Public T Cell Receptor β-Chains Are Not Advantaged during Positive Selection¹ Free

The cornerstones of adaptive immunity, i.e., exquisite specificity, versatility, and long-term memory, all rely on the generation and survival of functional lymphocyte repertoires. The mechanisms behind the formation, selection, and maintenance of the naive T cell repertoire are currently poorly understood. V(D)J recombination is an integral part of thymic T cell differentiation, producing diverse repertoires of αβ and γδ T cell receptors on immature thymocytes (1). It has been estimated that the theoretical maximum size of the T cell repertoire is 10¹⁵ αβTCR (2), many orders of magnitude higher than the actual number of T cells (3).

Thymic selection events dramatically reduce the size of the repertoire by removing cells that express out-of-frame β-chains at the β-selection checkpoint (4, 5) and subfunctional or autoreactive TCR during positive and negative selection (6). It is not fully understood how these events mold the vast preselection repertoire into a functional naive peripheral repertoire. Extrapolation of molecular data based on the frequency of TCR rearrangements has estimated that the murine naive repertoire contains only 2 × 10⁶ clones of ∼10 cells each (7). This implies an extreme narrowing of the repertoire at selection and would be predicted to lead to significant differences in the functional repertoires of genetically identical individuals. In support of this, ∼70% of the TCR repertoires of two immunologically naive DBA/2 mice were found to be private to each individual. However, these two animals shared 12 TCR sequences of 215 analyzed from mouse A and 130 analyzed from mouse B (8), implying that although naive repertoires are mainly nonoverlapping, a “public” component does exist. Public TCR can be defined as TCR gene rearrangements present, often at high frequency, in naive or immune T cell repertoires of different individuals. Public TCR are commonly observed in Ag-specific repertoires in humans (9, 10) and in mice (11, 12, 13), reflecting the limited repertoire available for the recognition of specific peptide/MHC complexes. It has recently been shown that TdT^−/− mice retain the ability to raise public T cell repertoires against a number of CD4⁺ and CD8⁺ T cell epitopes (14). These authors also noted a striking absence of nontemplate nucleotide N additions in the CDR3 of published public TCR and suggested that either the recombination machinery produces public TCR more efficiently than private TCR or that selection events favor public TCR. In addition to this, it has been proposed that repertoires devoid of N additions such as those of neonates and TdT^−/− strains may be highly MHC reactive (15, 16) and that TCR lacking N nucleotides interact predominantly with MHC helices and less with peptide (17).

The CD8⁺ T cell response to the minor histocompatibility Ag HYD^bSmcy displays a conserved TCR profile (18). Several D^bSmcy-specific T cell clones derived from different mice of the same inbred strain used a TCR β-chain rearrangement identical with that of the prototypic B6.2.16 TCR (19). Each clone also coexpressed the Vα9 gene segment but used differing CDR3α rearrangements. Selection of Vα9-chains in B6.2.16 β-chain transgenic mice has been shown to be MHC dependent, and in peripheral recognition this TCR β-chain makes important contacts with cognate peptide/MHC (18). We hypothesized that public TCR rearrangements like the B6.2.16 β-chain interact favorably with selecting self-peptide/MHC compared with private β-chains, facilitating efficient coselection with a broad range of α-chains (18). Single cell PCR studies have shown that cells expressing TCR using the same ancestral β-chain rearrangement are selected with differing TCR α-chains (20). Following β-selection at the double-negative (DN) 3 stage, thymocytes undergo up to seven rounds of proliferation before commencing α-chain gene rearrangement (21), leading to extensive β-chain sharing by cohorts of immature thymocytes before positive selection. β-Chain rearrangements able to mediate selection with a broad range of different α-chains would therefore be over-represented in the naive T cell repertoire. Further, if the rearrangement occurred commonly such TCR may be public, being present in the repertoires of different individuals sharing the same MHC haplotype. Several reports have suggested that the naive TCR repertoire is further molded after selection and exit from the thymus by contact with self-peptide/MHC in peripheral tissues (22, 23). It has been suggested that this involves contact with the same complex that instructed selection in the thymus (24). The naive repertoire may therefore be significantly modified and fine-tuned by TCR-dependent homeostatic or survival mechanisms. We therefore hypothesize that some β-chains may also mediate an advantage at the level of “peripheral repertoire selection” and that this and/or favored thymic selection may explain the presence of public TCR gene rearrangements in T cell repertoires.

By means of a large-scale sequencing study, we show that public β-chains are detectable in wild-type naive T cell repertoires. These β-chains were isolated from H2^b and H2^k CD4⁺ and CD8⁺ T cell repertoires, implying selection on a wide range of MHC class I and II self-peptide/MHC complexes. In addition to this, we investigate the selection of public and private β-chains of interest by means of a novel bone marrow (BM) competition chimera assay and show that β-chains are selected into the naive peripheral repertoire with similar efficiency irrespective of CDR3β sequence. The presence of public β-chains in naive repertoires is therefore not due to favored thymic selection but is likely to be due to peripheral events, possibly mediated by specific interactions of the TCR. These data also suggest that thymic selection does not narrow the TCR repertoire based on β-chain sequence and therefore implies a dominant role for the TCR α-chain in T cell selection.

