Evidence for Physical Interaction between the Immunoglobulin Heavy Chain Variable Region and the 3′ Regulatory Region*

B cell-specific expression of immunoglobulin heavy chain (IgH) genes utilizes two cis regulatory regions, the intronic enhancer (Eμ), located in the JH-Cμ intron, and a complex regulatory region that lies 3′ to the IgH gene cluster, 3′ RR. We hypothesized that the 3′ RR is involved in IgH gene transcription in plasma cells via physical interaction between distal 3′ RR enhancers and target VH sequences, with loop formation by intervening DNA. In support of this hypothesis we report sequence data at DNA recombination breakpoints as evidence for loop formation preceding DNA inversion in a plasma cell line. In addition, using the chromosome conformation capture technique, physical interactions between VH and 3′ RR were analyzed directly and detected in MPC11 plasma cells and variants and normal splenic B cells but not detected in splenic T cells or in non-B cells. VH-3′ RR interactions were present in the absence of Eμ, but when the hs1,2 enhancer was replaced by a NeoR gene in a variant cell line lacking Eμ, H chain expression was lost, and interactions between VH and 3′ RR and among the 3′ RR regulators themselves were severely disrupted. In addition, the chromosome conformation capture technique detected interactions between the myc promoter and 3′ RR elements in MPC11, which like other plasmacytomas contains a reciprocal translocation between the c-myc and the IgH locus. In sum, our data support a hypothesis that cis VH-3′ RR and myc-3′ RR interactions involve physical interactions between these DNA elements.


B cell-specific expression of immunoglobulin heavy chain (IgH) genes utilizes two cis regulatory regions, the intronic enhancer (E), located in the J H -C intron, and a complex regulatory region that lies 3 to the IgH gene cluster, 3 RR. We hypothesized that the 3 RR is involved in IgH gene transcription in plasma cells via physical interaction between distal 3 RR enhancers and target V H sequences, with loop formation by intervening DNA. In support of this hypothesis we report sequence data at DNA recombination breakpoints as evidence for loop formation preceding DNA inversion in a plasma cell line. In addition, using the chromosome conformation capture technique, physical interactions between V H and 3 RR were analyzed directly and detected in MPC11 plasma cells and variants and normal splenic B cells but not detected in splenic T cells or in non-B cells. V H -3 RR interactions were present in the absence of E, but when the hs1,2 enhancer was replaced by a Neo R gene in a variant cell line lacking E, H chain expression was lost, and interactions between V H and 3 RR and among the 3 RR regulators themselves were severely disrupted. In addition, the chromosome conformation capture technique detected interactions between the myc promoter and 3 RR elements in MPC11, which like other plasmacytomas contains a reciprocal translocation between the c-myc and the IgH locus. In sum, our data support a hypothesis that cis V H -3 RR and myc-3 RR interactions involve physical interactions between these DNA elements.
During B cell development, antibody heavy chain genes undergo sequential DNA recombination and mutation events. These include construction of the variable region gene (VDJ joining), which occurs very early in B cell development, synthesis of membrane Ig, class switch recombination, and somatic hypermutation at the mature B cell stage, and secretion of IgH from plasma cells. Key cis regulators of B cell-specific expression of immunoglobulin heavy chain (IgH) 5 genes are V H promoters, I region promoters located upstream of each constant region gene, and two sets of enhancers (the intronic enhancer (E) located in the J H -C intron and a complex regulatory region that lies 3Ј of the IgH gene locus (3Ј RR) (1)). The murine 3Ј RR contains four enhancers arrayed in two separate structural and functional units (2) together with a recently identified downstream extension, which contains additional DNase I hypersensitive sites, binding sites for CCCTC-binding factor (CTCF), and insulator sequences (3).
Initial insight into the contribution of 3Ј RR enhancers to high levels of IgH gene expression in plasma cells came from analysis of cell lines containing deletions of intronic or 3Ј RR enhancers. These data include 1) maintenance of IgH chain expression in plasma cell lines at high levels despite deletion of the intronic enhancer (4), 2) ϳ90% reduction in IgH expression in a plasma cell line that has a deletion of 3Ј RR regulators extending from hs3a to hs4 (5,6), and 3) complete loss of H chain expression from a plasma cell line that has a combination of a deletion of the intronic enhancer and substitution of the hs1,2 enhancer by the Neo R gene (7).
Knock-out mouse models have provided insight into the contribution of 3Ј enhancers to CSR; a combined deletion of hs3b and hs4 affects CSR to all isotypes except IgG1 (8), whereas a deletion of all 3Ј IgH sequences from a BAC transgenic mouse affected CSR to all isotypes (9). However, the contribution of the entire endogenous 3Ј RR to heavy chain expression in plasma cells has not been addressed by knock-out models, and no phenotype was detected upon examination of other 3Ј RR deletions in cell lines, including a combined deletion of hs3a and hs1,2 (10) and a specific deletion of hs4 (11).
