Peptide-Specific, TCR-α–Driven, Coreceptor-Independent Negative Selection in TCR α-Chain Transgenic Mice

During thymocyte differentiation, expression of the TCR loci is temporally separated. The β locus rearranges first at the CD4⁻CD8⁻ double-negative (DN) 3 stage where expressed β-chains pair with pre-Tα to form the pre-TCR (reviewed in 1). Spontaneous oligomerization of the pre-TCR initiates thymocyte proliferation and expression of the CD4 and CD8 coreceptors, characterizing the double-positive (DP) stage of thymocyte differentiation (2). The α locus is expressed at this stage, creating a broad αβ TCR repertoire from which positive and negative selection mold the mature CD4 single-positive (SP4) and CD8 single-positive (SP8) populations. As thymocytes differentiate, sensitivity to Ag declines. This is evident from constitutively dimerized CD3ε, which signals in DN stage thymocytes but not in more mature thymocytes (2). In the absence of pre-TCR, prematurely expressed transgenic (Tg) TCR α-chains as αβ or αγ heterodimers can direct DN to DP differentiation both in the absence and presence of a positively selecting MHC ligand (3–6).

Unlike αβ T cells, γδ T cells differentiate directly from DN thymocytes. Initiation of TCR γ and δ locus rearrangement precedes TCR-β rearrangement, although αβ and γδ T cells can carry rearranged γ and β loci, respectively (7, 8), suggesting lineage commitment is not tightly linked with TCR rearrangement. Rather, an adequate signal through the γδ TCR is required for γδ lineage commitment (9, 10). As well as functioning similarly to pre-TCR in initiating progression to the DP stage, premature expression of αβ TCR at the DN stage can have additional consequences. A feature of several αβ Tg strains is development of a population of αβTCR⁺, DN, “γδ wannabe” T cells (11, 12). Such populations have a γδ T cell-like transcriptional profile and are considered to arise due to the Tg αβ TCR directing a γδ program of differentiation (13, 14). In the presence of cognate, agonist ligand, the αβ TCR delivers a stronger signal, leading to negative selection at the immature DN stage as exemplified by the B6.2.16 HY-specific TCR (15). When the B6.2.16 TCR is expressed at the temporally appropriate, DP stage, negative selection is delayed until the SP stage (16). Engagement of agonist HYD^bSmcy ligand by B6.2.16 αβ TCR⁺ DN but not DP thymocytes also directs development of a large CD8α/α intraepithelial lymphocyte population (15, 16).

The impact of MHC-peptide recognition on thymocytes prematurely expressing TCR has been investigated in the MHC class I-restricted, 2C TCR Tg strain (17). Somewhat surprisingly, the K^b class I molecule, which directs conventional positive selection of DP to SP8 thymocytes, also mediates the negative selection of a proportion of DN 2C⁺ thymocytes. In the absence of K^b expression in TAP^−/− mice, the A^b class II molecule acts to promote the selection of DP 2C thymocytes. Immature thymocytes are thus uniquely MHC- peptide sensitive, as those K^b/self-peptide complexes mediating positive selection of DP thymocytes can also deliver a coreceptor-independent, negatively selecting signal at the DN stage (17).

T cells from mice Tg for a single TCR β- or α-chain coexpressed with an endogenous partner-chain repertoire are generally biased to the original specificity of the donor TCR (18). This is evident by analysis of Ag-specific responses and in some examples by staining of naive T cells with MHC tetramer. Such bias results from Ag specificity being partly encoded by the Tg TCR chain and by preferential positive selection of partner chains, which contribute to Ag recognition (19–21).

The HY-specific C6 CD8 T cell clone recognizes the class I molecule H2-K^k with the Smcy-derived peptide TENSGKDI (HYK^kSmcy) (22). As expected, thymic positive selection in female C6 αβ TCR (C6αβ) Tg mice is strongly skewed to the CD8 lineage (23). In this study, we have investigated the influence of the HYK^kSmcy complex on thymocytes expressing only the Tg C6 TCR α-chain (C6α) together with an endogenous TCR-β repertoire. On the original selecting H2^k MHC haplotype, the thymic skew to the CD8 lineage is preserved, indicating that preferential engagement with MHC class I is conferred by the C6 α-chain. In the presence of the HYK^kSmcy complex, deletion of a large proportion of C6α Tg DN thymocytes was observed. Deletion was not dependent on partner chains using the original β variable (BV) segment. By contrast, recognition of the HYK^kSmcy complex by mature SP8 thymocytes and peripheral CD8 T cells was strictly dependent on the original C6 partner β-chain. These data show that the increased sensitivity of immature thymocytes allows specific recognition of Ag at a lower signaling threshold with fewer molecular interactions required between the αβTCR⁺ immature thymocytes and the MHC-peptide ligand to initiate signal transduction in comparison with mature T cells.

