Chromosomal position of a VH gene segment determines its activation and inactivation as a substrate for V(D)J recombination.

Complete IgHC gene rearrangement occurs only in B cells in a stage-specific and ordered manner. We used gene targeting to reposition a distal V(H) gene segment to a region just 5' of the D(H) gene cluster and found its activation to be highly dependent on the chromosomal domain within which it resides. The targeted V(H) gene segment rearranged at a higher frequency than its endogenous counterpart, its rearrangement was no longer ordered, and its ability to be silenced by allelic exclusion was lost. Additionally, the targeted V(H) gene segment lost lineage specificity, as VDJ(H) rearrangement was observed in thymocytes. These data suggest that locus contraction, mimicked by proximal targeting, can override any regulation imposed by DNA sequences immediately surrounding V(H) gene segments.

V(D)J recombination underlies the remarkable diversity of antigen receptors in the immune system (for review see reference 1). A common recombinase dependent on the lymphocyte-specific gene products RAG1 and RAG2 recognizes and cleaves pairs of conserved recombination signal sequences (RSSs), which flank all Ig and TCR V, D, and J gene segments. Components of the nonhomologous end-joining double-stranded DNA (dsDNA) break repair system then catalyze the formation of coding exons by ligating the pair of broken coding ends to one another. Because RSSs at each of the seven rearranging loci (IgHC; κ- and λLC; and TCRα, -β, -γ, and -δ chains) are recognized by the same recombinase machinery, the cell-type and stage-specific regulation of rearrangement is thought to rely on the accessibility of the recombinase to specific rearranging loci within chromatin structure (2, 3).

V(D)J recombination is regulated in three types of ways. First, rearrangement occurs in a lineage-specific manner. Ig and TCR genes rearrange completely only in developing B and T cells, respectively. Second, rearrangement is ordered within a lineage with IgHC or TCRβ locus rearrangement preceding IgLC or TCRα locus rearrangement in B and T cells, respectively. In addition, D-to-J rearrangement precedes V-to-DJ rearrangement in both IgHC and TCRβ loci. Finally, recombination in the IgHC and TCRβ loci are subject to allelic exclusion: the observation that each developing lymphocyte assembles only one functional IgHC or TCRβ chain gene, contributing to the clonotypic specificity of antigen recognition.

The IgHC locus has been extensively studied in an attempt to decipher the molecular basis of regulated V(D)J recombination. As noted in the previous paragraph, V-to-DJ rearrangement invariably follows D-to-J rearrangement (4). Direct V-to-D rearrangement is not observed, even on alleles with a targeted deletion of the J cluster of gene segments (5). Perhaps more remarkably, a V gene segment will bypass intervening germline D segments to join to a partially rearranged DJ segment. The D-to-J step in IgHC gene assembly is not lineage specific. DJ alleles are found in up to 40% of T cells, but complete VDJ alleles are not seen in these cells (6). In addition, unlike V-to-DJ rearrangement, D-to-J rearrangement is not subject to allelic exclusion. IgHC transgenic mice contain endogenous DJ- but minimal VDJ-rearranged alleles (7, 8).

Several trans-acting factors and signaling pathways have been implicated in the regulation of V-to-DJ rearrangement. Mice deficient in the transcription factor Pax5, the histone methyltransferase Ezh2, or the IL-7Rα chain show defects at this step of IgHC gene assembly (9–13). In each of these instances, the defect is greater for the more distal V gene segments, suggesting that long-range chromosomal interactions may play an important role in this regulation. This idea is consistent with the results of fluorescent in situ hybridization experiments showing developmentally regulated chromosomal compaction or looping of distal V genes into juxtaposition with the D-J end of the IgHC locus (14–17). STAT5 (activated by IL-7Rα signaling), Pax5, and Ezh2 have each been shown to localize to V region sequences in vivo (11, 18, 19). Remarkably, forced expression of Pax5 in developing T cells results in the activation of V-to-DJ rearrangement and partial locus compaction in the “wrong” lineage (14, 20).

Rearrangement within the IgHC locus is influenced in cis by the intronic heavy-chain enhancer. Targeted deletion of this element results in a moderate decrease in D-to-J rearrangement but a near-complete absence of V-to-DJ rearrangement (21–23). Perhaps surprisingly, deletion of the most J-proximal D gene, DQ52, along with a promoter 5′ of this gene segment that is responsible for germline transcription of the J cluster of gene segments, has little effect on IgHC rearrangement (23, 24).

