β-Selection: Abundance of TCRβ–/γδ– CD44–CD25– (DN4) cells in the foetal thymus

Expression of TCRβ and pre-TCR signalling are essential for differentiation of CD4–CD8– double negative (DN) thymocytes to the CD4+CD8+ double-positive (DP) stage. Thymocyte development in adult Rag1, Rag2 or TCRβδ-deficient mice is arrested at the DN3 stage leading to the assumption that pre-TCR signalling and β-selection occur at, and are obligatory for, the transition from DN3 to DN4. We show that the majority of DN3 and DN4 cells that differentiate during early embryogenesis in wild-type mice do not express intracellular (ic) TCRβ/γδ. These foetal icTCRβ−/γδ− DN4 cells were T lineage as determined by expression of Thy1 and icCD3 and TCRβ DJ rearrangement. In addition, in the foetal Rag1–/– thymus, a normal percentage of DN4 cells were present. In wild-type mice after hydrocortisone-induced synchronisation of differentiation, the majority of DN4 cells that first emerged did not express icTCRβ/γδ, showing that adult thymocytes can also differentiate to the DN4 stage independently of pre-TCR signalling. Pre-TCR signalling induced expansion in the DN4 population, but lack of TCRβ/γδ expression did not immediately induce apoptosis. Our data demonstrate in vivo differentiation from DN3 to DN4 cell in the absence of TCRβ/γδ expression in the foetal thymus, and after hydrocortisone treatment of adult mice.


Introduction
Signalling through the pre-T cell receptor (TCR) is essential for differentiation of CD4 -CD8double-negative (DN) thymocytes to the CD4 + CD8 + double-positive (DP) stage [1,2]. The DN population can be sub-divided into four subsets, DN1-DN4, depending on the expression of the cell surface molecules CD44 and CD25. The earliest CD44 + CD25 -(DN1) cells give rise to the CD44 + CD25 + (DN2) population that progresses to the CD25 + CD44 -(DN3) stage, and then to the CD44 -CD25 -(DN4) population [3]. Expression of rearranged TCRb and formation of the pre-TCR complex leads to a series of events collectively known as b-selection, resulting in allelic exclusion of the TCRb locus, expansion and differentiation to the DP stage [4]. Several studies have indicated that thymocytes that successfully pass the b-selection checkpoint receive pre-TCR induced signals for proliferation [4][5][6] and survival [7,8]. Recently, BCL2A1 has been identified as an antiapoptotic agent that is up-regulated following pre-TCR signalling [9].
Rearrangement of the TCRb locus and expression of a functional TCRb chain are essential for differentiation, and the development of thymocytes in adult mice deficient in Rag1 [10], Rag2 [11] or TCRbd [12] is Correspondence: Tessa Crompton, Division of Cell and Molecular Biology, Imperial College London, Sir Alexander Fleming Building, South Kensington Campus, London SW7 2AZ, UK Fax: +44-207-5842056 e-mail: t.crompton@imperial.ac.uk arrested at the DN3 stage. The fact that T cell development in these mice is arrested at the DN3 stage has lead to the assumption that pre-TCR signalling and b-selection occur at the transition from DN3 to DN4 and are obligatory for differentiation beyond the DN3 stage of development. Contrary to this assumption, rare DN4 cells that do not express TCRb or cd have been detected in adult wild-type mice [7]. As cell death is increased in this TCRb -/cd -DN4 population, it has been suggested that cells that fail b-selection die at the DN4 stage [7]. Here we show that in the wild-type foetal thymus differentiation from DN3 to DN4 frequently occurs independently of TCRb/cd expression (and hence normal pre-TCR signalling).
In addition, in the foetal Rag1 -/thymus a normal percentage of DN4 cells were present, although no DP cells were observed. In the foetal thymus, T cell development occurs in a largely synchronized wave [13], allowing the study of thymocyte populations as they first emerge, and enabling the definition of requirements for the transition of cells from one population to the next. After synchronizing adult thymocyte development by hydrocortisone treatment, the majority of DN4 cells that emerge in vivo in the adult thymus do not express intracellular (ic) TCRb/cd.

