Tcra gene recombination is supported by a Tcra enhancer- and CTCF-dependent chromatin hub

Abstract

Antigen receptor locus V(D)J recombination requires interactions between widely separated variable (V), diversity (D), and joining (J) gene segments, but the mechanisms that generate these interactions are not well understood. Here we assessed mechanisms that direct developmental stage-specific long-distance interactions at the Tcra/Tcrd locus. The Tcra/Tcrd locus recombines Tcrd gene segments in CD4⁻CD8⁻ double-negative thymocytes and Tcra gene segments in CD4⁺CD8⁺ double-positive thymocytes. Initial V_α-to-J_α recombination occurs within a chromosomal domain that displays a contracted conformation in both thymocyte subsets. We used chromosome conformation capture to demonstrate that the Tcra enhancer (E_α) interacts directly with V_α and J_α gene segments distributed across this domain, specifically in double-positive thymocytes. Moreover, E_α promotes interactions between these V_α and J_α segments that should facilitate their synapsis. We found that the CCCTC-binding factor (CTCF) binds to E_α and to many locus promoters, biases E_α to interact with these promoters, and is required for efficient V_α–J_α recombination. Our data indicate that E_α and CTCF cooperate to create a developmentally regulated chromatin hub that supports V_α–J_α synapsis and recombination.

T and B cells produce diverse antigen receptors through the recombination of variable (V), diversity (D), and joining (J) gene segments at the T-cell receptor (Tcra, Tcrb, Tcrg, and Tcrd) and Ig (Igh, Igκ, and Igλ) loci. This V(D)J recombination is initiated by the lymphoid-specific recombination-activating gene-1 (RAG-1) and RAG-2 proteins, which recognize the recombination signal sequences (RSSs) that flank all V, D, and J gene segments and then cleave the DNA between the RSSs and the adjacent coding gene segments (1). A critical feature of the reaction is the assembly of a synaptic complex composed of two RSSs before the generation of RAG-dependent DNA double-strand breaks (DSBs). As such, lineage- and developmental stage-specific V(D)J recombination events can be regulated not only by changes in RAG protein expression and RSS accessibility to RAG proteins but also by the ability of those RSSs to undergo synapsis (2).

Conformational changes of antigen receptor loci are believed to support V(D)J recombination events because they can bring distant RSSs into proximity and therefore increase the probability of RSS synapsis (2, 3). Studies using 3D-FISH have demonstrated that lineage- and development stage-specific locus contraction marks the recombination windows at antigen receptor loci (3). For example, the 3-Mb Igh locus contracts specifically in pro-B cells to support V_H-to-D_HJ_H recombination (4–7). This contracted conformation brings distal and proximal V_H segments, which are separated by megabases in the linear DNA sequence, to the vicinity of the D_HJ_H cluster, presumably allowing all V_H segments a similar opportunity for recombination (8). In addition, the mapping of Igh locus DNA–DNA contacts by chromosome conformation capture (3C) and related methods identified three domains, each composed of multiple DNA loops (9). The Igh enhancer, E_μ, was found to promote DNA contacts within the 3′ domain and to promote large-scale contraction of the Igh locus, perhaps by mediating interdomain contacts. However, our understanding of the molecular mechanisms regulating locus contraction and long-distance DNA contacts within antigen receptor loci remains rudimentary.

The CCCTC-binding factor (CTCF) is a highly conserved multifunctional zinc finger protein (10). CTCF not only insulates gene activity by blocking enhancer–promoter interaction or demarcating boundaries between active and inactive chromatin but also functions as a chromatin organizer that mediates DNA contacts and loop formation (11). CTCF is essential for the formation of higher-order chromatin architecture at various loci, including the H19/Igf2 (12–14), β-globin (15, 16), MHC class II (17), and Ifng (18) loci. These conformations are believed to facilitate developmental stage- or cell activation-specific gene expression. Global mapping of CTCF-binding sites and CTCF-associated chromatin loops also suggests that CTCF can facilitate long-distance interactions for coordinated gene expression, demarcate chromatin boundaries, or facilitate enhancer–promoter communication (19–23). However, it remains difficult to predict how CTCF might function at any individual locus.

Recently, several studies have suggested a role for CTCF in the regulation of V(D)J recombination at antigen receptor loci (9, 24–28). For example, knockdown of CTCF in a pro–B-cell line reduced DNA contacts between regulatory elements that flank the D_HJ_HC_H region, increased antisense transcription through the D_H and V_H regions, and partially suppressed contraction of the Igh locus (25). Deletion of a pair of CTCF-binding sites located between V_H and D_H segments caused high-frequency rearrangement of the D_H-proximal V_H gene segments and led to developmentally inappropriate V_H gene recombination in B cells and thymocytes (24). In addition, conditional knockout of CTCF allowed the intronic Igk enhancer to interact more frequently with proximal Vκ segments and promoted a bias toward proximal Vκ recombination (26). These studies suggested that CTCF may influence V(D)J recombination by mediating DNA contacts that insulate or restrict enhancer activity to defined portions of the Igh and Igk loci. Whether CTCF impacts V(D)J recombination by playing a more direct role in facilitating RSS synapsis remains unclear.

The Tcra/Tcrd locus is a complex genetic locus that contains both Tcra and Tcrd gene segments that are regulated distinctly during thymocyte development (29). In CD4⁻CD8⁻ double-negative (DN) thymocytes, the Tcrd enhancer (E_δ) activates the D_δ and J_δ gene segments to promote V_δ-to-D_δ-to-J_δ rearrangement. In CD4⁺CD8⁺ double-positive (DP) thymocytes, the Tcra enhancer (E_α) activates 3′ V_α promoters and the TEA promoter (TEAp) at the 5′ end of the J_α array to initiate V_α-to-J_α rearrangement (Fig. 1A). Recent work revealed that V_α-to-J_α recombination is compromised in mice that are conditionally deficient in cohesin (30). Cohesin is known to regulate sister chromatid segregation during mitosis (31). It now is appreciated that cohesin has additional roles in nondividing cells, mediated in part by its ability to interact physically and functionally with CTCF at many genomic sites (32–34). Notably, cohesin was found to bind to E_α and the TEAp and to facilitate their long-distance interaction in DP thymocytes (30). Thus, cohesin appears to regulate Tcra recombination by promoting contacts between cis-regulatory elements that are needed to generate accessible J_α RSSs. Whether CTCF is an important regulator of Tcra locus recombination is currently unknown.

Fig. 1.