In the present paper, we describe experiments aimed at distinguishing whether DNA sequences immediately surrounding V gene segments are sufficient for the proper regulation of V-to-DJ rearrangement or whether the regulation of rearrangement depends on the chromosomal position and the context of a V gene segment. We found that targeting a distal V gene segment to a region ∼1 kb 5′ of DFL16.1 caused it to recruit activating chromatin modifications, to rearrange more frequently than its endogenous counterpart, to rearrange directly to unrearranged D gene segments, to violate allelic exclusion, and to lose lineage specificity. We conclude that chromosomal position profoundly affects the regulation of V gene segment rearrangement.

RESULTS

Targeted insertion of a distal V gene segment into the 5′ of D region of the IgHC locus

To test to what extent chromosomal proximity contributes to the regulation of V-to-D rearrangement, we used homologous recombination in embryonic stem cells to target a distal V family gene segment along with ∼1.3 kb of upstream promoter sequence and ∼500 bp of downstream sequence to a region ∼500 bp 5′ of DFL16.1 (Fig. 1). The distance from DFL16.1 to the RSS of the targeted V gene (termed V) is ∼1 kb. Both the conceptual translation of this V gene segment and its RSS closely match the consensus for the V gene family (unpublished data). In addition, its promoter sequence contains the canonical octamer binding site. Presumably, an identical copy of this V gene, referred to as its endogenous counterpart, lies in the distal region of the IgHC locus, although the existence of a V gene with this exact sequence was not demonstrated in a paper on the sequence of the IgHC locus from another mouse strain (25).

Figure 1.

Targeted insertion of the (A) A dot plot of a 160-kb nucleotide region of the IgHC locus spanning the V(D)J interval (y axis) against a 16-kb subregion surrounding DFL16.1 (x axis) reveals the repetitive nature of the D cluster. The diagonal line across the entire map represents the exact match of the subfragment with the locus itself, and the other diagonal lines depict repeated regions. Probes for Southern blotting recognized multiple regions of DNA that were differentiated by size after digestion with the restriction enzyme SpeI. The Southern probes P1 and P2 are indicated below the sequences they recognize, and the position of the targeted V gene segment embedded in a region of repetitive 5′ of D gene sequence is indicated by the arrow. P1 recognizes only two regions, both within the V-to-D interval, and P2 recognizes multiple sequences in the D gene repeats. (B) The endogenous locus with the Southern probes P1 and P2, DFL16.1 (DFL), and SpeI (S) and EcoRI (RI) sites. The SpeI sites at the endogenous locus give rise to the 16-kb segment, whereas targeting of the V gene segment introduces a third SpeI site between V and DFL16.1. (C) The targeted locus depicts the addition of V, the floxed Neo cassette, and the newly introduced SpeI site. (D) Upon Cre-mediated deletion of the floxed Neo, the EcoRI site within the Neo is removed, creating a 2.6-kb product, allowing the distinction of Neo-containing from Neo-deleted mice with P3. (E) Southern blot performed on SpeI-digested DNA and probed with P1 confirmed integration of the left arm (1, germline; 2, targeted). Digestion with SpeI and probing with P2 confirmed integration of the right arm (3, germline; 4, targeted). Mice positive for right- and left-arm targeting were mated onto Cre, and digestion with EcoRI and probing with P3 confirmed deletion of the Neo cassette.

V is frequently rearranged and expressed in knock-in mice

To measure the frequency of V rearrangement, we took advantage of the fact that the targeted V gene possesses an SspI restriction endonuclease site that only one other functional V gene is predicted to have (Fig. 2 A). We amplified complementary DNA (cDNA) synthesized from bone marrow, spleen, CD4+ CD8+ (double-positive [DP]) thymocyte, and IL-7–dependent pro–B cell culture RNA from wild-type and homozygous V mice with a degenerate V gene primer that is complementary to the leader sequence of most V family genes paired with a primer complementary to an IgHC constant region exon. RT-PCR products were subjected to digestion with SspI to assess what fraction of the products represent transcription of a rearranged V gene segment. The wild-type samples had minimal cleavage from the contribution of endogenous V genes, whereas the fraction of cleavable product in the targeted animals was significant (Fig. 2 B), implying that the targeted V gene segment rearranges very frequently. In IL-7–dependent pro–B cell cultures, cells are not subject to selective pressure for pre-BCR assembly. In this setting, about half of the total V family rearrangements involve the targeted V gene segment, and this does not appear to differ in primary cells from bone marrow and spleen of these animals. Thus, IgHC rearrangements involving the V gene segment can apparently undergo positive selection during B cell development and contribute to the B cell repertoire in knock-in mice. As expected, the DP thymocytes do not produce spliced rearranged transcripts.