Results
The majority of early foetal DN4 thymocytes do not express icTCRb or cd Analysis of TCRb expression in foetal thymocytes revealed a large population of CD3 -CD4 -CD8triple negative DN4 cells (CD44 -CD25 -) that did not express detectable icTCRb. We analysed the intracellular expression of TCRb in DN3 and DN4 cells of embryonic day (E) 15.5, E16.5 and adult wild-type thymi. DN3 and DN4 subsets that fell within the live FSC/SSC gate were identified by positive staining for anti-Thy1.2, presence or absence, respectively, of cell surface expression of CD25 and the absence of CD44, CD3, CD4, CD8 expression (Fig. 1A). In a typical experiment, on E15.5, 12.66% of the DN3 cells and 15.23% of the DN4 cells expressed icTCRb (Fig. 1B). The percentage of icTCRb + cells increased in both DN3 and DN4 subsets on E16.5 and in a typical experiment 18.41% of the DN3 and 54.59% of the DN4 cells expressed icTCRb (Fig. 1C). On E16.5, approximately 12% of icTCRb -DN4 cells stained positively for icTCRcd expression (Fig. 1C). Thus, 40% of DN4 cells on E16.5 did not express either icTCRb or TCRcd. As previously shown, in the adult thymus, 18.49% of DN3 and 90.45% of DN4 cells expressed icTCRb (Fig. 1D).
The fact that the majority of the E15.5 foetal DN4 thymocytes were icTCRbwas surprising as, in contrast to the adult thymus [5,7,14,15], in the foetal E15.5 thymus there was no enrichment for icTCRb + cells in the DN4 population compared to the DN3 subset ( Fig. 1), suggesting that the foetal TCRb + DN3 cells did not preferentially differentiate to the DN4 stage compared to the foetal TCRb -DN3 cells. Analysis of icCD3 expression confirmed that the icTCRb -DN4 cells were T lineage cells. Both icTCRband icTCRb + DN4 cells expressed icCD3 and 54% of the icTCRband 90% of the icTCRb + DN4 cells expressed icCD3 at high level (Fig. 1E).
Given the reduction in the proportion of icTCRb -DN4 cells present in the thymus from E15.5 to E16.5 to adult, it seemed likely that icTCRb -DN4 cells are gradually removed from the thymus from E15.5 onwards. We therefore calculated the absolute numbers of icTCRband icTCRb + DN4 cells in the thymus on E15.5, E16.5 and in adults. We found that the number of icTCRb -DN4 thymocytes in fact increases over time but to a lesser extent than the number of icTCRb + DN4 cells, accounting for the reduction in overall percentage (Fig. 1F).

DN4 cells develop in Rag1 -/foetal thymus
The presence of icTCRb -DN4 cells in the first wave of differentiation of wild-type foetal thymocytes led us to ask whether DN4 cells are present in Rag1 -/foetal thymi. We analysed Rag1 -/foetal thymi at E15.5, E16.5, E17.5, thymi at 3 weeks postnatally (juvenile) and adult thymi and observed a substantial DN4 population on all embryonic days tested ( Fig. 2A, B), whereas the percentage of DN4 thymocytes was extremely small in the adult thymus of Rag1 -/mice as previously reported [10]. In a typical experiment, 31.12% of the thymocytes were DN4 on E15.5 compared to only 0.76% of the adult thymocytes. The percentage of DN4 thymocytes gradually decreased over time until they were virtually undetectable in the adult thymus (Fig. 2B). Correspondingly, the percentage of DN3 cells increased substantially from E15.5 to E16.5 and thereafter the thymus consisted mainly of DN3 thymocytes (Fig. 2B). As expected, no TCRb expression or DP thymocytes were observed at any stage of embryonic or postnatal development in Rag1 -/mice (data not shown). Although the percentage of Rag1 -/-DN4 thymocytes gradually decreased over time after E15.5, the absolute number of DN4 cells did not change appreciably, whereas the number of DN3 cells increased (Fig. 2C).
The Rag1 -/-DN4 cells observed, as defined by the absence of CD44 and CD25 cell surface markers, were true T lineage DN4 cells as the majority stained positive for cell surface expression of HSA (94.22%) and expressed intracellular CD3 (Fig. 2D). In contrast, in  a typical experiment only 0.42% stained positive with anti-NK1.1 and 8.58% with anti-B220 antibodies (Fig. 2D).

