Stage-dependent reactivity of thymocytes to self-peptide–MHC complexes

Abstract

In mice that express a transgene for the 2C T cell antigen-receptor (TCR) and lack a recombinase-activating gene (2C⁺RAG^−/− mice) most of the peripheral T cells are CD8⁺, a few are CD4⁺, and a significant fraction are CD4⁻CD8⁻ [double negative (DN)]. The DN 2C cells, like DN T cells that are abundant in various other αβ TCR-transgenic mice, appear to be derived directly from DN thymocytes that prematurely express the TCR transgene. The DN 2C cells are virtually absent in mice deficient in major histocompatibility complex class II (MHC-II) but more abundant in mice deficient in MHC-I, suggesting that the DN 2C thymocytes are positively selected by self-peptide–MHC-II (pMHC-II) complexes and negatively selected by self-pMHC-I complexes. The pMHC-I complexes, however, positively select CD8⁺ 2C T cells in the same mice. The different effects of thymic pMHC-I on DN and CD8⁺ thymocytes are consistent with the finding that DN 2C thymocytes are more sensitive than more mature CD4⁺CD8⁺ [double positive (DP)] thymocytes to a weak pMHC-I agonist for the 2C TCR. Together with previous evidence that DP thymocytes respond more sensitively than T cells in the periphery to weak pMHC agonists, the findings suggest progressive decreases in responsiveness to self-pMHC-I complexes as thymocytes develop from DN to DP thymocytes and then to mature naïve T cells in the periphery.

In spleen, lymph nodes, and other peripheral lymphoid tissues, T lymphocytes that express an antigen-specific αβ receptor (TCR) also express one of the two mutually exclusive coreceptors, CD4 or CD8. The CD4⁺ T cells respond to peptide–MHC class II complexes (pMHC-II) and the CD8⁺ T cells to peptide–MHC class I complexes (pMHC-I). The preference of CD8 T cells for pMHC-I and CD4 T cells for pMHC-II derives in part from the ability of CD8 to bind to a nonpolymorphic site on MHC-I molecules and CD4 to bind, albeit more weakly, to a nonpolymorphic site on MHC-II molecules (1, 2). Among other effects, binding of the coreceptors stabilizes TCR–pMHC bonds and enhances the apparent affinity of TCR for pMHC (3).

Although the specificity of a TCR for class I or class II pMHC complexes is tightly linked in wild-type mice to the T cell expression of CD4 or CD8, mismatches are often seen in mice that express a TCR transgene. At least two mechanisms contribute to these mismatches (4, 5). In TCR-transgenic mice that express recombination-activating genes (RAG⁺), the absence of allelic exclusion of the TCR α genes can lead to expression in the same T cell of two TCRs, the transgenic αβ TCR and another TCR, consisting of the transgenic TCR β chain and an endogenous TCR α chain. The second TCR likely confers different specificities on the T cell, whose coreceptor lineage commitment may thus not match the specificity of the transgenic TCR. However, this mechanism is excluded in TCR-transgenic mice that are RAG^−/−. TCR–coreceptor mismatches found in TCR-transgenic RAG^−/− mice are particularly significant because they show that a single TCR can recognize both pMHC-I and pMHC-II complexes despite the many differences between peptides associated with class I and class II MHC molecules and the MHC molecules themselves (6). That this level of degeneracy in antigen (pMHC) recognition is not unusual is evident from its having been detected, to varying extents, in 22 of 24 different TCR-transgenic RAG^−/− mouse lines reported in the literature (5).

There is additional evidence for degeneracy in antigen recognition by the 2C TCR (7). The specificity of this TCR, originally described in a cytolytic CD8 T cell clone, was found to be restricted by several class I MHC molecules: initially by L^d, then by K^b and several K^bm variants, and recently by a yet-to-be identified nonclassical class I MHC (8–11). Furthermore, in 2C TCR-transgenic mice on the H-2^b [C57BL/6 (B6)] and RAG^−/− backgrounds (referred to as 2C⁺RAG^−/− mice), this receptor was found on some T cells that express only CD4. This mismatch implied that the 2C TCR might also recognize pMHC-II, and, indeed, the CD4⁺ 2C T cells failed to develop in 2C⁺RAG^−/− mice that were deficient in either class II MHC (I-A^b−/−) or class I MHC [transporters associated with antigen processing (TAP)^−/−], suggesting that pMHC-II as well as pMHC-I are recognized and that both are required for the development of the CD4⁺ 2C T cells (5).