Figure 2.

The frequency of rearrangement of the targeted A degenerate V family leader sequence primer was paired with a constant region primer (Cμ) to amplify random-primed cDNA synthesized using RNA purified from wild-type and homozygous V knock-in IL-7–dependent pro–B cell cultures, total bone marrow, total spleen, and CD4/CD8 DP thymocytes (DP-Thy). A restriction endonuclease, SspI, specific for the targeted V gene segment (as well as one endogenous V gene segment and four pseudogenes) was used to digest the PCR product. (A) A schematic of primers and restriction enzyme digest performed on cDNA. (B) Restriction enzyme–digested (+) and –undigested (−) amplified cDNAs from wild-type and V knock-in cells were analyzed on agarose gels. As a control, β-actin was amplified, and the products analyzed in every other lane to correspond with the amplified cDNA shown above. Numbers indicate fragment lengths in nucleotides.

The targeted V gene segment recruits activating chromatin modifications in thymocytes

Because the targeted V gene segment was inserted within a region containing developmentally regulated histone modifications in cell lines (Fig. 3 A) (26), we proceeded to compare the histone modifications surrounding the targeted V gene with those in the region 5′ of DFL16.1 on the unaltered allele in heterozygous V animals. We performed chromatin immunoprecipitations (ChIPs) on V heterozygous RAG bone marrow cultured in IL-7 and on V heterozygous RAG thymocytes using antibodies that recognize H3 acetylation and H3K4 dimethylation. We used RAG bone marrow to ensure that sufficient germline sequence would be available for the PCR reaction. The primer set called 5′V amplifies a region ∼2 kb upstream of the V gene segment sequence, and 3′V amplifies a region ∼400 bp downstream of the targeted V gene segment RSS (Fig. 3 B). Predictably, we found that the targeted V gene segment is modestly enriched for H3 acetylation and significantly enriched for H3K4 dimethylation in pro–B cells. To our surprise, we found that the 3′ end of the V insertion, but not the unperturbed allelic region 5′ of DFL16.1, was enriched for H3K4 dimethylation in thymocytes (Fig. 3 C). Thus, the V gene segment is capable of recruiting histone modifications not otherwise found 5′ of DFL16.1 in unperturbed thymocytes.

Figure 3.

ChIP detects modified histone H3 surrounding the targeted (A) Schematic of the PCR primers used to analyze DNA recovered from ChIP using normal rabbit serum (NRS), antiacetylated H3, and antidimethylated H3K4. PCR assays using primers annealing 5′ of V (1 and 2, 5′V), 3′of V (3 and 4, 3′V), and on the endogenous allele 5′ of DFL16.1 (1 and 4, 5′D) are depicted. (B) Enrichment of precipitated as compared with input DNA from IL-7–dependent pro–B cells from a RAG heterozygous (Het) mouse or thymocytes from a V heterozygous RAG mouse. Error bars represent one SD from an average of two experiments. β-globin (β-glo) was amplified as a negative control.

V undergoes rearrangement in DP thymocytes

Given our observation that the targeted V gene segment can recruit H3K4 methylation in T cells, we went on to ask whether the lineage specificity of V-to-DJ rearrangement was likewise perturbed by repositioning this V gene segment. We purified wild-type, heterozygous, and homozygous V genomic DNA from bone marrow, spleen, and sorted (>99% pure on reanalysis; unpublished data) DP thymocytes and used PCR to detect V-to-DJ and D-to-J rearrangements. As expected, we detected rearrangement of DFL16.1-to-J segments in both wild-type and mutant samples. In wild-type samples, V-, V-, and V-to-DJ rearrangement was limited to the bone marrow and spleen (Fig. 4). In contrast, abundant V–to–DJ rearrangements were observed in all three tissues, including DP T cells from both V heterozygous and homozygous animals. This demonstrates that T cells are capable of performing V–to–DJ rearrangement in a context where endogenous V gene segment rearrangement is prohibited, and that such rearrangement does not require Pax5. Indeed, V rearranges on a Pax5-null background in the bone marrow as well (Fig. S1, available at http://www.jem.org/cgi/content/full/jem.20071787/DC1). Thus, position within the IgHC locus contributes to the lineage specificity of V-to-DJ rearrangement, and T cells possess all of the factors necessary to use V gene segments in V(D)J recombination.

