classic MBP by TECs is sufficient to mediate tolerance (Fig. 4C). To determine which tolerogenic MBP family is synthesized by TECs, identical BM chimeras were generated by using golli^−/− thymus grafts such that only classic MBP could be synthesized by TECs. Positive selection was observed in these animals, indicating that TECs cannot synthesize classic MBP in sufficient quantities to mediate deletion of 8.6 thymocytes (Fig. 4D). This result also demonstrates that golli proteins mediate the negative selection seen in MBP^+/+ thymus grafts (Fig. 4C). These functional data demonstrating that TECs do not synthesize tolerogenic classic MBP are consistent with the observation that mTECS synthesize golli but not classic MBP (1, 17). Together with our previous data showing that TECs induce 8.6 thymocyte deletion by presenting classic MBP obtained from a nonBM-derived source (Fig. 3E), these data demonstrate that TECs acquire classic MBP and crosspresent it in the MHC class I pathway.

The fact that TECs do not acquire classic MBP from BM-derived cells (Fig. 2D) indicates that the antigen must be obtained from myelin. Because the thymus is innervated, classic MBP could be acquired from myelin within the thymus. Alternatively, degraded myelin from the periphery could enter the thymus either as a bloodborne antigen or via APCs that acquire the antigen in the periphery. To investigate these possibilities, we transplanted 8.6 β2m^−/−MBP^−/− BM into thymectomized MBP^+/+ mice containing a MBP^−/− thymus graft. Positive selection of 8.6 thymocytes was observed in these BM chimeras, indicating that TECs do not acquire sufficient MBP from the blood to induce tolerance (Fig. 4E). Although the MHC class I-deficient APCs could still acquire exogenous MBP from the blood or periphery in these chimeras, the positive selection observed confirms our earlier result that TECs do not mediate negative selection by using MBP obtained from APCs (Fig. 2D). We then generated identical experimental animals except that the 8.6 MBP^−/− BM was not β2m deficient. Positive selection of 8.6 thymocytes was again observed (Fig. 4F), indicating that central tolerance to MHC class I-restricted MBP-specific T cells is not induced by BM-derived cells presenting MBP acquired from the blood or periphery. Together, these data demonstrate that TECs use two redundant mechanisms to tolerize MHC class I-restricted MBP79–87-specific cells: they synthesize and directly present MBP79–87 derived from golli MBP, and they crosspresent MBP79–87 derived from classic MBP acquired from myelin in situ. This latter mechanism is disrupted in thymus grafts, likely because of the unmyelinated status of the graft obtained from embryonic or neonatal donors, suggesting that the anatomical location and/or amount of myelinated nerve fibers within a nonmanipulated thymus is critical for TECs to acquire classic MBP.

Mice.

C3HeB/FeJ (C3H), C3Fe.SWV-Mbp^shi/J (MBP^−/−), and C3.129P2(B6)-B2m^tm1Unc/J (β2m^−/−) mice were from The Jackson Laboratory. TCR-transgenic (8.6) mice have been described previously (11). Golli^−/− mice were a kind gift from A. Campagnoni (University of California, Los Angeles) and were backcrossed 10 generations onto the C3H background, except for the 6-week-old 8.6 golli^−/− mice in Fig. 1C (gen. 5), the 8.6 golli^−/− BM donors in Fig. 2E (gen. 5), and the golli^−/− recipients in Fig. 3B (gen. 4). All mice were bred and maintained in a specific pathogen-free facility at the University of Washington. All procedures were approved by the Institutional Animal Care and Use Committee at the University of Washington.

Flow Cytometry.

Thymocytes (1–5 × 10⁶) were stained with fluorescently conjugated antibodies specific for CD3, CD4, CD8, HSA (CD24), Vα8, Vβ6, or K^k and analyzed by four-color flow cytometry on a FACScan LSR or six-color flow cytometry on a FACSCanto (BD Biosciences). All antibodies were purchased from BD Biosciences except CD8-PETR (Caltag Laboratories).

BM Chimeras.

BM cells (1 × 10⁷) depleted of T and B cells by Dynal bead separation (Dynal) with anti-Thy1.2-, TCR-, and B220-biotin antibodies (BD Biosciences) were transferred into lethally irradiated mice (1,000 rads on day −1). Recipients were provided neomycin sulfate (2 mg/ml, Sigma) in their drinking water from day −2 to day 14 and were analyzed 6–8 weeks later. For BM chimeras involving MHC class I-deficient BM, NK cell depletion of recipient mice by i.v. administration of antiasialo GM1 antibody (25 μl, Wako Chemicals USA, Inc.) was performed 1 day before BM transplantation (on day −1).

Thymus Transplantation and Thymectomy.

One to two thymic lobes from embryonic day-15–21 fetuses or neonatal day-0–2 mice were inserted under the kidney capsule of 4- to 6-week-old recipient mice under anesthesia (isoflurane) and provided with postoperative analgesia (carprofen, 5 mg/kg) for a minimum of 48 h. Thymus transplantation was immediately followed by thymectomy. Mice were used as recipients to generate BM chimeras 2–4 weeks after surgery.

Statistical Analysis.

All P-values were calculated with a Student's t test.