Antagonist Peptide Selects Thymocytes Expressing a Class II Major Histocompatibility Complex–restricted T Cell Receptor into the CD8 Lineage

The C5 TCR transgenic mouse line A18 on the Rag1^−/− background (21) was crossed to invariant chain deficient mice (20) to generate Rag^−/−, Inv^−/− A18 transgenic mice. Genotyping was performed by PCR analysis.

The agonist peptide for the A18 T cell receptor is peptide 106–121 from mouse C5. Two antagonists were generated by replacement of lysine residue 113 with either isoleucine for antagonist 113I (K → I) or with valin for antagonist 113V (K → V). Antagonist 113V was used for the organ cultures.

Dendritic cells derived from cultures of bone marrow cells with GM-CSF as described (22) were pulsed with 10 nM A18 agonist peptide for 2 h. They were then washed and cultured with different doses of antagonist or control peptide in the presence of spleen T cells from A18 Rag1^−/− TCR transgenic mice. After 48 h of culture, supernatants were transferred to fresh wells with 5,000/well IL-2–dependent CTLL cells and proliferation of CTLL was assessed by uptake of [³H]thymidine.

NTOC were cut in half and placed on nucleopore membranes (0.8 μm pore size; Costar Corp., Cambridge, MA) floating on 1 ml Iscove's modified Dulbecco's medium (Life Technologies, Inc., Paisley, Scotland) supplemented with 10% heat-inactivated FCS, 5 × 10⁻⁵ M 2-mercaptoethanol, 2 × 10⁻³ M L-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin, and peptide when indicated. Medium containing peptides was changed daily. After 7 d of organ culture, thymic lobes were mechanically disaggregated into single cell suspensions.

Expression of cell surface antigens by thymocytes from NTOC was determined by cytofluorometry using a FACScan^® and Cell Quest software (Becton Dickinson & Co., Mountain View, CA). Dead cells were excluded by forward and side scatter characteristics. mAbs used were PE-labeled anti–CD4 (H129.19; Boehringer Mannheim, Mannheim, Germany), FITC-labeled anti–CD8α (YTS 169.4; reference 23), biotinylated anti–CD8β (KT112, provided by K. Tomonari, Fukui Medical School, Fukui, Japan), biotinylated anti–heat-stable antigen (HSA) (YBM 5.10; reference 24), or biotinylated anti–Vβ8.3 (25). Biotinylated mAb were detected with streptavidin-Red670 (GIBCO BRL, Gaithersburg, MD).

Thymocytes from NTOC were plated out in 96-well microtiter plates (5 × 10⁴/well) and treated with 10 μg/ml anti–CD4 (RL172.4), 10 μg/ml anti–CD8α (3.168), or no mAbs, before adding rabbit complement (Cedarlane Labs. Ltd., Hornby, Ontario, Canada) for 45 min at 37°C. Cells were washed three times, and then cultured with bone marrow dendritic cells (2 × 10⁴/well) in the presence of 1 μM C5 peptide 106–121 for 72 h. Supernatant was tested for the presence of IL-2 by its ability to support growth of IL-2–dependent CTLL indicator cells.

Exchange of lysine residue 113 in the C5 peptide recognized by the A18 TCR results in loss of stimulatory activity for C5-specific A18 T cells. Two such altered peptides with a change of K → I (peptide 113I) or K → V (peptide 113V) displayed antagonist activity as judged by their ability to inhibit activation of A18 T cells to the agonist A18 peptide. Both peptides are nonstimulatory for A18 T cells in a range of concentrations from 1 nM to 10 μM (data not shown). Fig. 1 shows that both 113I and 113V blocked the IL-2 response of A18 T cells to dendritic cells prepulsed with agonist peptide, whereas an unrelated H2-E^k binding peptide from hen egg lysozyme (HEL 1-18, 2G7; reference 26) did not influence the response.

