An Igh distal enhancer modulates antigen receptor diversity by determining locus conformation

Abstract

The mouse Igh locus is organized into a developmentally regulated topologically associated domain (TAD) that is divided into subTADs. Here we identify a series of distal V_H enhancers (E_VHs) that collaborate to configure the locus. E_VHs engage in a network of long-range interactions that interconnect the subTADs and the recombination center at the D_HJ_H gene cluster. Deletion of E_VH1 reduces V gene rearrangement in its vicinity and alters discrete chromatin loops and higher order locus conformation. Reduction in the rearrangement of the V_H11 gene used in anti-PtC responses is a likely cause of the observed reduced splenic B1 B cell compartment. E_VH1 appears to block long-range loop extrusion that in turn contributes to locus contraction and determines the proximity of distant V_H genes to the recombination center. E_VH1 is a critical architectural and regulatory element that coordinates chromatin conformational states that favor V(D)J rearrangement.

Here the authors show enhancers in the Igh locus play a major role in configuring locus architecture and in V(D)J recombination, and identify a link between enhancer hub formation, locus contraction and cohesin-mediated loop extrusion.

Introduction

Progenitor B cells must develop a diverse antibody receptor repertoire to provide protection against a wide range of antigens and pathogens. Each mature B cell has a unique Ig receptor, created via V(D)J recombination through an ordered set of rearrangements. One of the ~100 functional Igh locus V_H genes must recombine with a rearranged DJ_H element, which itself is assembled from one of 8–12 D_H and one of 4 J_H gene segments in each pro-B cell¹. The intergenic control region 1 (IGCR1) separates the D_H J_H clusters from the proximal V_H segments and creates an insulated domain enforcing ordered D to J_H followed by V_H to DJ_H recombination^2–4. Igh locus contraction^5–7, diffusion related mechanisms⁸, and RAG scanning⁹ bring V_H segments into closer spatial proximity to the RC permitting usage of the V_H from across the locus. Nevertheless, the contribution of specific regulatory elements to locus conformation and the rearrangement of distal V_H segments remains unclear.

Chromatin is organized at the Mb scale into topological associating domains (TADs) that encompass spatial neighborhoods of high-frequency chromatin interactions^10–13 that include contact loops anchored by pairs of convergent CTCF binding elements (CBEs) in association with cohesin^14,15. TAD organization, CBEs, and loop domains may reflect the functional partition of chromatin regions by physiological activities^{10–12,14,16}. In this regard, it is significant that the Igh locus is contained within a 2.9-Mb TAD divided into three sub-TADs in pro-B cells¹⁷. TADs are frequently anchored by motifs bound by the architectural protein CTCF and its interaction partner, cohesin^14,18. The observation that CBEs situated at TAD boundaries are often in a convergent orientation¹⁴ has led to the proposition that chromatin loops are formed by an extrusion mechanism mediated by cohesin^19–22.

More than 100 CBEs are distributed throughout the locus and are convergent with 10 CBEs (termed 3’ CBE) at the 3’ TAD boundary^2,23. IGCR1 is composed of a pair of divergent CBEs where one motif is convergent with V_H CBE and the second is convergent with the 3’CBE²⁴ setting the stage for extensive IGCR1 centric loop extrusion. Studies have implicated loop extrusion in generating Igh locus contraction^25,26 and as a mechanism for a RAG scanning model implicated in VDJ gene assembly⁹. Nevertheless, the Igh locus topology is configured by large chromatin loops that provide access of the D_H-distal and -proximal V_H gene segments with rearranged 3’ D_HJ_H in pro-B cells^17,27. These observations raise a question regarding the interplay of loop extrusion and the contribution of locus topology determined by cis-regulatory elements to V_H choice during V_H to DJ_H rearrangement.

We report the discovery of four Igh V_H enhancers (E_VH1-4). Deletion of E_VH1 has a structural and functional impact on the Igh locus in cell lines and mice. E_VH1 participates in an enhancer interactome, determines locus topology, and modulates Igh gene transcription and regional V_H gene usage. Our findings reveal that E_VH1 may constrain loop extrusion and thereby determines chromatin conformational states that influence V_H to DJ_H recombination.

Results

Four enhancers identified in the Igh locus

The Igh TAD is divided into three sub-TADs linked by chromatin interactions between Site I of sub-TAD A, Friend of Site Ia (FrOStIa) and FrOStIb of sub-TAD B, and Sites II, II.5 and III all in sub-TAD C¹⁷ (Fig. 1a, g). To define motifs that anchor Igh chromatin loops, Site I and FrOStIa were visually scanned for distinct transcriptional and epigenetic features. Site I contains the highly transcribed V_H14-2 gene promoter (Pr)²⁸ that is prominently decorated with histone 3 lysine 4 methyl 1 (H3K4me1) and H3K4me3. The presence of H3K4me3 which mark transcriptionally active Prs²⁹ and the relative absence of enhancer-associated H3K27Ac modifications, indicates that this element predominantly functions as a Pr (Supplementary Fig. 1a). It is also bound by the co-activator MED1 and TFs Pax5 and IRF4 and lacks CTCF and RAD21 (Fig. 1b). We also identified E_VH1, located in FrOStIa, which is marked by H3K27Ac and H3K4me1 and is bound by TFs PAX5, IRF4, IKAROS, E2A, PU.1, and transcriptional co-activators MED1, BRG1, and p300 (Fig. 1b). H3K4me1 is associated with active and poised enhancers while H3K27Ac distinguishes active enhancers (reviewed in³⁰). E_VH1 is flanked by a CBE that is co-bound by CTCF and RAD21 (Fig. 1b). Likewise, we visually identified a series of other E_VHs including E_VH2, E_VH3, and E_VH4 (red rectangles) with an epigenetic and TF binding profile similar to E_VH1 that are interspersed among the intermediate and distal V_H gene exons (Fig. 1a, c)(Supplementary Fig. 1). E_VH2 is located within FrOStIb in sub-TAD B, and E_VH3 and E_VH4 are positioned between Site II and Site II.5 in sub-TAD C (Fig. 1a, c). We consider the possibility that V_H14-2 Pr and the E_VHs play a role in V_H gene transcription, locus topology, and rearrangement of nearby V_H genes.