Mice were housed at Hammersmith Hospital animal facility with the appropriate Home Office and ethical authority. C6αβ TCR Tg mice express a TCR specific for the Y-chromosome–derived male Ag HYK^kSmcy (22, 23). Tg mice expressing the C6 TCR α-chain on a TCR-α^−/− background (C6α) were bred to the H2^b and H2^k MHC haplotypes. C6α H2^k mice were also crossed to a bacterial artificial chromosome (BAC) Tg strain overexpressing Smcy mRNA at ∼8-fold higher than the male gene (24) to obtain C6α, TCR-α^−/−, H2^k Smcy⁺ females. Tissues were harvested and live lymphocytes were counted using a hemacytometer and analyzed by flow cytometry.

The CDR1 and CDR2 germline-encoded CDRs of the C6 β-chain were mutated to alanine/glycine linkers using overlap PCR as previously described (25). The C6 wild-type (WT) and CDR1/2 mutant were cloned into the MIGR1 retroviral vector (26). Retroviral supernatants were produced using Phoenix packaging cells, which were a kind gift from Prof. Hans Stauss, London. C6α Tg TCR-α^−/− bone marrow hemopoietic stem cells (HSCs) were transduced and transferred into TCR-α^−/− (H2^k) recipient mice as previously described (25). Mice were sacrificed 8 wk after transfer.

All Abs were obtained from BD Pharmingen (San Diego, CA). Cells were stained in PBS with 10% FCS and 0.01% sodium azide. For tetramer staining, cells were incubated with 1 μl APC-conjugated HYK^kSmcy tetramer (Proimmune, Oxford, U.K.) in balanced salt solution with 10% FCS for 15 min at room temperature, washed, and subsequently costained with the appropriate Abs. Cells were washed and analyzed on a FACScalibur or FACScan flow cytometer using Cell Quest software (BD Biosciences, San Jose, CA).

Single-cell suspensions of H2^k splenocytes were irradiated (30 Gy) and resuspended (10⁷/ml) in complete RPMI (RPMI 1640 medium supplemented with 10% FCS, 2 mM l-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin, and 5 × 10⁻⁵ M 2-ME). Cells were plated at 50 μl/well in triplicate in a 96-well, flat-bottomed plate.

Single-cell suspensions of splenocytes from C6 αβTCR Tg mice and retrogenic mice (C6α⁺, TCRα^−/− Tg mice expressing retroviral vector encoded WT C6 β-chain or C6 CDR1, CDR2 double-mutant β-chain) were resuspended in complete RPMI and plated with APCs at 1:1, except for the C6αβ Tg, which was 10× diluted to give a comparable number of CD8 T cells. The HYK^kSmcy peptide (Smcy, TENSGKDI) was added to give a final concentration range of 1 nM to 10 mM. Plates were incubated at 37°C in an atmosphere of 5% CO₂ for 72 h and pulsed with tritiated thymidine (0.5 μCi) for the last 18 h, harvested, and counted (Wallac 1205β plate counter, Wallac Pharmacia, Milton Keynes, U.K.).

Thymocytes were harvested and stained with anti-CD4 and anti-CD8 and subject to cell sorting on a FACSAria cell sorter (BD Biosciences) to obtain DN, DP, SP4, and SP8 populations. RNA was extracted from cell populations using Trizol (Invitrogen, Carlsbad, CA) and cDNA synthesized using Superscript II RNase H- reverse transcriptase (Invitrogen) and random hexamers (Amersham Biosciences, Piscataway, NJ). BV6 and BV11 products were amplified using specific BV region BC primers and size distributions determined as described (27).

All values are shown as mean ± SEM, SD, or range as indicated. Groups were compared using the Student two-tailed unpaired t test, where significance is reached at p < 0.05 (Prism v2.01, Graph Pad, San Diego, CA).