Transcription of Rag1 and Rag2 is down-regulated in the foetal DN4 population
It has been shown that transcription of Rag1 and to a lesser extent Rag2 is down-regulated in adult DN4 cells relative to DN3 cells [16]. To test whether downregulation of the Rag genes in the DN4 relative to the DN3 population depended on the expression of TCRb/ cd, we analysed the transcription of Rag1 and Rag2 by quantitative RT-PCR in DN3 and DN4 thymocytes sorted from E15.5, E16.5 and adult wild-type thymi. Rag1 transcripts were two to four times less abundant in DN4 thymocytes compared to DN3 cells (Fig. 3A, left panel). More than 80% of DN4 thymocytes on day E15.5 did not express TCRb, suggesting that down-regulation of Rag1 in the DN4 subset on day E15.5 was independent of pre-TCR signalling. Rag2 transcription was also significantly decreased in foetal DN4 thymocytes compared to DN3 cells but the difference in Rag2 expression in the adult thymocyte population was not significant (Fig. 3A, right panel). When foetal and adult Rag1 -/thymocytes were treated with antiCD3 antibody for 24 h to mimic the pre-TCR signal, down-regulation of Rag2 transcription was even greater (6-12-fold) compared to down-regulation of Rag2 in foetal DN4 wild-type thymocytes (Fig. 3B). The latter observation suggests that, although pre-TCR was not necessary for down-regulation of the Rag genes at the DN4 stage, a strong signal can accelerate downregulation of Rag2.

D to J TCRb rearrangement in TCRb -/cd -DN4 thymocytes
To further assess the origin and lineage commitment of the TCRb -/cd -DN4 cells we analysed the extent of their DJ TCRb rearrangement. We sorted DN4 TCRb -/cdand TCRb/cd + cells from E16.5 foetal wild-type thymi (Fig. 3C, D) and performed a PCR-Southern assay to compare the extend of DJ TCRb rearrangement in the two cell types, as previously described [17]. We amplified the TCRb locus from equal amounts of DNA template from sorted TCRb -/cdand TCRb/cd + DN4 cells using Db2-and Jb2.7-specific primers. We observed similar level of DJ rearrangement in both cell types (Fig. 3F, G) demonstrating that the TCRb -/cd -DN4 cells are T lineage committed and suggesting that they originate from DN2/DN3 cells.
We also tested the possibility that the percentage of DN4 TCRb -/cdcells decreased due to active apoptosis in this subset. We analysed the intracellular expression of the active form of caspase-3, as it has recently been shown that pre-TCR up-regulates the anti-apoptotic gene BCL2A1 that rescues cells from caspase-3-mediated apoptosis [9]. We were able to detect induction of apoptosis in control, adult DP cells treated for 2 h with antiCD3 ( Fig. 4B), but the active form of caspase-3 was virtually absent in both icTCRb -/cdand icTCRb + foetal DN4 thymocytes (Fig. 4C) and among DN3 thymocytes (data not shown). In contrast in the adult thymus we were able to detect more apoptotic cells amongst the icTCRb -/cd -DN4 subset compared to TCRb + DN4 subset. In a typical experiment 2.91% of icTCRb -/cd -DN4 cells were caspase-3 positive compared to 0.25% of the icTCRb + DN4 cells (Fig. 4C). Our data indicate that in the foetal thymus, clonotypic TCRb expression promotes proliferation of DN thymocytes. Lack, however, of TCRb expression did not immediately induce cell death in foetal DN4 cells, as evidenced by the absence of active caspase-3-expressing cells in the DN4 icTCRb -/ cdpopulation. The increased apoptosis that we observed in the adult DN4 icTCRb -/cdpopulation might be attributed to the accumulation of ageing cells in this subset.