Developmentally regulated interactions between E_α and the J_α-C_α region. (A) The 1.6-Mb Tcra/Tcrd locus of strain 129 mice, with the 3′ 450 kb expanded to show several V segments and the D_δJ_δC_δ and J_αC_α regions. E_δ and E_α (ovals) and V and TEA promoters (bent arrows) are indicated. (B) Interactions between E_α and the J_α region. WT DN thymocytes (n = 3–8), WT DP thymocytes (n = 4–10), E_a^−/− DP thymocytes (n = 3–9), and B cells (n = 2–4) were analyzed by 3C using an E_α anchor. Interactions of the E_α-containing HindIII fragment (no. 14, black bar) with other HindIII fragments (nos.1–13 and 15–18, gray bars) are plotted as means ± SEM with normalization to the nearest neighbor fragment, no. 15. White circles indicate CTCF-binding sites. (C) WT DN thymocytes (n = 3), WT DP thymocytes (n = 2–4) E_a^−/− DP thymocytes (n = 2), and B cells (n = 2) were analyzed by 3C using the TEAp as an anchor. Interactions of the TEAp-containing HindIII fragment (no. 4, black bar) with other HindIII fragments (nos. 9–18, gray bars) are plotted with normalization as in B. All thymocyte preparations were from Rag2^−/− background mice.

We previously used 3D-FISH to show that the 5′ portion of the Tcra/Tcrd locus is contracted in DN thymocytes but extended in DP thymocytes, whereas the 3′ end of the locus is contracted similarly in both thymocyte populations (35). The unique 5′-extended and 3′-contracted conformation in DP thymocytes was hypothesized to be significant, in that it could bias initial Tcra recombination events to use 3′ V_α gene segments, thereby saving 5′ V_α gene segments for subsequent rounds of V_α-to-J_α recombination that replace the initial recombination events. Although the 3′ end of the Tcra/Tcrd locus is contracted in both DN and DP thymocytes, E_α activity and V_α-to-J_α recombination occur specifically in DP thymocytes. Therefore, we predicted that there must be DP thymocyte-specific DNA looping and DNA contacts within the contracted 3′ region that were not resolved by 3D-FISH. Here, we used 3C to document an E_α-dependent chromatin hub at the 3′ end of the Tcra/Tcrd locus that forms specifically in DP thymocytes. This chromatin hub brings 3′ V_α and 5′ J_α gene segments into proximity to facilitate initial Tcra rearrangement. We found that the Tcra/Tcrd locus is rich in CTCF-binding sites and that CTCF marks many locus cis-regulatory elements. Moreover, the loss of CTCF in DP thymocytes dysregulates long-distance interactions among these elements, suppresses chromatin hub formation, and impairs initial V_α-to-J_α rearrangement. Our data suggest that E_α and CTCF cooperate to organize a DP thymocyte-specific chromatin hub that sets the stage for synapsis and recombination of 3′ V_α and 5′ J_α gene segments.

Results

Identification of a DP Thymocyte-Specific Chromatin Hub at the 3′ End of the Tcra/Tcrd Locus.

Previous studies indicated that E_α regulates up to 500 kb of the Tcra/Tcrd locus, including the E_α-proximal 3′ V_α gene segments and the entire J_α array (36, 37). Activation of the TEAp at the 5′ end of the J_α array is believed to occur through direct enhancer–promoter interaction across 75 kb in DP thymocytes (30). However, it is unclear how the E_α–TEAp interaction is regulated during thymocyte development and whether E_α activates 3′ V_α gene segments through direct interactions as well. To answer these questions, we performed 3C assays to assess long-distance interactions across the E_α-proximal 400 kb of the Tcra/Tcrd locus (Fig. 1A). In 3C, interacting DNA sequences are trapped by formaldehyde cross-linking; after digestion with restriction enzymes, interacting DNA sequences are ligated and then detected by quantitative PCR (qPCR). To map interactions in cells carrying homogeneous, unrearranged Tcra/Tcrd alleles, we harvested DN thymocytes from recombinase-deficient Rag2^−/− mice in which thymocyte development is blocked at the CD44⁻CD25⁺ DN3 stage (38, 39). We harvested DP thymocytes by injecting Rag2^−/− mice with anti-CD3 antibody to stimulate DN-to-DP differentiation (40). To address the role of E_α, we analyzed E_α^−/−Rag2^−/− mice in similar fashion. As a control, we also analyzed splenic B cells.

We first measured DNA interactions across the J_α–C_α region, spanning 120 kb from J_δ2, a gene segment 90 kb upstream of E_α, to Dad1, a gene extending 30 kb downstream of E_α (Fig. 1 B and C). Using an E_α-containing HindIII fragment (Fig. 1B, fragment 14) as an anchor, in control splenic B cells (Fig. 1B, black dashed line) we detected strong signals from neighboring fragments (Fig. 1B, fragments 13 and 15) but very low signals at increasing distances from E_α, reflecting the absence of long-distance interactions in this cell type. In DP thymocytes (Fig. 1B, red line), E_α was found to interact strongly with the TEAp (Fig. 1B, fragment 4) and to display gradually diminishing interactions moving from 5′ to 3′ across the J_α region (Fig. 1B, fragments 6–11). Broad interactions between E_α and 5′ J_α gene segments likely reflect the presence of several minor J_α promoters that are distributed across this region (41). In addition to these interactions, we noted relatively strong interactions, above the B-cell background, at sites flanking the anchor (Fig. 1B, fragments 12 and 16). This result suggests that the whole region may be more compact in DP thymocytes than in B cells.

DN thymocytes displayed substantially reduced E_α interactions across the entire J_α array (Fig. 1B, blue line). Interactions with the TEAp region (Fig. 1B, fragments 3–5) were reduced but were clearly elevated above the B-cell background. To verify this point, we also analyzed interactions between E_α and the TEAp using the TEAp-containing fragment (Fig. 1C, fragment 4) as an anchor. Indeed, the TEAp interacted with E_α (Fig. 1 B and C, fragment 14) strongly in DP thymocytes and weakly in DN thymocytes compared with background interaction in B cells.

To assess the role of E_α in mediating interactions across the J_α–C_α region, we analyzed E_α^−/− DP and E_α^−/− DN thymocytes. In E_α^−/− DP thymocytes, all identified E_α interactions were reduced to the level of the B-cell background, suggesting that these interactions are strictly E_α dependent (Fig. 1 B and C, black line). In addition, the weak E_α–TEAp interaction detected in DN thymocytes was also E_α dependent (Fig. S1). Because E_α is not active in DN thymocytes, the weak E_α–TEAp interaction in these cells must be mediated by E_α-bound factors independent of ongoing transcription. Overall, our data demonstrate that E_α participates in a range of lineage- and developmental stage-specific long-distance interactions across the J_α array in DN and DP thymocytes.