The C6 TCR [Valpha8.3 (AV8), Jalpha27 (AJ27); Vbeta11 (BV11), Jbeta2.4 (BJ2.4)] is specific for the TENSGKDI peptide derived from the Y chromosome gene Smcy, presented by the MHC class I molecule H2-K^k (22). In female H2^k mice, Tg expression of the C6 TCR results in thymic positive selection into the CD8 lineage (Fig. 1A) (23). Negative selection of immature thymocytes is a feature of αβ TCR Tg mice that express the cognate peptide as a self-Ag (28). As expected, the presence of the cognate peptide for the C6αβ TCR in H2^k males mediates massive thymic deletion (Fig. 1B) (29). Tg expression of just the C6 α-chain with an endogenous TCR β repertoire also leads to skewed thymic positive selection to the CD8 lineage in females (Fig. 1C) and males (Fig. 1D), which is maintained in the periphery (Table I). Skewed selection to the CD8 compartment is not an intrinsic feature of the C6 α-chain but is dependent on MHC haplotype, as the normal skew in favor of the CD4 lineage is seen on an H2^b MHC background in both females and males (Fig. 1E, 1F).

Surprisingly, the C6α/β^endogenous repertoire also mediated substantial deletion in the thymus of H2^k males but not females (Fig. 1G; *p < 0.0001). Loss of cellularity reflected conventional MHC-restricted Ag recognition, as it was not seen when the C6 α-chain transgene was crossed onto the H2^b haplotype. This deletion resulted in a loss of over 40% of cells in the male thymus. In C6αβ male thymi, the deletion is more severe, where cellularity is <20% that of females (Fig. 1G). Given the dominant influence of the C6 α-chain on positive selection into the SP8 compartment, the deletion phenotype seen in the presence of the Y chromosome may reflect recognition of the same ligand (H2-K^k/TENSGKDI) as recognized by the C6αβ TCR. To test this hypothesis, an Smcy BAC transgene (24) was crossed onto the C6α strain. The deletion phenotype was reproduced in those female H2^k C6α Tg mice also expressing the Smcy transgene, where thymus cellularity was reduced by >50% (Fig. 1H; p < 0.0001). All crosses were also homozygous knockout for the endogenous TCR α locus, excluding any contribution from endogenous TCR α-chains (30). Poor pairing of the Tg C6 β-chain with the endogenous TCR α-chain repertoire (I. Bartok, S. Holland, H. Kessels, J. Silk, M. Alkhinji, and J. Dyson, manuscript in preparation) precluded analysis of Ag reactivity within the reciprocal (α^endogenous/C6β) repertoire.

To determine whether deletion is dependent on the BV segment used by the C6 TCR (BV11) or a limited subset of variable segments, which may create a hole in the male repertoire if they imparted reactivity to male Ag, we compared the BV repertoire of H2^k C6α females and males through T cell development using Vβ region specific Abs recognizing 15 of the defined BV segments (Fig. 2 and data not shown). Proportional use of BV11, the BV segment used by the C6 TCR, was comparable in H2^k C6α males and females among DN (Fig. 2A), DP (Fig. 2B), SP4 (Fig. 2C), SP8 (Fig. 2D), and peripheral CD4 (Fig. 2E, 2G) and CD8 (Fig. 2F, 2H) T cells. This indicates that the deletion observed in male mice is not restricted to a subset of cells expressing BV11. Even total deletion of BV11 representing around 5% of the preselection repertoire (31) could not account for the scale of the observed deletion in C6α males and C6α female Smcy⁺ Tg thymi. As expected, BV11 and BV5 were underrepresented in H2^k C6α male and female mice after the DP stage in development due to conventional negative selection mediated by recognition of endogenous superantigen encoded by integrated Mtv genomes (32, 33). Male and female AV8⁺ BV repertoires (Fig. 2A–F) are diverse at DN, DP, and SP stages: male repertoires do not exhibit loss of particular BV segments that could account for the large decrease in cellularity. Following selection to the mature SP stage and exit into the periphery, an overrepresentation of BV8.1/8.2 (SP4, SP8, CD4, CD8), BV6 (SP8, CD8), and BV8.3 (SP4, SP8, CD4) is seen. This occurs similarly in both male and female repertoires, again demonstrating similar BV usage in deleted (male) versus nondeleted (female) repertoires.