These various observations suggest a potential physical interaction between distal 3Ј RR enhancers and target IgH sequences, with intervening DNA sequences forming loops. The distances involved are as much as ϳ40 -180 kb. Previous experiments identified a plasma cell line in which DNA rearrangements detected by genomic Southern analysis were con-sistent with an inversion of sequences extending from V H to 3Ј RR (12). Here we report DNA sequence analysis of recombination breakpoints that explains this inversion by recombination between the V H gene and the 3Ј RR. To determine whether V H -3Ј RR interactions regularly occur, we have implemented the chromosome conformation capture (3C) technique, which captures transient interactions between otherwise linearly distant DNA sequences. This method has been used to analyze the effects of enhancers on target promoters in the ␤-globin and T H2 cytokine loci (13)(14)(15). Of particular pertinence to the regulation of antibody gene regulation is a recent study that has shown 3C interactions between DNA segments containing the intronic and 3Ј enhancers and the expressed V gene both in the MPC11 plasmacytoma cell line and in normal splenic B cells (16). This interaction likely reflects the impact of intronic and distal enhancers on transcriptional activation of expression.
The 3C experiments reported here have confirmed interactions in cis between V H sequences and 3Ј RR enhancers in the MPC11 plasma cell line and in normal resting B cells. Interactions in cis between the myc exon 2 promoter and 3Ј RR enhancers were also detected in MPC11, which like other plasmacytoma cell lines contains a reciprocal chromosomal translocation that brings the myc oncogene under the control of IgH regulatory sequences. Analysis of MPC11 variants with deletions of endogenous intronic or 3Ј enhancers has addressed the role of specific V H -3Ј RR interactions in heavy chain expression.

EXPERIMENTAL PROCEDURES
Cells-The mouse plasmacytoma cell line, MPC11 (BALB/c, IgG2b, ), synthesizes a 55-kDa heavy chain and has a single J H region and intronic enhancer associated with the expressed ␥2b chain (17). The IgH unexpressed allele has undergone a reciprocal chromosomal translocation with the myc oncogene (18). 9921 cells were derived from MPC11 via an intermediate (9.7.1) (19). These cells synthesize IgG2a instead of the parental IgG2b as a result of a deletion that juxtaposes the MPC11-rearranged V H gene to switch ␥2a sequences (17). In the course of the DNA recombination, the intronic enhancer and C␥2b sequences were deleted. B48 cells, obtained from Laurel Eckhardt, Hunter College, The City University of New York, were derived from 9921, differing only in the substitution of the Neo R gene for the hs1,2 enhancer on the expressed allele. H chain expression is completely absent from B48 cells (7). F5.5 cells were derived from MPC11 via an intermediate (ICR4.68) (20). Sequence analysis of F5.5 cDNA showed that the expressed 50-kDa ␥2b H chain resulted from a single base insertion leading to a premature termination (20). Genomic Southern analysis of F5.5 revealed a second J H -associated map and a C⑀ rearrangement (20). EL-4 cells are a T lymphoma; the IgH locus is in germline configuration (21). The mouse erythroleukemia cell line is derived from DBA mice, and the IgH locus is in germline configuration. All cell lines were maintained in Dulbecco's modification of Eagle's medium supplemented with 15% fetal bovine serum and 1% penicillin-streptomycin. All cells were grown at 37°C in an atmosphere of 5% CO 2 .
Resting B cells were isolated from spleens of C57BL/6 mice. Total splenocytes were depleted of CD43 ϩ -activated B cells and non-B cells by incubation with anti-CD43 microbeads (Milte-nyi Biotec), washing with cold phosphate-buffered saline containing 3% fetal bovine serum, and passage through AutoMACS (Miltenyi Biotec) according to the manufacturer's instructions. The purity of resting B cells, as detected by FACScan (BD Biosciences), was ϳ95% B220 ϩ . Splenic T cells (ϳ95% CD3 ϩ ) were isolated using a pan-T cell isolation kit (Miltenyi Biotec). The protocol for these studies was approved by the Animal Institute Committee of the Albert Einstein College of Medicine.
Isolation and Sequence Determination of the VDJ-3Ј RR and C⑀ Rearrangement Breakpoints from F5.5-Genomic Southern analysis of F5.5 identified an 8.1-kb BglII fragment that hybridized to both J H and 3Ј RR sequences (12). This fragment was isolated from a Charon 40 phage library prepared from F5.5. A 1.4-kb BamHI/BglII fragment containing the inversion breakpoint was subjected to partial DNA sequence analysis. A 10.1-kb SpeI fragment containing the rearranged C⑀ gene (12) was cloned into Charon 40 XbaI-digested arms. An 800-bp SacI/KpnI fragment containing the C⑀ rearrangement breakpoint (12) provided the template for initial DNA sequence analysis. Preliminary sequences obtained from clones of the breakpoints located the DNA segments involved in these rearrangements. Primers designed to analyze genomic sequences at the inversion breakpoints were selected accordingly. The first set of primers (3Ј RR/VDJ-forward (F), agactgtgagagtggtgcctt; 3Ј RR/VDJ-reverse (R), tttcccctcttatttcccca) amplified a ϳ500-bp segment from VDJ to 3Ј RR. The second set (VDJ/C⑀ m -F2, ctctagatggactaggtc; VDJ/C⑀ m -R, tcagagccttcctagagagcc) amplified an ϳ800-bp fragment from the VDJ to C⑀ region. The expressed V H gene of F5.5 was sequenced using the primers VDJ/C⑀ m -F2 and 3Ј RR/VDJ-F as above, found to be identical to that previously shown for MPC11 variants, 9.7.1 (GenBank TM accession number M17056), 9921 (GenBank TM accession number M17058), and 11.9.3 (GenBank TM accession number M17059) (22) and provided the foundation for identification of the VDJ inversion breakpoints in F5.5.