Fig. 1. Identification of Site I and FrOStIa loop anchors motifs.

a Schematic of the Igh locus with genomic coordinates (chr12, mm10) the directionality of which follows chromosome 12 showing the Igh sub-TADs A-C¹⁷ and DNA elements (3’Eα, red dot; CBEs, purple arrows with orientation indicated; Eγ, teal dot; Eµ, pink square; E_VH1 (chr12:114182400-114183200), E_VH2 (chr12:114609100-114609900), E_VH3 (chr12:115023400-115024300), E_VH4 (chr12:115257500-115258400), red squares; C_H region genes, green bars; D_H and J_H exons, black bars; V_H gene segments, blue bars. The 3C primer sites in Site I and FrOStIa are indicated at bottom. b Public Chip-seq and RNA-seq positive (+) and negative (-) strand analyses for V_H14-2 and E_VH1 and F.6 CBE were generated from Rag deficient pro-B cells (Supplementary Table 11). c ChIP-seq studies identify E_VH1, E_VH2, E_VH3 and E_VH4 (red rectangles). d–f 3C assays. Arcs (3C assays), primers are identified below the graphs, 3C probes (anchor symbol). Average crosslinking frequencies are from independent chromatin samples as indicated. Statistical comparisons are to T cells. P values from two-tail Student’s t test and SEMs. Source data are provided as a Source Data file. d 3C assays analyzing Site I anchored at F.6 CBE. Chromatin samples for Abl-t 445.3.11 line, n = 3; Rag2^−/− pro-B cells, n = 3; T cells: n = 3. e 3C assays analyzing FrOStIa and anchored at Site I.3. Chromatin samples for Abl-t 445.3.11 line, n = 2; Rag2^−/− pro-B cells, n = 2; T cells, n = 2. f (Upper panel) Schematic of subTAD A showing a subset of V_H genes and 3C primers (P) with genomic coordinates (chr12, mm10). (Lower panel) 3C assays using the Site I.3 (I.3) anchor and P.1-P.6 primers. Chromatin samples for Abl-t 445.3.11 line, n = 6; Rag2^−/− pro-B cells (n = 2); T cells (n = 2). g Schematic of the Igh locus with looping interactions between IGCR1:Site I (subTAD A; blue and orange dots), Site I:E_VH1 (subTAD A-B; orange and red dots), and between Sites I:II, Sites II:III and Sites I:III (all subTADs; red arcs)¹⁷.

Identification of Site I and FrOStIa loop anchors

We examined the spatial organization of Site I and FrOStIa in CD19⁺ Rag2^−/− pro-B cells, the Abelson transformed (Abl-t) Rag1 deficient pro-B cell line 445.3, and ConA activated splenic T cells using the 3C anchors, FrOStIa F.6 and E_VH1 (Fig. 1d) (Supplementary Fig. 2a, b). The F.6 3C fragment contains a CBE and is located ~15 kb upstream of E_VH1 (Fig. 1b). In both cases, looping interactions with 3C fragments Site I.1, I.2 and I.3 (V_H14-2 gene) were elevated in pro-B cell specific chromatin templates as compared to that found in T cells (Fig. 1d) (Supplementary Fig. 2b). Likewise, the Site I.3 anchor probe, which harbors the V_H14-2 gene, associates with F.2-F.3 (E_VH1) and F.6 (CBE) and F.8-F.9 in pro-B cell specific fashion (Fig. 1e). Thus, the Site I.3 fragment is a major interaction partner with E_VH1 and F.6 (CBE). It was difficult to determine candidate loop anchors in Sites I.1 and I.2 due to an unremarkable epigenetic landscape and TF binding profile.

To determine whether the Site I.3 fragment also associates with D_H proximal V_H genes we analyzed locations close to V_H segments (P.1-P.6) and IGCR1 in chromatin from Rag2^−/− pro-B cells, the 445.3 line, and ConA activated splenic T cells in 3C assays (Fig. 1f). IGCR1, composed of two CBEs in divergent orientation, is a boundary element that separates the D_H-J_H clusters from the most proximal V_H genes^3,4,31 (Fig. 1a). Site I.3 contacts P.1-P.5 and IGCR1 but not P.6 in pro-B cell specific chromatin and not in T cells (Fig. 1f). Thus, Site I.3 engages in a multiplicity of interactions spanning sub-TAD A (blue arc) and may directly or indirectly bridge IGCR1 and the D_H proximal V_H genes with E_VH1/F.6 located in subTAD B (Fig. 1g).

Deletion of a V_H promoter, E_VH or CBE diminish looping and transcription

To establish a cell culture model to study the involvement of specific transcriptional elements (TE) and CBE motifs in Igh locus function we employed genome editing to generate identical deletions on both alleles of the V_H14-2 promoter (Pr), E_VH1, F.6 CBE, E_VH2 and Site I.1 in the 445.3.11 sub-line (Fig. 2a)(Supplementary Fig. 3a–d). V_H Prs are composed of a polypyrimidine tract, heptamer and octamer and deletion of the octamer/heptamer is sufficient to abolish Pr activity³². CTCF binding at the F.6 CBE in the control and Site I.1 CBE KO lines was detected whereas binding was abolished in the F.6 CBE KO lines in ChIP assays (Fig. 2b). Because there was little epigenetic guidance to identify potential loop anchor motifs within Site I.1 we arbitrarily constructed deletions centered on the V_H2-8 exon and an adjacent CBE (Supplementary Fig. 3e). Site I.3:F.6 looping interactions were dependent on the integrity of the V_H14-2 Pr, F.6 CBE and E_VH1 motifs but not on E_VH2 or deletions around the V_H2-8 gene in Site I.1 (Fig. 2c) (Supplementary Fig. 2c, d). F.6:Site I.1 and Site I.3:Site I.1 contacts required intact F.6 CBE and V_H14-2 Pr, respectively, indicating the autonomy of each these interactions (Fig. 2c). Thus, the V_H14-2 Pr, E_VH1 and F.6 CBE coordinate Site I:FrOStIa looping.

Fig. 2. Igh locus function is dependent on a V_H promoter, an E_VH and CBE in Abl-t pro-B cells.