Additionally, in agreement with a previous study of TCR α-only Tg mice (19), the distribution of endogenous TCR β-chain variable segments expressed on splenic CD4 and CD8 T cells of C6α-only mice resembled that of non-Tg mice (Fig. 2G, 2H). In H2^k mice, the enhanced selection into the CD8 lineage directed by the C6 α-chain is thus dependent on MHC haplotype but does not appear to be associated with a skewed BV segment usage, suggesting the C6 α-chain plays a major role in the interaction with MHC class I.

Overall, usage of the BV segments we examined remains very similar throughout development in males and females. This indicates that in the male, cells are lost from each BV-expressing subset and therefore that the deletion occurs via interaction of K^kSmcy with TCR displaying the C6 α-chain in combination with a wide β-chain repertoire.

The C6 α-chain transgene is expressed from a human CD2 promoter/enhancer cassette resulting in premature expression of an αβ TCR repertoire at the DN stage (34). Thymocyte subset cellularity is shown in Fig. 3. In C6α males, the cellularity of the DN population is reduced compared with females, revealing deletion is already evident at this stage (Fig. 3A; p = 0.0002). This decrease in immature cells results in a proportionately similar loss of cellularity in the more mature DP (Fig. 3B; p < 0.0001) and SP subsets (Fig. 3C, 3D; p ≤ 0.001). As cells are lost at the earliest developmental stage in C6α males, we examined DN cellularity in C6α Smcy⁺ females. A significant decrease in the number of DN cells was also seen in female H2^k mice expressing the C6 α-chain and the male gene (Smcy⁺, Fig. 3E; p = 0.01). This decrease persists into all subsequent thymocyte subsets in these mice (data not shown). In both C6α males and C6α Smcy⁺ Tg females, presence of the cognate Ag tends to further increase the relative proportion of the DN subset (Fig. 3F and Table I), although this was not statistically significant (H2^k male versus female: p = 0.06; Smcy⁺ versus Smcy⁻: p = 0.08). As reported for other TCR α-chain Tg strains, a proportional increase in the DN subset most likely reflects reduced cellular expansion associated with the DN to DP transition when mediated by the αβ TCR in comparison with the pre-TCR. In terms of proportions of the other thymocyte subsets, there were no significant differences between the percentages of DP or SP4 cells in H2^k or H2^b male and female C6α mice. As discussed above, H2^k male and female mice displayed an increased proportion of SP8 cells compared with H2^b mice as cells expressing the C6 α-chain are preferentially selected on H2^k molecules to the CD8 lineage (Fig. 1 and Table I). All C6α mice also had lower percentages of SP4 thymocytes (∼3%) than non-Tg mice (Fig. 1).

Expression of CD25 and CD44 divides the DN compartment into four subsets corresponding to successive stages of differentiation (DN1: CD25^loCD44^hi; DN2: CD25^hiCD44^hi; DN3: CD25^hiCD44^lo; DN4: CD25^loCD44^lo). TCR β-chain rearrangement occurs at the DN3 stage where pre-TCR signaling initiates proliferation and transition to the DN4 and then DP stages. To confirm that mature TCR is expressed early on the surface of C6α Tg DN thymocytes, DN subsets were analyzed for cell surface expression of AV8. As expected, mature TCR was upregulated on thymocytes after the DN3 stage, following endogenous TCR β rearrangement (Fig. 4A, 4B). Analysis of DN3 and DN4 subsets showed that males and females exhibit similar DN3 cellularities (Fig. 4C), but that males showed significantly reduced cell number in the DN4 subset (Fig. 4D; p = 0.02), indicating loss of cells following mature TCR expression. In comparison with Smcy⁻ littermates, C6α Smcy⁺ females also had reduced cellularity in the DN4 (Fig. 4D; p ≤ 0.02) but not DN3 population (Fig. 4C), also consistent with defective transition between DN3 and DN4.

In order to assess the relative sensitivity of thymocytes in depleted (male) and nondepleted (female) repertoires, we examined CD5 levels on DN, DP, and SP cells. CD5 is upregulated on cells that have received strong Ag receptor signals; therefore, low levels of CD5 on thymocytes are associated with cells receiving weaker signals through the TCR (35). Interestingly, C6α male AV8⁺ DN4 thymocytes showed decreased levels of CD5 on the cell surface compared with female cells (Fig. 4E). This is consistent with the survival of comparatively lower affinity clones in the male. Potentially male-reactive cells, which would have received strong TCR signals and therefore should express high levels of CD5, are absent from the repertoire. This pattern of decreased CD5 staining was seen on total DN and DP subsets, whereas DN3 cells of males and females show very low levels of CD5 staining (data not shown). We also found that the level of Vα8 staining on DN4 thymocytes was lower in male versus female mice (Fig. 4F), which is likely to reflect the deletion of cells expressing male-reactive TCR and/or downregulation of the TCR following signaling.