All Abs were obtained from BD Pharmingen. For tetramer staining, cells were incubated with 0.2 μl of PE-conjugated D^bSmcy tetramer (ProImmune) for 15 min at room temperature, washed, and subsequently costained with the appropriate Abs. Cells were washed and analyzed on a FACScalibur flow cytometer using CellQuest software (BD Biosciences).

C57BL/6 (B6)⁴mice were vaccinated i.m. with a DNA preparation encoding the HYD^bSmcy epitope and a tetanus toxin helper determinant (25). Three D^bSmcy-specific T cell lines were derived separately from three female mice and cultured as previously described (26).

All mice used in this study were housed in specific pathogen-free (SPF) conditions at Imperial College Hammersmith (London, U.K.) animal facility with appropriate Home Office and ethical authority. Female B6 and B10.BR mice aged 6–8 wk were obtained from Harlan Olac. TCRα^−/− thymocytes were used to obtain a preselection repertoire. B6.2.16 β-chain transgenic mice have been described previously (19). BM donors were young adult TCRβ^−/−TCRδ^−/− (βδ^−/−) on a FVB strain background. Recipients of transduced BM hemopoietic stem cells (HSC) were B6 RAG^−/− mice.

CD8⁺ and CD4⁺ splenocytes were purified using anti-CD8 or anti-CD4 Dynabeads (Dynal Biotech) according to the manufacturer’s instructions. Thymocytes from B6 mouse no.14 were stained with anti-CD8-FITC and anti-CD4-PE (BD Pharmingen) and subjected to FACS to obtain single-positive (SP) CD8 (SP8) and SP CD4 (SP4) thymocyte populations. Sorting was performed by the FACS facility at the Medical Research Council Clinical Sciences Centre, Hammersmith Hospital (London, U.K.) on a FACSVantage sorter (BD Biosciences).

Purified or sorted lymphocytes were subject to RNA extraction using TRIzol reagent (Invitrogen Life Technologies). cDNA was synthesized using SuperScript II RNase H⁻ reverse transcriptase (Invitrogen Life Technologies) and random hexamers (Amersham Biosciences) using 50–250 ng of RNA in a final volume of 60 μl.

Synthesized cDNA was amplified by PCR using the Vβ8.2F (5′-GGTGACATTGAGCTGTAAT-3′) and Jβ2.3R (5′-AGTCAGTCTGGTTCCTGAG-3′) primers (Sigma-Genosys). Purified PCR products were cloned into bacteria using the pCR2.1 vector in the TA cloning kit (Invitrogen Life Technologies). Further PCR were conducted on individual colonies again using the Vβ8.2F and Jβ2.3R primers, and products were sequenced using the FBV8.2S (5′-TTCATATGGTGCTGGCAGC-3′) primer. Vα9-Cα-chains were cloned and sequenced as previously described (18). Samples were analyzed separately using equipment reserved only for RT-PCR or PCR. PCR reagents and mixes were aliquoted and prepared under sterile conditions. Appropriate blanks were included at all stages.

The B6.2.16 β-chain was cloned from B6.2.16 β-chain transgenic splenocytes (product P1). The end primers B6F1 (forward: 5′-GGAGATCTACCACCATGGCTAACACTGCCTTCCCTG-3′) and B6R2 (reverse: 5′-CCGAATTCGCTCAGGAATTTTTTTTCTTGACCATGG-3′) overlapped with the B6.2.16 β-chain start and stop codon, respectively, and contained restriction enzyme sites (B6F1: BglII; B6R2: EcoRI). B6F1 and R1 (reverse: 5′-CACAGAAGTACACTGATGTC-3′) primers were used in PCR with P1 cDNA to generate product P2. In parallel, F2 (forward: 5′-ACGCTGTATTTTGGCTCAG-3′) and B6R2 primers generated product P4. Products P2 and P4 were designed to overlap with ∼280-bp sequences encoding the CDR3 region obtained during the sequencing study (products P0-APP or P0-GGG). P2 was hybridized to P0 and amplified using the primer pair B6F1 and Jβ2.3R, giving rise to product P3. Further, P3 and P4 were subjected to annealing and extension and amplified using the end primers B6F1 and B6R2, giving rise to a full-length β-chain with the desired CDR3 sequence. For these PCR, annealing temperatures of 54°C were used with 30 cycles. For annealing and extension reactions, an annealing temperature of 56°C was used and primers were added to the reaction mix after 10 of the 30 cycles were complete. β-Chains were cloned into pCR2.1 and sequenced using the internal primers SEQF1 (5′-TCATATGGTGCTGGCAG-3′), SEQF2 (5′-GTCTCTGCTACCTTCTG-3′), and SEQR (5′-GATGGCTCAAACAAGGAG-3′).

The MIGR1 retrovirus has been described elsewhere (27, 28). Full-length β-chains were excised from pCR2.1 and cloned into MIGR1. The GFP gene was removed from an aliquot of each construct by the digestion of 10 μg of DNA with HindIII to excise a single fragment encoding the entire GFP gene and part of the internal ribosome entry site. Compatible ends were closed using T4 ligase (Invitrogen Life Technologies). This provided two sets of MIGR1 constructs encoding the same β-chains either with or without the GFP gene.

