A Barrier-Type Insulator Forms a Boundary between Active and Inactive Chromatin at the Murine TCRβ Locus Free

On their cell surface, αβ T lymphocytes express a unique TCR composed of TCRα- and TCRβ-chains that is acquired during T cell development in the thymus. The genetic loci encoding these chains (Tcra and Tcrb) consist of variable (V), diversity (D), and joining (J) gene segments that are assembled through V(D)J recombination. This recombination reaction is initiated by the lymphoid-specific recombinase proteins RAG1 and RAG2, which recognize and generate DNA double-strand breaks at recombination signal sequences that flank V, D, and J gene segments (1). V(D)J recombination occurs in a developmental stage-specific manner, with developmental control exerted in large measure through regulated access of the recombinase to selected recombination signal sequences in chromatin (2–4). An accessible chromatin configuration, in turn, is directed by TCR locus promoter and enhancer elements.

The murine Tcrb locus is roughly 700 kb in length and contains 21 functional V_β gene segments, all but one of which is located upstream of two tandem D_β-J_β-C_β gene segment clusters (Fig. 1A). The sole exception is the V_β14 gene segment, which lies downstream of the C_β2 gene segment at the very 3′ end of the locus. A special feature of the Tcrb locus is the arrangement of 19 functional V_β gene segments in a central cluster that is separated from the 5′ distal V_β2 gene segments by ∼150 kb containing seven trypsinogen genes and gene fragments, and from the D_β-J_β-C_β clusters by a 250-kb region comprising 13 trypsinogen genes and gene fragments. Notably, whereas V_β, D_β, and J_β gene segments are all transcribed and undergo recombination in CD4⁻CD8⁻ double-negative (DN) thymocytes, trypsinogen genes are not expressed and adopt a relatively closed chromatin conformation in the same cells (5). Hence the locus displays a modular structure with alternating regions of active and inactive chromatin in DN thymocytes.

A Tcrb enhancer (E_β) located between C_β2 and V_β14 works in collaboration with promoters associated with D_β1 and D_β2 to activate the chromatin associated with D_β and J_β gene segments (6–10). Deletion of this enhancer severely reduces D_β-J_β-C_β germline transcription and chromatin modifications, and nearly abolishes Tcrb locus recombination, including the initial D_β-to-J_β step and the subsequent V_β-to-D_βJ_β step (6, 7, 9, 10). However, deletion of E_β does not influence transcription and chromatin structure of unrearranged V_β gene segments, which are activated independently (10). Similarly, E_β does not appear to influence transcription and chromatin structure of the trypsinogen genes, because these genes are inactive in DN thymocytes (5). Rather, the influence of E_β is limited to no more than 25 kb encompassing the two D_β-J_β-C_β clusters near the 3′ end of the Tcrb locus (10).

In eukaryotic cells, active and inactive genes can coexist and maintain differential regulation despite their proximity to multiple sets of regulatory elements. Insulators prevent the misregulation of neighboring genes by delimiting the chromatin domains over which regulatory elements can function. Two types of insulator elements are involved in the establishment of genomic domains: enhancer-blocking elements that interfere with the ability of enhancers to activate the promoters of neighboring inactive genes, and barrier elements that block the spread of repressive heterochromatin into regions containing active genes (11, 12). In vertebrates, enhancer-blocking is mediated by the DNA-binding protein CTCF (CCCTC-binding factor), which is thought to function by promoting the formation of chromatin loops (13–15). Several mechanisms have been proposed to explain barrier activity, but a common theme is the recruitment of histone-modifying enzymes that deliver activating chromatin modifications that can interrupt the propagation of silent chromatin (16–19). Although transcriptional activity has been demonstrated for many yeast insulators with barrier function, transcription is not an absolute requirement for this function (20, 21).