The data described above show that HYK^kSmcy binding by C6α DN thymocytes does not have strict BV segment requirements, indicating the complex can be recognized by TCR expressing β-chains that differ at CDR1, 2, and 3 from the cognate C6 β-chain. One explanation for this observation is that the β-chain plays a negligible role in Ag recognition by the C6 TCR. We therefore examined whether mature T cells of C6α females, which have not been depleted of HYK^kSmcy reactivity, are able to recognize the complex. MHC class I HYK^kSmcy tetramer was used to stain splenocytes from C6αβ female, C6α female, and male H2^k mice (Fig. 5A–C). Staining was detectable only on C6αβ Tg female CD8 T cells (Fig. 5A). Lack of staining within the female C6α CD8⁺ subset (Fig. 5B) shows that the cognate C6 β-chain is required to achieve the level of affinity necessary for tetramer binding. Male cells showed no tetramer binding, as expected (Fig. 5C). We could not assess tetramer binding in immature thymocytes, as MHC-peptide tetramers require the presence of a T cell coreceptor for adequate binding to Ag-specific TCR.

As tetramer binding may also not detect mature T cells with low affinity for HYK^kSmcy, we assessed whether the female C6α T cells could be activated by APCs pulsed with high levels of the TENSGKDI peptide. Female C6αβ TCR Tg splenocytes proliferated in response to the peptide, whereas male and female C6α Tg splenocytes made no response. These cells are however capable of activation and proliferation in response to Con A (data not shown). These data confirm the requirement for the C6 β-chain in imparting mature T cell reactivity to the male peptide. This is in contrast to the HYK^kSmcy-mediated deletion operating at the DN stage, which does not require the cognate β-chain.

Analysis of the TENSGKDI peptide containing amino acid substitutions at nonanchor residues showed both amino- and carboxy-terminal substitutions led to dramatic loss of proliferation by the C6 T cell clone (22), implicating both the α and β CDR3 regions in recognition. We therefore tested whether the C6 TCR containing a β-chain retaining its CDR3 region but lacking the germline-encoded CDR1/2 regions might facilitate Ag recognition by mature T cells. The central part of the germline BV11 CDR1/2 loops were replaced with glycine/alanine linkers by PCR, and the modified C6 β-chain was cloned into the MIGR1 retroviral vector (26). The WT and mutant C6 β-chains were then introduced into C6α HSC, which were transferred to irradiated female TCR α-deficient recipients. Both WT and CDR1/2 mutant β-chains were expressed by a high proportion of peripheral CD8 T cells in the resulting chimeras, which were used to assess Smcy peptide-stimulated proliferation. The WT C6 β-chain restored proliferation of female C6α peripheral T cells to the level of C6αβ Tg cells, whereas those expressing the C6 α-chain paired with the mutant β-chain were not responsive to the Smcy peptide (Fig. 5D). These data indicate that activation of peripheral CD8 T cells by the HYK^kSmcy complex requires contributions from the germline-encoded BV11 CDR regions.

We considered that although BV segment usage is not skewed, the high level of HYK^kSmcy reactivity within the DN C6α population may arise from pairing constraints imposed by the C6 α-chain, which may favor assembly of C6α/β^endogenous TCR with structural characteristics, facilitating recognition of HYK^kSmcy. To investigate this possibility, we used spectratyping to analyze TCR β CDR3 diversity within the DN, DP, and SP populations of female and male thymocytes. All BV segments showed Gaussian distributions of CDR3 length. Representative spectratypes are shown for BV6 and BV11 (Fig. 6A, 6B). Considering the absence of TCR α revision in C6α mice, which normally provides β rearrangements with multiple sequential opportunities for audition, CDR3β diversity in the male and female C6α thymocyte populations was surprisingly high, suggesting the C6 α-chain can pair and mediate positive selection with a broad range of endogenous β-chains.