Figure 4.

Genomic DNA was purified from total bone marrow, total spleen (Spl), and sorted CD4/CD8 DP thymocytes (DP-Thy) from wild-type (1), heterozygous (2), and homozygous (3) V mice. Two mice representing each genotype are shown. Products from a PCR rearrangement assay using degenerate primers for the V, V, and V gene families (as well as DFL16.1 as a control) paired with a JH3 reverse primer were analyzed on an agarose gel, Southern blotted, and probed with an internal JH3 probe. The last lane in each set of assays (separated by the vertical black bar) is a PCR assay programmed with water in place of template. A phosphorimage is shown. APRT was amplified and run on an agarose gel as a loading control. Numbers indicate fragment lengths in nucleotides.

The order of recombination is not tightly regulated in V mice

In developing B cells, D gene segments almost invariably rearrange to J gene segments before V-to-DJ rearrangement (4). To test whether the V gene segment undergoes normally ordered rearrangement, we assayed genomic DNA from wild-type, heterozygous, and homozygous V mice for direct V-to-D rearrangement over a distance of ∼60,000 nucleotides (Fig. 5 A). As expected, we failed to detect V-to-D rearrangements of either V or V family V gene segments in wild-type bone marrow. V-to-D rearrangements were detected sporadically and at low levels in wild-type spleen and thymus DNA samples. In contrast, we detected significant levels of such rearrangements in V DNA samples from all three tissues in the targeted mice (Fig. 5 B). These rearrangements occurred by deletion and not inversion, because the downstream primer was 3′ of the DQ52 gene segment and not within the D gene segment itself. Thus, the ordered regulation of IgHC gene assembly is dependent on the position of V gene segments within the locus.

Figure 5.

The order of DNA purified from wild-type, heterozygous (Het), or homozygous (Homo) V bone marrow, spleen (Spl), or thymus (Thy) was analyzed for direct V-to-D rearrangement by pairing degenerate V gene segment–family primers (V or V) with a reverse primer downstream of D (A). The V primer anneals to the V leader sequence, whereas the V primer anneals within FR3, resulting in PCR products of different sizes. (B) Each pair of lanes represents DNA from two different mice. The Southern blot analysis of PCR assays is shown (left), probed with an internal primer downstream of DQ52. The APRT gene was amplified from the same samples as a control and was visualized on an agarose gel (right). Numbers indicate fragment lengths in nucleotides.

The targeted V gene segment violates allelic exclusion at the levels of rearrangement and protein expression

It has been suggested that ordered IgHC assembly might be necessary for effective allelic exclusion (3). Because we observed that positioning V proximal to DFL16.1 resulted in an increased frequency of direct V-to-D rearrangement, we went on to ask whether IgHC allelic exclusion was disrupted by this mutation as well. We addressed this possibility in two ways. First, we assayed genomic DNA purified from FACS-sorted wild-type and V bone marrow pro– and pre–B cells and thymocytes for dsDNA breaks at various RSSs (Fig. 6 A). Such dsDNA breaks are reaction intermediates in V(D)J recombination and indicate active rearrangement of the gene segment under study at the time of DNA isolation (27). Pro–B cells were defined as B220+CD43+ intracellular μ− (icμ−) and pre–B cells as B220+CD43−icμ+. Our analysis revealed that pro–B cells contain dsDNA breaks at RSSs 5′ of DFL16.1 but not at those 5′ of Jκ5. In contrast, wild-type pre–B cells possess abundant breaks at RSSs 5′ of Jκ5, but not at those 5′ of DFL16.1. Remarkably, in the targeted animals, RSS breaks are easily detectable in pre–B cell DNA both 5′ of DFL16.1 and 3′ of V, indicating active V–to–D rearrangement in violation of allelic exclusion (Fig. 6 A). The primers used to amplify 3′of V signal end breaks were not degenerate and, therefore, did not amplify other members of the V gene family. Breaks at the V endogenous counterpart in wild-type pro–B cells were below the level of detection. However, we could not discern whether V or 5′ of DFL16.1 RSS breaks occur on unrearranged or DJ-rearranged knock-in alleles. Nonetheless, we conclude that the targeted V gene is able to undergo rearrangement into the pre–B cell stage, when endogenous V-to-DJ recombination is silenced by allelic exclusion.