Here, we describe the development and relative abundance of CD4⁻CD8⁻ [double negative (DN)] 2C T cells in peripheral lymphoid tissues of 2C⁺RAG^−/− mice. Consistent with findings in other TCR-transgenic mice (12–14), our evidence indicates that these DN 2C T cells derive directly from DN thymocytes rather than from CD4⁺CD8⁺ [double positive (DP)] or CD4⁺ or CD8⁺ [single positive (SP)] thymocytes that stop expressing CD4 or CD8 coreceptors. The DN 2C cells were virtually absent in 2C⁺RAG^−/− mice that lacked pMHC-II (I-A^b−/−) and more abundant in 2C⁺RAG^−/− mice that were deficient in pMHC-I (TAP^−/−), suggesting that DN 2C thymocytes, which express the 2C TCR prematurely, are positively selected on pMHC-II and negatively selected on pMHC-I. In accord with evidence that they are subject to negative selection, the DN thymocytes responded more sensitively than the DP thymocytes to a weak pMHC-I agonist for the 2C TCR. Together with evidence that DP thymocytes are more responsive than lymph node T cells to weak pMHC agonists (15, 16), the findings imply that as TCR⁺ thymocytes go through successive stages of development, they become progressively less responsive to self-pMHC complexes. The present findings are also notable in showing that some T cells, at least as TCR⁺ DN thymocytes, can respond to naturally occurring (endogenous) self-pMHC complexes in the absence of CD4 and CD8.

Results

Abundance of CD4⁻CD8⁻ 2C T Cells in the Spleen of 2C⁺RAG^−/− Mice.

Although most of the 2C TCR⁺ cells in the spleen and lymph nodes of 2C⁺RAG^−/− mice are CD8⁺, many are CD4⁻CD8⁻ (DN) (average frequency of 24%; Fig. 1A; see also Fig. 3A). In comparison, DN T cells are much less frequent in the spleen of B6 mice and OT1⁺RAG^−/− mice on the B6 background (Fig. 1A). Based on their surface phenotype, the DN 2C T cells appear to be bona fide mature naïve T cells: (i) Like CD8⁺ 2C cells, the DN 2C cells in the spleen are CD62L^hi, CD44^lo, CD25⁻, and CD69⁻ (Fig. 1B). (ii) As expected, the DN 2C cells express a slightly lower level of CD5 than CD8⁺ 2C cells because CD5 levels reflect the strength of TCR interactions with pMHC complexes (17), and the absence of CD8 coreceptor on DN 2C cells doubtless results in weaker TCR–self-pMHC interactions. (iii) The DN 2C cells do not express B220 and are thus unlike the abundant DN T cells in lpr/lpr mice (Fig. 1B) (18). (iv) The DN 2C cells do not express natural killer (NK) cell markers DX5 and NK1.1, excluding the possibility that they are derived from NKT cell lineage (19).

Fig. 1.

Properties of DN 2C T cells in 2C⁺RAG1^−/− mice. (A) Frequency of DN T cells in splenocytes from B6, 2C⁺RAG1^−/−, and OT1⁺RAG1^−/− mice. CD4 versus CD8 staining profiles are shown for TCR⁺ cells. Numbers are percentages of DN cells. (B) Comparison of cell surface phenotype between DN 2C cells (solid line) and CD8⁺ 2C cells (shaded histogram) from spleen of 2C⁺RAG1^−/− mice. Splenocytes were stained for 2C TCR, CD4, CD8 plus CD5, CD25, CD44, CD62L, CD69, B220, DX5, or NK1.1. (C) Response of DN and CD8⁺ 2C cells to the QL9/L^d complex. Splenocytes from 2C⁺RAG1^−/− mice were incubated with medium alone (M), T2-L^d cells, or T2-L^d cells in the presence of 1 μM QL9 peptide for 6 h. Expression of the 2C TCR and CD69 was measured by gating on DN 2C cells (solid line) and CD8⁺ 2C cells (shaded histogram). (D) Comparison of IFN-γ production by DN and CD8⁺ 2C T cells. Splenocytes from 2C⁺RAG^−/− mice were incubated with 1 μM SIY or immobilized anti-CD3 and anti-CD28 antibodies for 24 h. Intracellular IFN-γ expression was measured for DN 2C cells (solid line) and CD8⁺ 2C cells (shaded histogram).

Fig. 3.

