A Role for Histone Acetylation in the Developmental Regulation of V(D)J Recombination

See allHide authors and affiliations

Science  21 Jan 2000:
Vol. 287, Issue 5452, pp. 495-498
DOI: 10.1126/science.287.5452.495


V(D)J recombination is developmentally regulated in vivo by enhancer-dependent changes in the accessibility of chromosomal recombination signal sequences to the recombinase, but the molecular nature of these changes is unknown. Here histone H3 acetylation was measured along versions of a transgenic V(D)J recombination reporter and the endogenous T cell receptor α/δ locus. Enhancer activity was shown to impart long-range, developmentally regulated changes in H3 acetylation, and H3 acetylation status was tightly linked to V(D)J recombination. H3 hyperacetylation is proposed as a molecular mechanism coupling enhancer activity to accessibility for V(D)J recombination.

V(D)J recombination is initiated by recombinase activating gene–1 (RAG-1)– and RAG-2–mediated cleavage between T cell receptor (TCR) and immunoglobulin coding gene segments (V, D, and J) and flanking recombination signal sequences (RSSs) (1). Chromosomal and nucleosomal RSSs are refractory to RAG-mediated cleavage relative to naked DNA (2, 3), and V(D)J recombination is thought to be regulated in vivo by enhancer- and promoter-dependent changes in chromatin structure that provide RAG proteins access to specific RSSs (4). However, the nature of chromatin structural modifications associated with accessibility to RAG proteins is not known. Recent studies indicate that enhancers and promoters can direct the hyperacetylation of core histones (5) due to histone acetyltransferase activity of transcriptional coactivators (6). Moreover, histone hyperacetylation alters chromatin structure, as suggested by increased general sensitivity to endonucleases (7) and increased binding of transcription factors (8). Hyperacetylation of histones therefore provides a potential mechanism linking enhancer and promoter activity to RSS accessibility to RAG proteins.

We initially addressed the relation between histone acetylation and accessibility for V(D)J recombination by studying a TCRδ minilocus V(D)J recombination reporter in thymocytes of transgenic mice (Fig. 1, A and B). This reporter contains unrearranged human Vδ, Dδ, Jδ, and Cδ gene segments. With a functional enhancer in the Jδ3-Cδ intron, V, D, and J gene segments are all accessible to RAG proteins, and fully rearranged products (VDJ) are efficiently generated (9, 10). With the enhancer deleted or mutated, only V-to-D rearrangement is observed (9, 11, 12). The failure of VD-to-J rearrangement without an enhancer reflects an inability of RAG proteins to access and cleave J segment RSSs (13). Because V and D accessibility is maintained in an enhancer-independent manner, minilocus accessibility is sharply divided between an enhancer-independent 5′ region and an enhancer-dependent 3′ region.

Figure 1

Diagrammatic depictions of thymocyte development and V(D)J recombination substrates analyzed. (A) DN, DP, and SP represent consecutive stages of T cell development that typically comprise 1 to 5%, 80 to 90%, and 10 to 15%, respectively, of total thymocytes. Developmental blockades in RAG-2−/−, RAG-2−/− × TCRβ and Eα−/− mice are depicted by dashed lines. (B) Human TCRδ minilocus constructs vary in the enhancer included in the Jδ3-Cδ intron. Rearrangement and accessibility phenotypes are shown. Eδ line A and Eα line J undergo both V to D and VD to J rearrangement (9, 10), whereas E- line H and EδmCore line Z undergo only V to D rearrangement (9,12). Filled circles locate segments analyzed by CHIP. (C) Murine TCRα/δ locus. Typical TCRδ and TCRα rearrangement events (occurring in DN and DP thymocytes, re- spectively) are shown. Arrows identify transcriptional promoters. Filled circles locate segments analyzed by CHIP. BEAD, blocking element alpha delta; and TEA, T early alpha promoter and exon.