Figure 6.

The targeted (A) LM-PCR assay for RAG-induced dsDNA breaks at RSSs downstream of V and upstream of DFL16.1, Jκ5, and Dβ1 in DNA purified from sorted wild-type or heterozygous (Het) V bone marrow from pro– (CD43+icμ−) and pre– (CD43−icμ+) B cells or thymocytes. Four to six littermate or age-matched mice of each genotype were pooled for staining and sorting. Southern blot analysis of the LM-PCR is shown probed with internal gene-specific primers. Numbers indicate fragment lengths in nucleotides. (B) V is rearranged and expressed on splenic B cells from hμ-transgenic mice. Spleens were harvested from wild-type, wild-type × hμ-transgenic, and V × hμ-transgenic mice and stained with fluorochrome-labeled anti-B220 and anti–mouse IgM and IgD antibodies. Mouse IgM/IgD staining is displayed on a histogram after gating on B220+ cells. Wild-type × hμ-transgenic mice (continuous line) have 2.51 ± 0.16% IgM/IgD+ splenic B cells (horizontal bar indicates gate; n = 4), whereas V heterozygous × hμ-transgenic mice (dashed line) have 9.98 ± 1.11% IgM/IgD+ splenic B cells (n = 4). A wild-type mouse (dotted line) is shown as a reference for mouse IgM/IgD expression among B220+ splenic B cells.

To further examine IgHC allelic exclusion, we bred a well-characterized human μ (hμ) transgene onto either a wild-type or V heterozygous genetic background. Expression of a transgenic hμ protein in developing B cells inhibits endogenous V-to-DJ rearrangement and surface expression of mouse IgHC on splenic B cells (Fig. 6 B, continuous line) (7). In contrast, the hμ transgene has far less of an effect on the expression of mouse IgHC protein in V mice (Fig. 6 B, dashed line). Flow cytometric analysis of surface mouse IgM/IgD expression on B220+ splenocytes in hμ-transgenic mice revealed a fourfold increase in mouse μ expression when one targeted V allele is present (Fig. 6 B).

We went on to examine the frequency of V(D)J-rearranged alleles in bone marrow cDNA from wild-type and V mice in the presence of the hμ transgene. This analysis confirmed that the targeted V gene was indeed abundantly rearranged in the bone marrow of targeted animals on the hμ transgenic background (Fig. S2, available at http://www.jem.org/cgi/content/full/jem.20071787/DC1). Collectively, these experiments show that repositioning a V gene segment to a location proximal to DFL16.1 results in the disruption of IgHC locus allelic exclusion.

DISCUSSION

We have demonstrated that the chromosomal position of a V gene segment within the IgHC locus rather than V gene–associated sequences greatly influences the frequency, order, and cell-type specificity of its rearrangement. Eμ, the only known enhancer in the J region of the locus, is required to promote optimal accessibility of the IgHC locus for both D-to-J and V-to-DJ rearrangement, yet D-to-J rearrangement consistently precedes V-to-DJ rearrangement (4, 21–23). Moreover, several recent studies have demonstrated that locus contraction occurs in correlation with but independently of V-to-DJ rearrangement (15–17). We hypothesized that differential regulation of the clusters of D and V gene segments depends on the distance between these sequences within the nucleus. Through the repositioning of a normally distal V gene segment, this is precisely what we observed.

The frequency with which the V gene segment rearranges is greatly enhanced by repositioning, which was quite unexpected considering similar experiments in T cells at the TCRβ locus (28). The TCRβ locus is much like the IgHC locus in that it has V, D, and J gene segments. It is the first locus to undergo rearrangement in T cells and, thus, like the IgHC locus, requires silencing to enforce allelic exclusion during TCRα rearrangement. Vβ gene segments are separated from Dβ and Jβ gene segments by ∼350 kb of DNA, with the exception of Vβ14, which lies ∼10 kb downstream of the DJCβ clusters on the far side of the only known enhancer in the locus, Eβ (29). The TCRβ locus undergoes ordered and cell type–specific rearrangement: Dβ-to-Jβ is followed by Vβ-to-DJβ rearrangement only in T cells and is strictly dependent on Eβ. Eβ deletion results in the absence of germline transcription that normally precedes any rearrangement, and homozygous Eβ-deleted animals lack αβ T cells altogether (30, 31). The domain of chromatin structure regulated by Eβ, as identified by a restriction enzyme accessibility assay, extends only 2 kb upstream of Dβ1 (32).