Source data are provided as a Source Data file. All p values are from two-tail Student’s t test. a (Upper panel) The Igh locus from the V_H14-2 gene to E_VH2. (Lower panel) Biallelic identical deletions constructed for the V_H14-2 promoter (V_H14-2 Pr, yellow rectangle), F.6 CBE (F.6_CBE, purple triangle), E_VH1 (red square), and E_VH2 (white square) in the Abl-t 445.3.11 cell line. b ChIP assays for CTCF binding at the F.6 CBE in the control (n = 3), Site I.1- (n = 3) and F.6 CBE (n = 3) KO lines. c 3C looping assays. (Upper panel) Map of genomic distances. (Lower panel) Normalized 3C crosslinking frequencies from four experiments in which the 455.3.11 (n = 10) or 445.3.8 (n = 2) matched controls were set to 1.0 and represented once. The crosslinking frequencies were averaged for chromatins from each KO type (E_VH1 KO, n = 9; E_VH2 KO, n = 6; F.6_CBE KO (n = 6); F.6_CBE KO (n = 8); VH14-2 Pr KO (n = 6) and normalized to the control. d QRT-PCR assays for the V_H14-2 gene using 18 S RNA as a loading control. Samples were from control (n = 14) VH14-2 Pr KO (n = 8), E_VH1 KO (n = 6), F.6_CBE KO (n = 9). e RT-PCR of V_H81X and V_H2-5 GLTs and the 18 S RNA loading control were harvested at 32 and 28 cycles, respectively. Cells were STI-571 (2.5 mM) treated for 48 h. f Schematic of the D_HJ_H clusters and proximal V_H gene segments (vertical bars), with primer positions (colored arrows) used to analyze D-J and V-DJ rearrangements. g, h Lines were stably complemented with RAG1 and STI-517 treated. g D-J rearrangements in gDNA were analyzed by qPCR with DFL16.1 and J_H1 primers. Samples were from 455.3.11 (n = 4), 455.3.8 (n = 5), V_H14-2 Pr KO (n = 4), E_VH1 KO (n = 6), F.6_CBE KO (n = 4). h Normalized qRT-PCR data for V- > DJ rearrangement using V_H81X and V_H7183 primers in combination with J_H1. V_H81X-J_H1 analyses were from control 455.3.11 (n = 4), 455.3.8 (n = 3), E_VH1 KO (n = 6),V_H14-2 Pr KO (n = 3), F.6_CBE KO (n = 4). Samples numbers for VH7183-JH1 analyses were identical to that of V_H81X-JH1 except for 455.3.11 (n = 12), V_H14-2 Pr KO.1 (n = 5) and F.6_CBE KO.2 (n = 4).

To validate E_VH1 enhancer function we tested the transcriptional effects of deletion on V_H germline transcript (GLT) expression. These GLTs are ncRNAs that are transcribed from V_H genes poised to undergo V(D)J rearrangement at the pro-B cell stage. Unlike most V_H GLTs, the high level of V_H14-2 GLT expression was evident in the absence of STI-571 induction that is used to induce Ig gene transcription³³. V_H14-2 GLTs were abolished or significantly diminished in the V_H14-2 Pr-, E_VH1- and F.6-KO lines, respectively, as compared to the control demonstrating that the transcription of this GLT is both Pr- and E_VH1- dependent (Fig. 2d). We studied two additional V_H genes in 445.3.11 cells that were arrested in G1 phase by STI-571 treatment. V_H81X is a member of the V_H7183 family and the first functional D_H-proximal V_H gene and V_H2-5 is a member of the V_HQ52 family (V_HQ52.7.18) and located in sub-TAD A (Fig. 1a). V_H GLTs, composed of a 5’ leader sequence and V_H exon are often found as unspliced and spliced (Fig. 2e). The V_H81X (unspliced, 300 bp; spliced, 163 bp) and V_H2-5 (unspliced, 338 bp) GLTs are clearly present in the 445.3.11 control and in the E_VH2 KO lines (Fig. 2e). However, expression of these transcripts is abolished in E_VH1 KO cells, and diminished in V_H14-2 Pr- or F.6_CBE KO lines indicating that 1) E_VH1 is essential for transcription of at least several V_H genes and 2) that the V_H14-2 Pr and F.6 CBE elements may contribute to E_VH1 function by facilitating its spatial proximity to V_H exons in subTAD A.

V(D)J recombination requires specific loop anchors in Abl-t lines

To investigate the influence of TE and CBE KOs on V(D)J recombination we studied D- > J_H and V_H- > DJ_H rearrangements in Rag1^−/− Abl-t 445.3.11 control and KO lines. D- > J_H recombination reflects the activity of the Eμ proximal recombination center (RC) that forms over the J_H cluster and to which RAG1/2 is recruited³⁴. Cells were stably complemented with a RAG1 expression construct and treated with STI-571 to induce endogenous RAG2 and VDJ recombination³⁵. D- > J_H rearrangements, assessed in gDNA using the DFL.16 and J_H1 primers, were largely comparable in control and KO lines indicating that the first step in V(D)J recombination is intact and independent of TE and CBE deletions (Fig. 2f, g). Slightly reduced D- > J_H recombination in E_VH1 KO.1 may be attributable to subclone variability (Fig. 2f, g).

V_H81X, is the most highly used V_H gene in Abl-t pro-B cell lines³⁶ and in mouse C57Bl/6 pro-B cells³⁷ whereas distal V_HJ558- > DJ_H recombination is rare in Abl-t pro-B cell lines³⁶. V_H7183- > DJ_H rearrangements were analyzed in qRT-PCR using two forward primer sets specific for V_H81X (gray arrow), and a pan specific V_H7183 primer (red arrow) in combination with the reverse J_H1 primer (blue arrow) (Fig. 2f). Although V_H7183 rearrangement levels were strikingly reduced in all Abl-t pro-B cell lines harboring TE and CBE deletions, V_H81X rearrangements were significantly diminished only in E_VH1 KO lines (Fig. 2h). It is possible that the differences we detect for V_H81X and V_H7183 rearrangements in the KO lines are related to a sensitivity limitation of the assay due to the relatively low level of V_H7183 recombination and GLT expression. Therefore, we asked if V_H81X GLT expression was correlated with the propensity to engage in rearrangement. Indeed, V_H81X GLTs were undetectable in the E_VH1 KO lines consistent with the significant reduction in VDJ recombination but only partially diminished in V_H14-2 Pr- and F.6 CBE KO line suggesting that even low-level transcription may contribute to V_H81X- > DJ_H recombination (Fig. 2e). Thus, E_VH1 deletion led to the deepest deficits for both transcription and recombination relative to other KOs (Fig. 2e, h).