Abundance of adult TCRb -/cd -DN4 cells after hydrocortisone treatment
Differentiation of foetal thymocytes occurs in a largely synchronized wave. In adults, however, thymocyte production is continuous and the adult thymus has reached steady state with pre-existing cells at all stages of development and homeostasis between thymocyte populations. The difference in the expression of TCRb in the DN4 subset that we observe between foetal and adult thymocytes could be due to a foetal restricted capability of thymocytes to differentiate to the DN4 stage Histogram (E) shows the quantification of the TCRb -/cdand TCRb/cd + DN4 thymocyte DNA preparations as assessed by real time PCR. Southern hybridisation was performed on PCR products amplified using specific Db2 and Jb2.7 primers from equal amounts of DNA preparations from each thymocyte subset (F). The probe used for the southern hybridisation corresponded to the full-length, germline (GL) TCRb locus. Quantification of the extent of Db2-Jb2.7 rearrangement relative to the TCRb -/ cdthymocytes, calculated for each cell type as the ratio of the intensity of the band corresponding to the GL sequence divided by the total intensity of the bands corresponding to the six rearranged TCRb, is also shown (G).
independently of normal pre-TCR signalling, either as a result of differences between foetal and adult hematopoietic progenitors [22,23] or between the foetal and adult thymic microenvironment [24]. Alternatively, it is possible that adult TCRb/cdthymocytes differentiate to the DN4 stage as efficiently as foetal thymocytes but that their percentage is very low due to the pre-existing high number of proliferating TCRb + DN4 thymocytes that mask their presence. To distinguish between these two hypotheses, we synchronized the differentiation of immature thymocytes by treating adult mice with hydrocortisone (HC). HC treatment induces apoptosis of immature DP thymocytes [25] but mature single positive cells survive treatment and the thymus then grows exponentially to recover its normal size and subset distribution a week following the treatment. HC enabled us to study the differentiation of thymocytes in the adult thymus in a synchronized wave and in a rapidly expanding thymus that had not yet reached homeostasis. At 3 days after HC treatment, the thymus contained only 1.27% DP thymocytes (Fig. 5A). Analysis of intracellular expression of TCRb in DN3 and DN4 cells, identified by the absence of surface expression of CD3, CD4, CD8, CD44 and B220 markers, the presence of Thy1.2, and presence or absence, respectively, of CD25 (Fig. 5B) showed that the percentage of cells expressing TCRb was equivalent between DN3 (11.59%) and DN4 (10.13%) populations (Fig. 5C, top panel). At 4 days after HC treatment, the percentage of icTCRb + cells in the DN4 population significantly increased to 48.49% compared to 13.32% of DN3 cells (Fig. 5C, middle panel). At 7 days after treatment, the percentage of TCRb + DN4 cells had reverted back to the percentage observed in non-treated wild-type adult mice with typically 15.37% of DN3 and 90.02% of DN4 cells staining positive for icTCRb + (Fig. 5C, lower panel), and 90.36% of thymocytes were CD4 + CD8 + DP, indicating that the thymus had fully recovered (data not shown). Although the absolute number of DN3, DN4 TCRb -/cdand DN4 TCRb + cells increased overtime, there was preferential expansion of the DN3 and TCRb + DN4 populations (Fig. 5D).
To assess cell cycle status in the DN4 population, we compared cyclin B1 expression in the TCRb -/cdand TCRb + DN4 thymocytes at 3 and 4 days after HC treatment (Fig. 6A). Approximately four times more cells in the TCRb + DN4 compartment showed high intensity cyclin B1 staining (Fig. 6A), indicative of cells in the S/G2/M stages of the cell cycle, than in the TCRb -/cd -DN4 population, indicating that TCRb expression (and hence pre-TCR signalling) induces expansion.
In summary, in the synchronised and rapidly expanding HC-treated adult thymus, we were able to detect a large percentage of TCRb -/cdadult DN4 thymocytes, similar to that observed in the foetal thymus. Thus, the rarity of TCRb -/cd -DN4 cells in the normal adult thymus cannot be attributed to an intrinsic inability of either adult lymphoid or stromal compartments to support pre-TCR independent differentiation to DN4. Our experiments suggest that the decrease in the ratio of TCRb -/cd -:TCRb + DN4 cells after treatment is due to pre-TCR induced expansion of the TCRb + subset rather than apoptosis of the TCRb -/ cd -DN4 population. The detection of significantly more apoptotic cells in the TCRb -/cd -DN4 subset in nontreated adults and 7 days after HC treatment seems to reflect ageing cells accumulating in this subset rather than immediate induction of apoptosis due to lack of pre-TCR signalling.