Insertion of a targeted Vβ gene segment ∼7 kb upstream of Dβ1 did not increase its frequency of rearrangement (26). However, deleting Dβ1 along with the 350-kb Vβ-Dβ interval resulted in a significant increase in the frequency of rearrangement of those Vβ gene segments now positioned much closer to Eβ (33). Thus, only when the entire Vβ-to-Dβ interval was deleted did the frequency of Dβ gene segment–proximal Vβ genes segments increase. This could be explained by the presence of a boundary element in the Vβ-to-Dβ interval that prevents the spreading of open chromatin from extending to the Vβ gene segments even when they are brought much closer to the Dβ and Jβ gene segments. Similarly, it is possible that the increase in rearrangement frequency of V is caused by its position within the realm of accessibility potentially limited by an analogous chromosomal boundary.

Lineage specificity and the role of transcription factors and histone modifications in V-to-DJ rearrangement

We found that the chromosomal context of a V gene segment influences the lineage specificity of its rearrangement. Previous studies had shown that transgenic expression of Pax5 in thymocytes was sufficient to activate V-to-DJ rearrangement and cause compaction of the distal and proximal regions of the IgHC locus (14, 20). More recently, it was shown that Pax5 can bind directly to a subset of V gene segments and can recruit the recombinase to these V gene segments via a protein–protein interaction (19). Pax5 had also been shown to be necessary for removal of inhibitory histone methylation around the distal V gene segments (34). The 1.3 kb of 5′ of V promoter region sequence upstream of the targeted V gene does contain potential Pax5 binding sites, but V–to–DJ rearrangement was independent of Pax5 in thymocytes and in IL-7–dependent bone marrow culture, calling into question an obligatory role for this transcription factor in IgHC V(D)J recombination (Fig. S1). Our results are more consistent with Pax5 regulating V-to-DJ rearrangement by bringing distal V gene segments into proximity with D segments (compaction).

IL-7 signaling has been proposed to play a role in IgHC allelic exclusion by regulating STAT5 binding to V gene segment promoters (18, 35, 36). This idea was recently challenged by the observation that allelic exclusion is intact in the presence of a constitutively active STAT5 (37). Our data also argue against a role for V gene promoters in establishing allelic exclusion because the targeted V gene promoter is not sufficient to enforce allelic exclusion. It remains possible, however, that IL-7 signaling is required for V gene activation, because IL-7 signaling does play a role in T cell development (38, 39). Indeed, in STAT5ab mice, Pax5 and Ezh2 expression are normal and chromosomal contraction occurs, but rearrangement is nonetheless impaired (36). Thus, STAT5 binding to V gene promoters and subsequent histone acetylation may be necessary to promote but not sufficient to properly regulate V gene segment activation. V gene segment promoters and RSSs from RAG IL-7–dependent pro–B cells are H3K4 methylated, but sorted double-negative thymocyte V gene segment promoters and RSSs remain unmethylated at H3K4, suggesting a role for this modification in the activation of V gene segment recombination (34). In agreement with this, we see recruitment of H3K4 methylation to the targeted V gene segment RSS in both IL-7–dependent bone marrow culture from RAG heterozygous V mice and primary thymocytes from RAG heterozygous V mice.

Ordered rearrangement

Various mechanisms have been proposed to explain the ordered nature of IgHC gene assembly. One such mechanism involves the preferential binding of RAG complexes to 3′ of D RSSs, limiting the accessibility of the RAGs to the 5′ of D RSSs until after D-to-J rearrangement deletes the 3′ of D RSS. This, however, cannot be the case, because we see direct V to D joining on the targeted locus. It is worth noting that we only observed direct V–to–D rearrangements involving the DQ52 gene segment; no direct V–to–DSP2 family gene segment rearrangement was observed (unpublished data). Promoters upstream of the D gene segments become active upon D-to-J rearrangement (40, 41), and it may be that rearrangement-induced transcription attracts V genes to rearranged DJ gene segments. Although not all D gene segments were tested, the observation that DQ52 is noticeably available for direct V–to–D rearrangement could be a reflection of the promoter upstream of DQ52 (driving the μ° germline transcript), which is active before D-to-J rearrangement (40, 42). The failure of V genes in their normal chromosomal positions to rearrange to the accessible DQ52 might be caused by boundary element activity or simply distance within the nucleus.