Retroviral supernatant was produced using Phoenix packaging cells as previously described (29, 30, 31). Donor βδ^−/− mice were pretreated with 150 mg/kg 5-fluorouracil (Invivogen) 5 days before BM harvest. BM was harvested from long bones of several animals and pooled. Cells (2 × 10⁶) were plated per well in a 24-well plate in 2 ml of supplemented IMDM plus 20 ng/ml recombinant mouse IL-3, 10 ng/ml IL-6, 50 ng/ml stem cell factor, and 50 ng/ml recombinant human Flt-3 ligand (all from BioSource International) and cultured for 48 h at 37°C with 5% CO₂. Cells (1.5 × 10⁶) were then resuspended in 1.5 ml of viral supernatant with 8 μg/ml Polybrene (Sigma-Aldrich) per well in a 24-well plate and subjected to spin infection by centrifugation at 600 × g for 90 min at 25°C. Infected cells were cultured for a further 48 h with the growth factors listed above. Just before transfer into recipients, HSC were assessed for efficiency of transduction by intracellular staining with anti-Vβ8.1/8.2 Ab (BD Pharmingen) by using the eBioscience intracellular staining kit. Cell number and transduction efficiency were used to calculate the total number of transduced cells. Two HSC populations, each expressing a different β-chain, were adjusted to equal transduced cell number and were mixed. Each HSC mix was washed in PBS and divided for transfer into two recipients to provide pairs of chimeras, each expressing the same β-chains.

B6 RAG^−/− recipient mice were given 200 μg of anti-NK1.1 Ab (clone PK136) i.p. 48 h before the adoptive transfer of transduced HSC and sublethally irradiated by the administration of 600 rad 24 h later. The following day, the infected HSC mix was transferred into recipient mice by i.v. injection. Eight weeks after transfer, chimeras were analyzed by flow cytometry or TCR repertoire sequencing.

Several D^bSmcy-specific T cell clones derived from female mice immunized with male cells have previously been shown to use an invariant β-chain rearrangement with an identical nucleotide sequence to that of the prototypic B6.2.16 β-chain (18, 19). We wanted to investigate whether other D^bSmcy-specific T cells also used this gene rearrangement and to determine whether this is a case when Ag is administered via DNA vaccine rather than by traditional immunization with male cells. The p.DOM-Smcy/D^b DNA vaccine expresses a tripartite fusion protein (BCL₁ leader/N-terminal domain of tetanus toxin C fragment/Smcy epitope) (25). Three independent T cell lines (F2, F6, and M1) were derived from the spleens of three different B6 female mice given the DNA vaccine. All lines showed a striking bias toward the use of Vβ8.1/8.2 within the D^bSmcy-specific population (Fig. 1,A). TCR expressing the Vβ8.2 and Jβ2.3 gene segments were sequenced from each of the three lines. We found that virtually all (≥99% lines F2 and F6) or all (line M1) of the Vβ8.2-Jβ2.3⁺ TCR used the GDNSAETL rearrangement (Fig. 1,B), confirming that this β-chain is highly favored and also dominates the response stimulated by this form of immunization. Other CDR3β loops present at very low frequency in lines F2 and F6 were highly related to the B6.2.16 β-chain, differing by only one or two amino acids as underlined: GDNGAETL (line F2) and SDNGAETL (line F6). We reasoned that it would be likely that these TCR used the Vα9 gene segment as shown in the previously defined T cell clones (18). Because there are no Abs available to assess Vα9 expression, we sequenced TCR from the lines using primers hybridizing to Vα9 and the Cα gene segment by RT-PCR. All three lines were found to use Vα9 but used differing Jα segments and distinct CDR3α loops (Fig. 1,B), indicating that cross-contamination had not occurred between the lines. Variability was low within each line, suggesting that the lines were losing diversity in vitro. It seems that although limited in diversity the three lines had not become monoclonal because not all CD8⁺ cells expressed Vβ8.1/8.2 (Fig. 1 A) and not all Vβ8.2⁺ TCR used the B6.2.16 CDR3 rearrangement. Nevertheless, these data confirm that the B6.2.16 β-chain dominates the response to D^bSmcy and also indicate that this β-chain was rearranged and survived selection events in a further three mice in addition to those previously analyzed (18) and is therefore highly public.