Reduced proliferation, increased apoptosis in foetal Rag1 -/-DN4 versus DN3 To account for the virtual absence of DN4 cells in Rag1 -/adult thymus compared to the readily detectable population in the Rag1 -/foetal thymus, we analysed proliferation and apoptosis of Rag1 -/-DN3 and DN4 cells during embryonic development and in the adult Rag1 -/thymus. On all embryonic days tested the percentage of DN4 cells that were in cell cycle, as assessed by the intensity of cyclin B1 staining, was lower than the corresponding percentage of DN3 cells, and the percentage of cycling cells decreased over time in both DN3 and DN4 subsets (Fig. 7A). Interestingly, the ratio of cycling cells in the DN3 relative to the DN4 population increased sixfold between foetal and adult thymus, accounting for the decline in the DN4 population (Fig. 7B).
Staining for the active form of caspase-3 revealed that, contrary to the data obtained from the wild-type foetal thymus, apoptotic cells were present in Rag1 -/foetal DN3 and DN4 subsets and their percentage was higher in the DN4 subset (Fig. 7C, D). In a typical experiment the percentage of apoptotic cells in the DN3 subset varied from 0.31% and 0.63% for E15.5-17.5 and from 0.89% to 1.58% in the DN4 subset (Fig. 7C). In the adult Rag1 -/thymus, we were only able to detect a very small percentage of active caspase-3-positive cells in both the DN3 (0.02%) and the DN4 (0.14%) subsets (Fig. 7C) and there was no significant difference in the percentage of apoptotic cells between the two subsets (Fig. 7D). The observed increase in apoptosis in the foetal DN4 subset relative to the DN3 subset during ontogeny provides an additional explanation for the decline of the DN4 population.

Discussion
Here we showed that in the foetal thymus downregulation of CD25 cell surface expression and differentiation to the DN4 stage can occur independently of TCRb expression and hence pre-TCR signalling. Rare TCRb -/cd -DN4 cells in adult mice have been previously reported [7,15], but we showed that in contrast to the adult thymus, most DN4 thymocytes in the E15.5 foetal thymus were icTCRb -/cd -. These TCRb -/cd -CD25 -CD44thymocytes were T lineage, based on expression of Thy1 and icCD3, and TCRb DJ rearrangement. Likewise, we found a normal percentage of DN4 cells in the early Rag1 -/foetal thymus, despite their inherent inability for TCRb/cd rearrangement and expression.
Most DN4 cells that emerged after HC treatment of adult mice were also icTCRb -/cd -. The abundance of TCRb -/cd -DN4 cells in the HC-treated adult thymus demonstrated that their production was not foetalrestricted and that both adult lymphoid and stromal elements are capable of supporting differentiation to DN4 in the absence of TCRb/cd expression. This similarity with the foetal thymus could reflect the synchronicity of the transition in both cases (enabling the identification of normally undetectable TCRb -/cdthat would physiologically make the transition from DN3 to DN4 in the adult thymus). Alternatively, the production of TCRb -/cd -DN4 cells in the adult HCtreated thymus could be a direct consequence of the treatment, as regeneration and rapid growth are induced. During embryogenesis, the foetal thymus also expands exponentially. It therefore remains to be determined to what extent pre-TCR independent mechanisms of differentiation from DN3 to DN4 occur in the adult thymus under physiological conditions, but clearly TCRb/cd expression independent differentiation from DN3 to DN4 is important in the foetal thymus. It likewise remains to be determined if low-level signalling through other components of the pre-TCR complex in the absence of TCRb expression, such as CD3 polypeptides, would be sufficient to drive down-regulation of CD25.
The fact that DN4 thymocytes are found in the foetal Rag1 -/thymus is important for the interpretation of the phenotype of other genetically altered mice that have arrested thymocyte development. For example, particularly in the foetus, the appearance of DN4 cells may have lead to the assumption that a mutation affected processes following pre-TCR signalling when in fact this might not necessarily be the case.