To ask whether insulator activity helps to maintain the integrity of adjacent regulatory domains of the Tcrb locus, we mapped chromatin modifications to define the border between active D_β-J_β-C_β and inactive trypsinogen chromatin. Remarkably, this border, situated ∼6 kb upstream of promoter D_β1 (PD_β1), corresponded to an endogenous long terminal repeat (LTR). This element was shown to have bona fide barrier activity, because it could insulate an integrated reporter construct from chromosomal position effects in transfected cells. We discuss the implications of our observations in the context of Tcrb locus activation during thymocyte development.

Rag2^−/− mice (22), Rag1^−/− E_β^−/− mice (10) (kind gift from P. Ferrier, Centre d’Immunologie Marseille-Luminy, Marseille, France), and Rag2^−/− × Tcrb transgenic mice (23) were used in accordance with protocols approved by the Duke University Animal Care and Use Committee.

Histone modifications were analyzed by chromatin immunoprecipitation (ChIP) using antiacetylated histone H3 (anti-H3ac; no. 06599; Upstate Biotechnology-Millipore), antidimethylated H3 lysine 4 (anti-H3K4me2; no. 07030; Upstate Biotechnology-Millipore), antitrimethylated H3 lysine 4 (anti-H3K4me3; no. 04745; Upstate Biotechnology-Millipore), and antidimethylated histone H3 lysine 9 (anti-H3K9me2; no. Ab1220; Abcam). ChIP with rabbit IgG (no. AB-105-C; R&D Systems) served as a control. Chromatin was prepared by one of two methods. For some experiments, chromatin was prepared in large scale by micrococcal nuclease digestion and purification of mononucleosomes or a mixture of dinucleosomes and trinucleosomes by sucrose gradient fractionation as described previously (24). For other experiments, chromatin was prepared in small scale by micrococcal nuclease digestion as follows: 1 × 10⁷ thymocytes were washed and lysed as described previously (24). Nuclei were then resuspended in 200 μl digestion buffer, and 5 U micrococcal nuclease (Worthington) was added for 5 min at 37°C to generate a mixture consisting primarily of mononucleosomes with a small fraction of dinucleosomes. The reaction was stopped by addition of 300 μl of 10 mM Tris-HCl, pH 8.0, 5 mM EDTA, 10 mM sodium butyrate, after which the sample was sonicated on ice for eight 15-s intervals using a Sonicator 3000 (Misonix) with the output set to 3.0. After centrifugation, the supernatant was harvested and Triton X-100 was added to a concentration of 1% (v/v). Chromatin was precleared with protein A-Sepharose/salmon sperm DNA slurry (Upstate Biotechnology-Millipore) and was incubated with the appropriate specific Ab overnight at 4°C followed by addition of protein A-Sepharose/salmon sperm DNA slurry for 1 h. Immunoprecipitates were washed vigorously and DNA was purified. Bound and input fractions were quantified by quantitative real-time PCR using a LightCycler and FastStart DNA Master SYBR Green I kit (Roche) or a QuantiFast SYBR Green PCR kit (Qiagen). Ratios of bound to input for modifications H3ac and H3K4me2 were normalized to those for β₂-microglobulin; ratios for H3K9me2 were normalized to those for MageA2. PCR reaction conditions were as follows: 5 min at 95°C, followed by 45 cycles of 1 s at 95°C, 5 s at 60–64°C (depending of the primer pair), 7 s at 72°C. PCR primers (Supplemental Table I) matched perfectly only a single Tcrb locus site in the C57BL/6 genome and were judged to amplify unique, single-copy sequences on the basis of melting curve analysis and amplification efficiency from input chromatin.

For analysis of CTCF binding, cross-linked chromatin was prepared and used for ChIP as described previously (25) using 5 μg anti-CTCF Ab (no. 07-729; Upstate Biotechnology-Millipore). Bound and input materials were quantified as described earlier. A known CTCF binding site in the c-myc locus insulator (26) was used as a positive control. PCR primers are shown in Supplemental Table II.