Allelic exclusion and V gene segment position

Rearrangement of the targeted V allele was not subject to allelic exclusion imposed by an hμ transgene. This observation is consistent with previously published results showing that endogenous D-proximal V gene segments continue to rearrange at a low but detectable frequency in IgHC transgenic mice (8). What is surprising is that the targeted V gene segment was frequently expressed, increasing the fraction of cells concomitantly expressing both mouse and human IgHC by fourfold compared with wild-type hμ transgenic mice. This is in contrast to the inserted Vβ gene segment within the TCRβ locus, where in the presence of a transgenic TCRβ the targeted gene was not subject to allelic exclusion at the level of rearrangement, but protein expression was inhibited (28).

The targeted V gene segment also escaped allelic exclusion in nontransgenic mice. We detected dsDNA RSS breaks 3′ of V as well as 5′ of DFL16.1 in sorted pre–B cell DNA from heterozygous knock-in but not wild-type animals. We detected a much stronger dsDNA RSS break signal from the targeted V gene segment as compared with endogenous V gene segments in the pro–B cell samples, which can be explained by the increased frequency of rearrangement of the targeted gene segment (Fig. 2). The persistence of 5′ of DFL16.1 breaks in pre–B cells and thymocytes from V knock-in but not wild-type mice speaks more accurately to the contrast between the wild-type and heterozygous animals. Thus, targeting a V gene segment to this chromosomal position affects the inactivation as well as the activation of the V gene segment rearrangement.

The chromosomal domain model

Changes in the proximity of V gene segments to Eμ upon D-to-J rearrangement does not adequately explain the order, frequency, and cell-type specificity of V gene segment rearrangement in wild-type mice. DQ52 rearrangement to J gene segments deletes as little as 700 bp of DNA, hardly enough to significantly alter the configuration of a 2-mb locus. Additionally, the targeted V gene segment can rearrange directly to DQ52, which is over 80 kb away, roughly the same length of DNA that normally separates V and DFL16.1 (∼90 kb). This argues that distance alone might not account for such radical changes in the regulation of V gene segment recombination. We have demonstrated that D-to-J rearrangement itself does not account for the activation of the V gene segment, because direct V–to–D rearrangements are observed. It is possible that the region 5′ of DFL16.1 into which we inserted VH-KI has enhancer or promoter activity or that the region of H3K4 methylation 5′ of DFL16.1 in B cells (Fig. 3, A and B) (26) stimulates V(D)J recombination, but this cannot account for the activation of V gene segment rearrangement in thymocytes. Of the two histone modifications we explored, the only modifications seen in thymocytes at this locus are those recruited by the targeted V gene, not by the endogenous locus (Fig. 3).

What might explain readily detectable V rearrangement in thymocytes? We favor the possibility that V and D gene segments normally occupy distinct chromosomal domains, but the targeted V gene segment now lies within the D domain. This D domain is normally accessible to the recombinase in both developing B and T cells. Our results lead us to two possible models. The V genes themselves can undergo activation in either B or T cells, but the timing, frequency, and lineage specificity of V-to-DJ rearrangement may be dependent on either (a) locus compaction or (b) a V domain control element that is unable to affect the repositioned V gene. Both of these models require that the V and D domains remain functionally separate, suggesting the existence of a boundary element. We are in the process of targeting the V gene segment to a series of locations progressively further from DFL16.1.

MATERIALS AND METHODS

Cell lines.

63-12 (RAG2null Abelson murine leukemia virus [AMuLV]–transformed pro–B cell [43]; provided by F. Alt, Harvard Medical School, Boston, MA), P5424 (RAG2null, p53null pro–T cell [44]; provided by P. Mombaerts, The Rockefeller University, New York, NY), and Pax5null (AMuLV-transformed pro–B cell line derived from mice containing LacZ in place of exon 2 of Pax5 [45]; provided by M. Busslinger, Research Institute of Molecular Pathology, Vienna, Austria) cell lines were grown at 37°C/5% CO2 in RPMI 1640 supplemented with 5–10% FCS, antibiotics, and 50 μM β-mercaptoethanol.