Overall, 10 pmol of each primer, 2 units of HiFi Taq polymerase, and 100 ng of purified F5.5 DNA were used in a standard 50-l PCR reaction. The PCR reaction was performed using 1.5 min of extension time, 68°C annealing temperature, and 30 -35 cycles in a ThermoHybaid PCR machine. PCR products were examined on a 0.8% agarose gel and purified using the QIAquick PCR purification kit. DNA was sequenced in the Albert Einstein College of Medicine sequencing facility using the primers used for PCR amplification. Sequences were analyzed via BLAST; C⑀ and 3Ј RR sequences were compared with 129Sv germline sequences contained in two BAC clones, i.e. ⑀ (Gen-Bank TM accession number AJ851868) and 3Ј RR (GenBank TM accession number AF450245), respectively. MPC11 and its variants, like F5.5, derive from BALB/c, which is more similar to 129Sv than to the C57Bl/6 strain selected for the mouse genome project.
3C Assay-Except for minor modifications, the procedure was performed according to the method of Splinter et al. (23). Briefly, 10 7 cells were subjected to formaldehyde cross-linking for 10 min at room temperature with tumbling, after which glycine was added to quench the formaldehyde. Nuclei were prepared, and non-cross-linked protein was removed from chromatin DNA by incubation with SDS. Triton X-100 was added to sequester the SDS, after which samples were digested with 2000 units of HindIII overnight at 37°C with shaking. The digest was terminated with the addition of SDS and incubation at 65°C. Digested nuclei (10 6 nuclei) were then diluted (ϳ1:15) with 1ϫ T4 DNA ligase buffer to a concentration of ϳ3 ng of DNA/l, and SDS was sequestered by Triton X-100. Ligation was promoted by T4 DNA ligase and terminated with the addition of proteinase K, and cross-links were reversed by incubation at 65°C overnight. Samples were treated with RNaseA, and DNA was purified by phenol:chloroform extraction and isopropanol precipitation. The concentration of dsDNA was determined using picogreen (Molecular probes kit) by the FLUOstar OPTIMA fluorescence reader (BMG Lab Technologies).
One hundred ng of ligated DNA samples were analyzed by 32 cycles of PCR (94°C/30 s, 60°C/20 s, and 72°C/30 s) in a 25-l reaction volume. Analysis of J H3 association with hs1,2 and hs5-7 was determined using 62°C as the annealing temperature. Taq polymerase was obtained from New England Biolabs. PCR products were separated by 2% agarose gel. Gel pictures were captured by Gene Snap Software and then analyzed by Gene Tools Software (SYNGENE GENE genius BioImaging System).
To exclude the biased restriction digestion of one site over another in different cell types, we checked the digestion efficiency of each designated HindIII cleavage site by PCR (27 cycles of 94°C/30 s, 58°C/20 s, and 72°C/30 s) using primers flanking the cut sites (supplemental Table S1) and 100 ng of template DNA extracted from undigested and HindIII-digested nuclei. Restriction digestion efficiencies were calculated by comparing PCR signals from digested and undigested samples and ranged from 66 to 89%, comparable with other studies (16,24). The identity of PCR fragments was confirmed by sequence determination.
To correct for differences in amplification efficiency between different primer sets, we generated an equimolar mixture of two BACs, namely BAC199M11 (AF450245, containing all the 3Ј RR elements: HS3a, HS1,2, HS3b, HS4, and HS5,6,7) and RP109B20 (AC079180, extending from D H segments through to the first exon of ␥2a), a plasmid (M11-myc 3Ј) containing the MPC11 myc translocation breakpoint with switch ␥2a sequences (18), two calreticulin fragments that contain the selected HindIII restriction sites (Ensembl: ENSMUST 00000003912), and a pGK-Neo R PCR fragment containing a HindIII cleavage site (7). The control template, containing all possible ligation products, was then subjected to HindIII digestion followed by random ligation with DNA ligase. Primers for 3C analysis were upstream (forward) of HindIII cleavage sites (supplemental Table S1).
Variations in efficiency of cross-linking, restriction digestion, ligation, and amplification in different cell types were normalized via the cross-linking signal derived from two non-adjacent DNA fragments of the calreticulin housekeeping gene (fragments A and B, supplemental Table S1) as used by others (13). Reverse transcription-PCR analysis showed that calreticulin transcription levels were at similar levels in the cell lines we used and, independently, at comparable levels in primary B and T cells.
3C experiments were performed on three independent cell preparations of each source using multiple independent PCR reactions. Means and S.D. were calculated, and Student's t test was conducted to determine the statistical significance of the differences between two samples.