5′ RACE was performed on thymocyte RNA isolated from a Rag2^−/− Tcrb transgenic mouse using a GeneRacer Kit (Invitrogen) according to the manufacturer's instructions. Strand-specific primers were used for PCR amplification of cDNA products. PCR products were purified by agarose gel electrophoresis and cloned using a TOPO TA Cloning Kit (Invitrogen). The anti-sense primer tested was 5′-AGCCACAGTGTGTGGTCTGCTACAG-3′. Two sense primers (5′-CAAAGACTCAGAGAAGGTTGTTCACTC-3′ and 5′-GATCAGTTGGAGTAACAGGAAAGAGAG-3′) were also tested but did not amplify specific products.

RNA was extracted using TRIzol reagent (Invitrogen) following the manufacturer’s specifications. Three micrograms RNA were digested with 3 U RNase-free DNase (Sigma) in 33 μl digestion buffer (20 mM Tris-HCl, pH 8.0, 50 mM KCl, 2 mM MgCl₂) for 15 min at 23°C. After addition of EDTA to a final concentration of 2.5 mM and incubation at 68°C for 10 min, the RNA was split in two aliquots that were incubated with SuperScript III (Invitrogen) or mock incubated, according to the manufacturer’s instructions. The resulting cDNAs were diluted 1:10 (or 1:100 for Actb analysis) and used for conventional PCR using primers P1 (5′-GGCTATCTGCAACACAGATCTT-3′) and P3 (5′-TGATGGCTCAAACAAGGAGAC-3′) or quantitative real-time PCR using primers P1 and P2 (5′-AAGCCACAGTGTGTGGTCTG-3′). Experimental PCR values were normalized to those for Actb (primers 5′-TCGTAGATGGGCACAGTGTG-3′ and 5′-GGTCAGAAGGACTCCTATGTG-3′). To quantify transcripts detected by the P1-P2 primer set relative to total germline Tcrb transcripts, we conducted amplifications with P1-P2 and amplifications with a C_β primer set (5′-AAAAGGCTACCCTCGTGTGC-3′ and 5′-CTGTGGACCTCCTTGCCATT-3′) from both cDNA and genomic DNA. Percent germline transcripts using the 470-bp exon was estimated as follows: (P1-P2_cDNA/P1-P2_{genomic DNA})/(C_{β cDNA}/C_{β genomic DNA}) × 100.

To create a reporter construct that would be sensitive to chromosomal position effects, we generated a pGL3 (Promega)-based vector that expressed GFP under the control of a weak promoter (human V_δ1 promoter [PV_δ1]) and enhancer (human Tcrd enhancer [E_δ]). The 1.8-kb PV_δ1 fragment was excised from plasmid EVNS’ (27) with SalI and HindIII, and cloned into the pGL3-promoter vector digested with XhoI and HindIII. The luciferase gene was then excised by XbaI digestion, followed by Klenow treatment to obtain a blunt end, and subsequent digestion with NcoI. The GFP gene, isolated from plasmid MSCV-IRES-GFP by HindIII digestion, Klenow treatment, and digestion with NcoI, was then introduced into the pGL3 vector in place of luciferase. A 380-bp E_δ fragment, obtained from plasmid EVNS’ by digestion with XbaI and EcoRI, followed by Klenow treatment, was then cloned downstream of GFP by ligation into a BamHI site rendered blunt by Klenow treatment. The resulting plasmid, pGL3-PV_δ1-GFP-E_δ, was used to assay fragments for barrier activity, with two copies of each test fragment introduced both upstream of the promoter and downstream of the enhancer. Test fragments were amplified by PCR to introduce terminal MluI and NheI sites or NheI and XhoI sites, and the two PCR fragments were then subcloned stepwise into the multiple cloning site of pGL3-promoter. For cloning into the upstream site, the fragments were excised as a dimer by XhoI digestion, Klenow treatment, and MluI digestion. The excised dimer was then cloned into MluI and SmaI digested pGL3-PV_δ1-GFP-E_δ. For cloning into the downstream site, the dimer was excised by digestion with XhoI and MluI, followed by Klenow treatment, and then introduced into the AfeI site of the plasmid. Test fragments were: HS4, a 1.2-kb fragment containing the chicken β-globin 5′ hypersensitive site (HS) 4 insulator (positive control, amplified using 5′-TGAGAACGCGTCCGCGGATCTCACGGGGACAG-3′ and 5′-AGAGAGCTAGCTCACTGACTCCGTCC-3′); fragment A, a 1.5-kb fragment spanning from 7.6 to 6.1 kb upstream of D_β1 (MMAE000665 145657–147165, amplified using primers 5′-ATGCACGCGTACGCATCAGAAGAGGGCATC-3′ and 5′-ATGCGCTAGCGATCTGCAGTTAGCTCACTGG-3′); and fragment B, a 1.5-kb fragment spanning from 6.9 to 5.4 kb upstream of D_β1 (MMAE000665 146357–147868, amplified using primers 5′-ATGCACGCGTACTCAGAGAAGGTTGTTCACTC-3′ and 5′-ATGCGCTAGCAAGCCACAGTGTGTGGTCTG-3′).