Mice expressing a Tg, monoclonal TCR β-chain generally select T cell repertoires biased to the original specificity (18). This is not surprising, as the contribution of the Tg chain to recognition of a specific MHC-peptide complex can be maintained with multiple partner chains. The process of positive selection can also favor partner chains that are similar to the original. For example, B6.2.16 TCR β Tg mice select a high proportion of α-chains using the AV9 gene segment into the CD8 lineage (21). Many AV9 chains reconstitute HY specificity, underlying the strong bias.

In general, TCR β Tg mice display markedly restricted AV repertoires (18, 20, 36, 37). By contrast, the endogenous β-chain repertoire coselected with the C6 α-chain was not strongly skewed and resembled the WT β repertoire. Tg mice expressing the α-chain of the CD8 lymphocytic choriomeningitis virus-specific P14 TCR, in the absence of endogenous α-chain expression, also selected a broad BV repertoire (19). Further, the C6α chain selected similar CD4 and CD8 BV repertoires on both H2^k and H2^b backgrounds (data not shown). This may suggest differential roles for the contribution of TCR α- and β-chains during selection. It is possible that many β-chains are able to mediate selection with a fixed α-chain, but in a reciprocal situation when the TCR β-chain is invariant, fewer α-chain variable segments are compatible with selection. Interestingly, recent crystal structures have shown that TCR α-chains seem to dominate in interaction with MHC-peptide structures (38). It will be of interest to examine BV repertoires in other TCR α-chain Tg strains, as there are few published examples.

Recognition of agonist MHC-peptide complexes causes deletion when a Tg MHC class I-restricted αβTCR is prematurely expressed by DN thymocytes, leading to severely reduced thymic cellularity (15, 28). This is the case for C6αβ TCR Tg males (Fig. 1G) in which the agonist HYK^kSmcy complex is expressed. In C6 TCR α-only Tg mice on a TCR α knockout background, endogenous β-chain expression allows a C6α/β^endogenous repertoire to be expressed after the DN3 stage (Fig. 4B). DN3 to DN4 thymocytes also express the pre-Tα–chain, although more efficient pairing of the αβ heterodimer will limit the amount of surface pre-TCR. As for other TCR α Tg strains (39), premature expression of the C6 α-chain leads to a high DN/DP ratio, again highlighting the specialized function of the pre-Tα–chain in providing the optimal pre-TCR signal for β selection. Expression of mature TCR on immature thymocytes resulted in the deletion of a large proportion of DN thymocytes. This was seen in both males and Smcy⁺ Tg females and was dependent on the H2^k haplotype, demonstrating Ag-dependent, MHC-restricted, coreceptor-independent recognition. Additionally, the presence of the Tg TCR α-chain in all mice leads to low percentages of SP4 thymocytes compared with non-Tg mice (Fig. 1 and data not shown), a phenomenon reported in other TCR α Tg strains (39). This may reflect the absence of TCR α revision, severely limiting the number of αβ complexes auditioning for selection.

Deletion at the DN stage was not, however, dependent on use of the original BV11 variable segment. DN thymocytes are thus able to respond to the HYK^kSmcy complex without any of the original C6 β-chain CDRs. Conversely, recognition of the HYK^kSmcy complex by mature CD8 T cells is strictly dependent on the contribution of the C6 β-chain. Data from many MHC-peptide/TCR structures show the TCR β-chain CDR3 region tends to be positioned over the carboxy-terminal region of the peptide. A T > G substitution at P7 of the Smcy peptide resulted in an ∼1000-fold loss of peptide sensitivity by the C6 T cell clone, suggesting the β-chain makes important contact with the HYK^kSmcy complex (22). A variant of the C6 β-chain with the CDR3 region intact but missing the CDR1 and CDR2 regions was also incapable of recognizing the HYK^kSmcy complex over a wide range of peptide concentration (Fig. 5D). TCR α-chain–dependent deletion of DN thymocytes on the HYK^kSmcy complex is thus achieved without the CD4 and CD8 coreceptors and without specific contributions from the β CDR regions, which are required for recognition by mature T cells.