There are several studies indicating that pre-TCR signalling and b-selection induce proliferation [5, 14, 18 -21]. We also demonstrated directly, both by PI and anti-cyclin B1 staining that four times more TCRb + cells were dividing in both the foetal and HC-treated adult thymus, compared to the TCRb -/cd -DN4 population. We found more proliferating cells in both foetal TCRb + and TCRb -/cd -DN4 cells than in their adult counterparts, consistent with the exponential increase in thymocyte number during embryogenesis. In addition, it is worth noting that we detected some cell division in all thymocyte populations examined, including the TCRb -/cd -DN4 population, suggesting the existence of pre-TCR independent mechanisms of proliferation, as previously reported [7,26].
It has been shown that the pre-TCR can promote cell survival by induction of anti-apoptotic agents [9] and that cell death is increase in the adult TCRb -DN4 population relative to the adult TCRb + DN4 [7]. The experiments of Falk and others [7] correlated cell surface CD69 expression with absence of icTCRb/cd in the DN4 population, and then demonstrated relatively more apoptosis in the CD69 + DN4 population than in the CD69 -DN4 cells. Consistent with this study, our analysis confirms directly an increase in apoptosis of TCRb -/cd -DN4 cells relative to TCRb + DN4 cells in the adult thymus (Fig. 4C). Our experiments suggest, however, that absence of the pre-TCR complex does not immediately induce apoptosis as we did not observe apoptosis amongst the foetal or recently differentiated adult DN4 cells either by caspase-3 or annexin-V staining ( Fig. 4 and 6, and data not shown), regardless of the presence or absence of TCRb/cd. The increase in apoptosis in the TCRb -/cd -DN4 population on successive days after HC treatment is consistent with accumulation of ageing cells and absence of pre-TCR induced pro-survival signals that are up-regulated in the TCRb + DN4 population [9]. The proportion but not the number of icTCRb -/cd -DN4 cells greatly decreased over time, and was very low in the fully developed adult thymus. Taken together our experiments indicate that the age-dependent decline in the proportion of TCRb -DN4 cells is due to pre-TCR induced proliferation of the TCRb + DN4 subset, rather than immediate induction of apoptosis in cells that fail to pass the b-selection checkpoint at the DN3 stage.
The fate of the TCRb -/cd -DN4 cells is unclear. In the foetal and HC-treated adult thymus, there was no enrichment for TCRb + cells in the DN4 subset compared to the DN3 population. It is possible that some wild-type TCRb -/cd -DN4 cells eventually rearrange and express TCRb or TCRcd and continue their maturation, but the foetal Rag1 -/-DN4 thymocytes clearly cannot progress beyond the DN4 stage as neither DP nor SP cells are observed in either the foetal or adult Rag1 -/thymi. We showed that the foetal Rag1 -/-DN4 population is disadvantaged compared to the DN3 population due to both reduced proliferation and increased apoptosis, explaining the virtual absence of DN4 cells in the adult thymus. In addition, the presence of apoptotic Rag1 -/-DN3 and DN4 cells, not detectable in the equivalent wild-type populations and the reduced proliferation in the Rag1 -/foetal compared to wild-type thymocyte populations seem to reflect differences in the thymic microenvironment of Rag1 -/and wild-type thymus. This is consistent with a previous report showing that the reduced proliferation of adult Rag2 -/compared to wildtype thymocytes was at least partly due to differences in the thymus environment rather than the absence of cell autonomous, pre-TCR related proliferation signals in the Rag2 -/thymocytes [26].
In conclusion, we have shown that in the foetal thymus the transition from the DN3 to the DN4 stage is frequently independent of pre-TCR signalling. Our data also suggest that pre-TCR signalling induces prolifera-tion and that lack of it does not immediately induce apoptosis.

Mice
Wild-type mice were C57BL/6, purchased from B & K Universal Ltd (UK) and Rag1 -/- [10] mice were purchased from Jackson Laboratories, USA. All mice were bred and maintained at the Central Biomedical Services unit at Imperial College London according to UK Home Office regulations.
Timed mates were performed by mating a male with two females overnight and monitoring the females for plugs. The day the plug was found was counted as E0.5.
Cell suspensions were stained with the antibodies for 30 min on ice in 50 ll Dulbecco's modified medium (Life Technologies), supplemented with 5% FCS and 0.01% sodium azide. Cells were washed in this medium between incubations and prior to analysis on the FACScan (Becton Dickinson). Events were collected in list mode using CellQuest software and data analysed using CellQuest Pro software. Live cells were gated according to their FSC and SSC profiles. Data are representative of at least three experiments.