Jurkat cells were maintained in RPMI 1640 medium supplemented with 10% FBS, 100 U/ml penicillin, and 0.1 mg/ml streptomycin. The cells (1 × 10⁶) were transfected with 1 pmol linearized test plasmid plus 0.33 pmol of a linearized puromycin resistance plasmid by electroporation using an Amaxa Nucleofector system (Lonza), as instructed by the manufacturer. Transfected Jurkat cells were then cultured and expanded in nonselective medium. After 6 d of culture, cells expressing intermediate levels of GFP were sorted on a Beckman-Coulter MoFlo and plated at a density of 1× 10⁵/ml in selective medium containing 0.2 μg/ml puromycin. After 2 wk of selection, GFP⁺ cells were resorted using a BD Biosciences FACSVantage (defined as day 0) and cultured in the presence or absence of puromycin for up to 42 d to monitor the loss of GFP expression using a BD Biosciences FACSCanto II.

Posttranslational modifications of histone tails can reflect the transcriptional status and overall configuration of chromatin: open and transcriptionally active euchromatin is typically marked by acetylation of histone H3 on lysines 9 and 14 and dimethylation of histone H3 on lysine 4, whereas condensed and transcriptionally silent heterochromatin is typically enriched in histone H3 dimethylated at lysine 9 (17, 28–30). To define the edge of the E_β-regulated chromatin domain, we used ChIP to analyze an 11-kb region spanning from the silent Prss2 gene (the most 3′ trypsinogen gene) to the active PD_β1 in DN thymocytes of Rag2^−/− mice (Fig. 1A, 1B). We detected a signature of highly active chromatin at PD_β1 (H3ac and H3K4me2) and of inactive chromatin at Prss2 (H3K9me2). Notably, there was a sharp transition from H3K9me2 to H3ac over a 1-kb region situated between 5.5 and 6.5 kb upstream of D_β1 and between 2 and 3 kb downstream of the Prss2 polyadenylation signal. Not surprisingly, we detected elevated H3K4me2 that mapped to PD_β1 (31). However, we also noted a peak of H3K4me2, centered 6.2 kb upstream of D_β1, that mapped to the transition between H3K9me2 and H3ac. This suggested that chromatin-modifying enzymes may be recruited to this site by an active cis-regulatory element.

Repressive chromatin, as reflected by the H3K9me2 mark, may spread until opposed by chromatin-modifying activities that deliver activating histone modifications such as H3ac and H3K4me2 (32, 33). If this is the case at the Prss2-D_β1 boundary, the activity of E_β itself could be critical to prevent the encroachment of heterochromatin into the D_β-J_β-C_β region. Previous studies showed that, in the absence of E_β, the D_β-J_β-C_β region is characterized by a loss of germline transcription that is associated with reduced histone acetylation, increased DNA methylation, and resistance to nuclease digestion (10). Increased H3K9me2 was also detected at both D_β1 and C_β1 (31).