The unique sensitivity of DN thymocytes to Ag engagement presumably underlies the recognition of the HYK^kSmcy complex by many C6α/β^endogenous TCR combinations. This may explain the lack of requirement for the C6 β CDR3 region and its contribution to the interaction with the HYK^kSmcy complex, resulting in a loss of affinity that takes it below the threshold for peripheral T cells but not their more Ag-sensitive progenitors. It might be expected that many endogenous β CDR3 loops would disrupt docking of the TCR onto the HYK^kSmcy complex. However, as the C6 β-chain CDR3 region is 14 aas long, most endogenous rearrangements will be shorter and so unlikely to cause steric clash. The spectratypes of BV6 and BV11 C6α⁺ TCR indicate that there is little restriction of diversity in the selected CDR3 repertoire (Fig. 6).

The most striking observation is the extent of deletion and maintenance of the pattern of BV segment use between males and females, which shows that the deletion is not dependent on a single or small set of BV segments. The repertoire remains diverse and comparable on male and female thymocytes from the DN to DP transition (Fig. 2A, 2B). As DN4 is the first stage at which mature TCR is expressed on the cell surface in these mice (Fig. 4B), the cells staining positive for AV8 and BV in the DN fraction in Fig. 2A will represent the DN4 cells. This is the stage at which the cellularity is decreased in C6α H2^k males and C6α H2^k Smcy⁺ females. We assume that the predeletion BV repertoires (i.e., at DN3, immediately following β-chain gene rearrangement) would look similar in terms of diversity, although this analysis is precluded by lack of cell surface TCR at this stage.

The β-chain carries the germline-encoded CDR1 and CDR2 regions, which are used in interaction with MHC α helices. The importance of these interactions in both positive selection and peripheral T cell activation has been established from crystallographic, biophysical, and genetic approaches (40–42). The behavior of the C6α/β^endogenous TCR repertoire at the DN stage suggests that coreceptor-independent, specific recognition of MHC-peptide complexes is TCR α-chain–dependent and can occur promiscuously, where a range of β-chain germline-encoded CDR1 and CDR2 regions allow docking with the HYK^kSmcy complex. Regardless of the presence of the Smcy peptide, the H2^k haplotype skews the C6α/β^endogenous repertoire to the CD8 lineage, which may reflect favorable recognition of the H2-K^k class I molecule by the AV8 CDR1/2 regions. This skew to the CD8 lineage affects all BV segments examined (Fig. 2). These data suggest, in this example, that the AV8 TCR α-chain has a stronger influence than the β-chain on recognition of MHC. Indeed, alternative β CDR1/2 regions do not direct alternate docking positions, which prevents either enhanced CD8 T cell selection in females or DN thymocyte deletion in the presence of the HYK^kSmcy complex.

In the non-Tg thymus, the timing of TCR α-chain expression is critical for normal thymocyte development (16). TCR α-chain Tg mice do not reflect this, instead highlighting this need for ordered TCR gene rearrangement. We are not aware of any reports of physiological expression of αβTCR on DN cells. However, signaling events required for γδ T cell development occur at the DN stage, raising the possibility that they may also involve low affinity interactions of the γδ TCR with self-ligands. The data presented in this study confirm that thymocytes display variable signaling thresholds for activation and indicate that DN thymocytes are hypersensitive to peptide, despite lacking coreceptors, which are required for optimal MHC-restricted TCR signaling in more mature cells. Pre-TCR signaling is thought to be ligand independent, and it is not fully understood how DN3 thymocytes receive and interpret this signal, or signals, which are clearly important in the selection of suitable TCR β-chains. Pre-TCR and TCR activation share common downstream signaling pathways (43), and signaling through both receptors induces the expression of CD5 in a signal-strength–dependent manner (35, 44). One may speculate that if the pre-TCR were able to interact weakly with self-ligands in the normal non-Tg thymus, a signal could be transduced at a relatively low activation threshold. DN cells interacting with low affinity may transduce a survival signal or preferentially expand during the proliferative burst associated with β selection. Interestingly, sequencing studies have shown that TCR β-chain repertoires expressed on immature preselection DP thymocytes are not as diverse as expected (25, 45). It remains to be clarified whether such observations result from nonrandom TCR gene recombination, from preferential expansion of cells expressing particular β-chains at DN3-DN4, or by another mechanism.

Further work aiming to understand how thymocytes interpret TCR signals at different thresholds during distinct developmental stages and the mechanisms behind coreceptor-independent signaling will undoubtedly enhance our understanding of the fundamental processes involved in the selection of healthy and autoimmune T cell repertoires.

We acknowledge the help and advice of Dr. Maggie Millrain in tetramer staining.

Disclosures The authors have no financial conflicts of interest.