To test whether accessibility reflects the acetylation status of minilocus chromatin, we measured histone H3 acetylation by a chromatin immunoprecipitation (CHIP) assay (14) using αAcH3 antiserum. Serially diluted DNA samples isolated from the input, αAcH3-bound, and -unbound fractions were tested by polymerase chain reaction (PCR) (Table 1) (15) to analyze specific sites along the minilocus. We initially compared a minilocus containing Eα [transgenic line J (10)] to one lacking an enhancer [transgenic line H (9)]. Eα turns on at the double positive (DP) stage and is therefore active in >95% of thymocytes (Fig. 1A) (10, 16). Both Vδ1 and Cδ were hyperacetylated in unfractionated thymocytes of Eα line J, because they were heavily enriched in the αAcH3-bound fraction as compared with either the unbound fraction, the input fraction, or the bound fraction of a control immunoprecipitate (IP) (Fig. 2A). In contrast, the B cell–specific gene Oct-2 was hypoacetylated, because it was poorly represented in the αAcH3-bound fraction and displayed a signal nearly equivalent to that in the bound fraction of the control. Acetylation at both Vδ1 and Cδ was sharply reduced in thymocytes of E- line H. Thus, Eα promotes H3 hyperacetylation at distant sites in the transgene. However, even without Eα, acetylation of Vδ1 was substantially above that of Oct-2. Thus, a component of Vδ1 acetylation is enhancer-independent.

Figure 2

H3 acetylation within a human TCRδ minilocus. (A) H3 acetylation was measured in total thymocytes of mice carrying minilocus constructs that include (J) or lack (H) Eα. CHIP was performed with anti-AcH3 or, as a control, no antibody, and serial threefold dilutions of bound (B), unbound (U), and input fractions were analyzed by PCR. (B) H3 acetylation was measured as in (A) in the DN thymocytes of RAG-2−/− mice carrying a minilocus with either a wild-type (R×A) or a mutant (R×Z) Eδ. Acetylation is plotted as B/U, calculated as described (18). (C) H3 acetylation was measured as in (A) in the DP thymocytes of RAG-2−/− × TCRβ transgenic mice carrying a minilocus with a wild-type Eδ (R×A×β). Control CHIP was with normal rabbit serum.

Table 1

Primers and probes.

View this table:

To further analyze enhancer-dependent and -independent aspects of minilocus acetylation, we compared a minilocus containing a wild-type Eδ [transgenic line A (9)] to one in which Eδ is inactivated by a 3–base pair (bp) mutation in the binding site for CBF/PEBP2 [transgenic line Z (12)]. Because Eδ is only occupied and active in double negative (DN) thymocytes (16), both transgenes were bred onto the RAG-2−/− background (generating lines R×A and R×Z), in which thymocyte development is blocked at the DN stage (17). The lack of V(D)J recombination on this background also allowed analysis of a minilocus that was unrearranged in all thymocytes.

In DN thymocytes of Eδ line R×A, the minilocus was hyperacetylated along its entire length, with acetylation 25-fold overOct-2 for Vδ1; >100-fold over Oct-2 for Dδ3, Jδ1, and Eδ; and 10-fold over Oct-2 for Cδ (Fig. 2B) (18). This profile was markedly different in thymocytes of EδmCore line R×Z. Acetylation at Vδ1 and Dδ3 was 25- and 18-fold, respectively, over Oct-2. However, acetylation at Jδ1, Eδ, and Cδ was much lower. Thus, the histone H3 acetylation patterns of the Eδ and EδmCore miniloci are fully consistent with the previously determined patterns of V(D)J recombination and accessibility to RAG proteins. Concordance of the acetylation and accessibility profiles leads us to propose that enhancer-independent acetylation of V and D gene segments allows for enhancer-independent minilocus V-to-D rearrangement, whereas enhancer-dependent acetylation of J gene segments allows for enhancer-dependent minilocus VD-to-J rearrangement.

Eδ is inactivated upon transit of thymocytes from DN to DP due to loss of occupancy at binding sites for CBF/PEBP2 and c-Myb (16). To investigate changes in histone H3 acetylation across this developmental transition, we bred a rearranged TCRβ transgene into Eδ line R×A to generate R×A×β mice, in which nearly 100% of thymocytes are DP (19). Enhancer-dependent acetylation was largely extinguished across the transgene, with the exception of some residual acetylation over Eδ itself (Fig. 2C). Thus, developmental loss of Eδ occupancy causes a marked reduction in minilocus H3 acetylation. Enhancer-independent acetylation at Vδ1 was maintained in DP thymocytes, but that at Dδ3 was extinguished. Enhancer-independent acetylation of these elements appears to be under distinct control.