We wanted to investigate the diversity of the unstimulated, naive peripheral T cell repertoire to determine whether public β-chains are detectable before Ag-driven expansion. We developed a large-scale RT-PCR, cloning, and sequencing strategy to analyze TCR repertoires with the Vβ8.2-Jβ2.3 gene rearrangement as used by the B6.2.16 β-chain. We fixed the V-J rearrangement to increase the likelihood of finding public rearrangements but did not limit the analysis to one CDR3 length, as we wanted to investigate a more global repertoire within the Vβ8.2-Jβ2.3⁺ population. CD4⁺ and CD8⁺ T cell repertoires were analyzed from one B10.BR (H2^k) and three B6 (H2^b) female mice. We also investigated the diversity of β-chain rearrangements from TCRα^−/− thymocytes. These cells do not express mature αβTCR and thus cannot undergo positive or negative selection, therefore provide a glimpse at the rearranged preselection TCR β-chain repertoire. Approximately 100 CDR3β sequences were analyzed per sample. Results showed that, as expected, thymic selection narrows the repertoire because the preselection repertoire contained fewer repeated sequences than the naive peripheral repertoires (Table I and Fig. 2). On average, 29% of the sequences in the postselection peripheral repertoires were repeats compared with 11% in the preselection repertoire. However, the existence of repeats in the preselection repertoire, with one sequence seen three times and five seen twice (Fig. 2), demonstrated that even before selection the diversity of the repertoire is limited. This mirrors an earlier study of Vβ2-Jβ2.2 TCR in TCRα^−/− thymocytes, where repeated sequences were seen even in a small sample (4). Generally, the B10.BR repertoires appeared more diverse than the respective B6 repertoires (Table I and Fig. 2), possibly implying differing extents of β-chain selection by different MHC molecules. Some samples, particularly the B6 CD8⁺ repertoires in mice no.10 and no.14, displayed several dominant sequences that appeared repeatedly. In each repertoire, the “dominant” β-chains were different, demonstrating that cross-contamination had not occurred between samples leading to over-representation of the same sequences in multiple repertoires. The B6.2.16 β-chain was not found among the naive H2^b repertoires despite being highly public in B6 CD8⁺ repertoires as revealed by the HY immunization analysis (Fig. 1). The repertoires all showed a mean/mode CDR3 length of 9–10 amino acids (Table I), and across all repertoires CDR3 length ranged from four to 16 amino acids (data not shown). The preselection repertoire displayed a statistically longer mean CDR3β length than any of the postselection repertoires (p < 0.05). This agrees with a previous report (32) where CDR3β rearrangement analyses in human double-positive (DP) and SP thymocyte and peripheral blood T cell populations showed that thymocytes with shorter CDR3 loops are selected from the DP to the SP stage and exported into the periphery.

To assess the effect of peripheral events on the naive repertoire, postselection splenocyte and SP4 and SP8 thymocyte Vβ8.2-Jβ2.3 repertoires were compared from the same animal (Fig. 3). Interestingly, the SP4 and CD4⁺ repertoires were very similar in terms of diversity (16 and 12% repeats, respectively), whereas the SP8 and CD8⁺ repertoires were very different (8 and 35% repeats, respectively) in complexity. When we compared the sequences in the SP4 vs CD4⁺ and the SP8 vs CD8⁺ repertoires, we found striking differences. The CD4⁺ repertoires shared six common sequences, displayed by the arrows labeled A–F (Fig. 3, A and C). In contrast, there were no common sequences in the CD8⁺ repertoires. This suggests differences in the way the CD4⁺ and CD8⁺ T cell repertoires are established and/or maintained.

Nine hundred and forty-seven CDR3β sequences from all 11 naive repertoires were collated and screened for overlapping sequences. Fifty-six sequences were found to be common to two or more repertoires, but because this included some found in the CD4⁺ and CD8⁺ repertoires of the same animal, public sequences were therefore defined as those appearing in three or more repertoires. The 23 public β-chains that met these criteria are shown in Table II. The sequences are grouped in order of ascending CDR3 length. As in the whole repertoires (Table I), most of the public CDR3β are 9–10 amino acids in length, suggesting that CDR3β loop length is not a vital factor in the selection of public Vβ8.2-Jβ2.3 β-chains. Some CDR3β sequences were common to defined populations, such as GDSAETL seen in 5/5 of the peripheral CD4⁺ repertoires, or GDAGGSAETL, seen in 3/3 peripheral CD8⁺ but not CD4⁺ B6 repertoires. These two chains may therefore be preferentially selected on MHC class II and H2^b class I molecules, respectively. Interestingly, most (19/23) of the public CDR3 sequences were shared by CD4⁺ and CD8⁺ cells, and almost half (11/23) were shared between H2^b and H2^k haplotypes (Table III). This implies a high degree of promiscuity during selection. CDR3β loops such as APPAGAETL are present in multiple repertoires and survived selection on H2^b and H2^k and by both MHC class I and II molecules. This particular CDR3β displays an unusual protein sequence, with a double proline motif encoded by an atypical gene rearrangement containing several N additions, Dβ1 (CAGGG), and/or an inverted Dβ2 (GCCCCCCC) gene segment (Fig. 4,A). Atypical recombination is likely to occur rarely, suggesting that this chain may be particularly advantaged to be represented so widely. Such public chains may mediate selection on features conserved in multiple self-peptide/MHC complexes, leading to an advantage and subsequent over-representation in several repertoires. Another highly public chain, GDWGGAETL, was seen in six repertoires and additionally in the preselection repertoire, possibly indicating preferential recombination and successful selection. Interestingly, this CDR3β rearrangement contains no N additions, being produced by a direct join of Vβ with DJβ (Fig. 4 B) as is GDAGGAETL, another common chain appearing in four repertoires. These rearrangements are similar to the TCR of MHC-reactive neonatal and TdT^−/− repertoires. The appearance of multiple public β-chains in naive repertoires lends weight to the hypothesis that some β-chains are advantaged during thymic and/or peripheral selection.