Effect of MHC-I and MHC-II deficiency on the development of DN 2C T cells. (A) Percentages of DN and CD8⁺ 2C T cells in spleens from 2C⁺RAG^−/−TAP1^+/+ and 2C⁺RAG^−/−TAP1^−/− mice. Each symbol represents one mouse. Numbers below the x axis are average number of 2C cells in the spleen in each category. ∗, P value of <0.05 between DN 2C cells (TAP⁺ vs. TAP⁻) or between CD8⁺ 2C cells from 2C⁺RAG^−/−IA^b+/+ and 2C⁺RAG^−/−IA^b−/− mice. (B) Percentages of DN 2C cells in peripheral blood lymphocytes from 2C⁺RAG^−/−IA^b+/+, 2C⁺RAG^−/−IA^b+/−, and 2C⁺RAG^−/−IA^b−/− mice.

To investigate DN 2C T cell function, we compared the responses of the DN and CD8⁺ 2C cells to various stimuli, including the QLSPFPFDL (QL9) peptide associated with MHC-I L^d (QL9/L^d) and the SIYRYYGL (SIY) peptide associated with MHC-I K^b (SIY/K^b). Like CD8⁺ 2C cells, the DN 2C cells were activated to express CD69 and down-regulate TCR by QL9/L^d (Fig. 1C), which strongly activates 2C T cells even when CD8 is absent or blocked by an anti-CD8 antibody (20–22). Unlike the CD8⁺ 2C T cells, however, the DN 2C cells were not stimulated with SIY/K^b to express IFN-γ (Fig. 1D), in keeping with much evidence that the response of 2C T cells to the SIY/K^b complex is highly dependent on CD8 (5, 21). As expected, both the CD8⁺ and DN 2C cells expressed IFN-γ in response to anti-CD3 plus anti-CD28 antibodies. Together, these results show that DN 2C cells in the peripheral lymphoid organs of 2C⁺RAG^−/− mice are functional.

Development of CD4⁻CD8⁻ 2C Cells in the Thymus.

DN thymocytes were more abundant in the thymus of 2C⁺RAG^−/− mice than OTI⁺RAG^−/− or wild-type B6 mice (Fig. 2A). To determine whether the DN 2C thymocytes can directly differentiate into mature DN 2C T cells, we analyzed the expression of surface markers associated with thymocyte selection and maturation. The most immature DN thymocytes (CD25⁺) did not express TCR in B6 mice, but in 2C⁺RAG^−/− mice, the majority of CD25⁺ DN thymocytes already expressed high levels of 2C TCR [Fig. 2 B and C and supporting information (SI) Fig. 6] and therefore possessed the necessary basis for positive and negative selection. Consistent with their premature expression of the TCR, the DN thymocytes expressed a much higher level of CD5 in 2C⁺RAG^−/− mice than in B6 mice (Fig. 2C). Similarly, CD25⁺ DN thymocytes of OT1⁺RAG^−/− mice expressed the OT1 TCR, although less uniformly than in 2C⁺RAG^−/− mice and also had a high level of CD5. Furthermore, in 2C⁺RAG^−/− mice ≈13% of the TCR⁺ thymocytes with markers that indicated they have been selected (CD69⁺ or HSA^loCD62L^hi) were DN, whereas the corresponding percentages were far lower in B6 mice (1–3%) and OT1⁺RAG^−/− mice (3–5%) (SI Fig. 7). Thus, the surface features of many DN 2C thymocytes (TCR^hiCD5^hiCD69⁺HSA^loCD62L^hi) indicate that they are mature DN cells, consistent with evidence in other αβ TCR-transgenic mice that many DN thymocytes develop directly into mature thymocytes and serve as precursors of DN T cells in the periphery without passing through the DP stage of thymocyte development (12–14).

Fig. 2.

Development of DN 2C T cells in 2C⁺RAG1^−/− mice. (A) Comparison of CD4 versus CD8 staining profiles of thymocytes from B6, 2C⁺RAG1^−/−, and OTI⁺RAG1^−/− mice. Numbers are percentages of DN cells. (B and C) Thymocytes from B6, 2C⁺RAG1^−/−, and OTI⁺RAG1^−/− mice were stained with an antibody mixture (APC-conjugated anti-CD4, CD8, CD11b, CD11c, and B220) plus antibodies to CD25, CD44, either TCRβ or CD5. (B) CD25 versus CD44 expression profiles for APC-negative cells. (C) TCRβ and CD5 expression for CD25⁺ APC-negative cells.