Inversion/Deletion in F5.5 Is Indicative of V H -3Ј RR
Interaction-The results of genomic Southern analysis of F5.5, a murine myeloma cell line derived from MPC11, are consistent with an inversion between the expressed V H gene and the 3Ј RR (12,20). A likely mechanism to account for this inversion is physical interaction between these distally located segments accompanied by DNA breaks and rejoining (Fig. 1A). Such an interaction would bring the expressed V H gene into contact with distally located 3Ј RR sequences that have been predicted to influence its expression.
To obtain additional evidence for this model by identifying the sequences at the breakpoints, we compared DNA rearrangements in F5.5 with wild-type genomic sequences of the 3Ј RR (GenBank TM accession number AF450245 (25)). The VDJ-3Ј RR inversion breakpoint in F5.5 cells was contained in an 8.1-kb BglII fragment, whereas the predicted reciprocal C⑀ breakpoint was analyzed from a 10.1-kb SpeI fragment. These fragments were initially isolated from phage libraries, confirmed by Southern analysis, subcloned, and then subjected to partial sequence analysis (data not shown). We completed the sequences of the inversion breakpoints on F5.5 genomic DNA using a PCR strategy ( Fig. 1A and supplemental Figs. S1A and S1B). Consistent with an inversion model, VDJ sequences located on the 5Ј to 3Ј non-coding strand were associated with the 5Ј to 3Ј coding strand of the 3Ј RR to generate the VDJ-3Ј RR breakpoint (supplemental Fig. S1A). Sequence analysis of the C⑀ rearrangement showed that VDJ 5Ј coding sequences were joined through a 50-bp junction of 3Ј RR sequences to C⑀ membrane exon 1 sequences, both from the non-coding strand (supplemental Fig. S1A). This implied that in addition to the inversion, there was an internal ϳ40-kb deletion that began in C⑀ membrane exon 1 and continued through C␣ and enhancers hs3a and hs1,2 of the 3Ј RR (Fig. 1B). The 3Ј end point of this deletion, marked by a TGGG motif in 3Ј RR sequences, was located ϳ6 kb downstream of hs1,2.
The DNA sequences of the inversion breakpoints provide direct evidence for the reciprocal recombination between expressed VDJ and 3Ј RR sequences (Fig. 1C). The VDJ breakpoint occurred immediately after the codon for N14 (numbering from the N terminus of the mature MPC11 H chain, Gen-Bank TM accession number AAA38329), whereas the 3Ј RR breakpoint occurred ϳmidway between hs1,2 and hs3b enhancers. Sequences at the breakpoints (VDJ: 5Ј to 3Ј coding strand depicted) match their respective 3Ј RR and V H MPC11 counterparts except for a three-base pair insertion (AGA) in the 3Ј RR segment immediately upstream of the crossover point. Upstream of the VDJ-3Ј RR breakpoint, there is sequence identity (10/12 bases) between germline V H MPC11 and 3Ј RR sequences, whereas downstream sequences show little homology. The crossover point itself is within a CCT triplet contained in both V H MPC11 and 3Ј RR at the 3Ј limit of the region of homology. The V H MPC11 breakpoint, including the CCT, is a palindrome (AGGCCT). The inversion breakpoint is located only 50 bp downstream of the 3Ј RR deletion end point. These data imply that the inversion of DNA sequences extending from V H to 3Ј RR in F5.5 has occurred through recombination after physical interaction between V H and 3Ј RR sequences.
3C Assay-Dynamic interactions between the 3Ј RR and its predicted V H target sequence might generate only a transient intervening loop without any detectable downstream DNA rearrangements. To detect these predicted V H -3Ј RR interactions, we have implemented the 3C technique (Fig. 2). The MPC11 cell line and several of its variants provide unique resources for analysis of interactions between the 3Ј RR elements and their predicted V H gene target sequences. MPC11 has only one J H region and a single intronic enhancer, both of which are located on the single expressed IgH locus (17). Hence, J H sequences used as a probe or primer source can exclusively track the single IgH expressed allele in MPC11 and its variants. As in other plasmacytomas, the second IgH locus in MPC11 has undergone reciprocal translocation with the myc gene (18), and the translocated IgH:myc allele has been amplified to three copies (26).