To directly assess the relative efficiencies of thymic selection of the β-chains identified during the sequencing study, a novel BM chimera competition assay was established. Full-length TCR β-chain cDNAs encoding the CDR3 regions of interest were constructed and cloned into the MIGR1 retroviral vector. Viral supernatants were then used to infect TCRβδ^−/− HSC, which were adoptively transferred into irradiated RAG-deficient B6 recipients. The use of βδ^−/− HSC limited β-chain expression to the transduced chain only while allowing development of an endogenous α-chain repertoire. Two populations of transduced βδ^−/− HSC expressing different β-chains were mixed (1:1 ratio) and transferred into recipients. After allowing T cell reconstitution for 8 wk, the proportion of cells expressing either β-chain in the thymocyte and peripheral T cell compartments was analyzed by cell sorting and TCR sequencing. The B6.2.16 β-chain rearrangement (Fig. 4,C) was used as a reference because it is known to be selected into the peripheral CD8⁺ T cell pool of all B6 mice analyzed (Fig. 1 and Ref. 18). Secondly, we used the APPAGAETL (APP) β-chain (Fig. 4,A), which was highly public in naive repertoires (Table II) despite arising from an unusual and presumably rare rearrangement. Finally, we chose a sequence seen twice in the preselection repertoire but not in any of the peripheral repertoires, GGGLGGRAETL (GGG, Fig. 4,D), representing a frequently rearranged chain that may be poorly selected in comparison. Chimeras were made where these β-chains were placed in competition in varying combinations. All chimeras were made in duplicate from the same BM mix (Table IV).

Initially we sought to establish whether this approach was technically viable for comparing the progress of two populations of transduced HSC through development to mature peripheral T cells. We constructed two pairs of chimeras expressing the B6.2.16 and APP β-chains or the B6.2.16 and GGG β-chains (chimeras 1–4). All four chimeras displayed DN, DP, and SP thymocytes and peripheral T cells 8 wk after transfer (data not shown). DN, DP, and SP4 thymocyte and CD4⁺ splenocyte populations were sorted from each chimera and subjected to TCR sequencing analysis. Both TCR β-chains were represented in all T cell subsets examined, derived from the transduced populations BM1 or BM2 (Fig. 5). In chimeras 2 (Fig. 5,B) and 4 (Fig. 5,D) there was a roughly 1:1 ratio of the two β-chain sequences at the DN stage, indicating that the method allowed successful reconstitution of the immature thymocyte compartment with cells expressing either B6.2.16 and APP or GGG β-chains. Chimeras 1 (Fig. 5,A) and 3 (Fig. 5,C) showed a bias at the DN stage toward cells expressing BM1 (B6.2.16β in both cases), although this must be due to recipient variation because chimeras 1 and 2 and chimeras 3 and 4 were duplicates (see Table IV) and therefore received the same mix. Because the DP cells in these chimeras constituted a mixture of preselection and postselection thymocytes, we concentrated on the ratios of the two β-chains at the DN (preselection) stage and subsequently at the SP4 (postselection) stage to investigate whether cells expressing one β-chain were advantaged during thymic selection. Remarkably, in all four chimeras the proportion of cells expressing either β-chain remained reasonably constant. For example, in chimera 2, 48% of the TCR sequenced from the DN subset derived from BM1 (B6.2.16β⁺) and 43% of the sequences from the SP4 population used this β-chain. Furthermore, there was little change in the CD4⁺ splenocyte subset, where 37% of the TCR used the B6.2.16 β-chain in chimera 2 (Fig. 5 B). The same general pattern was also seen in the other three chimeras. These data indicate that individual β-chains direct thymocyte selection into the mature SP4 and CD4 compartments with similar efficiencies.

To confirm these results and to further investigate the successive stages of T cell development in competition chimeras, a further eight chimeras were constructed as four pairs. To avoid the laborious process of cell sorting, TCR cloning, and sequencing, we devised a GFP^+/− readout system where one HSC population expressed GFP (BM1) while the other (BM2) did not. Because the cells were obtained from TCRβδ^−/− donors and the recipient mice were RAG deficient, only transduced cells could develop into mature T cells and any GFP⁻ T cells would derive only from the BM2 population. We placed varying combinations of the B6.2.16, APP and GGG β-chains in competition and included two chimeras where both transduced populations expressed B6.2.16β as a control for artifacts mediated by the introduction of GFP into the system.

In all eight GFP^+/− chimeras the DN, DP, and SP populations were observed in the thymus 8 wk after reconstitution, again indicating that the experimental approach recapitulated normal T cell development (Fig. 6,A). The SP4 and SP8 populations were observed at a normal ratio, indicating that the experimental system produced T cells and that selection of TCR was occurring on both MHC class I and II (Fig. 6,B). All but one chimera (number 11) displayed a large DN subset and small DP subset (Fig. 6 B), probably indicating a partial block at the DN to DP transition likely to be due to the presence of host RAG^−/− and nontransduced donor-derived DN1–3 thymocytes.

Preselection DP and postselection SP thymocytes were analyzed for the presence or absence of GFP and therefore the expression of β-chain 1 or β-chain 2 (Fig. 7). Preselection thymocytes were defined as TCR^lowCD4⁺CD8⁺ to exclude host RAG^−/−GFP⁻ DN1–3 thymocytes and positively selected TCR^high DP cells from the analysis (R3; Fig. 7,C). Chimeras 5 and 6 received BM transduced with B6.2.16βGFP⁺ and B6.2.16βGFP⁻ as a control. Fig. 7,D shows that there was ∼a 1:1 ratio of GFP⁺ to GFP⁻ cells in the defined preselection subset of chimera 5, reflecting the input ratio and indicating that cells transduced with both constructs fared equally well in thymic reconstitution, mirroring the sequencing results from chimeras 2 and 4. However, this was not the case for partner chimera 6, which received exactly the same mix (Table V). Here, virtually all TCR^low DP cells (99.6%) were GFP⁻, indicating a severe bias toward cells deriving from the BM2 population. Similar results were seen in the experimental chimeras 7–12, where the TCR^low DP subset was dominated by one population (Table V). This was not due to an artifact resulting from the expression of GFP in developing thymocytes, as the dominant population expressed GFP in four of seven chimeras.