Positive and Negative Selection of the CD4⁻CD8⁻ 2C Cells.

To determine whether premature TCR expression by DN 2C thymocytes renders them susceptible to selection by self-pMHC complexes, we generated 2C⁺RAG^−/− mice that were deficient in either pMHC-I, by introducing the TAP^−/− mutation, or pMHC-II, by introducing the IA^b−/− mutation (mice on the B6 background are naturally deficient in IE). As shown in Fig. 3A, the proportion of CD8⁺ 2C T cells was greatly diminished in the absence of MHC-I, from an average of ≈70% in TAP^+/+ mice to ≈10% in TAP^−/− mice, in agreement with previous evidence that positive selection of CD8⁺ 2C T cells depends on K^b (23). In contrast, the proportion of the DN 2C cells was greatly increased in the absence of MHC-I, from an average of ≈24% in TAP^+/+ mice to ≈90% in TAP^−/− mice. The absolute numbers of DN 2C cells were also greater in the spleen and thymus of 2C⁺RAG^−/−TAP^−/− than 2C⁺RAG^−/−TAP^+/+ mice (Fig. 3A and Table 1). In addition, the DP thymocytes were far more numerous in 2C⁺RAG^−/−TAP^−/− than in 2C⁺RAG^−/−TAP^+/+ mice (Table 1).

Table 1.

Effect of TAP1 deficiency on thymocytes in 2C⁺RAG^−/− mice

Thymocyte subset	TAP1+/+ cell no.	TAP1−/− cell no.	P values* (TAP+ vs. TAP−)
DN	5.7 ± 1.7	15.2 ± 3.0	0.01
DP	8.2 ± 6.8	136.7 ± 19.4	0.01
CD4 SP	2.6 ± 3.9	0.7 ± 1.2	0.33
CD8 SP	3.3 ± 0.4	2.5 ± 0.6	0.003

Values are average cell numbers (× 10⁶) ± SD for five TAP^+/+ and four TAP^−/− mice.

*Student's ttest, two-tailed.

Conversely, in the MHC-II-deficient mice (2C⁺RAG^−/−IA^b−/−) mice, the DN 2C cells were almost absent (Fig. 3B). To verify that IA^b is required for maturation of DN 2C thymocytes, we generated chimeric mice by injecting bone marrow cells from 2C⁺RAG^−/− mice into lethally irradiated wild-type B6 mice (IA^b+) or B6 IA^b−/− mice. Analysis of the chimeric thymuses 5 weeks later revealed that mature DN thymocytes (HSA^lo) were much more abundant in the IA^b+ than in the IA^b−/− recipients (SI Fig. 8).

Together, the results with MHC-I-deficient and MHC-II-deficient mice indicate that DN 2C thymocytes in 2C⁺RAG^−/− mice are subject to positive selection by pMHC-II and negative selection by pMHC-I. When both pMHC-II and pMHC-I were present (2C⁺RAG^−/− TAP⁺IA^b+), the outcome of their opposing effects seemed to be the production of an intermediate level of DN 2C T cells, which varied among individual 2C⁺RAG^−/− mice, from a low of ≈5% to a high of ≈50% of all T cells in the periphery. Similar variability among individual 2C⁺RAG^−/− mice has also been seen with CD4⁺ 2C T cells (5) and attributed to the heterogeneity of TCR–pMHC interactions in the thymus, possibly because of the nonuniform distribution of thymic antigen-presenting cells (APCs) and heterogeneity of the pMHC they present (24). This heterogeneity may be exaggerated by the distorted thymic architecture generally found in TCR-transgenic mice (SI Fig. 9) (24). That αβ TCR expression in some DN thymocytes and T cells has been described in normal (nontransgenic) mice (25) indicates that the DN thymocyte selection described here may not be limited to TCR-transgenic mice.

Stage-Dependent Sensitivity of 2C Thymocytes to Self-pMHC Complexes.

The positive selection of CD8⁺ 2C T cells by pMHC-I (K^b) complexes probably occurs in thymocytes at the DP stage (26). But, as noted above, pMHC-I complexes have the opposite selective effect on DN 2C thymocytes. A possible explanation for this disparity is that the DN and DP thymocytes differ in sensitivity to self-pMHC-I complexes.