We asked whether E_α similarly regulates the 3′ V gene segments through direct physical contacts in DP thymocytes. Using the E_α fragment as an anchor, we found that E_α interacted with several V gene segments (TRAV19, TRAV21, TRDV1, TRDV2-1, and TRDV2-2) at distances of 280–400 kb (Fig. 2 A and B, sites i, iii, iv, vi, and vii, respectively). All these interactions were greatly diminished in DN thymocytes, in E_α^−/− DP thymocytes, and in B cells, indicating that the interactions were lineage specific, developmental stage specific, and E_α dependent. Similar results were obtained by analysis using TRAV21 or TRDV2-2 as an anchor (Fig. 2C). However, we failed to detect interactions between E_α and V_α gene segments located at the extreme 5′ end of the Tcra/Tcrd locus (Fig. S2). We conclude that in DP thymocytes E_α regulates the proximal 400 kb of the Tcra/Tcrd locus through direct interactions.

Fig. 2.

Developmentally regulated interactions among E_α, the TEAp, and 3′ V gene segments. (A) Distribution of 3′ V gene segments spanning 280–400 kb upstream of E_α. Filled and open rectangles denote functional V gene segments and pseudogenes, respectively. Analyzed restriction fragments are identified (i–vii). White circles indicate CTCF-binding sites. (B and C) Interactions between E_α and 3′ V gene segments. WT DN thymocytes (n = 3), WT DP thymocytes (n = 4–7), E_a^−/− DP thymocytes (n = 2–4), and B cells (n = 2–4) were analyzed by 3C using E_α (B), TRAV21 (C, Left), or TRDV2-2 (C, Right) HindIII fragments as anchors. (D and E) Interactions between the TEAp and 3′ V gene segments. WT DN thymocytes (n = 3), WT DP thymocytes (n = 2–5), E_a^−/− DP thymocytes (n = 2–4), and B cells (n = 2–4) were analyzed as above using TEAp (D), TRAV21 (E, Left), or TRDV2-2 (E, Right) HindIII fragments as anchors. (F) Interactions between 3′ V segments. WT DN thymocytes (n = 3), WT DP thymocytes (n = 4), E_a^−/− DP thymocytes (n = 2), and B cells (n = 2–4) were analyzed using TRAV21 (Left) or TRDV2-2 (Right) HindIII fragments as anchors. All thymocyte preparations were from Rag2^−/− background mice. Data were plotted as means ± SEM with normalization as in Fig. 1B.

E_α might interact in pairwise fashion with the TEAp and V segments that are widely separated in nuclear space or might interact with these sites when all are in close physical proximity. To test these alternatives, we analyzed interactions between the TEAp and 3′ V segments. Indeed, contacts between the TEAp and 3′ V segments (TRAV19, TRAV21, and TRDV2-2) (Fig. 2 D and E) and between two 3′ V segments (TRAV21 and TRDV2-2) (Fig. 2F) were high in DP thymocytes but low in DN thymocytes and B cells, suggesting that these elements are in proximity in DP thymocytes. Importantly, all these interactions were E_α dependent, indicating that E_α not only interacts with 3′ V segments and the TEAp through direct pairwise interactions but also tethers them together to form a multicomponent chromatin hub.

Notably, both E_α and the TEAp interacted less well with intergenic regions (Fig. 2 A, B, and D, sites ii and v). E_α also interacted poorly with Tcrd gene segments upstream of the TEAp (Fig. 1B). Therefore, inactive regions are looped-out from this E_α-dependent, DP stage-specific chromatin hub. We propose that this chromatin hub functions to approximate V_α and J_α RSSs to increase the probability of RSS synapsis, in this way supporting primary V_α-to-J_α rearrangement in DP thymocytes.

TEAp Suppresses Tcrd Gene Segments in DP Thymocytes.

Because the TEAp region interacts with E_α and multiple V segments in the DP stage-specific chromatin hub, we asked whether the TEAp regulates long-distance interactions among these elements. To answer this question, we collected DP thymocytes from anti-CD3–injected TEAp^−/−Rag2^−/− mice (Fig. 3A) and performed 3C assays using the E_α fragment as an anchor. Compared with WT alleles (Fig. 3B, red line), TEAp-deleted alleles (Fig. 3B, blue line) demonstrated reduced interactions between E_α and 5′ J_α segments (Fig. 3B, fragments 5 and 6) and increased E_α interactions with central J_α segments (Fig. 3B, fragments 8 and 9). This shift parallels the documented down-regulation of 5′ J_α promoters and up-regulation of central J_α promoters on TEAp-deleted alleles (Fig. 3A) (41). Notably, we detected only a slightly reduced E_α–TEAp interaction on TEAp-deleted alleles (Fig. 3B, fragment 4) but dramatically increased interactions between E_α and sites upstream of the TEAp in the region containing J_δ2, E_δ, C_δ, and TRDV5 (Fig. 3B, fragments 1–3). However, increased interactions did not extend as far as the 3′ V gene segments (Fig. 3 C and D). We suspect that increased interaction between E_α and TRDV5 obscured a diminished E_α–TEAp interaction on TEAp-deleted alleles. Further analysis also revealed that interactions between the TEAp and 3′ V gene segments (TRAV21 and TRDV2-2) were reduced (Fig. 3E), but that interaction between a Tcrd gene segment (J_δ2) and a 3′ V gene segment (TRDV2-2) was elevated in TEAp-deleted DP thymocytes (Fig. 3F). Taken together, our data indicate that the TEAp region loops out from the E_α-dependent chromatin hub in TEAp-deleted DP thymocytes and that nearby Tcrd gene segments are included more frequently in that hub.

Fig. 3.

Regulation of E_α contacts by the TEAp. (A) WT TEA promoter-deleted (TEAp⁻), and TEA-terminator (TEA-T) alleles are diagrammed, with bent arrows denoting active promoters and gray shading denoting regions of reduced accessibility and promoter activity on mutant alleles. (B) Interactions between E_α and J_α regions. WT (data from Fig. 1B), TEAp^−/− (n = 4–11), and TEA-T (n = 3–5) DP thymocytes were analyzed by 3C using an E_α anchor. (C and D) Interactions between E_α and 3′ V segments. WT (n = 5) and TEAp^−/− (n = 4) DP thymocytes were analyzed by 3C using E_α (C), TRAV21 (D, Left), or TRDV2-2 (D, Right) as anchors. (E) Interactions between 3′ V genes and the TEAp. WT (n = 5) and TEAp^−/− (n = 4) DP thymocytes were analyzed by 3C using TRAV21 (Left) or TRDV2-2 (Right) as anchors. (F) Interaction between TRDV2-2 and J_δ2. WT (n = 4) and TEAp^−/− (n = 4) DP thymocytes were analyzed by 3C using the TRDV2-2 HindIII fragment as an anchor. All thymocyte preparations were from Rag2^−/− background mice. Data are plotted as means ± SEM with normalization as in Fig. 1B.