In the 3C method, formaldehyde is added to live cells to fix DNA-protein interactions that may accompany long-range regulation involving linearly distant chromatin fragments. Restriction digestion of chromatin in fixed nuclei followed by ligation in dilute solution captures these interactions in a covalent form, which provides a template for PCR amplification. HindIII was chosen based on the position of its restriction sites with respect to IgH regulatory elements ( Fig. 2A), its enzymatic activity in the presence of SDS, and its being subject to inactivation at 65°C. Because chromatin limits the accessibility of restriction enzymes to their specific sites (Fig. 2B), we compared the relative efficiency of HindIII cleavage in the cell sources we analyzed. Individual sites were cleaved at comparable levels ranging from 66 to 89% in both EL-4 and MPC11 and in FIGURE 1. A, map of the inverted IgH locus in F5.5 before and after rearrangement. The IgH locus in F5.5 has undergone an inversion extending from a position within V H coding sequences to the 3Ј RR together with a deletion of C␣ and a portion of 3Ј RR sequences. Inversion and deletion breakpoints were initially identified by genomic Southern analysis then isolated from phage libraries, subcloned, and subjected to partial sequence analysis. Two sets of primers (short arrows) were designed to amplify the region of the breakpoints to complete the sequence determination using genomic F5.5 DNA as a template. An ϳ800-bp fragment contained the V H /3Ј RR-⑀ breakpoint (supplemental Fig. S1A), and an ϳ500-bp fragment contained the J H -3Ј RR breakpoint (supplemental Fig. S1B). Fifty bp of the 3Ј RR are retained between VDJ and C⑀ after the inversion. B, deletion of C␣, hs3a, and hs1,2 in F5.5. DNA recombination in F5.5 as compared with its germline unrearranged counterparts (GenBank TM accession numbers AJ851868 (C⑀) and AF450245 (3Ј RR)). Ovals show switch-like motifs. C, comparison of the inversion breakpoints (bracketed segments from supplemental Figs. S1, A and B) shows that they derive from reciprocal recombination between VDJ and 3Ј RR sequences. The reverse complement of the sequence in supplemental Fig. S1B, i.e. coding with respect to V H MPC11, was used in this comparison. The corresponding amino acid sequences in the vicinity of the recombination breakpoint are shown, and the position of recombination is immediately between codons for amino acids 14 and 15 of the mature MPC11 heavy chain (GenBank TM accession number AAA38329). Switch-like motifs are circled. A hot spot for activation-induced cytidine deaminase is circled with a dotted line. Sequences that are identical between VDJ and 3Ј RR are boxed. The aga sequence (complementary to the underlined tct in supplemental Fig. 1B) is introduced as a non-templated insertion to maintain homology at the recombination breakpoint. VDJ sequences are shaded. 3Ј RR sequences are shown in smaller letters. The complementary strand of the 3Ј RR sequence at the inversion breakpoint is shown in italics. D, depiction of inversion and deletion rearrangements in the IgH locus in F5.5 mediated by reciprocal recombination between VDJ and 3Ј RR sequences. The inverted sequence includes a portion of the VDJ region, the intronic enhancer (E), constant region sequences C␥2b, C␥2a, and C⑀, and 50 bp fragment of the 3Ј RR. A second loop beginning downstream of C⑀ and extending through C␣, hs3a, and hs1,2 was deleted. The inversion and deletion endpoints within the 3Ј regulatory region are located 50 bp from each other.
other cells we examined (data not shown) that were similar to levels reported in other studies (16).
Relative primer pair efficiency was measured (Fig. 2C, control template, first lane) using a mixture of naked DNA from BACs and plasmids in equimolar amounts as templates for all possible interactions between these otherwise linearly distant fragments. The mixture was subjected to restriction enzyme cleavage and random DNA ligation. Interactions between restriction fragments in MPC11 cells were not observed unless DNA ligase was added (Fig. 2C). Individual primer pairs showed differences in amplification efficiency that were normalized to the signal generated by interactions between two non-adjacent fragments affiliated with the calreticulin housekeeping gene. As shown by others (13), we observed that calreticulin was expressed at similar levels in the various cell sources we have analyzed (data not shown). 3C analysis used DNA amounts in the linear range of the assay, as shown for CalR (Fig. 2D) and for V H -hs1,2, V H -hs3b-4,and V H -hs5-7 (supplemental Fig. S2, A-C). The 3C signal obtained from cells was normalized to the endogenous calreticulin signal (Fig. 2E).
Interactions between Expressed V H and 3Ј RR in MPC11-We specifically wanted to test whether 3Ј RR regulators could interact with sequences at the 5Ј end of the expression unit, namely the expressed VDJ gene and/or the intronic enhancer, E. The expressed MPC11 VDJ H2 gene ( Fig. 2A) is located within a 3.7-kb HindIII restriction fragment, which begins upstream of the V H promoter and terminates downstream of J H3 , whereas unrearranged germline J H2 and J H3 genes are contained in a 0.8-kb HindIII fragment. A primer for J H3 sequences tracks the expressed V H gene from MPC11 and corresponding J H2 and J H3 germline sequences in non-B cells, such as EL-4. Interactions between VDJ-associated fragments and 3Ј RR regulators were substantially stronger in MPC11 than in EL-4 (Fig. 3, A and  B). In MPC11, the fragment containing the expressed VDJ gene interacted with 3Ј RR regulators, namely, hs3a, hs1,2, hs3b-hs4, and hs5-7, whereas the analogous unrearranged germline fragment from EL-4 T cells (or from mouse erythroleukemia cells, data not shown) showed only low levels of interactions with 3Ј RR fragments (Fig. 3B). Other studies have also shown similar background bands in cell sources selected to be biological controls (16,27). Under our  NOVEMBER 30, 2007 • VOLUME 282 • NUMBER 48 experimental conditions, only background interactions were observed between the expressed VDJ gene and "hole" (transmembrane protein 121), the closest neighboring gene 3Ј of the IgH locus, located ϳ26 kb downstream of hs7 (28). Hole is not expressed in mature B cells (28) or plasma cells. 6 Furthermore, there were no interactions between J H3 and the CalR gene, which is located on a separate chromosome ( Fig. 2C and 3B). These observations suggest that physical interactions between the 5Ј end of the IgH transcription unit and the 3Ј RR are specific to the IgH locus and to the plasma cell line that we have analyzed.