To investigate this possibility, we compared the responses of DN, DP, and SP 2C thymocytes to two pMHC-I complexes, QL9/L^d and SIY/K^b. All of these cells responded equally well to QL9/L^d (Fig. 1C and data not shown), the most potent agonist known for the 2C TCR (27). However, a weaker agonist for this receptor, such as SIY/K^b, is likely more representative of self-pMHC. Because CD8 enhances the responses of CD8⁺ T cells, especially when TCR–pMHC interactions are weak (22), we compared the responses of DN, DP, and SP (CD8⁺) 2C thymocytes to SIY-K^b in the absence and presence of a blocking anti-CD8 antibody. As shown by CD69 up-regulation (Fig. 4A), in the presence of the antibody, DN 2C thymocytes responded more vigorously than DP and CD8⁺ SP 2C thymocytes and even more than DN and CD8⁺ 2C T cells from the lymph nodes. In the absence of the anti-CD8 antibody, where CD8 was free to exert its effects, the DN thymocytes were still at least as responsive as the DP and CD8⁺ SP thymocytes (Fig. 4B). Small differences in the 2C TCR level on the various 2C thymocytes (SI Fig. 6) seem unlikely to account for the marked differences in their reactivity to SIY-K^b when the anti-CD8 antibody was present. Instead, the findings likely reflect differences in sensitivity of 2C thymocytes at different stages of development to TCR ligation by weakly reactive pMHC-I. In the presence of the anti-CD8 antibody, DN 2C cells from lymph nodes were also more responsive than CD8⁺ (SP) 2C thymocytes, although not as responsive as the DN thymocytes (Fig. 4A).

Fig. 4.

Comparison of responses by DN, DP, and CD8⁺ 2C thymocytes and DN and CD8⁺ 2C cells from the lymph nodes of 2C⁺RAG^−/− mice to pMHC (SIY/K^b) complexes. T2-K^b cells were incubated with the indicated concentrations of SIY peptide for 1 h, then mixed with the various 2C cells in the presence or absence of 40 μg/ml CD8-blocking antibody 2.43. Three hours later, CD69 expression was measured by flow cytometry. (A) Titration of the SIY peptide in presence of the anti-CD8 antibody. (B) Effect of CD8. Values are those obtained in A with 1 μM SIY in the presence of the anti-CD8 antibody (filled bars) or absence of this antibody (open bars). All values are averages of triplicates (±SD in B). Representative data from one of the three experiments are shown.

In the absence of anti-CD8 antibody, >80% of CD8⁺ 2C T cells from lymph nodes up-regulated CD69 (Fig. 4B), a significantly greater response than the CD8⁺ (SP) thymocytes. The slightly lower level of CD8 in CD8⁺ thymocytes than in lymph node CD8⁺ 2C T cells (SI Fig. 6) may have contributed to this difference but warrants further investigation. The much greater effect of CD8 on CD8⁺ T cells in lymph nodes than in the thymus would not have been predicted from the changes in CD8 glycosylation that accompany thymocyte maturation and that are associated with stronger, TCR-independent binding of CD8 to MHC-I (28, 29).

Lineage of CD4⁻CD8⁻ 2C Cells.

The DN αβ TCR T cells that are abundant in various TCR-transgenic mice have a number of features in common with γδ lineage T cells (13, 30, 31). Based on their positive selection by pMHC-II complexes, DN 2C cells might also be expected to resemble αβ T cells of the CD4 lineage. Their negative selection by pMHC-I complexes suggests, on the other hand, that they are more similar to αβ T cells of the CD8 lineage. Studies have shown that transcription factors GATA3 and ZFP67 play critical roles in CD4 T cell development, whereas Runx3 and Runx1 play critical roles in CD8 T cell development (32). As a preliminary assessment of the DN 2C cell lineage, we compared by real-time PCR the levels of GATA3, ZFP67, Runx3, and Runx1 transcripts in purified immature (CD25⁺) DN thymocytes from 2C⁺RAG^−/−, OT1⁺RAG^−/−, and B6 mice. The relative levels of GATA3 and ZFP67 were significantly lower in CD25⁺ DN thymocytes of 2C⁺RAG^−/− and OT1⁺RAG^−/− mice than B6 mice, whereas the relative levels of Runx3 and Runx1 were more similar in B6, 2C⁺RAG^−/−, and OT1⁺RAG^−/− thymocytes (Fig. 5A). To examine this issue further, we isolated splenic DN and CD8⁺ 2C T cells from 2C⁺RAG^−/− mice and splenic CD4⁺ and CD8⁺ T cells from B6 mice and measured by RT-PCR the expression of CD8-specific genes Eomes and perforin (33). As shown in Fig. 5B, Eomes and perforin were expressed at similarly high levels in DN and CD8⁺ 2C cells from the 2C⁺RAG^−/− mice and CD8⁺ T cells from B6 mice, whereas they were hardly expressed in the control CD4⁺ T cells. Thus, the DN 2C T cells that developed in 2C⁺RAG^−/− mice more closely resemble T cells of the CD8 lineage than the CD4 lineage, in accord with the stronger responses of these cells to self-pMHC-I than self-pMHC-II complexes and with the original 2C clone having been a CD8⁺ cytotoxic T cell (8).