TEA promoter deletion could impact long-distance interactions through the loss of TEA transcription or through the loss of TEAp-bound factors. To distinguish these possibilities, we analyzed TEA-terminator (TEA-T) DP thymocytes, which have an active TEAp but carry an introduced transcription terminator immediately downstream (Fig. 3A). TEA-T alleles mimic TEAp-deleted alleles with respect to the shift in promoter activity from 5′ J_αs to central J_αs, suggesting that TEA transcription rather than the TEAp itself targets E_α to different regions of the J_α array (42). We found that on TEA-T alleles, as compared with WT, the E_α–TEAp interaction was unperturbed (Fig. 3B, black line, fragment 4), an E_α–5′ J_α interaction (Fig. 3B, fragment 6) was suppressed, and an E_α–central J_α interaction (Fig. 3B, fragment 8) was increased substantially. We suspect that reduced interactions with 5′ J_α segments were partially obscured (Fig. 3B, fragment 5) by high-level interaction with the TEAp (Fig. 3B, fragment 4). However, in other respects, E_α interactions across the J_α array were quite similar on TEAp-deleted and TEA-T alleles. Importantly, TEA-T alleles did not display the strong E_α–Tcrd gene interactions that were apparent on TEAp-deleted alleles. This finding indicates that TEAp-bound factors, rather than TEAp-directed transcription, limit E_α from interacting with upstream Tcrd gene segments. Our data demonstrate that the TEAp limits E_α interactions with Tcrd gene segments and with central J_α gene segments through distinct mechanisms.

To evaluate the biological consequence of increased interactions between E_α and Tcrd gene segments on TEAp-deleted alleles, we analyzed transcriptional activity and chromatin accessibility. Interestingly, TEAp-deleted alleles revealed two- to fivefold increases in transcription at Tcrd gene segments located upstream of the TEAp, including TRDV4, D_δ1, J_δ1, J_δ2, and TRDV5 (Fig. 4A). However, increased transcription did not extend to V segments further upstream (TRAV17, TRAV19, TRAV21, and TRDV2-2), suggesting that the impact of the TEAp is regional. We also analyzed histone H3 acetylation, a marker related to chromatin accessibility. Indeed, H3 acetylation also was increased at Tcrd gene segments immediately upstream of the TEAp (J_δ1, J_δ2, and TRDV5) but not at more distant sites (Fig. 4B). To evaluate whether loss of the TEAp also influences Tcrd gene segment recombination at the DP stage, we compared the frequency of TRAV17-J_α49 and TRDV5-J_α49 recombination in WT and TEAp^−/− DP thymocytes (Fig. 4C). Remarkably, TRDV5-J_α49 recombination increased more than fourfold in TEAp^−/− DP thymocytes, whereas TRAV17-J_α49 recombination was unchanged. Hence, the TEAp represses activation and recombination of the Tcrd gene segments in DP thymocytes.

Fig. 4.

Regulation of Tcrd gene segments by the TEAp. (A) qRT-PCR analysis of germ-line transcription in WT (n = 5) and TEAp^−/− (n = 3) DP thymocytes (both on a Rag2^−/− background). (B) ChIP and qPCR analysis of histone H3 acetylation in WT (n = 2) and TEAp^−/− (n = 2) DP thymocytes (both on a Rag2^−/− background). (C) qPCR analysis of TRAV17-J_α49 and TRDV5-J_α49 rearrangement in WT (n = 2) and TEAp^−/− (n = 2) recombinase-sufficient CD71⁻ DP thymocytes. Results are expressed as means ± SEM with normalization to recombination in WT thymocytes.

Role for CTCF in the Regulation of Tcra Locus Recombination and Conformation.

Because CTCF has been implicated in long-distance DNA interactions (11), we asked whether CTCF contributes to DNA contacts within the 3′ portion of the Tcra/Tcrd locus. We used ChIP sequencing (ChIP-Seq) to assess the distribution of Tcra/Tcrd locus CTCF-binding sites in Rag1^−/− DN thymocytes, Rag1^−/− Tcrb transgenic DP thymocytes, and Rag1^−/− pro-B cells. Because of the duplicated nature of the central portion of the Tcra/Tcrd locus (Fig. 1A), we focused our analysis on CTCF sites at the 5′ and 3′ ends of the locus. Within these regions, we identified CTCF binding at many sites (Fig. 5 A and B). In DP thymocytes, CTCF binding appeared to mark many V segment promoters (Fig. 2A, white circles, and Fig. 5 A and B), and in every case there was an identifiable CTCF consensus sequence (43) within 200 bp of the transcription start site. Notably, among the three CTCF-free proximal V gene segments (TRAV22, TRDV4, and TRDV2-1), one (TRAV22) is a pseudogene, and another (TRDV4) is not used in adult thymocytes. E_α and the TEAp, which were shown to bind cohesin subunit Rad21 in a previous study (30), also displayed strong CTCF binding; notably, the E_α and TEAp CTCF sites are eliminated on E_α-deleted and TEAp-deleted alleles, respectively (Fig. 1B, white circles, and Fig. 5B). The localization of CTCF at many regulatory elements suggests that CTCF may mediate interactions among these elements and contribute to Tcra/Tcrd locus regulation.

Fig. 5.

CTCF binding to the Tcra/Tcrd locus. Rag1^−/− DN thymocytes, Rag1^−/− Tcrb transgene DP thymocytes, and Rag1^−/− pro-B cells (all on a C57BL/6 background) were analyzed by ChIP-seq. Results for the 5′ 200-kb of the locus are depicted in A, and results for the 3′ 450-kb portion are depicted in B. (C) CTCF binding to strain 129 Tcra/Tcrd locus sites was analyzed by ChIP-qPCR using Rag2^−/− DN thymocytes (WT DN, n = 3–4), DP thymocytes prepared from anti-CD3–injected Rag2^−/− (WT DP, n = 5–6) and Rag2^−/−E_α^−/− (E_α^−/− DP, n = 4) thymocytes. TRAV17-a and -b are two neighboring CTCF sites upstream of TRAV17. INT1 and INT2 are two intergenic sites between TRDV3 and TRDV4. c-Myc and PDβ served as positive and negative controls, respectively. The data are plotted as means ± SEM.