A Eμ-V_HPr-E_VH1 chromatin hub mediates Igh locus architecture

Efficient V_H- > DJ_H recombination, transcription and chromatin loop formation all require intact V_H14-2 Pr, E_VH1 and F.6 CBE motifs suggesting that these elements organize essential structure/function relationships in the Igh locus. V_H14-2Pr participates in a multiplicity of contacts in subTAD A including IGCR1 and with E_VH1 in subTAD B indicating a high degree of connectivity (Fig. 1e, f). To explore the role of V_H Pr-E interactions we evaluated the overall topology of the locus spanning Eμ to E_VH1 and the formation of a regulatory hub using high resolution 3D DNA FISH and short probes (4.8-7.5 kb) for Eμ (green), 3’Eα/Site I.3 (3’Eα/V_H14-2 Pr; red) and E_VH1 (blue) (Fig. 3a). To explore the topology of the Igh locus we measured mean spatial distances³⁸ between FISH probes (Fig. 3a, b). The mean spatial distances separating the Eμ anchor probe from the 3’Eα and Site I.3 probes increased as a function of genomic distance and then leveled off between Site I.3 and E_VH1 in controls indicating locus compaction (Fig. 3c). In contrast, overall mean spatial distances increased in E_VH1 and F.6_CBE KO lines indicating substantial decompaction in subTAD A-B upon loss of these elements (Fig. 3c, d). The increased spatial distance separating Eμ and 3’Eα was confirmed in 3C studies which showed reduced crosslinking frequency for Eμ:3’Eα in the E_VH1- and F.6_CBE KO lines as compared to the control line (Fig. 3e). Deletion of the V_H14-2 Pr led to a dramatic increase of average spatial distance between Eμ and E_VH1 whereas Eμ and Site I.3 were more compacted relative to the control in FISH assays (Fig. 3c, d). Thus, Eμ:Site I.3 association is not mediated by direct interaction with the V_H14-2 Pr and must therefore be supported by another element. We conclude that E_VH1 and F.6_CBE have a profound influence on locus compaction in subTADs A-B and that the V_H14-2 Pr:E_VH1 interaction serves to bridge E_VH1 to Eμ possibly through its interaction with IGCR1 (Fig. 1f).

Fig. 3. Transcriptional elements regulate locus compaction in Abl-t pro-B cells.

Source data are provided as a Source Data file. a The Igh locus from 3’Eα to FrOStIa and the position of FISH probes. Chromatin loops detected in 3C studies from pro-B cells (red arcs; Fig. 1d, f) and previous work (black arcs)^4,57,78. FISH probes are separated by 220 kb (Eμ-3’Eα), 568 kb (Eμ-Site I.3) and 758 kb (Eμ-E_VH1). b Representative nuclei from fixed 445.3.11 and KO lines were simultaneously hybridized in duplicate with the indicated probes (y-axis). Short probes were labeled with AlexaFluor 488 (Eμ, green), AlexaFluor 555 (Site I.3 or 3’Eα, red), and AlexaFluor 647 (E_VH1, blue). Scale = 2 μm. c Mean spatial distances (y-axis) were plotted as a function of genomic distance (x-axis) with lines indicating connectivity only, for 404 alleles from two independent experiments for each probe combination²⁷. Vertical dashed line shows the Eμ anchor. d N = 404 alleles/genotype over two independent experiments. Box plots represent the distribution of spatial distances with medians (middle line), 25^th and 75^th percentile (box), and 10^th and 90^th percentile (whiskers), and outliers (single points) shown. P values from two-tailed nonparametric Mann-Whitney U test. e 3C crosslinking frequencies for Eμ:3’Eα looping interactions anchored at Eμ, analyzed in chromatin samples from the 445.3.11 control (n = 4) and KO (n = 4) lines except for E_VH1 KO (n = 7) with SEMs. P values from unpaired two tailed Student t test. f Quantitation of 3D probe configurations in three-color DNA FISH. Pairwise and three-way distances between the red, green, and blue DNA FISH signals were divided into three categories: <0.3, 0.31–0.5, >0.5 μm. Nine probe configurations (upper panel) and their frequencies (lower panel) are shown. Inset a: Three-way overlap with all probes <0.3 μm from each other, insets b–d: beads on a string configuration with probes 1-2 and 2-3 or 1–3 < 0.3 μm from each other, insets e–g: pairwise probe configurations with two probes <0.3 μm from each other, inset h: all probes distanced by 0.31–0.5 μm, inset i: all probes distanced by >0.5 μm. P values, two tailed Chi square tests. g Representation of Igh allele configurations.

Molecular contacts between chromatin elements are indicated by superimposed FISH probes (≤0.3 μM). Segregation of probe contacts into structural categories allows identification of pure pairwise and three-way overlap interaction frequencies in single cells. Importantly, pure pairwise Site I.3- E_VH1 interactions were significantly diminished in all three KO lines in accord with our  3C analyses (Figs. 2c, 3f). Probes for Eμ, Site I.3 and E_VH1 were superimposed or overlapped as beads on a string in varying order in ~20% of control alleles whereas the frequency of these probe configurations trended lower- or were significantly reduced in all three KO lines (Fig. 3f, inset a-d). Thus, formation of a chromatin hub requires the presence of the V_H14-2 Pr-, E_VH1- and F.6 CBE. Significantly, the frequency of distantly spaced probes (≥0.31 μM) in which no overlap occurs greatly increased from ~20% of alleles in controls to 40-50% in the KO lines demonstrating an overall destabilization of subTAD A-B conformation (Fig. 3f inset h-i). The reduced presence of three-way interactions and the increased topological instability are consistent with reduced V- > DJ recombination found for these KO lines. Eμ-Site I.3 looping increased upon V_H14-2 Pr KO suggesting that this Pr blocks these interactions, perhaps by binding with alternative interaction targets (Fig. 3f inset e). The Eμ- E_VH1 configuration was largely impervious to TE deletion indicating that this pairwise conformation may be formed through other interactions (Fig. 3c). An ensemble of topological configurations detected in single cells including the Eμ-V_H14-2 E_VH1/F.6 CBE chromatin hub are schematically represented and together represent the population average locus conformation (Fig. 3g). Hence, the V_H14-2 Pr, E_VH1 and F.6_CBE have a profound influence on 3’Igh locus compaction and the spatial accessibility of proximal V_H genes to the RC.

V_H gene usage is altered in mice lacking E_VH1

To assess the influence of E_VH1 on Igh repertoire and locus conformation in a physiological setting, we deleted 515 bp spanning E_VH1 using a genome editing strategy in mice (Fig. 4a). To analyze the preselected Igh repertoire we FACS purified pro-B cells from WT and E_VH1^−/− mice, and performed VDJ-seq, an unbiased assessment method for V_H segment usage in VDJ_H junctions in gDNA (Supplementary Fig. 4a)³⁹. The overall profile for V_H gene usage in our WT samples compares well to previously reported findings³⁹ (Fig. 4b). In contrast, deletion of the E_VH1 element led to significantly reduced V_H gene usage, in a 335 kb domain which we refer to as the E_VH1 zone of influence (ZOI), extending from 5’ of V_H81X (V_H7183.2.3) to the V_HJ606 genes that mark the end of the intermediate genes and are located approximately halfway through subTAD B (Fig. 4c, d). V_H genes closest to E_VH1 are more severely reduced, in general, by the absence of E_VH1. V_H gene usage outside the ZOI was sporadically altered with increased usage near E_VH2 and E_VH3 implying widespread structure/function changes in E_VH1^−/− alleles (Fig. 4b, c).

Fig. 4. E_VH1 regulates regional V_H gene usage in mice.