To determine whether these changes reflected the spreading of heterochromatin from the 5′ boundary, we compared histone modifications across this region in DN thymocytes of Rag2^−/− mice in the presence or absence of E_β. E_β deletion abolished H3ac in the immediate vicinity of D_β1 and partially reduced H3ac 5.5 and 3.4 kb upstream of D_β1 (Fig. 2A). E_β deletion also abolished H3K4me2 at D_β1, but this modification was unperturbed at the 5′ boundary (Fig. 2B). Of note, E_β-deficient alleles displayed a small increase in H3K9me2 at PD_β1, but no increase in H3K9me2 in the region between the 5′ boundary and PD_β1 (Fig. 2C). This indicates that E_β plays no role in halting the spread of heterochromatin. Rather, if such an activity is present, it is likely mediated by an autonomous element that maintains an active chromatin configuration at the boundary.

Although a barrier effect may be achieved by disruption of a nucleosome array (34) or attachment to nuclear structures (35), many of the best described barriers including the silent mating-type loci HMR and HML in yeast (16, 36) and chicken β-globin 5′ HS4 (18) share the property that they are highly enriched in histone modifications associated with transcriptional activation (H3ac, H4ac, and H3K4me2). However, transcription itself has been shown not to be necessary for barrier activity (21).

Because the heterochromatin-to-euchromatin transition at the 5′ boundary of the E_β regulatory domain coincided with a discrete H3K4me2 peak, we asked whether a transcription unit mapped to this region. Accordingly, we searched for transcripts that could initiate in the region by analyzing thymocyte RNA in a 5′-RACE assay using a variety of specific primers. We were unable to amplify PCR fragments using primers designed to detect anti-sense (relative to the orientation of D_β1-J_β1-C_β1) transcripts (data not shown). However, we did detect and clone several PCR products indicative of sense transcription initiating within a 50-bp region ∼5.7 kb upstream of D_β1 (Fig. 3A).

Analysis of the DNA sequence around the start sites revealed the presence of an endogenous retroviral LTR (LTR BglII family endogenous retrovirus K [ERVK]) (Fig. 3A). This retroviral LTR is 397 bp in length and contains several putative CCAAT and TATA boxes. It is flanked by a 750-bp long interspersed element (LINE) Mur3 L1 fragment that is affixed at its 3′ end to another 365-bp LTR11 family ERVK. The detected transcripts initiate near the junction between the 5′ LTR and the LINE element, whereas the peak of H3K4me2 is centered further upstream at the 5′ edge of the 5′ LTR.

To more fully characterize the structure of these transcripts, we carried out RT-PCR using a forward primer (P1) that anneals across the LTR-LINE junction (to ensure specificity) and a reverse primer (P3) that anneals to C_β1 (Fig. 3B). We identified two major products that were shown by cloning and sequencing to correspond to differentially spliced versions of primary transcripts that run from the LTR at least 11 kb through C_β1 (Fig. 3B, 3C). The transcripts include three novel exons of 470, 780, and 72 bp, which are defined by splice sites that conform well to the 5′ (^A/_CAG-GT^A/_GAGT) and 3′ (CAG-G) splice-site consensus sequences (Fig. 3B, 3C). The 3′ end of the 780-bp exon aligns with Jβ1.2; to the best of our knowledge, the remaining spice sites have not been described previously.