Although CTCF binding was observed at many locations in thymocytes and B cells, some sites appeared to display binding that was stronger in or unique to DP thymocytes, including sites at TRAV18, TRAV19, TRAV20, TRAV21, TRDV3, and J_α42 (Fig. 5B). Moreover, the ChIP-Seq experiments described above were conducted using samples that were of C57BL/6 origin, whereas our 3C experiments analyzed strain 129 Tcra/Tcrd locus alleles. Therefore, we used ChIP-qPCR to confirm both constitutive and DP stage-specific CTCF binding to strain 129 alleles (Fig. 5C). Indeed, compared with a negative control site, PDβ, all tested C57BL/6 CTCF-binding sites were positive by ChIP-qPCR analysis of strain 129 DP thymocytes. Also consistent with the ChIP-Seq data, ChIP-qPCR revealed CTCF binding to be developmentally regulated at many sites in the 3′ 450 kb of the locus, including TRAV17-b, TRAV18, TRAV19, TRAV20, TRAV21, TRDV1, TRDV3, J_α42, and Dad1. In contrast, developmentally regulated binding was not apparent at sites at the 5′ end of the locus (TRAV3-1, TRAV6-1, and TRAV7-1) or at some sites at the 3′ end (TRAV17-a, INT1, INT2, TRDV5, TEAp, and E_α). To determine whether DP stage-specific CTCF binding is E_α dependent, we compared WT and E_α^−/− DP thymocytes (Fig. 5C). Interestingly, almost all sites displaying elevated CTCF binding in DP thymocytes were E_α dependent, whereas sites displaying constitutive CTCF binding were not. Thus, E_α stimulates CTCF binding to multiple sites across the 3′ portion of the Tcra/Tcrd locus in DP thymocytes. Because CTCF-binding sites colocalized with regions that interact physically with E_α (Figs. 1B and 2 B and C) and with each other (Fig. 2 D–F and Fig. S3) across the 3′ portion of the locus, we hypothesized that CTCF is situated at the base of DNA loops and contributes to the formation of the E_α-dependent chromatin hub.

To examine whether CTCF regulates the Tcra/Tcrd locus, we analyzed DP thymocytes of Ctcf conditional-knockout mice (Fig. 6A). To assess the potential influence of CTCF on Tcra recombination, we bred Ctcf ^f/f mice with Lck-Cre mice to initiate Ctcf deletion in DN thymocytes. We also bred these alleles onto a Rag2-deficient background to study changes in germ-line transcription and locus conformation in the absence of CTCF. In Ctcf^f/f Lck-Cre mice, thymocyte development was partially blocked at the immature single-positive stage during the DN-to-DP transition (Fig. 6B), as described previously (44). We sorted CTCF-deficient DP thymocytes based on their expression of a LacZ marker and compared them with DP thymocytes sorted from their WT (Lck-Cre⁻) littermates (Fig. 6B). We further used CD71 staining to distinguish early CD71⁺ from later CD71⁻ DP thymocytes (30). CTCF-deficient DP thymocyte preparations typically displayed >90% loss of Ctcf alleles and substantially reduced CTCF protein (Fig. 6 C and D).

Fig. 6.

Regulation of V_α-J_α recombination by CTCF. (A) The Ctcf floxed and deleted alleles are diagrammed, with Ctcf exons numbered, loxP sites represented by triangles, and PGK-puromycin (puro^r) and splice acceptor-lacZ (SA-LacZ) cassettes indicated. Ctcf-deleted alleles produce a Ctcf-LacZ fusion protein. (B) Representative flow cytometric analysis of Ctcf^f/f (WT) and Ctcf^f/f Lck-Cre (KO) littermates. Thymocytes were gated for CD4⁺CD8⁺ (Left) and CD71⁺ or CD71⁻ (Center) and were incubated with lacZ substrate FDG to identify Ctcf-deficient CD71⁺ or CD71⁻ DP thymocytes (Right; WT, gray shading; KO, black line). (C) qPCR analysis of Ctcf genomic DNA in CD71⁺ (Left) and CD71⁻ (Right) DP thymocytes sorted from anti-CD3–injected WT (n = 4) and Ctcf-deficient (KO, n = 3) mice. (D) Representative Western blot analysis of CTCF protein in sorted thymocytes used in C. Analysis of PLCγ1 controlled for total protein. (E) Genomic DNA isolated from CD71⁺ (early) and CD71⁻ (late) DP thymocytes of WT (n = 4) and Ctcf-deficient (KO, n = 4) mice were analyzed by qPCR to measure V_α-to-J_α recombination using TRAV17 and TRAV19 primers paired with J_α61, J_α56, and J_α49 primers. Results are expressed as means ± SEM with normalization to recombination in unfractionated WT thymocytes.

To assess an effect of Ctcf deletion on early V_α-to-J_α recombination, we quantified Tcra coding joints by qPCR using primers specific for 3′ V_α (TRAV17 and TRAV19) and 5′ J_α (J_α61, J_α56, and J_α49) gene segments (Fig. 6E). Analysis of WT DP thymocytes revealed low-level, J_α61-biased recombination in early (CD71⁺) DP thymocytes, with elevated and more uniformly distributed recombination in late (CD71⁻) DP thymocytes. This result is consistent with the notion that early rearrangements are preferentially targeted to J_α61 and that these rearrangements often are replaced by subsequent rearrangements to downstream J_α gene segments. Notably, we observed substantial reductions in all these V_α-J_α recombination events in both early and late CTCF-deficient DP thymocytes (Fig. 6E). Thus, CTCF is essential for efficient recombination of 3′ V_α and 5′ J_α gene segments.

To determine whether CTCF regulates early Tcra rearrangement by mediating loop formation within the 3′ portion of Tcra/Tcrd locus, we used 3C to analyze interactions within the DP stage-specific chromatin hub in LacZ⁺ DP thymocytes from anti-CD3 antibody–stimulated Ctcf ^f/f Rag2^−/− Lck-Cre mice. Indeed, we observed that loss of CTCF reduced E_α contacts with the TEAp (Fig. 7 A, fragment 4, and Fig. 7B), with J_α segments (Fig. 7 A, fragments 5–9, and Fig. 7B), and with certain 3′ V gene segments (TRAV19, TRAV21, and TRDV1) (Fig. 7 C and D). In addition, the loss of CTCF also reduced TEAp contacts with certain 3′ V gene segments (TRAV19 and TRAV21) (Fig. 7 E and F), indicating that the E_α-dependent chromatin hub is partially but not completely disrupted. However, loss of CTCF increased E_α contacts with Tcrd gene segments (Fig. 7A, fragments 1 and 2), so that the E_α bias toward 5′ J_α over Tcrd gene segments that characterizes WT DP thymocytes was eliminated in CTCF-deficient DP thymocytes. We suggest that CTCF promotes primary Tcra gene recombination by targeting E_α to 3′ V_α and 5′ J_α segments and increasing the probability of physical interactions between these gene segments.

Fig. 7.