To further probe the relation between H3 acetylation and accessibility for V(D)J recombination, we measured developmental changes in acetylation at the endogenous TCRα/δ locus (Fig. 1C). This locus contains distinct sets of gene segments that are activated for V(D)J recombination at different stages of T cell development (20). TCRδ rearrangement (Vδ-Dδ-Jδ) occurs in DN thymocytes and TCRα rearrangement (Vα-Jα) in DP thymocytes (Fig. 1A). Eδ activity at the DN stage (10,16) is partly responsible for TCRδ accessibility, because TCRδ rearrangement is partially inhibited in Eδ−/− mice (21). TCRα rearrangement is markedly impaired in Eα−/− mice (22), indicating that Eα activity at the DP stage (10,16) is critical for accessibility across the 70-kb Jα region. Eα is also required for transcription from the germ line T early alpha (TEA) promoter 5′ of Jα segments, and up-regulates TCRδ transcription as well (22).

We analyzed RAG-2−/− mice (17) to assess acetylation in DN thymocytes, RAG2−/− × TCRβ (R×β) mice (19) to assess acetylation in DP thymocytes, and Eα−/− mice [which, like R×β mice, are blocked at DP (22)] to assess the effect of Eα. The DN compartment (RAG-2−/−) is characterized by a hyperacetylated TCRδ region [Eδ, Cδ (23), and Vδ5] and a hypoacetylated TCRα region (TEA, Jα25, and Cα) (Fig. 3). The blocking element alpha delta (BEAD) region spanning Vδ5 to TEA, previously shown to have enhancer-blocking activity (24), appears to be a transition zone. Notably, although the Jα-Cα region is hypoacetylated in DN thymocytes, Eα itself is hyperacetylated. Because Eα is occupied but inactive in DN thymocytes (16), local acetylation over Eα likely relates to its occupancy, rather than activity.

Figure 3

H3 acetylation within the endogenous murine TCRα/δ locus. H3 acetylation measured in the DN thymocytes of RAG-2−/− mice (27), the DP thymocytes of RAG-2−/− × TCRβ (R×β) mice, and the DP thymocytes of Eα−/− mice. Control CHIP was with normal rabbit serum. Acetylation is plotted as in Fig. 2B. NA, not applicable.

The transition to DP (R×β) was marked by a pronounced shift to a hyperacetylated state over the entire Jα-Cα region, and an increase in acetylation over Vδ5 (Fig. 3) and Cδ (23). These changes reflect activation of Eα, because they were not detected in DP thymocytes of Eα−/− mice. Thus, Eα modulates histone acetylation over at least 85 kb, paralleling its effects on both V(D)J recombination and transcription. Although Eδ itself is inactivated on transition to DP (16), H3 acetylation in the TCRδ region was maintained in Eα−/− DP thymocytes. Thus, TCRδ acetylation is determined, at least in part, by elements other than Eδ, a notion consistent with the residual TCRδ rearrangement in Eδ−/− mice (21).

In summary, analysis of both a transgenic V(D)J recombination reporter substrate and the endogenous TCRα/δ locus indicate Eδ and Eα to be long-range developmental regulators of H3 acetylation. H3 hyperacetylation is a consequence of enhancer occupancy, because a 3-bp Eδ mutation that prevents CBF/PEBP2 binding prevents enhancer-dependent minilocus H3 hyperacetylation. Nevertheless, the precise mechanism that translates enhancer occupancy into long-range changes in histone acetylation is not known. In both the minilocus and the endogenous locus, regions displaying H3 hyperacetylation were invariably those displaying accessibility to the recombinase. Because H3 hyperacetylation is highly predictive of gene segment accessibility yet exists in the absence of recombination (i.e., on a Rag-2−/− background), we propose that it plays a primary role in establishing accessibility for V(D)J recombination. Histone hyperacetylation was recently reported not to affect accessibility of mononucleosomal RSSs to recombinant RAG proteins in a simple in vitro system (3). Acetylation-dependent accessibility may therefore require more-complex chromosomal substrates or remodeling activities that are present in vivo but not yet reproduced in vitro. We suggest a model for V(D)J recombination in which cis-regulatory elements direct access to RAG proteins in vivo by inducing the region- and developmental stage–specific hyperacetylation of histone H3.

  • * To whom correspondence should be addressed. E-mail: krang001{at}


View Abstract

Navigate This Article