Source data are provided as a Source Data file. a E_VH1 was deleted in mice using genome editing. Guide RNAs (blue) with PAM motifs (red) are indicated with the E_VH1 deletion. b, f–k P values from unpaired two-tailed Student’s t test. b VDJ-seq analysis of gDNA from WT and E_VH1^−/− pro-B cells. WT (n = 2), E_VH1^−/− (n = 2). V_H gene reads from two independent experiments were normalized. c Ratio of E_VH1^−/−/WT normalized reads. Chromatin loop anchors (grey vertical bars) and V_HJ606 genes (red line) are indicated. d VDJ-seq cumulative fold change (E_VH1^−/−/WT) in ZOI (red) and distal (blue) regions (c) using the two tailed Mann Whitney Wilcoxon test for p values. Box plots show the mean (middle line), 25th and 75th percentile (box), 10th and 90th percentile (whiskers) and outliers (single points). e Schematic of the D_HJ_H clusters and proximal V_H gene segments (vertical bars), with J_H and pan-specific V_H7183 primer positions (colored arrows) used to analyze D-J and V-DJ rearrangements in qPCR assays as described⁴⁰. f–h Genomic DNA was prepared from primary pro-B cells isolated by FACS from WT (green bars) and E_VH1^−/− (blue bars) mice (Supplementary Fig. 4b). WT (n = 2), E_VH1^−/− (n = 2). Each sample was pooled from three mice. f D-J rearrangements were analyzed with the degenerate DFL16.1/DSP2 primer in combination with each of four J_H primers and normalized values were set to 1.0 for WT. g, h V-DJ rearrangements were amplified using the V_H7183 (g) or the V_HJ558 (h) primer in combination with each of four J_H primers and normalized values were set to 1.0 for WT. i–k GDNA were prepared from control (n = 2) (gray bars) and E_VH1 KO.1 and KO.2 (n = 2) lines (red bars). i, j, k D-J (i), V_H7183-DJ (j), and V_H81X-DJ (k) were analyzed in combination with each of four J_H primers and normalized values were set to 1.0 for the control. P values from unpaired two-tailed Students t test.

We observe that V_H81X usage is not diminished in the VDJ-seq analysis of E_VH1^−/− pro-B cells in contrast to what was observed in the E_VH1 KO Abl-t lines (Fig. 2h). It should be noted that the repertoire of rearrangements is very limited in Abl-t cell lines, with V_H81X being the dominant V_H gene used for rearrangement³⁶, presumably due to the absence of locus contraction⁹ so that the repertoires are quite different in these two cell types. More relevant to this issue, VDJ-seq in pro-B cells allows analysis of the relative ratio of V_H gene usage within normalized data sets but cannot determine if the absolute level of rearrangement is lower in one sample compared to another. Therefore, to address the possibility that the overall frequency of proximal V_H gene rearrangements was lower in the E_VH1^−/− pro-B cells despite the relative usage of V_H7183 genes being similar, we used pan-specific V_H7183 and V_HJ558 primers with reverse primers for each J_H segment⁴⁰ and analyzed VDJ rearrangements in gDNA of WT and E_VH1 KO cells in both primary pro-B cells and the Abl-t lines (Supplementary Fig. 4b) (Fig. 4f–k). V_H7183 rearrangements in conjunction with each J_H segment are significantly diminished or trended lower in E_VH1^−/− pro-B cells as compared to WT whereas D-J recombination is unaffected (Fig. 4f, g). Even more pronounced, V_H7183 and additionally the V_H81X gene rearrangements are greatly reduced in the Abl-t E_VH1 KO line whereas D-J recombination is intact in accord with our findings (Fig. 2g, h) (Fig. 4i, j, k). Hence, there is an overall similarity in the impact of E_VH1 deletion on V_H7183 gene usage in pro-B cells and Abl-t pro-B lines. In contrast, there are variable effects of E_VH1 deletion on V_HJ558 rearrangements with different J_H segments in pro-B cells which may reflect the variable impact of E_VH1 deletion on V_HJ558 usage found in the VDJ-seq study (Fig. 4c, h). We conclude that the frequency of V(D)J rearrangement is reduced for the proximal V_H genes in the E_VH1^−/− mice. Collectively, these findings indicate a consistent diminution of VH7183 usage upon loss of E_VH1.

Splenic B1a cells binding Ptc are diminished in E_VH1^−/− mice

The E_VH1 ZOI encompasses the D_H proximal V_H5 (V_H7183) and V_H2 (V_HQ52) families and the small intermediate gene families including the V_H11 and V_H12 exons that are frequently expressed in B1 B cells^41,42. V_H11.2 in particular is highly over-represented in B1a B cells, and such antibodies are reactive with PtC⁴¹. Because V_H11.2 rearrangement was reduced 2-fold in E_VH1^−/− as compared to WT pro-B cells (Fig. 4b, c), we examined the presence of B cell compartments and the frequency of B1a- and PtC binding cells. E_VH1 deletion had little effect on B cell development as comparable proportions of Hardy fractions (B-C’, pro-B; D, pre-B; E, immature B; F, mature B) in the BM and marginal zone and follicular B cells in the spleen are detected in WT and E_VH1^−/− mice (Supplementary Fig. 5a–g). Although normal proportions of peritoneal B1 B cells were detected in E_VH1^−/− (Supplementary Fig. 5f, g), splenic B1 B cells expressing surface CD19⁺B220^loCD23^-IgM^hiIgD^lo were significantly reduced in E_VH1^−/− relative to WT mice (Fig. 5a, b). Moreover, B1a cells binding PtC are depleted in E_VH1^−/− as compared to WT mice (Fig. 5c, d). The reduced rearrangement of V_H11.2 in the E_VH1^−/− pro-B cells is therefore a possible explanation for the reduction of splenic B1a cells and those binding PtC. Thus, perturbation of the preselected Igh repertoire in pro-B cells has potential ramifications for the peripheral repertoire.

Fig. 5. Phosphatidylcholine binding splenic B1a cells are reduced in E_VH1^−/− mice.

a Representative flow cytometry plots of splenocytes gated for B1 B cells were assessed using FVS-510 and antibodies (CD19-PerCp, B220-APC Cy7, CD23-FITC, IgM-e450, IgD-AF700). b, d Frequencies of B1 B cells were averaged and each symbol represents results from one mouse. b Average frequency of B1 B cells (CD19⁺B220^loCD23^-IgM^hiIgD^lo) was assessed in three independent experiments. WT (n = 35), E_VH1^−/− (n = 19) mice. P values from unpaired two-tailed Student’s t test. c Representative flow cytometry plots of splenocytes gated for B1 B cells using FVS-510 and antibodies (CD19-PerCp, B220-APC Cy7, CD43-APC, IgM-e450, IgD AF700) and PtC liposome (FITC). d Average frequency of B1 B cells binding PtC (CD19⁺B220^loCD43⁺IgM^hiIgD^loPtC⁺). WT (n = 5), E_VH1^−/− (n = 5) mice. P values from unpaired two-tailed Student’s t test.