Interaction between the IgH Variable Region and 3 RR
The Intronic Enhancer Is Dispensable for V H -3Ј RR Interaction-Immediately downstream and adjacent to the HindIII fragment containing the expressed V H gene in MPC11 is a 2.2-kb HindIII fragment that contains J H4 and E. Specific interactions between V H and E could not be analyzed because the HindIII fragments that contain these elements are adjacent. In a flexible chromatin context, adjacent fragments are especially subject to random collisions, resulting in nonspecific interactions even when monitored with unidirectional primers (27). We therefore chose to analyze the contribution of the intronic enhancer to 3C interactions by assaying 9921 (Fig. 2A). The IgG2a expression levels in 9921 (which lacks the intronic enhancer) are comparable not only to levels of IgG2b in MPC11 but also to levels of IgG2a expressed in MPC11 switch variants that retain the intronic enhancer (17,29). These data have contributed to the general hypothesis that high levels of IgH expression in murine plasma cells that occur in the absence of the intronic enhancer depend on the 3Ј RR (30). In fact, 3C experiments in 9921 cells (Fig. 3C) have detected interactions between the expressed V H gene and each of the three HindIII fragments containing 3Ј RR enhancers at the same general levels as shown in MPC11. These observations were confirmed for hs3a and hs5-7 with primers designed in the opposite orientation (supplemental Fig. S3, A and B). The binding of V H to hs3a, although slightly reduced, is significantly higher than in EL-4 T cells. These experiments show that E is not required for V H -3Ј RR interactions that occur in a plasma cell line synthesizing large levels of antibody and suggest that 3Ј RR interactions with the expressed V H gene may be important for H chain expression. 6 Z. Ju, unpublished observations.

FIGURE 3. Analysis of PCR amplification of 3 RR interactions with J H3 sequences as an anchor in EL-4 thymoma cells and MPC11 and MPC11
variants. 9921 has deleted the intronic enhancer, whereas B48 has, in addition, sustained a substitution of Neo R for hs1,2 from the expressed IgH allele. A, samples were analyzed in duplicate, with each lane as an independent PCR reaction. The CalR signal represents the interaction between two fragments associated with the CalR locus and is used as a normalization signal (see "Experimental Procedures"). No interactions were detected between V H (J H3 as anchor primer) and hole or CalR. Ctrl identifies the PCR signal detected after HindIII digestion and random ligation in the control template and is used to normalize for primer pair efficiency (see "Experimental Procedures"). B, graphical presentation of interactions between V H with J H3 as an anchor primer and 3Ј RR elements in MPC11 compared with EL-4. C, graphical presentation of interactions between V H and 3Ј RR in 9921 compared with EL-4. D, graphical presentation of interactions between V H and 3Ј RR in B48 compared with EL-4. X indicates a disruption of interactions. E, analysis of PCR amplification of 3Ј RR interactions with a Neo R primer as an anchor in B48 compared with 9921. F, graphical presentation of interactions between a Neo R primer and 3Ј RR elements in B48 compared with 9921. X indicates a disruption of interactions. No signal was detected in 9921, indicated by a small black square at the base line. Anchor primers are denoted by triangles.

Replacement of hs1,2 on the Expressed IgH Allele by Neo R Disrupts V H -3Ј RR
Interactions-The B48 cell line ( Fig. 2A) differs from 9921 only by a substitution of the Neo R gene for hs1,2 on the expressed H chain allele; this entirely cripples heavy chain expression (7). To determine whether the Neo R substitution for hs1,2 affects interactions between the V H gene and the 3Ј RR, we carried out 3C on B48 cells. These experiments (Fig.  3D) showed a significantly enhanced interaction between the expressed V H gene and hs3a; this was coupled with a severe reduction in interactions between V H and 3Ј RR elements downstream of the Neo R insertion, including hs1,2 downstream sequences, hs3b-4 and hs5-7 fragments, to the same background levels as observed in EL-4 thymoma cells. Interaction of the V H gene with hs3a but not with hs 5-7 in B48 cells was also observed with primers of the opposite orientation designed for sequences near the upstream HindIII sites of hs3a and hs5-7 (supplemental Fig. S3C). These same primers detected interaction between the V H gene and both hs3a and hs 5-7 in MPC11 and 9921 cells. Repetitive sequences near the 5Ј HindIII sites of hs1,2 and of hs3b-4 precluded the design of primers for these specific segments. Our experiments showed that the retention of interaction of V H with hs3a but not with downstream 3Ј RR elements in B48 was unlikely to derive from a bias in primer design or orientation. These experiments provide evidence for the contributions of 3Ј RR regulators to H chain expression.