To decrease presentation of endogenous peptides and enable preferential loading of exogenously added peptides onto MHC class II molecules, we crossed class II-restricted TCR-transgenic A18 mice on the RAG-1^−/− background to invariant chain–deficient (Ii^−/−) mice. MHC class II expression in Ii^−/− mice has been reported to be reduced but not abolished and the absence of invariant chain interferes with CD4 T cell development of many, but not all, specificities (20, 27). Neonatal thymus organ culture of A18 Ii^−/− mice showed no differentiation of mature CD4⁺ cells (Fig. 2 A, medium), placing A18 in the former category. The few CD4⁺ cells that were found in NTOC displayed an immature phenotype (TCR^lo-int and HSA^hi). Thymus cortical epithelial cells, which are responsible for the induction of positive selection of T cells (28), show a twofold reduction in MHC class II levels (data not shown). Since we know that the A18 TCR is positively selected in F1 mice with the nonselecting H2-^q background (Barthlott,T., R.J. Wright, and B. Stockinger, manuscript submitted for publication), we can exclude that this degree of reduction in MHC class II expression is the reason for the lack of mature CD4⁺HSA^lo cells in A18 Ii^−/− lobes. An alternative and more likely explanation for the defect in positive selection of A18 cells is an impairment in the processing of the normally positively selecting peptide or its proper assembly with class II, due to the absence of Ii. The greatly increased peptide binding capacity of Ii^−/− antigen-presenting cells allowed us to test the effect of exogenously added peptides on positive selection of A18 T cells.

Antagonist peptides have been shown to be successful in generating class I-restricted CD8 T cells. We asked whether modified A18 peptides with antagonist activity for the nominal C5 peptide would rescue the development of CD4 T cells in NTOC of invariant chain negative A18 thymocytes. However, thymus organ culture in the presence of antagonist peptide did not result in the generation of mature CD4⁺ cells at any concentration tested. Fig. 2 shows antagonist peptide 113V as an example, antagonist 113I behaved in the same way (data not shown). Instead of expected CD4⁺ cells, CD8⁺ cells with the characteristics of mature T cells (TCR^hi HSA^lo) were generated (Fig. 2,A, 1 μM antagonist and 10 μM antagonist). These cells express CD8αβ heterodimers on the surface (Fig. 3). The increased percentage of CD8⁺ cells was reproducibly reflected in a 5–10-fold increase in absolute cell numbers as shown in Fig. 4. These CD8⁺ cells could not have been derived by endogenous receptor rearrangements leading to expression of a class I-restricted TCR because of the absence of RAG-1 protein in those mice. Therefore, it clearly demonstrated that the antagonist peptide caused T cells with the class II-restricted A18 TCR to differentiate into the CD8 lineage. The antagonist peptide is presented on MHC class II since anti–class II antibodies block the generation of CD8 cells in NTOC (Fig. 3). The antigenic C5 peptide was also tested for its influence on positive selection in NTOC. However, it mostly led to deletion of A18 thymocytes. Very inefficient selection of CD8 and CD4 cells occurred in some but not all experiments with C5 peptide at picomolar concentrations (data not shown).

Thymus lobes from newborn A18 Ii^+/− mice were cultured under the same conditions as described above for A18 Ii^−/− lobes, to test whether the antagonist would influence normal positive selection. As expected, mature CD4⁺ TCR^hi HSA^lo cells, but no mature (HSA^lo) CD8⁺ cells were generated from lobes cultured in medium alone (Fig. 2,B, medium). In contrast, CD4⁺ cell maturation did not take place in the presence of antagonist peptide and, as seen with Ii^−/− lobes, CD8⁺ cell maturation was induced (Fig. 2 B, 1 μM antagonist and 10 μM antagonist). Prevention of CD4 lineage commitment by antagonist peptide confirms data from Spain et al. (18) showing a block in the differentiation of CD4⁺ 2B4 TCR transgenic thymocytes in thymus organ culture in the presence of a cytochrome c antagonist variant. Spain et al. (18) did not report an enhanced differentiation into the CD8 lineage, presumably because their mice were not deficient for RAG and therefore contained significant numbers of CD8⁺ cells in the absence of antagonist peptide.