E_VH1 is an Igh locus architectural element in mice

To define the contribution of E_VH1 to locus architecture, in situ Hi-C data sets¹⁴ were generated and biological replicates were merged (Supplementary Fig. 6). Rag deficient pro-B cells are used to ensure that the Igh locus has not undergone V(D)J recombination and are compared with those from mouse embryonic fibroblasts (MEF) cells in which the Igh locus is inactive⁴³. Pooled Hi-C samples recapitulated reported chromatin structures with high reproducibility scores and similar data quality (Supplementary Table 10) (Supplementary Fig. 6a)¹⁴. The Igh 2.9 Mb TAD (yellow box) in pro-B cells displays strong locus boundaries (green/blue arrows) and rarely interacts with other compartments (black brackets) as previously observed (Supplementary Fig. 6a, b)¹⁷. In contrast, the 5’ Igh TAD boundary in MEF is subsumed into an adjacent TAD and interacts with other compartments indicating a B cell specific difference in TAD structure (Supplementary Fig. 6a, b).

To identify pro-B cell specific interactions we created Hi-C difference heatmaps by subtracting MEF- from pro-B- derived contacts, a strategy used in our earlier studies¹⁷ and adapted to Hi-C data sets (Methods). The Rag1^−/− minus MEF difference map reveals nested self-interacting domains (vertical dashed lines) spanning hundreds of kb along the locus that have been schematically summarized (Fig. 6a, bi, iii). We detect stripes (blue arrowheads) originating from the 3’RR and IGCR1 as previously observed^23,44 that are absent in MEF (Fig. 6bi) and are also visible in the unsubtracted Hi-C data (black arrows) (Supplementary Fig. 6b, c). Architectural stripes reflect interactions between a single anchor and a continuum of genomic elements, are considered evidence of cohesin mediated loop extrusion^19,44 which likely facilitates V(D)J joining in pro-B cells⁹. These observations support a major role of IGCR1 as a downstream anchor of loop extrusion for the V_H locus rather than Eμ as previously suggested⁹. We observe a series of stripes (black arrowheads) extending in a 5’->3’ direction from the corners of multiple self-interacting domains (Fig. 6bi). Several of these stripes meet off the diagonal to form “dots” indicating the presence of IGCR1-E_VH1 and IGCR1- E_VH2 (marked a and b), E_VH1-E_VH2 (marked c), and V_HJ606 genes-E_VH2 (marked d) anchored loops indicating that E_VH- E_VH and E_VH-Pr contacts contribute to Igh locus conformation (Fig. 6biii). Corner dots are most frequently anchored by CBE in a convergent orientation, bound by the CTCF architectural protein and represent persistent loops^14,45. All CBEs bound by CTCF in the V_H domain are convergent with the IGCR1 CBEs²³. Each of the E_VHs is paired with a nearby CBE bound by CTCF and the cohesin component, RAD21 (Fig. 1b) (Supplementary Fig. 1). The F.6 CBE flanking E_VH1 interacts with Site I.3, and Site I.I and these contacts were abolished upon F.6_CBE deletion in Abl-t pro-B lines (Fig. 2c) (Supplementary Fig. 2c, d). These observations together suggest that the Hi-C dot domains were formed by loop extrusion that is impeded by CTCF/cohesion as noted^19,21,46. Another possibility is that active enhancers, such as E_VH1s can function as barriers to loop extrusion^47,48.

Fig. 6. E_VH1 modulates Igh locus topology and compaction in pro-B cells.

a Schematic of the Igh locus. Eμ is situated between J_H and C_H genes and 3’Eα. The J_H and D_H segments are located between the constant region genes (C_H) and V_H7183 (V_H81X) segments. b Regulatory elements (colored dots) identified in a arrayed along the Hi-C diagonal. Hi-C difference heatmaps at 10 kb resolution from purified CD19⁺ pro-B cells from Rag1^−/− (n = 2) (i) and Rag1^−/−E_VH1^−/− (n = 2) (ii) mice were generated (see Methods). Nested loop domains (dashed vertical lines), stripes (3’->5’, blue triangles; 5’->3’ black triangles), dots (black circles), lost dots (gray circles). Rag1^−/−E_VH1^−/− specific stripes (purple triangles) and dots (purple circles). iii, iv) Schematic of nested loop domains in Hi-C difference heatmaps from Rag1^−/− (iii), and Rag1^−/−E_VH1^−/− (iv) pro-B cells. c The Igh locus includes the interspersed V_H J558/3609 (V_H1-86) families at the 5’ end, the V_H7183 (V_H81X) family at the 3’ end and the intermediate V_H segments (V_H14-2) include the S107 family with nine smaller V_H families. Representative V_H genes are positioned above the locus (chr12, mm10), V_H81X (113578506 − 113578799), V_H14-2 (113994469-113994762), V_H1-2 (114469075-114469371) V_H1-86 (116000028-1160000321). BAC probes, H14, Eμ, RI, RII, RIII are indicated by their colors and are shown with the distances between them. d–f Source data are provided as a Source Data file. d Representative nuclei from fixed Rag2^−/− and Rag2^−/−E_VH1^−/− pro-B cells hybridized with three labeled probe combinations in two biologically independent experiments. Scale bar, 2 μm. Rag2^−/− pro-B alleles for each probe pair: n = 430 (H14-RI), n = 541 (RI-RII), n = 922 (Eμ-RI, RI-RIII, Eμ-RIII). Rag2^−/−E_VH1^−/− pro-B alleles for probe pairs: n = 1213 (H14-RI), n = 400 (RI-RII), n = 566 (Eμ-RI, RI-RIII, Eμ-RIII). e Mean inter-probe distances as a function of genomic distance. Lines indicate connectivity only. The RI anchor probe position, vertical dashed line. f Box plots show the median (middle line), 25th and 75th percentile (box) and 10th and 90th percentile (whiskers) and outliers (single points) of spatial distances between probes and anchor R1. P values from unpaired two-tailed nonparametric Mann–Whitney U test.