3Ј RR Elements Interact with Each Other and with E-3C experiments in MPC11 and 9921 detected interactions between the upstream expressed V H anchor fragment and each of the fragments containing 3Ј RR enhancers. We, therefore, wanted to determine whether the 3Ј RR regulators interacted with each other. Experiments showed that hs3a interacted at similar levels with the other 3Ј regulators as well as with E in MPC11 cells, but not in EL-4 T cells, and the interaction did not extend to the hole gene (Fig. 4). Similar observations were detected for normal splenic B cells and T cells, respectively (data not shown). It should be noted that we cannot distinguish the interactions between 3Ј regulators that may occur on the non-IgHexpressed alleles that are involved in the reciprocal myc chro-mosomal translocation from those occurring on the IgH expressed allele. These observations of 3Ј RR interactions in B cells are relevant to our studies of B48 where, using the Neo R gene as an anchor (Fig. 3, E and F), we observed its strong interaction only with hs3a but not with any of the downstream 3Ј RR enhancers nor with the upstream expressed V H gene. These findings suggested that a closed loop between Neo R and hs3a precluded additional 3Ј RR interactions. However, we cannot eliminate the possibility that nonspecific 3C interactions that are known to be able to occur between adjacent or nearby restriction fragments (27) may contribute at least in part to the Neo R interaction with hs3a. The replacement of hs1,2 (5.1 kb) by the Neo R gene (1.9 kb) results in a shorter distance between Neo R and hs3a (ϳ7.5 kb) than between hs1,2 and hs3a (ϳ10.3 kb) in cells with a germline configuration, and both distances are shorter than the distance between Neo R (or hs1,2) and hs3b (13.7 kb). In addition, between hs3a and the Neo R gene, there is a single intervening ϳ5-kb HindIII fragment, whereas between Neo R and hs3b, there are four HindIII fragments. Both features would favor nonspecific hs3a-Neo R interactions. We can conclude, however, that the Neo R replacement of hs1,2 has severely disrupted V H interactions with downstream 3Ј RR elements.
Detection of Interactions between myc Promoter and 3Ј RR-The chromosomal translocation breakpoint in MPC11 joins myc exon 2 to switch ␥2a sequences (18), and various experiments have suggested that myc is under the control of the 3Ј RR (31). 3C experiments (Fig. 5) showed that the myc promoter interacted with the 3Ј RR regulators. In accord with the detection of only cis interactions by the 3C assay conditions, no interaction was detected with the intronic enhancer, which is located only on the IgH expressed, nontranslocated chromosome.

Similar V H -3ЈRR Interactions Are Observed in Primary B
Cells-We wanted to determine whether interactions between the 3Ј RR and target IgH sequences could also be identified in normal B cells. We analyzed HindIII fragments containing J H3 , which in B cells would track all VDJ rearrangements that used J H2 or J H3 segments; rearrangements to J H1 or J H4 would each be located on separate HindIII fragments and are not analyzed. 3C analysis (Fig. 6) showed that in splenic B cells, but not in splenic T cells (or in mouse erythroleukemia cells; data not shown), a J H3 -containing fragment was associated with 3Ј RR enhancers, hs3a, hs1,2, hs3b-4, and hs5-7. These experiments show that rearranged VDJ sequences in resting splenic B cells are associated with multiple 3Ј RR regulators.

DISCUSSION
The results we report here document physical interaction between the expressed IgH V H gene and 3Ј RR sequences by two approaches, 1) characterization of an inversion of the expressed IgH locus in the F5.5 plasmacytoma cell line and 2) 3C analysis of the MPC11 plasmacytoma cell line that is the parent of F5.5, MPC11 variants that contain targeted deletions of IgH regulatory elements, and normal splenic B cells. The distance separating expressed VDJ and 3Ј RR sequences is ϳ80 kb in MPC11 and F5.5, the two ␥2b, -producing cell lines that we have studied, and ϳ180 kb in IgM-producing splenic B cells.
Sequence analysis of the F5.5 inversion/deletion breakpoints provides evidence for interaction between the expressed V H gene and the 3Ј RR (see model, Fig. 1D). The inversion can be explained by reciprocal recombination between VDJ coding sequences immediately downstream of the codon for N-14, i.e. the 14th amino acid from the N terminus of the mature MPC11 IgH protein, and 3Ј RR sequences at a position located ϳhalfway between hs1,2 and hs3b. DNA interactions in F5.5 could be facilitated by the microhomology observed upstream of the inversion breakpoint between V H MPC11, a member of the populous IGHV J558 family, and 3Ј RR sequences (22). The inverted allele in F5.5 also contains an ϳ40-kb deletion that encompasses C␣ together with two of the 3Ј RR enhancers, i.e. hs3a and hs1,2. A similar deletion has been characterized in the 70Z/3 pre-B cell line (10), suggesting that specific IgH DNA segments may regularly interact with each other.