Intracellular staining for TCRb, TCRc/d, CD3, cyclin B1 and caspase-3 was performed on cells stained for surface markers as above following fixation and permeabilisation with the Cytofix/Cytoperm TM solutions (BD biosciences) according to the manufacturer's instructions. PI (Sigma) staining was carried out on sorted thymocytes treated with 100 lg/mL RNase (Sigma) and permeabilised in 0.1% Triton X-100, as described previously [27]. Cell cycle analysis was carried out using a doublet-discrimination module on the FACScan. C57BL/6J E15.5 and E16.5 foetal and adult thymocytes were sorted on a Modular Flow Cytometer (MoFlo, Cytomation, Inc., Fort Collins, CO) at the Cancer Research UK FACS laboratory. For purification of DN3 and DN4 populations for the real time RT-PCR analysis of Rag1 and Rag2 transcription, cells falling within the FSC/SSC live gate, >98% of which were CD45.2 + , were sorted using antibodies directed against CD25 FITC , CD3/CD4/CD8 PE and CD44 Cychrome . TCRb -/cdand TCRb + DN4 thymocytes were isolated as described in Fig. 3C, D, falling within the FSC/SSC live gate and staining negative for CD3/CD4/CD8/CD25/CD44/B220 and positive for Thy1.2 and negative or positive for icTCRb/cd or icTCRb antibody staining, respectively.

Foetal and adult thymus organ cultures
Foetal and adult thymi were cultured on 8-lm pore size Millipore filters (Millipore) in AIM-V serum free medium (Life Technologies) at 37 C and 5% CO 2 with 1 lg/mL azide-free anti-CD3 (BD Pharmingen) as previously described [17].
Real-time RT-PCR analysis RNA was extracted from the sorted thymocytes with the Absolutely RNA miniprep kit (Stratagene) and cDNA was synthesized with Superscript III (Invitrogen, Carlsbad, CA). The cDNA samples were analysed in triplicate by real-time PCR on an iCycler (Bio-Rad Laboratories) using the iQ TM SYBR J Green Supermix (Bio-Rad) according to the manufacturer's instructions and as previously described [17]. At least one primer for each amplification pair was designed to span exonexon boundaries to avoid amplification of genomic DNA. For the amplification of Rag1 transcripts, we used primers Rag1 exon 2 CCTTTGAGGTTTCTCTAAC and Rag1 exon 2-3 TCCTGCTACTAAAATCTCC, and for the amplification of Rag2 transcripts we used Rag2 forward CTGGCTTG-GCCGAAAGG and Rag2 reverse CTGCTTGTGGATGTGAAA-TACTCT.

Quantification of TCRb locus rearrangements
TCRb -/cdand TCRb/cd + DN4 thymocytes were sorted from E16.5 foetal thymi as described above, DNA was extracted using the DNeasy tissue kit (Qiagen) and D to J PCR-Southern hybridisation assay was carried out as previously described [17]. In brief, 24 cycles of PCR amplification for the D to J TCRb rearrangements were performed using the following primers: Db2 (5 0 ) GTAGGCACCTGTGGGGAAGAAACT and Jb2.7 (3 0 ) TGAGAGCTGTCTCCTACTATCGATT, and the amplified products were quantified by Southern blotting using a TCRb probe amplified from germline (unrearranged) TCRb locus with Db2 and Jb2.7 primers, corresponding to the genomic region between the TCRDb2 and TCRJb2.7 as previously described [17].
The genomic DNA concentration of the templates for the D to J PCR-Southern assay were quantified by real time PCR on an iCycler using primers specific for a single copy genomic locus 5 0 G2FAGAACCTGAAGACACACCTGCG and 3 0 G2R GAGGCATTGGAGAAGGCTTTG.

HC treatment
For the deletion of immature thymocytes and synchronization of thymocyte development in adult thymus, 4-8-week-old C57BL/6 mice were injected intraperitoneally with 0.4 mg/g HC sodium phosphate (Sigma), diluted in sterile PBS [25].