Regulation of DP thymocyte-specific DNA contacts by CTCF. (A) Interactions between E_α and J_α regions. DP thymocytes isolated from Ctcf^f/f Lck-Cre⁻Rag2^−/− (WT; n = 2–3) and Ctcf^f/f Lck-Cre⁺Rag2^−/− (KO; n = 4–6) mice were analyzed by 3C using an E_α anchor. (B) Interaction between E_α and the TEAp. WT (n = 6) and KO (n = 6) DP thymocytes were analyzed using the TEAp fragment as an anchor. (C) Interactions between E_α and 3′ V segments (fragment nomenclature as in Fig. 2A). WT (n = 6) and KO (n = 6) DP thymocytes were analyzed using the E_α fragment as an anchor. (D) Interactions between E_α and 3′ V gene segments. WT (n = 5) and KO (n = 5) DP thymocytes were analyzed using TRAV21 (Left) and TRDV2-2 (Right) fragments as anchors. (E and F) Interactions between 3′ V gene segments and the TEAp. WT (n = 5) and KO (n = 5) DP thymocytes were analyzed using (E) the TEAp fragment or (F) TRAV21 (Left) or TRDV2-2 (Right) fragments as anchors. Data are plotted as means ± SEM. (G) 3D-FISH was used to measure the distance between foci detected by a probe hybridizing to central V gene segments and a probe hybridizing downstream of C_α. Scatter-plots display distances between the centers of probe hybridization in DP thymocytes of Ctcf^f/f Lck-Cre⁻Rag2^−/− (WT) and Ctcf^f/f Lck-Cre⁺Rag2^−/− (KO) mice carrying either the 129 or the C57BL/6 Tcra/Tcrd locus. Median values are indicated by horizontal lines. Data are from one experiment for each cell type (66–104 alleles).

Because CTCF supports normal chromatin contacts within the 3′ portion of the Tcra/Tcrd locus, we asked whether CTCF also is involved in the contraction of the 3′ end in DP thymocytes. We performed 3D-FISH to measure the distance between foci detected by two probes, one hybridizing to V segments in the central portion of the V gene array and one hybridizing downstream of C_α. These probes had been used previously to detect contraction of the 3′ end in thymocytes compared with B cells (35, 45). In fact, we observed slightly greater contraction in CTCF-deficient than in WT DP thymocytes (Fig. 7G). Hence, CTCF is not required for Tcra/Tcrd locus contraction.

We then asked whether the disruption of CTCF-dependent DNA contacts also impacts Tcra/Tcrd locus germ-line transcription in CTCF-deficient DP thymocytes. As expected, loss of CTCF caused reduced transcription at TEA (Fig. 8A). Remarkably, transcription of Tcrd gene segments (J_δ1, J_δ2, and TRDV5) was increased substantially (Fig. 8A), in a manner nearly identical to that documented on TEAp-deleted alleles (Fig. 4A). Increases in E_α-Tcrd gene segment contacts and Tcrd germ-line transcription in both TEAp-deleted and CTCF-deficient DP thymocytes suggest that Tcrd gene segments are activated by E_α through the loss of CTCF binding at the TEAp.

Fig. 8.

Regulation of Tcrd gene transcription and recombination in DP thymocytes by CTCF. (A) qRT-PCR analysis of Tcra/Tcrd locus transcripts in DP thymocytes isolated from Ctcf^f/f Lck-Cre⁻Rag2^−/− (WT, n = 2–3) and Ctcf^f/f Lck-Cre⁺Rag2^−/− (KO, n = 4–6) mice. (B and C) LM-qPCR analysis of DSBs in genomic DNA of CD71⁺ and CD71⁻ DP thymocytes sorted from Ctcf^f/f Lck-Cre⁻ (WT, n = 3) and Ctcf^f/f Lck-Cre⁺ (KO, n = 3) mice. Specific primers were used to detect DSBs at the RSS 5′ of D_δ1 (B) and at the J_δ1 RSS (C). DSB frequency was normalized to the level in Lat^−/− DN thymocytes. DP thymocytes isolated from Rag2^−/− Tcrb transgene mice served as a negative control (Ctrl). Data are plotted as means ± SEM.

These data imply that CTCF normally may function to limit Tcrd gene activation in DP thymocytes. Because thymocytes newly entering the DP compartment may not have undergone V_δ-to-D_δ rearrangement on as many as 30% of alleles (Fig. S4), CTCF also may play an important role in limiting Tcrd gene recombination in DP thymocytes. To test this notion, we used ligation-mediated qPCR (LM-qPCR) to detect RAG-mediated DSBs at the 5′ D_δ1 and J_δ1 RSSs. Indeed, we observed increased DSBs in both early (CD71⁺) and late (CD71⁻) CTCF-deficient DP thymocytes at both sites (Fig. 8 B and C). This result supports the idea that CTCF specifies E_α targets to regulate Tcra/Tcrd locus recombination. By simultaneously facilitating E_α–TEAp and suppressing E_α–Tcrd gene segment interactions, CTCF helps commit the locus to V-to-J_α rather than V-to-D_δJ_δ recombination in DP thymocytes.

Discussion

We previously used 3D-FISH to show that the 3′ end of the Tcra/Tcrd locus is contracted similarly in DN and DP thymocytes, as compared with its extended configuration in B cells (35). Here we used 3C to detect a network of interactions between cis-regulatory elements at the 3′ end of the Tcra/Tcrd locus specifically in DP thymocytes, revealing the formation of a developmental stage-specific chromatin hub within the contracted portion of the locus (Fig. 9). We found that all detected DNA contacts within this chromatin hub depend critically on E_α, a potent enhancer that is known to regulate Tcra/Tcrd locus transcription and histone modifications across 500 kb in DP thymocytes. Our data reveal that E_α regulates individual V_α and J_α gene segments through long-distance physical interactions. Moreover, E_α recruits V_α and J_α gene segments into proximity in a manner that should facilitate synapsis of V_α and J_α RSSs.

Fig. 9.

An E_α- and CTCF-dependent chromatin hub within the contracted 3′ portion of the Tcra/Tcrd locus. In WT DP thymocytes, E_α and CTCF cooperate to nucleate interactions (thick lines) among V promoters and the TEAp within the contracted region. E_α-deleted alleles remain contracted, but long-distance DNA contacts that compose the hub are disrupted. In CTCF-deficient DP thymocytes, E_α interactions with some V and J_α segments are reduced (thin lines), whereas interactions with D_δ and J_δ segments are increased (thin lines), leading to reduced Tcra rearrangement but increased Tcrd rearrangement in DP thymocytes.

Tcra recombination occurs in multiple cycles starting with primary recombination events that preferentially use 3′ V_α and 5′ J_α gene segments. Because these gene segments are recruited selectively into the initial chromatin hub by E_α, our data suggest that this hub serves as a platform to support ordered Tcra recombination. The 5′ J_α segments previously were shown to support high-level RAG binding to form a recombination center (46). The chromatin hub described here provides direct evidence for the recruitment of distant V segments into this recombination center, thereby offering strong support for the recombination center model (2).