To evaluate the regulation of transcript expression, we analyzed cDNAs prepared from several RNA sources by quantitative real-time PCR using primers P1 and P2 (Fig. 3B, 3D). Transcripts were detected in DN thymocytes of Rag2^−/− mice and in CD4⁺CD8⁺ double-positive thymocytes of Rag2^−/− mice that express a Tcrb transgene (Rag2^−/−Tcrb), but were detected at much reduced levels in DN thymocytes of Rag1^−/− E_β^−/− mice (ΔE_β DN) and in splenocytes of Rag2^−/− Tcrb mice (non-T). Thus, the transcripts are both T cell specific and E_β dependent. However, even in DN thymocytes of Rag2^−/− mice, the abundance of these transcripts was low relative to standard germline Tcrb transcripts; by real-time PCR, transcripts detected by P1-P2 were only ∼0.1% as abundant as those detected by a pair of C_β primers (data not shown). One caveat is that because primer P1 spans the LTR-LINE junction, it can detect only a fraction of the transcripts initiating in this region (Fig. 3A). However, consistent with low-level transcription, the H3K4me3 modification, whose deposition is tightly linked to transcription (37), was readily detected at PD_β1 but was barely detected above background in the LTR-LINE region even on wild-type Tcrb alleles (Fig. 3E). We conclude that H3K4me2 and H3K4me3 deposition are regulated independently in this region, and that neither H3K4me3 nor transcription appears to be involved in forming the identified chromatin boundary.

To determine whether the boundary region contains bona fide barrier activity, we developed a chromosomal position effect assay similar to that used in previous studies (38, 39). This assay relies on a base plasmid in which GFP expression is driven by the human TCRδ enhancer (E_δ) and PV_δ1, with test fragments cloned as dimers both upstream and downstream of the GFP expression cassette (Fig. 4A). Test plasmids were cotransfected into Jurkat cells together with a plasmid conferring puromycin resistance. After selection and sorting for GFP expression, transfected cells were cultured for up to 6 wk in the presence or absence of puromycin. Negative chromosomal position effects were then scored as the loss of GFP expression in a fraction of the transfected cells. Because selection and test plasmids should cointegrate, GFP loss should not be observed with culture in the presence of puromycin. However, position effects should be revealed during culture in the absence of puromycin, because there should be no selection against silencing at the integration site under these conditions. As expected, in the absence of selection, we found that GFP expression was diminished over time in a fraction of the transfectants when no insulator was present, but was stably maintained when the expression cassette was flanked by the positive control insulator chicken β-globin 5′-HS4 (Fig. 4B, 4C). GFP expression was also lost over time when the expression cassette was flanked by a 1.5-kb test fragment situated upstream of LTR BglII ERVK (fragment A). However, GFP expression was stably maintained when the cassette was flanked by an overlapping 1.5-kb test fragment that included the LTR (fragment B). Notably, this fragment was as effective as chicken β-globin 5′-HS4. These data support the argument that the LTR functions as a barrier-type insulator that can prevent repressive chromosomal position effects.

Tcrb locus organization is striking in the sense that it presents alternating blocks of active chromatin composed of Tcrb gene segments and inactive chromatin composed of unrelated Prss genes and gene fragments. This unique organization poses intrinsic regulatory problems, particularly at the borders between active and inactive chromatin. Such is the case at the 3′ end of the locus, where a 250-kb domain of inactive trypsinogen chromatin is juxtaposed to a 25-kb domain of active D_β-J_β-C_β chromatin that is regulated by E_β. In this article, we identified an LTR at the border between these chromatin domains, and showed that LTR-containing DNA functions as a barrier-type insulator that can protect a transgene from negative chromosomal position effects. This LTR colocalizes with a peak of histone modification H3K4me2, suggesting that it may share with other barrier-type insulators the capacity to recruit histone-modifying enzymes that can block the propagation of silent chromatin (16–19). We suggest that by functioning in this way, this LTR helps to maintain the integrity of the E_β-regulated chromatin domain in the murine Tcrb locus.

Previous studies have revealed that some retroviral LTRs are transcriptionally active, and that they can activate the expression of neighboring or distant genes (40–42). However, we are unaware of previous instance in which a retroviral LTR was implicated in chromatin barrier activity. We note that although the overall organization of the human Tcrb locus is similar to that of mouse, interspecies sequence conservation in this region is restricted to the Prss2 gene and the PD_β1 only. Nevertheless, the human locus carries a 695-bp LTR (LTR8, family ERV1) that is located 4.6 kb upstream of D_β1. This LTR may serve a similar function in the human locus.