Despite the heavy biasing preselection, the “disadvantaged” population recovered into the SP subsets (Table V), where both GFP⁺ and GFP⁻ Vβ8.2⁺ SP8 (Fig. 7, E and F) and SP4 (Fig. 7, G and H) thymocytes were present in all chimeras, demonstrating that both β-chains mediated selection. Generally, the GFP⁺SP4 and GFP⁺SP8 populations were smaller than the equivalent GFP⁻ population irrespective of the companion β-chain. This would seem to indicate that GFP expression may have some influence on thymocyte development or survival. Because GFP^+/− representation in the SP populations was not strongly influenced whether the same (chimeras 3 and 4) or different CDR3β loops (chimeras 5–8) auditioned for selection, we conclude that CDR3 composition was not a strong influence on the outcome of thymic selection, confirming the results from the non-GFP chimeras (1, 2, 3, 4).

The GFP^+/− competition chimera data were then analyzed to determine whether particular β-chains were advantaged in directing selection into the SP4 or SP8 subsets. For each β-chain in each chimera (n = 16), the ratio of the percentage of cells expressing the chain within the SP4 and SP8 compartments was calculated and expressed as a proportional skew to either lineage (Fig. 8). This is a very stringent analysis, because bias will be introduced if one or both of the chains are preferentially selected into either lineage. Further, the analysis is largely independent of differences in the relative proportions of transduction, as the comparisons are within the GFP⁺ and GFP⁻ compartments. Strikingly, cells expressing any one of the three β-chains show a marked absence of skew to any lineage. The few outliers seen arise from distortion attributable to low rates of transduction.

The proportion of GFP⁺ to GFP⁻ cells was analyzed in Vβ8.2⁺CD4⁺ and Vβ8.2⁺CD8⁺ populations in chimera spleens (Table V). T cells expressing both β-chains were present in the spleen, indicating that cells derived from each BM population were successfully recruited into the periphery. Cells with the B6.2.16, APP, or GGG β-chain were therefore able to migrate into the periphery and receive the necessary homeostatic signals for survival.

We noted that in most chimeras the proportion of GFP⁺ cells was higher in splenocyte T cell subsets compared with the equivalent SP thymocyte population. This indicates that GFP expression does not inhibit the proliferation capabilities of peripheral T cells but may further suggest that GFP can affect cell population dynamics. Nevertheless, the purpose of the competition chimera experiments was to investigate whether the “public” β-chains found in naive repertoires are strongly advantaged during thymic selection. This is clearly not the case, because thymocytes expressing such β-chains (APP) were not advantaged during selection into either the CD4⁺ or the CD8⁺ T cell lineage, and cells expressing each β-chain successfully colonized peripheral tissues.

Use of the B6.2.16 TCR β-chain rearrangement in the response to D^bSmcy represents an unusual phenomenon by being dominant in 5/5 individual animals that we (this report and Ref. 18) and others (19) have examined, demonstrating a remarkable conservation. The absolute invariance of the β-chain sequence indicates that this rearrangement, partnered with Vα9, is extremely important for optimal interaction with D^bSmcy and subsequent signaling. The B6.2.16 β-chain was raised when mice were immunized with male cells or with a DNA vaccine encoding the D^bSmcy epitope coupled to the C terminus of the C fragment of tetanus toxin (25), despite the fact that the Smcy peptide originated from different cellular and polypeptide contexts. These data show that the mode of Ag delivery does not affect the TCR profile of this response. The B6.2.16 β-chain rearrangement has only one site of diversification (V-J) because it does not contain a diversity segment (Fig. 4 C). The N addition at this junction encodes for the third residue (asparagine) of the CDR3 loop. Of interest, independent rearrangements use the same codon (AAC) and not the alternative (AAT) residue, suggesting that the recombination machinery may operate in a biased fashion as previously suggested (33, 34). The B6.2.16 rearrangement may arise from a direct V-J joining event, although this is a rare event (35). Alternatively, an inserted D segment could have been removed by nucleotide deletion, but again this is likely to occur infrequently. The B6.2.16 rearrangement was not sufficiently frequent in any of the naive peripheral repertoires to be detected, consistent with this unusual structure. We may have detected this rearrangement in naive peripheral repertoires had we restricted our target repertoire and concentrated only on sequencing a CDR3 of eight amino acids in length. The B6.2.16 rearrangement is thus an example of a rare but also public β-chain. In contrast, the public chains detected here in multiple naive repertoires have high frequencies, being found in relatively small sample sizes (∼100).

The Vα9 chains used by the D^bSmcy-specific T cell lines described here are all type I α-chain rearrangements as defined by Ferreira and colleagues (18) and are therefore likely to make limited contact with peptide. The three Jα segments used by the lines, Jα45, Jα56, and Jα57 are located in an area of high frequency of recombination with Vα9 (36) and thus are likely to be common in the preselection repertoire. The use of such α-chain rearrangements in these postselection repertoires and the reported coselection of a broad range of α-chains with B6.2.16β (18, 37) lends weight to our previous work showing that the Vα9-chain CDR3 loop does not play a major role in the selection of TCR using the B6.2.16 β-chain on H2^b molecules (18).