CTCF was a strong candidate for regulating the DP thymocyte-specific chromatin hub because it has been shown to facilitate DNA looping and contacts in other instances and because we found it to bind E_α, the TEAp, and many V gene segment promoters (Fig. 9). Indeed, the E_α–TEAp interaction was partially diminished in Ctcf conditional-knockout DP thymocytes, suggesting that CTCF binding at these two sites may contribute to their interaction. In addition, loss of CTCF reduced interactions between E_α and 3′ V gene segments and between the TEAp and 3′ V gene segments. Reductions in these various DNA contacts likely explain impaired Tcra gene transcription and recombination in CTCF-deficient DP thymocytes. Notably, CTCF binding to E_α and to the TEAp is not developmentally regulated, whereas the E_α–TEAp interaction and TEA transcription clearly are. We propose that enhancer- and promoter-bound transactivators other than CTCF provide developmental-stage specificity and that they cooperate with bound CTCF to mediate these developmentally regulated events. Thus, the inefficient E_α–TEAp interaction in DN thymocytes may be mediated primarily by E_α-bound CTCF, with efficient E_α–TEAp interaction in DP thymocytes dependent not only on CTCF but also on additional E_α–bound transactivators induced at the DN-to-DP transition (47). Viewed in this way, CTCF may prime the TEAp to be targeted by E_α when E_α is activated in DP thymocytes. CTCF may prime the promoters of 3′ V gene segments similarly. However, some 3′ V segment promoters display induced and E_α-dependent CTCF binding in DP thymocytes, suggesting that CTCF recruitment and stabilization may depend on enhancer–promoter interaction as well.

CTCF and cohesin have been shown to colocalize and function coordinately at many genomic sites (32–34). Because the reduction of E_α–TEAp interaction and TEA expression in CTCF-deficient DP thymocytes was similar to the phenotype of cohesin subunit Rad21-deficient DP thymocytes, it appears that CTCF and cohesin regulate the Tcra/Tcrd locus through the same pathway (30). Notably, Cd4-Cre–triggered deletion of Rad21 caused a defect only in secondary Tcra recombination, whereas Lck-Cre–triggered deletion of Ctcf caused a defect in primary Tcra recombination. This difference likely reflects the fact that the Cd4-Cre–mediated deletion of Rad21 occurs too late to impact primary Tcra recombination events (30). Nevertheless, the two results, taken together, argue that CTCF and cohesin are required continuously throughout the course of primary and secondary Tcra recombination events in DP thymocytes.

Although CTCF was shown previously to be needed for complete Igh locus contraction (25), we did not observe impaired Tcra/Tcrd locus 3′ end contraction in Ctcf-knockout DP thymocytes (Fig. 9). Moreover, we previously observed no loss of 3′ end contraction on E_α-deleted Tcra/Tcrd locus alleles (35). Thus, neither CTCF nor E_α is required for 3′end contraction, and the mechanisms of Tcra/Tcrd locus contraction remain unknown. We speculate, however, that 3′end contraction provides an initial layer of organization that is essential to allow CTCF and E_α to mediate the DNA contacts needed for the activation, synapsis, and recombination of 3′ V_α and 5′J_α gene segments. This concept would explain why E_α-dependent DNA contacts and germ-line transcription are detectable at the 3′ but not the 5′ end of the locus, even though CTCF binds to both. It also would explain why the E_α–TEAp interaction is not detectable in B cells, even at low levels, although CTCF binds to both elements in these cells.

Although the ability of E_α to interact with upstream CTCF-bound promoters was compromised in CTCF-deficient DP thymocytes, interactions with Tcrd gene segments that lack CTCF-binding sites were elevated (Fig. 9). As a result, Tcrd genes underwent elevated transcription and recombination in CTCF-deficient as compared with WT DP thymocytes. We conclude that CTCF is not essential for long-distance contacts by E_α. Rather, CTCF biases these contacts to discrete sites in the locus and in this manner suppresses aberrant interactions.

Finally, we note that the unique distribution of CTCF sites at the Tcra/Tcrd locus may dictate a fundamentally different role for CTCF in the Tcra/Tcrd locus than in the Igh and Igk loci. At Igh and Igk, CTCF sites are generally intergenic except for those in the proximal one-third of the V_H array, which are downstream of V_H RSSs; moreover, CTCF sites are not associated with the major Igh and Igk enhancer elements (25, 26, 48–50). At these loci, CTCF promotes long-distance interactions between CTCF-binding elements (24–26), but these looping interactions appear primarily to restrict or insulate the activities of the major enhancer elements. As examples, CTCF-binding regions situated between V_H and D_HJ_H and between V_κ and J_κ segments appear to block the enhancers from activating relatively proximal V segments, thus promoting the formation of Igh and Igk repertoires in which natural biases to proximal V segments are suppressed (24, 26, 50). Our data suggest that CTCF binding at the TEA promoter inhibits E_α from activating Tcrd gene segments immediately upstream; thus, by analogy to Igh and Igk, the TEAp CTCF site could be considered an insulator that limits Tcrd gene activation in DP thymocytes. However, unlike Igh and Igk, CTCF sites at the Tcra/Tcrd locus mark E_α and many locus-promoter elements. In this scenario, the ability of CTCF to promote long-distance interactions, rather than insulating promoters from enhancers, appears to direct E_α to CTCF-marked promoters. Viewed in this light, the Tcrd activation phenotype reflects not an insulation defect but, more directly, a targeting defect. We suggest that CTCF may influence enhancer–promoter communication at antigen receptor loci through predominant insulating or targeting mechanisms, with the distribution of CTCF-binding sites being the critical determinant.

Materials and Methods

Mice.

E_α^−/− mice (37), TEAp^−/− mice (41), TEA-T mice (42), and Ctcf ^f/f mice (44) have been described previously. To induce Ctcf deletion in early DN thymocytes, Ctcf ^f/f mice were bred with Lck-Cre mice kindly provided by Jeffrey Rathmell (Duke University, Durham NC). Mice carrying modified Tcra/Tcrd and Ctcf alleles also were bred with Rag2^−/− mice (38) for analysis on a recombinase-deficient background. E_α^−/−Rag2^−/−, TEAp^−/−Rag2^−/−, TEA-T Rag2^−/−, and Ctcf^f/f Lck-Cre mice were of mixed genetic background but carried the 129 Tcra/Tcrd locus. Ctcf^f/f Lck-Cre Rag2^−/− mice were either on a mixed background with a 129 Tcra/Tcrd locus or on a pure C57BL/6 background. C57BL/6 background Rag1^−/− and Rag1^−/− Tcrb transgene mice were used for ChIP-seq. All mice were used in accordance with protocols approved by the Duke University Animal Care and Use Committee.