The structural profile of the Igh locus was substantially altered in Rag1^−/−E_VH1^−/− derived Hi-C difference maps (Fig. 6bi, ii) (Supplementary Fig. 7a, bi, ci). Many loops, represented by corner dots, involving E_VH1 were abolished including those between IGCR1-E_VH1 (marked a, grey) and E_VH1- E_VH2 (marked c, grey), whereas those involving the E_VH2 anchor such as IGCR1-E_VH2 (marked b), and V_HJ606 genes-E_VH2 (marked d) are retained in the Rag1^−/−E_VH1^−/− pro-B cells (Fig. 6bii, iv) (Supplementary Fig. 7bi, ci, d). Deletion of E_VH1 also led to the new appearance of IGCR1-Site I (marked e, purple circle) interactions, the formation of the δ loop (dashed box) configured by the association of Site I-V_HJ606 genes and a series of stripes and chevrons (purple arrowheads) (Fig. 6bii, iv) (Supplementary Fig. 7bi, ci, d). Active enhancers can function as TAD boundary elements and spatial proximity with promoters can be achieved during the extrusion process^47,48. In this context, it is evident that E_VH1 (green dot) directly participates in loop extrusion as stripes emanating from this position and moving toward IGCR1 disappear from the Hi-C difference map in Rag1^−/−E_VH1^−/− pro-B cells (Fig. 6bi, ii). Notably, impairment of Igh locus structure conferred by E_VH1^−/− extends from IGCR1 to the J606 genes and maps to the E_VH1 ZOI, thus, linking Igh spatial architecture with V_H gene utilization.

Loss of E_VH1 leads to locus hypercontraction

To explore the contribution of E_VH1 to the higher order locus conformation we compared the spatial distances between BAC FISH probes H14, Eμ-IGCR1 (referred to as Eμ) and the RI, RII, and RIII probes, corresponding to Sites I, II and III, hybridized in Rag2^−/− and Rag2^−/− E_VH1^−/− pro-B cells (Fig. 6c, d)¹⁷. We used probe RI as a reference and determined spatial distances with all other probes. Although spread over a large genomic region the average spatial distance separating probes RI from Eμ, RII, and RIII were similar whereas spatial distances to the H14 probe outside the locus were large indicating that most V_H genes are at comparable distances to the RC in Rag2^−/− pro-B cells, as previously observed (Fig. 6e, f)²⁷. In contrast, closer spatial distances are detected between the Eμ and RI probes with RIII in Rag2^−/− E_VH1^−/− pro-B cells reflecting hyper-compaction with the distal V_H gene segments relative to the control (Fig. 6e, f). Likewise, closer spatial distances are detected between the RI anchor with RII and RIII probes and greater distance to H14 in the Abl-t E_VH1 KO line reflecting increased locus compaction in the distal regions of the Igh locus relative to the control (Supplementary Fig. 8). Thus, E_VH1 deletion is correlated with increased overall locus compaction arguing for a link between long-distance loop extrusion and locus compaction.

E_VH1 coordinates an enhancer network

To explore the scope of E_VH interactions and their impact on Igh locus topology we reduced Hi-C dimensionality by deriving virtual 4C datasets. Using E_VH1, Eμ, and IGCR1 viewpoints we identified high-frequency interactions defined as in the top 15 (star) or 25 (circle) percent of all contacts in both biological replicates for Rag1^−/− (black symbols) and Rag1^−/−E_VH1^−/− (red symbols) pro-B cells (Fig. 7a). E_VH1 was highly interactive with its own flanking region, Eμ, E_VH2, E_VH3, and E_VH4 in Rag1^−/− pro-B cells (Fig. 7a). Although E_VH3 associates with E_VH1, its interaction frequency did not pass the threshold for high-frequency contacts in both replicates (Fig. 6A). Deletion of E_VH1 (red dashed line) led to loss of contact with all other E_VHs and with Eμ and the emergence of two new peaks (red arrows) of contact in Rag1^−/−E_VH1^−/− pro-B cells and implies that E_VH1 coordinates an enhancer network (Fig. 7a).

Fig. 7. E_VH interactome identified.

a Virtual 4C interactions were extracted from KR normalized Hi-C data sets from Rag1^−/− (black lines) and Rag1^−/−E_VH1^−/− (dashed red lines) CD19⁺ pro-B cells. Top: Genomic coordinates (mm10) with the E_VH1 ZOI and Sites (S) I, II, II.5 and III. 4C viewpoints (dashed vertical line) in a 30 kb running window analysis with 10 kb steps from merged biological replicates with E_VH1 interactions (black arcs) shown. Stars (top 15%), circles (top 25%) of locus-wide interactions. New peaks (red arrows) and lost peaks (black arrows) in Rag1^−/−E_VH1^−/− pro-B cells. b Diagram of the Igh locus with genomic distances and FISH probe indicated. Chromatin loops detected in pro-B cells (see Fig. 3a). c–e Source data are provided as a Source Data file. c Representative nuclei from Rag2^−/− and Rag2^−/−E_VH1^−/− CD19⁺ pro-B cells. Short FISH probes, Alexa Fluor 488 (Eμ, green), and Alexa Fluor 647 (E_VH1, blue), Alexa Fluor 555 (E_VH2, red) were hybridized simultaneously to fixed cells in duplicate. Scale bar, 2 μm. d Quantitation of 3D probe configurations in three-color DNA FISH. N = 400 alleles/genotype in two independent experiments. (Upper panel) Nine probe configurations. (Lower panel) The frequency of each probe configuration. P values from two tailed Chi square test. e Median spatial distance was derived for probes engaged in 3-way overlaps shown in panel d inset a. Rag2^−/− (n = 33), Rag2^−/−E_VH1^−/− (n = 54) alleles. Box plots show median (middle line) spatial distances, 25th and 75th percentile (box) and 10th and 90th percentile (whiskers), and outliers (single points) between. P values from two-tailed Wilcoxon signed rank sum test. f Graphically displayed median spatial distances between probes in three-way overlaps from Rag2^−/− (black arrows) and Rag2^−/−E_VH1^−/− (gray arrows) pro-B cells. g Representation of Igh alleles in distinct spatial configurations (a–c, e–g, h, i).

Examination of the Eμ viewpoint confirmed interaction with 3’Eα hs4, Eγ, and IGCR1 as previously reported^3,4,9,49 and with E_VH1 (Fig. 6A). Eμ:E_VH1 contacts were apparently lost and new or increased Eμ interactions (red arrows) with Eγ, Site I.3, and E_VH2 were detected in Rag1^−/−E_VH1^−/− pro-B cells suggesting that E_VH1 competes with Eγ and E_VH2 for interaction with Eμ (Fig. 7a). Finally, analysis of the IGCR1 viewpoint revealed interactions with 3’Eα hs4, Eμ, V_H81X and E_VH1 but not with Eγ or with other E_VHs indicating that IGCR1 could bridge Eμ to E_VH1 (Fig. 7a). We infer that E_VH1 may integrate contacts with Eμ, IGCR1 and the distal E_VHs.