Fig. 5.

Lineage of DN 2C T cells in 2C⁺RAG^−/− mice. (A) Relative levels of GATA3, Runx1, Runx3, and ZFP67 transcripts in purified CD25⁺ DN thymocytes from B6 mice (filled bars), 2C⁺RAG^−/− mice (open bars), or OT-1⁺RAG^−/− mice (hatched bars) were measured by real-time RT-PCR. Values are average of triplicates (±SD) normalized to the level of GAPDH transcript. P values for pair-wise comparison are <0.05 (*) or <0.01 (**). (B) Expression of perforin and Eomes transcripts in purified DN and CD8⁺ T cells (from spleen and lymph nodes) from 2C⁺RAG^−/− and B6 mice. RNA from various cell populations was diluted 3-fold and used for RT-PCR followed by agarose gel electrophoresis. A representative result from two experiments is shown.

Discussion

Many features of the DN TCR⁺ T cells that are abundant in αβ TCR-transgenic mice resemble those of T cells in the γδ lineage, including (i) bypassing the DP and SP stages in their development, (ii) the coexpression under some circumstances of an endogenous γδ TCR along with the αβ TCR transgene, and (iii) a propensity to develop into CD8αα⁺ intestinal intraepithelial cells (13, 30, 31). These unusual features have been shown recently by Baldwin et al. (34) to arise from the premature expression of the TCR in early (HSA^hiCD25⁺) DN thymocytes. Using a Cre/lox-based strategy, they delayed expression of the HY TCR α chain from the immature DN thymocyte stage, where it is expressed in conventional HY TCR-transgenic mice, to the DP stage and saw that the delay eliminated the anomalous maturation pathway and reduced DN T cell abundance. The premature expression of the αβ TCR, which is needed for DN thymocytes to mature without becoming DP cells, is associated with decreased formation of the pre-TCR and blocked transition from DN to DP thymocytes (14). As shown by the great increase in the number of DP thymocytes in TAP^−/− mice (Table 1), the impaired transition in 2C/RAG mice likely arises from the pronounced negative impact of 2C TCR interactions on DN thymocytes with self-pMHC-I. A similar large increase in the number of DP thymocytes has been seen in 2C TCR-transgenic mice that were deficient in β₂-microglobulin (14) or class Ia MHC (Kb⁻ Db⁻) (11). These effects may be especially pronounced in 2C TCR-transgenic mice because (i) this TCR transgene is uniformly expressed at a higher level in CD25⁺ DN thymocytes than are some other TCR transgenes (Fig. 2 and data not shown) (14) and (ii) there are differences in the intensity with which various TCRs react with self-pMHC complexes. Thus, 2C T cells and thymocytes recognize ubiquitously expressed mitochondrial peptide–K^b complexes (35–38), and, like male-specific HY TCR⁺ cells in male mice (39), they may be more reactive with self-pMHC than T cells that express other TCR transgenes (5).

Although no phenotypic markers distinguish αβ from γδ lineage T cells (except for the TCR), αβ and γδ T cells differ considerably in antigen recognition. Unlike αβ T cells, most γδ T cells are not MHC-restricted, and MHC selection is not involved in their development (40–42), except for a special subset (43). Although we have not systematically examined antigen recognition by the DN 2C T cells, the responses of these cells to some representative pMHC complexes (SIY/K^b and QL9/L^d) clearly indicate that they are as MHC-restricted as other T cells that express this TCR. Moreover, from differences in the abundance of DN 2C thymocytes and T cells in MHC-I-deficient and MHC-II-deficient mice, these cells seem to be subject to selection by pMHC. In support of our evidence for their negative selection, previous studies have noted an increased frequency in Annexin V⁺ DN thymocytes in TCR-transgenic mice compared with wild-type mice (34, 44). Others, however, have found no evidence for MHC selection of these DN T cells (13, 45). Different conclusions about MHC selection could arise from difficulties in measuring small changes in thymocyte abundance.