The mechanisms that result in DNA breaks and recombination in F5.5 are not known. Motifs in the vicinity of the inversion and deletion breakpoints include AGCC, a known activation-induced cytidine deaminase hotspot (32) (Fig. 1), GAGCT "switch-like" motifs, and TGGG. However, IgH segments that contain F5.5 breakpoints are not known to undergo recombination associated with CSR, and there is no additional evidence to support the involvement of activation-induced cytidine deaminase or CSR machinery. The inversion in F5.5 is most simply explained as an unusual sequel to ongoing physical interaction between VDJ and 3Ј RR sequences that is predicted to occur regularly as part of the transcriptional regulation of the IgH locus in plasma cells.
To test this prediction, we implemented 3C analysis on a progressive series of MPC11 variants. Our studies detected interactions between fragments mapping to the 5Ј end of the MPC11 IgH transcription unit (V H , J H , and E) with enhancers and insulators at the 3Ј end of the entire IgH locus (3Ј RR and hs5-7). As shown in Fig. 3, we found that 5Ј fragments could bind to several different 3Ј fragments. We do not know, however, whether in individual cells these interactions occur in single pairs or in varying constellations of multiple interactions; 3C analysis captures the totality of interactions that occur in the population of cells analyzed and does not discriminate subpopulations.
Our observations have not only revealed loop formation between V H and 3Ј RR regulators but also underscored its intimate association with H chain expression (see the model in Fig.  7). In accord with previous studies in which the deletion of the endogenous intronic enhancer (9921) had no effect on levels of IgH expression, the studies reported here found essentially no modifications in 3C interactions between V H and 3Ј RR regulators, except for a slight reduction in interaction between V H and hs3a. In contrast, examination of B48, which upon substitution of hs1,2 by the Neo R gene suffered complete loss of H chain expression, revealed an accompanying disruption of the loop structures present in MPC11 and 9921. In B48 the hs3a enhancer was the focus of increased interaction with upstream V H sequences on the one hand and downstream Neo R sequences on the other, coupled with a loss of interactions between V H and 3Ј RR regulators downstream of the Neo R insertion. As discussed under "Results," we cannot eliminate the possibility that the deletion of genomic sequences during the construction of B48 may have contributed a nonspecific component to the hs3a-Neo R signal. In any event it seems unlikely that hs3a would interact with both V H and Neo R at the same time because there was no detectable interaction between V H and Neo R . One explanation of these findings would then be a competition between V H and Neo R promoters for the hs3a enhancer. These data show that the enhanced interaction detected between the expressed V H gene and hs3a is inadequate to support any H chain expression in the absence of the intronic enhancer, instead suggesting that interactions of V H with hs1,2 and/or with other downstream enhancers are essential. Mouse knockouts accompanied by the introduction of the Neo R gene into the H chain locus have frequently resulted in a phenotype that is not apparent upon analysis of the "clean deletion." For example, substitution of the hs1,2 enhancer (and independently, hs3a) by Neo R to generate the mouse knock-out resulted in a loss of CSR similar to that occurring after clean deletion of hs3b and hs4 (33). The basis for the "Neo effect" is not known but has been suggested to result from promoter competition, in this case, the prevention of the influence of hs3b and/or hs4 on target I region sequences (33). Our experiments provide evidence for a link between an alteration in long-distance interactions between regulatory regions and the deleterious effect of Neo R on H chain expression. In turn, these experiments imply that these long-range interactions are essential for H chain expression in plasma cells.
Whether long-range interactions are similar at various stages of B cell development and how they might contribute to IgH expression and rearrangements are not known. Studies (34,35) using fluorescence in situ hybridization have reported loops involving unrearranged V H proximal and distal genes and C␣ genes in both pro-B and, although less frequently, also in CD8 ϩ T cells, some of which appear to be related to the onset of VDJ joining. Our observations of 3C interactions between J H3 and 3Ј RR sequences in plasma cells and resting splenic B cells but only at background levels in splenic T cells, EL-4 thymoma cells, or in mouse erythroleukemia cells imply cell type-specific longrange interactions involving 3Ј RR sequences. It should be noted that the interactions extend through the 3Ј RR to the hs5-7 region, which has insulator activity (3). Interactions are not evident with the hole gene located ϳ30 kb downstream of hs7 or with the CalR gene located on a different chromosome. This implies that interactions between J H and/or E with 3Ј RR elements are limited to the IgH locus in plasma cells and B cells.
Recent studies on the ␤-globin locus have suggested that interactions between locus regulators and distal target sequences may help target the locus to transcription factories; hence, facilitating gene expression (36). Such physical interaction is consistent with the prediction that the 3Ј RR is required for high levels of IgH expression in plasma cells (5)(6)(7)37). Similar interactions in normal splenic B cells provide a potential scaffold by which the 3Ј enhancers can facilitate germline transcription, which is a prerequisite for class switch recombination. Identification of 3C interactions in the locus between promoters, intronic, and 3Ј enhancers (16) invokes a similar mechanism for both H and L expression.
The 3Ј RR has also been implicated in the dysregulation of the myc oncogene upon its translocation with the IgH locus in murine plasmacytomas and Burkitt's lymphomas. Involvement of the myc promoter in cis with the multiple elements of the 3Ј RR is analogous to the interactions in cis of the V H gene with the 3Ј RR. This suggests that the 3Ј RR may work similarly in its influence on target elements within the IgH locus or on other loci, such as those involved in malignant transformation.