The presence of the B6.2.16 β-chain in the Ag-specific repertoires of all mice examined suggested that the individual repertoires were highly related. We sought to investigate whether the preimmune repertoires of genetically identical or distinct mice were related. By means of a large-scale sequencing study, we identified public, shared β-chains in naive peripheral T cell repertoires. We cannot formally rule out an amplification bias of particular CDR3β rearrangements. However, this seems extremely unlikely because we concentrated on a single Vβ-Jβ rearrangement, allowing us to focus on a reasonable subset of the TCR repertoire, and the oligonucleotide primers were located well outside the CDR3 region, where the products differ. Dominant β-chains accounting for upwards of 5% of an individual Vβ8.2-Jβ2.3 repertoire were distinct in each sample; therefore, one chain did not consistently dominate several repertoires at high frequencies. Further, the preselection and thymic SP repertoires did not contain sequences repeated frequently, indicating that the method can represent diverse repertoires and is therefore not inherently biased. The identification of highly related public sequences, such as those using the GDAGGXX motif, and the low frequencies of public chains in individual repertoires (most are seen once only) indicate that these data represent an accurate picture of the naive T cell repertoire. Finally, our data broadly agrees with previous studies of preselection (4) and naive repertoires, where repeated sequences were also identified in spleen (7, 8, 38). We also found that mean CDR3β loop length was generally shorter in peripheral repertoires compared with the preselection repertoire (Table I), in agreement with previously published data (32). Interestingly we also found that selection shortened the CDR3β loops of SP4 cells but not SP8 cells (data not shown), also in agreement with a more recent report from Yassai and colleagues (39).

The sequencing study allows us to divide common Vβ8.2-Jβ2.3 β-chain rearrangements into two groups: dominant β-chains, which are those seen at high frequencies within a given individual repertoire, and public β-chains, which are common to distinct repertoires of different individuals. The dominant β-chains are common in one repertoire but not seen in others. The absence of dominant β-chains in SP repertoires suggests that the survival and maintenance of T cells expressing such β-chains are determined largely by peripheral mechanisms. It could be argued that these may include Ag-driven expansion in response to environmental Ags, including nonpathogenic microorganisms and commensal flora. However, comparison of repertoires from SPF and germfree animals has suggested that the naive T cell repertoire is not substantially modified by the presence of commensals (38). Indeed, it is unlikely that the naive wild-type nonimmunized SPF mice used in this study made responses to environmental Ags using TCR Vβ8.2-Jβ2.3 rearrangements consistently during the extended period of repertoire sampling, especially as some public β-chains were not restricted to the MHC isotype or haplotype. Therefore it seems that the over-representation of some β-chains in peripheral T cell repertoires vs SP thymocyte repertoires is due to homeostatic mechanisms and competition for survival signals. This in turn implies that signals via the specific TCR are involved, presumably by contacting self-peptide/MHC, which would amount to an added level of postthymic repertoire selection. Several reports have suggested that peripheral T cell survival depends on contact with peptide/MHC as well as with lymphokines and cytokines (24, 40, 41, 42). Notably, naive CD8⁺ and CD4⁺ T cells have been shown to survive as resting cells that only cycle very slowly (22). Contact with peptide/MHC via TCR on peripheral naive cells can therefore promote survival rather than activation/expansion signals and may explain the over-representation of particular β-chains in naive repertoires if those TCR preferentially provide survival signals. In addition to this, TCR affinity has been shown to regulate naive T cell homeostasis, where TCR with higher affinities for peptide/MHC give rise to a survival advantage (43). Our data favor the hypothesis that some β-chains mediate a selective advantage through preferential interactions with self-peptide/MHC.

The intriguing similarities between CD4⁺ and SP4 repertoires and the striking difference between CD8⁺ and SP8 repertoires further suggests that postthymic modification of the repertoire differs depending on the expressed coreceptor. In most mouse strains, CD4⁺ cells outnumber CD8⁺ cells by several fold, implying that the formation of the two repertoires differs. It has been previously suggested that selection and differentiation of CD4⁺ cells is a “default” pathway (44), that CD8⁺ cells are selected by weaker signaling through the TCR (45), and that a shorter duration of TCR signaling encourages the development of CD8⁺ cells (46). Others have shown that coreceptor signaling events influence lineage choice. CD4 binds Lck with a greater avidity than CD8 (47), and Lck is particularly important in the selection of CD4⁺ cells (48, 49), again suggesting differing pathways for the establishment of these T cell subsets. The more stringent requirements for CD8⁺ T cell selection operating in the thymus may continue in the periphery, influencing the degree of peripheral expansion/survival and resulting in a narrower, optimized repertoire. Interestingly, studies of postselection thymocyte kinetics have shown that ∼3-fold more CD8⁺ than CD4⁺ thymocytes incorporated BrdU 30 min after pulsing (50). This indicates that CD8⁺ cells proliferate at a greater rate, further suggesting that peripheral CD8⁺ and CD4⁺ repertoires are differentially modified.