Primary pro–B cell culture.

Bone marrow was harvested from the femurs of 4–8-wk-old mice and rid of red blood cells by ACK lysis. Cells were cultured on S17 stromal cells in the presence of 10% IL-7–containing culture supernatant in RPMI 1640 with 10% FCS, antibiotics and 50 μM β-mercaptoethanol. After culture, >95% of cells stained positively for surface CD19 or B220. Primary thymocytes were purified by centrifugation with HISTOPAQUE-1083 (Sigma-Aldrich), after which they were >90% CD4/CD8 positive. Animal experimentation procedures were approved by the University of California, Berkeley Animal Care and Use Committee.

ChIP.

ChIP was performed as previously described (46). 10 μg of antidimethyl H3K4 antibody (Millipore), 15 μl of antisera against acetylated histone H3 (Millipore), or 10 μg of normal rabbit IgG (Santa Cruz Biotechnology, Inc.) were used for IP. The ratio of immunoprecipitated to input DNA for a given genomic region (IP/input) was defined as 2^(Ctinput-CtIP), where Ct is the cycle threshold for real-time PCR with SYBR green technology. Chromatin from cell lines was precleared with protein A/G–sepharose (Millipore) blocked with sheared salmon sperm DNA and BSA, whereas chromatin from primary cells was precleared with unblocked sepharose before IP. Error bars represent one SD from an average of two experiments. 5 × 106 cell equivalents were used per IP for cell lines, and 2.5 × 106 cell equivalents were used for primary cells. PCR primers are listed in Supplemental materials and methods (available at http://www.jem.org/cgi/content/full/jem.20071787/DC1).

Targeting construct and V mutant mice.

The targeting vector consisted of a left arm of 5.6 kb (from 74623–80260 of the bacterial artificial chromosome [BAC] available from GenBank/EMBL/DDBJ under accession no. AC073553), the V gene segment in the sense orientation, a loxP-flanked neomycin resistance cassette in the opposite transcriptional direction, and a right arm of 2.5 kb (from 80260–82746 of the same BAC) of homologous sequence. The V gene was targeted to a position ∼500 bp 5′ of DFL16.1 (∼80700 of the same BAC) in the same transcriptional orientation as DFL16.1. Cloning of the V gene segment and probes used for Southern blot analyses are described in Supplemental materials and methods.

Cell staining and sorting.

Spleen and thymus were strained through a 40-μm cell strainer, whereas femurs and tibias were dissected from mice and crushed with a mortar and pestle or flushed with a syringe. Lymphocytes were isolated by density centrifugation using HISTOPAQUE-1083. Antibodies used for sorting DP thymocytes were anti-CD8α–FITC (BD Biosciences) and CD4-PE (BD Biosciences). For transgenic hμ analyses, splenocytes were stained with B220-PE (BD Biosciences), IgM-biotin (clone II/41; BD Biosciences), and IgD-biotin (clone 11–26; eBioscience) with streptavidin-Cy5 (BD Biosciences). Antibodies used for sorting cells for ligation-mediated PCR (LM-PCR) were rat anti–mouse IgM-biotin (clone 1B4B1; SouthernBiotech), IgD-biotin (for magnetic bead depletion; SouthernBiotech), CD43-biotin (clone S7; BD Biosciences) with streptavidin-cychrome (BD Biosciences), B220-PE (BD Biosciences), and anti–mouse IgM-FITC (clone II/41; BD Biosciences), for intracellular staining after fixation in 1% paraformaldehyde and permeabilization with 0.1% saponin.

LM-PCR.

LM-PCR for detection of in vivo–generated broken signal ends was performed as previously described (47, 48). Primers are listed in Supplemental materials and methods.

Online supplemental material.

Fig. S1 shows that Pax5 is not necessary for V rearrangement in developing B cells. Fig. S2 shows that V is not subject to allelic exclusion. Supplemental materials and methods provides information about targeting construct cloning and probes for Southern screening, as well as primers used for PCR assays, including ChIP, frequency of V gene rearrangement, recombination assays, and LM-PCR. Online supplemental material is available at http://www.jem.org/cgi/content/full/jem.20071787/DC1.