The CD8⁺ cells generated with antagonist peptide in NTOC from A18 Ii^−/− mice displayed a mature phenotype according to their high expression of the TCR and downregulation of expression as shown in Fig. 2. To investigate the functional ability of these cells, we cultured them with dendritic cells presenting the C5 agonist peptide and assayed for IL-2 production. As expected, given that they contained no mature single positive cells, thymocytes recovered from NTOC from A18 Ii^−/− mice did not react to C5 peptide (Fig. 5, medium). In contrast, lobes cultured in the presence of antagonist peptide gave rise to thymocytes responding to C5 peptide (Fig. 5, 1 μM antagonist and 10 μM antagonist). This response was mediated by the CD8⁺ population, as depletion of CD8⁺ cells before the functional assay ablated C5 reactivity. In contrast, depletion of CD4⁺ cells did not affect C5 reactivity. These results confirm the maturity of the CD8⁺ cells and also their TCR specificity for C5 peptide. C5 peptide/class II complex recognition by the A18 TCR in the absence of CD4 was observed previously in CD4-negative T cell hybrids (data not shown).

We have demonstrated that antagonist peptide can alter lineage decision from CD4⁺ to CD8⁺ cells. Previously, positive selection of CD8⁺ cells was achieved with peptide analogues of the antigenic peptide or low concentrations of the nominal antigen presented by the selecting MHC class I molecule (3–5). In contrast, positive selection of CD4⁺ cells could be induced through coengagement of the TCR/CD3 complex with a variety of thymocyte surface molecules such as CD2, CD4, CD5, CD8, CD24, CD28, CD49d, CD81, or TSA-1 (29–33). In addition, cross-linking of TCRβ or CD3γε with mAb (34), as well as targeting thymocytes to thymic cortical epithelium via anti–TCR/CDR-1 hybrid antibodies (31), resulted in the exclusive generation of CD4⁺ cells even in the absence of MHC molecules. Thus, the signals for CD4 differentiation seemed to be promiscuous in comparison to signals for CD8 differentiation and it was suggested that development into the CD4 lineage follows a “default” pathway (32). However, we show here that CD8⁺ rather than CD4⁺ cells developed, even without the involvement of either CD8 or class I-specific signals. Instead of a default model for either lineage, the involvement of distinct signals seems to be more likely.

In terms of the effect of antagonist peptides, it has been shown that binding of the TCR to MHC molecules occupied by antagonist peptide results in a higher off rate (35, 36). A shorter interaction time between the TCR and its ligand might not allow sufficient time for coreceptor binding and consequently for the recruitment of the tyrosine kinase p56^lck (37). Lack of lck recruitment is presumably more debilitating for CD4 lineage cells since a much larger fraction of CD4 than CD8 molecules is associated with lck (38, 39), implying a more prominent role for lck in CD4⁺ cell development. Therefore, the generation of class II-restricted CD8⁺ cells in NTOC might be the consequence of insufficient lck recruitment in the presence of antagonist peptide (40). In support of this, class II-restricted T cells choose the CD8 pathway in mice lacking the CD4 molecule (41). Recruitment of lck to the TCR complex was first implied as a major player in CD4 lineage decision by Itano et al. (42). By introducing a hybrid protein consisting of the extracellular and transmembrane domain of CD8 and the cytoplasmic part of CD4, they could generate a large number of MHC class I-restricted CD4⁺ T cells in transgenic mice. The only known difference to CD8 transgenic mice was more efficient lck recruitment by the CD4 cytoplasmic domain. Basson et al. (43) directed differentiation into the CD8 lineage through TCR engagement independently of MHC specificity by using CD3ε-specific F(ab′)₂ antibodies. The CD3-F(ab′)₂ reagent was unable to activate mature T cells and instead resembled an antagonist peptide in terms of downstream signaling and inhibitory effect on agonist peptide responses. Based on this, they hypothesized that CD8 cell development is favored by antagonist-like signals, which have limited participation of lck signals. On the other hand, CD4 differentiation would require a stronger lck signal in relation to the TCR signal. Our results are consistent with these models and demonstrate for the first time the ability of a single peptide to convert CD4/CD8 lineage decision.

We thank Rose Zamoyska, Dimitris Kioussis, and David Raulet for critical comments on the manuscript.

HSA

heat-stable antigen

NTOC

neonatal thymic lobes