To investigate whether the appearance of public β-chains in naive repertoires was due to preferential thymic selection, we established a novel BM chimera competition assay to track the selection of competing β-chains. By expressing β-chains of interest in TCRβδ^−/− HSC and reconstituting RAG^−/− recipients, we ensured that αβ T cells could only develop from transduced HSC. The input of HSC was calibrated to achieve a 1:1 starting ratio of the two transduced populations. The consistent appearance of SP thymocytes and peripheral T cells in all chimeras indicated that transduced HSC differentiated normally into the T lymphocyte lineage. The use of a GFP^+/− system to identify cells expressing either β-chain provided a simple basis for the evaluation of relative selection efficiency, although the data gave rise to some concern as to whether the expression of GFP affects cell viability. However, the presence of GFP⁺ mature SP and peripheral T cells indicated that GFP⁺ cells were not dying or suffering from a major proliferation defect. Interestingly, we noted that Vβ8 expression was lower on the SP4, SP8, CD4⁺, and CD8⁺ cells that expressed GFP compared with those that did not (Fig. 7, F and H, and data not shown). This suggests that T cells express lower levels of TCR when GFP is present, which may influence TCR signaling thresholds and could therefore alter the capacity of a GFP⁺ cell to receive selection or survival signals. However, this did not confound the outcome of the experiment. Chimeras 1–4 were made without GFP, and all three β-chains survived selection into CD4⁺ and CD8⁺ T cell repertoires. These data were confirmed by the construction of GFP^+/− chimeras, where in all cases cells expressing both β-chains were present in mature T cell subsets despite the bias apparent in the preselection DP subset.

In all 12 chimeras, HSC expressing either construct differentiated into mature peripheral T cell populations, indicating that in all cases cells expressing B6.2.16, APP, or GGG β-chains survived selection and migrated into the periphery. In chimeras 1–4, the B6.2.16 β-chain was most consistently represented in all subsets, which was unsurprising as we know that this chain is selected by the H2^b haplotype. However, the proportion of cells deriving from BM1 or BM2 was consistent from DN to SP4 in these chimeras, indicating that selection did not bias the repertoire based on the β-chain CDR3 sequence. However, the DP subset was noticeably biased, which was mirrored by the intriguing observation in the GFP^+/− chimeras where DP cells were either dominated by GFP⁺ cells (chimeras 8, 9, 10, and 12) or GFP⁻ cells (chimeras 6, 7, and 11) irrespective of β-chain and not consistently between pairs of chimeras (chimeras 5 and 6, chimeras 7 and 8, etc.). This may imply that the DP population is subject to different population kinetics compared with other thymocytes and suggests that this is not an artifact dependent on the expression of GFP. However, the aim of the experiment was to determine whether cells expressing a public β-chain had an advantage during thymic selection over other “less public” or private chains. We conclude that this is not the case but rather that selection into the mature thymocyte population occurs with similar efficiency irrespective of the TCR β-chain. Of course, these data do not rule out the possibility that certain β-chain rearrangements would be disadvantaged during positive selection, and we tested three representative chains only. Sequential TCR α-chain rearrangement and audition for selection is likely to compensate for variable contributions to peptide/MHC binding from individual β-chains. Indeed the importance of sequential α-chain revision has been recently reported (51). Our data imply that thymocyte selection has evolved so that many different β-chains can mediate selection into the mature CD4/CD8 compartments, which in turn implies that the TCR α-chain has an important role in determining peptide/MHC specificity. The first stage in the development of the αβ TCR repertoire is the formation of an in-frame β-locus rearrangement and the subsequent expression of the pre-TCR, which signals to trigger extensive proliferation. This investment is made without any “quality control” of β-chain rearrangements and would be wasteful if the CDR3β composition had a strong influence on “selectability.” Indeed, these data suggest that CDR3β composition is not a major influence on thymic selection. This notion is further strengthened by the analysis of the equivalence of selection of different β-chains into the SP4 and SP8 subsets (Fig. 8).

Crystal structures of TCR:peptide/MHC complexes have demonstrated a dominant role for the Vα domain in peptide/MHC recognition (52, 53, 54, 55). Furthermore, the same β-chain variable segment is differently positioned on the same peptide/MHC ligand when paired with a different α-chain (54). With this structural insight, it is tempting to speculate that DP thymocytes sharing identical β-chains are functionally distinct because of the dominant role of their distinct α-chains in positioning of the TCR on peptide/MHC. Distinct architectures of each αβ TCR within β-clonal cohorts may broaden specificity and underlie the observed equivalence when auditioning for positive selection. Unraveling the relative contributions of the TCR α- and β-chains to peptide/MHC recognition is likely to be an important aspect in the understanding of TCR selection and Ag recognition.

Our data suggest that an adequate β-chain may be enough to sustain selection. If thymic selection does not favor the maturation of cells expressing public β-chains, we propose that such cells possess enhanced survival capacities, perhaps by making preferential contact with self-peptide/MHC in peripheral tissues. This work opens questions regarding the extent to which the naive repertoire is modified following thymic selection, the definitive stage in the development of the T cell repertoire.

We thank Prof. Adrian Hayday and Colin Roberts for the βδ^−/− mouse strain, Zoe Webster for rederivation, Dr. Shao-An Xue for help in establishing retroviral transduction, Prof. Hans Stauss for providing Phoenix cells, Dr. Jian-Guo Chai for performing i.v. injections, and Dr. Shohei Hori for the MIGR1-GFP retroviral vector.

The authors have no financial conflict of interest.