The findings that DN 2C cells in 2C⁺RAG^−/− mice are MHC-restricted in antigen recognition and subject to selection by MHC are more consistent with the αβ than γδ lineage. When, however, DN 2C thymocytes and lymph node cells were stimulated intensely for 3–4 days with plate-bound anti-TCR antibody (1B2) and high concentrations of IL-1 and IL-2 (13) CD8αα was expressed by ≈60% of the thymocytes and ≈14% of the lymph node cells, and a still smaller proportion of the lymph node cells expressed CD8αβ (data not shown). Because the CD8αα response may be indicative of γδ lineage, it is possible that the lineage of the DN 2C cells in 2C⁺RAG^−/− mice is mixed, although it is not apparent what a “γδ lineage” means in mice whose TCR γ and δ gene rearrangements are precluded by RAG deficiency (46).

The divergent selective effects of pMHC-I on CD8⁺ and DN 2C thymocytes in 2C⁺ RAG^−/− mice (Fig. 3 and Table 1) (9) seem counterintuitive because of (i) the rule that weak TCR–pMHC interactions in the thymus lead to positive selection and strong interactions to negative selection and (ii) the well known strengthening effect of CD8 on these interactions. How can self-pMHC-I complexes negatively select DN 2C thymocytes and positively select CD8⁺ (DP) 2C thymocytes? A partial answer is suggested by the differing responses of thymocytes at various stages of development to a relatively weak pMHC-I ligand (SIY/K^b) for the 2C TCR when the comparison was carried out in the presence of an anti-CD8 antibody to “level the playing field” between CD8⁺ and CD8⁻ 2C cells (Fig. 4): the greater responsiveness of the DN than DP thymocytes under these conditions is consistent with negative selection of the former and positive selection of the latter by self-pMHC-I. However, in the absence of CD8-blocking antibody, the DN 2C thymocytes, although still highly reactive, were not more reactive than the DP 2C thymocytes. Hence, the basis for the different selective effects of self-pMHC-I on the DN and DP thymocytes is not entirely clear.

Aside from its significance for T cell development, the responses of DN thymocytes to self-pMHC is particularly interesting in light of evidence that TCR-mediated signaling is greatly enhanced by CD4- and CD8-dependent cooperation between self-pMHC and agonist pMHC (47–49). Nevertheless, mature 2C T cells that lack CD4 and CD8 can make robust pMHC-specific responses to APC that present highly potent pMHC agonists (Fig. 1C) (21). Moreover, after infection with various pathogens, CD4^−/− mice produce CD8⁺ T cells that react specifically with strong pMHC-II agonists (50, 51). Similarly, CD4⁺ 2C T cells respond to a potent pMHC-I agonist (QL9-L^d) for the 2C TCR (5). Because CD8 probably does not bind significantly to IA^b, or CD4 to L^d, a T cell with a mismatch between its coreceptor and TCR specificity is essentially “double negative” and functionally similar to the overtly DN 2C T cells described here. It may be that with highly potent pMHC ligands the TCR–pMHC interaction is sufficiently long-lived to stimulate signal transduction without participation of coreceptor molecules. Nevertheless, the question still remains: how do thymocytes that lack CD4 and CD8 respond to APC that present only self-pMHC in the absence of an obvious potent agonist? Given the highly responsive status of some DN thymocytes, it may be that their steady-state levels of phosphorylation on critical signal-transducing molecules is so high that stochastic fluctuations in TCR–self-pMHC bond lifetimes can activate these DN thymocytes.

DP thymocytes have been shown to be more responsive than SP thymocytes or T cells in peripheral lymphoid tissues to weak pMHC (15, 16). The difference is consistent with the view that enhanced reactivity to self-pMHC of some DP thymocytes, leading to their negative selection, eliminates potentially autoreactive T cells before they enter the naïve T cell pool in peripheral lymphoid tissues. That DN thymocytes, when they express a TCR, are more susceptible than DP thymocytes to negative selection expands the variety of T cells whose responses to self-pMHC complexes can range widely, from negative to positive selection in the thymus, and beyond to survival and proliferative signals in naïve and memory cells in the periphery. This wide range of responses to TCR ligation by self-pMHC complexes emphasizes the critical character of the T cell's stage of development and status in determining the outcome of antigen recognition.