Insulators can encode two separable activities: barrier activity that prevents the spread of repressive chromatin and enhancer-blocking activity that prevents an enhancer on one side of the insulator from activating a promoter on the other (11, 12). The well-studied chicken 5′-HS4 insulator possesses both activities, with barrier activity mediated by USF1 and enhancer-blocking activity mediated by CTCF. Although our results provide evidence of barrier activity in the Prss2-D_β1 interval, they do not address directly the potential for enhancer-blocking activity. The only vertebrate enhancer-blocking protein identified to date is CTCF (13). We used a CTCF consensus binding-site algorithm (43, 44) to search for potential binding sites in the Prss2-D_β1 interval and found two possible candidates near DNase I HSs 10 and 11 located 2.4 and 3.4 kb upstream of D_β1 (45). However, as compared with positive controls including c-myc (26) and a site between E_β and V_β14, ChIP analysis revealed no significant binding of CTCF to these sites or to several additional points tested across the chromatin boundary defined in this study (J. Carabana and M.S. Krangel, unpublished observations). This suggests that the region is unlikely to contain a classical enhancer-blocking element. In contrast, two CTCF binding sites with demonstrated enhancer-blocking activity have been defined just upstream of DFL16.1 in the murine Igh locus (46).

Previous studies of barrier-type insulators HMR and HML in yeast (20, 21) and chicken β-globin 5′HS4 (18) have demonstrated that transcription is not required for barrier activity. Nevertheless, we detected low-level, E_β-dependent and T lineage-specific transcriptional activity associated with the barrier region, with transcripts extending for at least 11 kb across the D_β1-J_β1-C_β1 cluster. We do not know whether this transcription, which initiates at the LTR-LINE junction, depends on LTR or LINE sequences. Both elements can have active promoters; moreover, LINE transcripts are unusual because they initiate at the 5′ end of the LINE and run through a downstream LINE promoter (47, 48). Further work will be required to clarify this issue. Regardless, we suspect that this transcription is not important for barrier activity, because repressive H3K9me2 chromatin fails to spread in E_β-deficient mice, even though transcription, which is low on wild-type alleles, is even lower on E_β-deficient alleles.

We speculate that transcription initiating in the boundary region may serve a distinct and complementary function by promoting opening of the E_β-regulated chromatin domain. Previous work from Oltz and colleagues (9) demonstrated that maximal chromatin accessibility and histone modifications in the D_β1-J_β1-C_β1 region depend on functional interaction between E_β and PD_β1, but that E_β possesses an intrinsic chromatin opening function that is apparent in the absence of PD_β1. Chromatin opening might then proceed in stepwise fashion, with the initial step dependent on E_β only, and a subsequent step, resulting in recombinase accessibility, that requires activation of PD_β1 by E_β. Because LTR-LINE transcription across the D_β1-J_β1-C_β1 region is E_β dependent, this transcription may provide PD_β1-independent chromatin remodeling activity that might otherwise be attributed to E_β only, and that might support developmental activation of PD_β1. In this regard, recent studies have implicated low-level transcription initiating at a retroviral LTR situated at a distance from the human globin genes in their developmental activation (42). Thus, the boundary region may predispose D_β-J_β-C_β chromatin to become accessible in two very different ways: it may prevent invasion of repressive histone modifications that would act to oppose chromatin opening, and it may provide a chromatin opening function as well. That said, a chromatin opening function has yet to be demonstrated, and it remains to be proved whether the barrier activity exhibited in our rather artificial Jurkat position effect assay is predictive of similar activity in the context of endogenous cis-regulatory elements at the murine Tcrb locus. Additional experiments will be required to dissect the different functional activities of this element and to assess their biological significance in vivo.

We thank Hrisavgi Kondilis-Mangum and Han-Yu Shih for critical review of the manuscript and Zanchun Huang for technical assistance. We also thank Nancy Martin and Lynn Martinek (Duke University Cancer Center Flow Cytometry Shared Resource) for help with cell sorting.

The authors have no financial conflicts of interest.