Cell Collection.

DN thymocytes were obtained from 3-wk-old Rag2^−/−, E_α^−/− Rag2^−/−, TEAp^−/− Rag2^−/−, and TEA-T Rag2^−/− mice. DP thymocytes were obtained from these mice 10 d after a single i.p. injection of 150 μg anti-CD3 antibody (145-2C11; BioLegend). DP thymocytes were sorted from Ctcf ^f/f Lck-Cre⁺ mice or Ctcf ^f/f Rag2^−/− Lck-Cre⁺ mice or their Lck-Cre⁻ littermates 7 d after a single i.p. injection of 75 μg anti-CD3 antibody, using anti–CD71-PE (RI7217), anti–CD4-PE/Cy7 (GK1.5), and anti–CD8-APC/Cy7 (53-6.7). Because Ctcf deletion induces LacZ expression, thymocytes were incubated with LacZ substrate fluorescein di β-d-galactopyranoside (FDG) from a FluoReporter lacZ flow cytometry kit (Invitrogen) before sorting to identify LacZ⁺ cells. Splenic B lymphocytes were sorted from WT 129 mice by using anti–B220-PE/Cy7 (RA3-6B2) together with anti–CD3-PE5 (145-2C11), anti–CD4-PE5 (GK1.5), and anti–CD8-PE5 (53-6.7) to exclude T cells. All antibodies were obtained from BioLegend.

3C.

The 3C assays were performed as described previously with slight modifications (51). In brief, 10 million cells were cross-linked in 8 mL RPMI containing 10% FBS (vol/vol) and 2% paraformaldehyde (vol/vol) for 10 min on ice, followed by the addition of glycine to 0.125 M to stop the cross-linking reaction. After washing with 1× PBS, cells were lysed in 10 mM Tris (pH8.0), 10 mM NaCl, and 0.2% (vol/vol) Nonidet P-40 for 10 min on ice. Nuclei were pelleted by centrifugation and resuspended with 1.1× restriction enzyme buffer (NEBuffer 2; New England BioLabs) containing 0.3% (wt/vol) SDS. After 1 h of incubation at 37 °C, Triton-X was added to a final concentration of 2% (vol/vol) for an additional hour of incubation at 37 °C to neutralize the SDS. Chromatin then was digested by the addition of 200 U HindIII (New England BioLabs) for overnight incubation at 37 °C, followed by a second addition of 200 U HindIII for an additional 4 h. Digestion efficiency averaged 97%. Digestion was stopped by the addition of SDS to 0.8% (wt/vol) and heat inactivation at 68 °C for 10 min. Digested chromatin was purified from cellular proteins by centrifugation for 16 h at 35,000 rpm through 8 M urea in a Beckman SW40Ti rotor at 10 °C. The cross-linked chromatin pellet was resuspended in 2 mL of 30 mM Tris⋅HCl (pH 7.4) and 10 mM MgCl₂ and was dialyzed against the same buffer overnight to remove urea. Purified chromatin then was diluted by the addition of 5 mL of the same buffer followed by DTT to 10 mM and ATP to 1 mM. The chromatin then was ligated by the addition of 200 U T4 ligase (New England BioLabs) for overnight incubation at 16 °C, followed by the addition of another 200 U of T4 ligase for 4 h additional incubation. Ligated DNA was collected after overnight incubation at 65 °C with 10 μg/mL proteinase K and purification by phenol/chloroform extraction and ethanol precipitation. 3C products were quantified by Taqman-based real-time qPCR assays using a LightCycler 480 probe master kit (Roche) and a LightCycler 480 Real-Time PCR system (Roche). The following PCR program was used: 95 °C for 10 min, followed by 48 cycles of 95 °C for 10 s and 65 °C for 30 s. All PCR reactions were run in duplicate. The sequences of probes and PCR primers are shown in Table S1. An unbiased pool of 3C products generated by digestion and religation of equally mixed bMQ-440L6 and bMQ-206H21 BACs was used to generate standard curves. bMQ-440L6 spans proximal V_α/δ segments from TRAV19 to downstream of TRDV2-2, whereas bMQ-206H21 spans from J_δ2 to downstream of Dad1. Samples were normalized by setting the 3C signal between E_α and its 3′ neighbor fragment to one.

Germ-Line Transcription and Tcra Recombination.

RNA was isolated using TRIzol reagent (Invitrogen) and was converted to cDNA using SuperScript III (Invitrogen) and random hexamers according to the manufacturer’s instructions. Genomic DNA was isolated from sorted DP thymocytes by standard procedures. Tcra transcripts and rearranged DNA were quantified by real-time PCR using a QuantiFast SYBR Green PCR kit (Qiagen). All PCR reactions were run in duplicate using the following amplification program: 95 °C for 5 min, followed by 45 cycles of 95 °C for 10 s and 62 °C for 30 s. Primers for transcript analysis are shown in Table S2. Samples were normalized to signals for Actb. Primers for rearrangement analysis are shown in Table S3. Samples were normalized to signals for B2m.

ChIP.

Chromatin to be used for ChIP of acetylated histone H3 was prepared and immunoprecipitated using rabbit antiserum to acetylated H3 histone (06-599; Millipore) or control rabbit IgG (ab-105-c; R&D Systems) as previously described (52). Chromatin to be used for ChIP of CTCF was prepared by formaldehyde cross-linking as described (53) and was immunoprecipitated using rabbit antiserum to CTCF (07-729; Millipore) or control rabbit IgG (ab-105-c; R&D Systems). In both cases, samples were analyzed by real-time qPCR using the QuantiFast SYBR Green PCR kit. All PCR reactions were run in duplicate using the following amplification program: 95 °C for 5 min, followed by 45 cycles of 95 °C for 10 s and 62 °C for 30 s. PCR primers for anti-AcH3 ChIP are shown in Table S4. PCR signals were normalized first to the amount of input DNA and then to acetylation of Actb. MageA2 served as a negative control. Primers for CTCF ChIP are shown in Table S5. Signals were normalized first to input DNA and then to c-Myc for a positive control.

ChIP-Seq.

CTCF ChIP-seq was performed as described in ref. 52.

3D-FISH.

3D-FISH was performed using BAC clones RP24-334B8 and RP23-10K20 as described in ref. 35.

LM-qPCR.

Genomic DNA was prepared from sorted DP thymocytes, and linker ligation was performed as described in refs. 54 and 55. Ligated signal ends were detected by Taqman-based real-time qPCR assays. PCR reactions were performed using the program described for 3C analysis. Primers and probes are shown in Table S6.