Materials and Methods

Mice.

The 2C TCR-transgenic RAG1^−/− mice (2C⁺RAG^−/−) were backcrossed to the C57BL/6 (B6, H-2^b) background successively for 13 generations. Similarly, OT1 TCR-transgenic RAG1^−/− mice (OT1⁺RAG^−/−) were backcrossed onto the B6 background for >10 generations. B6 mice deficient in IA^b or TAP-I were from The Jackson Laboratory (Bar Harbor, ME) and backcrossed to the 2C⁺RAG^−/− mice to produce 2C⁺RAG^−/− mice deficient in lA^b or TAP-I. Mice were kept under specific pathogen-free facilities and used at 2–6 months of age.

Antibodies and Flow Cytometry Analysis.

Antibodies to CD8, CD25, CD44, CD62L, CD69, HSA, TCR β (Cβ), CD11b, and CD11c were purchased as conjugates from BD Biosciences (San Jose, CA). Clonotypic antibody 1B2, specific for the 2C TCR, was purified and conjugated with biotin. Single cell suspensions were prepared from thymus, lymph nodes, and spleen. Red blood cells were lysed by treating splenocytes with ammonium chloride (lysis buffer). Peripheral mononuclear cells were prepared from tail blood samples after eliminating red blood cells. Cells were stained in the presence of 2.5 μg/ml anti-FcR antibody in PBS containing 0.1% BSA and 0.1% NaN₃ and analyzed on a FACScaliber or a FACSAria, collecting 10,000–1,000,000 live cells per sample, by using Flowjo software (Tree Stars, Inc., Ashland, OR).

To detect intracellular IFN-γ, splenocytes were cultured in complete RPMI medium 1640 supplemented with 0.1 μM SIY peptide or cultured on plates coated with antibodies to CD3 and CD28 (immobilized at 10 μg/ml). The cultured cells were incubated with 10 μg/ml brefeldin A for 1 h before they were harvested and surface-stained with antibodies to CD4, CD8, and TCR β. After fixation with 4% paraformaldehyde, cells were stained with anti-IFN-γ antibody in 0.1% saponin in PBS.

Various monoclonal anti-CD8 antibodies differ in their effects on TCR interactions with pMHC-I (52). The anti-CD8 antibody used here (2.43) was shown to decrease markedly a cloned 2C T cell cytolytic response to a relatively weak agonist (SIY/K^b) but hardly to reduce this response to a strong agonist (QL9/L^d) (21).

In Vitro Stimulation of 2C Thymocytes and T Cells.

Thymocytes and lymph node cells were stimulated by incubating them with an equal number of T2-K^b or T2-L^d cells and peptides (SIY or QL9) at various concentrations. After various times (3–24 h), the cells were stained with antibodies to CD69, CD4, and CD8 and analyzed by flow cytometry.

Cell Sorting and RT-PCR.

DN thymocytes were enriched by negative selection. Total thymocytes were labeled with biotinylated antibodies against B220, CD11b, CD11c, CD4, and CD8, mixed with streptavidin-conjugated microbeads followed by AUTOMACS depletion. Enriched DN thymocytes were sorted with FACSAria by gating on CD25⁺ B220⁻ CD11b⁻ CD11c⁻ CD4⁻ CD8⁻ cells. RNA from the sorted CD25⁺ DN thymocytes was purified with TRIzol (Invitrogen, Carlsbad, CA), and cDNA synthesis and quantitative PCR were performed according to the manufacturers' instructions (Qiagen, Valencia, CA, and Applied Biosystems, Foster City, CA). For purification of T cells, lymphocytes from lymph nodes and spleen were stained with antibodies against CD4, CD8, and TCRβ, then sorted by FACSAria. Semiquantitative RT-PCR was performed by 3-fold dilution of cDNA according to instructions from Invitrogen and Qiagen.

Abbreviations

APC

antigen-presenting cell

DN

double negative (CD4⁻CD8⁻)

DP

double positive (CD4⁺CD8⁺)

NK

natural killer

pMHC

peptide–MHC complex

QL9

QLSPFPFDL peptide

RAG

recombination-activating gene

SIY

SIYRYYGL peptide

SP

single positive (CD4⁺ or CD8⁺)

TAP

transporters associated with antigen processing

TCR

T cell receptor.