To independently confirm that Eμ interactions shift from E_VH1 to E_VH2 upon E_VH1 deletion we performed FISH and examined inter-probe distance distributions using short probes for Eμ (green), E_VH1 (blue), and E_VH2 (red) in Rag2^−/− and Rag2^−/−E_VH1^−/− pro-B cells and segregated all nine possible FISH probe configurations (Fig. 7b, c). Indeed, the median spatial distances for pure pairwise Eμ:E_VH2 contacts were significantly increased, and both E_VH1:E_VH2 and Eμ:E_VH1 probe distances were modestly decreased in Rag2^−/−E_VH1^−/− pro-B cells, consistent with the virtual 4C profile (Fig. 7d).

We explored the proposition that E_VH1 participates in a multi-enhancer network by examining three-way probe overlaps (≤ 0.3 μM) that identify Eμ-E_VH1-E_VH2 molecular contacts (Fig. 7d, inset a-d). Two types of linear arrays (Eμ-E_VH1-E_VH2; Eμ-E_VH2-E_VH1) were significantly diminished in Rag2^−/−E_VH1^−/− pro-B cells indicating that these configurations are dependent upon E_VH1 (Fig. 7d, insets b, c). The E_VH2-Eμ-E_VH1 configuration occurs at very low frequencies and was not further considered (Fig. 7d, insets d). Although the frequency of superimposed Eμ-E_VH1-E_VH2 probe overlaps increased, the internal median spatial distances between probes were perturbed in Rag2^−/−E_VH1^−/− pro-B cells suggesting a deformed topology (Fig. 7d, inset a, 7e–g). These studies confirm that in the absence of E_VH1, 1) E_VH2 becomes the preferred Eμ interaction partner and 2) the enhancer hub frequency is decreased or its internal conformation is altered.

E_VH1 is a regional regulator of Igh gene expression

To assess E_VH1 enhancer function we evaluated transcriptional activity for a set of Igh index genes in Rag2^−/− and Rag2^−/−E_VH1^−/− pro-B cells. C_H germline transcripts (GLTs) (IμCμ, Iγ2bCγ2b), V_H GLTs (V_H14-2, V_HJ558), antisense (AS) transcripts from intergenic regions (DQ52, 81X, 7183, J606, J558), and the PAIR4 lncRNA were analyzed in RT-PCR assays (Fig. 8a)^40,50,51. As in the 445.11 E_VH1 KO lines, the V_H14-2 GLT was significantly reduced in Rag2^−/− E_VH1^−/− pro-B cells (Fig. 8b). Likewise, intergenic AS transcripts DQ52, V_H81X, and 7183 were significantly diminished in Rag2^−/−E_VH1^−/− pro-B cells indicating E_VH1 dependency (Fig. 8b). In contrast, the expression of other genes was unchanged (Iγ2bCγ2b GLT, PAIR 4 lncRNA) or elevated (sense V_HJ558, intergenic AS J606, J558) in Rag2^−/−E_VH1^−/− pro-B cells (Fig. 8b, c). The Iγ2bCγ2b GLT and PAIR 4 lncRNA are located downstream of Eμ and upstream of E_VH3 are apparently outside the E_VH1 ZOI (Fig. 8b). Elevated expression of sense spliced and unspliced V_HJ558 transcripts and AS intergenic J606 and J558 transcripts were detected relative to the 18 S RNA loading control in Rag2^−/−E_VH1^−/− pro-B cells indicating dysregulated transcription (Fig. 8b, c). Reduced expression of GLTs in the E_VH1 ZOI correlates with the decreased V_H to DJ_H rearrangement. The upregulated transcripts are all located upstream of the E_VH1 ZOI and in a region of the locus which experiences hyper-contraction and increased proximity with Eμ providing a plausible explanation for increased transcription (Fig. 6e, f). However, because the number of genes tested for expression is relatively small the correlation of transcription regulation with the E_VH1 ZOI is an inference.

Fig. 8. E_VH1 regulates Igh gene expression.

a Igh genes with primers for sense (green arrows) and intergenic antisense (black arrows) transcripts. b, c Source data are provided as a Source Data file. P values from unpaired two tailed Student’s t test. b QRT-PCR analyses from Rag2^−/−- and Rag2^−/−E_VH1^−/−- pro-B cells. RNA samples were isolated from each mouse and two cDNAs were synthesized from each RNA, tested in duplicate or triplicate and averaged. +/-SEMs are shown. RNA samples for Rag2^−/−: primers IμCμ, PAIR4, 7183 (n = 7); 81X, VH14-2, DQ52 (n = 6); J606, J558 (n = 5); Iγ2bCγ2b (n = 4) mice. Rag2^−/−E_VH1^−/−: Primers Iγ2bCγ2b, VH14-2, DQ52, J558 (n = 6); IμCμ, PAIR4, 81X, 7183, J606 (n = 4) mice. c Representative semi-quantitative RT-PCR for sense V_HJ558 (unspliced (US), 465 bp; spliced (S) 382 bp) (upper panel) using an 18 S RNA loading control (lower left panel). No reverse transcription (-RT). V_HJ558 PCR products were quantitated by densitometry. The values for Rag2^−/− samples were normalized to 100. +/-SEMs. Samples from Rag2^−/− and Rag2^−/−E_VH1^−/− (n = 3) mice. d Comparison of the distribution of normalized rearranged V_H genes (upper panel, VDJ-seq analysis; Fig. 4b) with normalized virtual  4C reads for E_VH1 (middle) and Eμ (bottom) viewpoints (Fig. 6a) shown as the Rag1^−/−E_VH1^−/−/ Rag1^−/− log₂ fold change (FC). Eighty-seven rearranged V_H genes mapped into 82 bins of 10 kb. E_VH1 ZOI (dashed red box). e Cumulative frequency of virtual 4C contacts in the ZOI and distal region is shown in box plots indicating the mean (middle line), 25th and 75th percentile (box) and 10th and 90th percentile (whiskers), and outliers (single points). P value from unpaired two-tailed Mann Whitney U test.

E_VH1 restricts chromatin contacts in the ZOI

To determine whether E_VH1 anchored chromatin contacts influence V_H gene usage in the ZOI, we divided the locus into 10 kb bins and mapped all 87 rearranging V_H genes identified in the VDJ-seq analysis into these bins. Next, virtual 4C contacts using the E_VH1 and Eμ baits were mapped and those contacts that fall within these same 10 kb bins were identified. Comparison of rearranged V_Hs (top panel) to virtual 4C contact profiles revealed that the cumulative frequency of E_VH1 anchored contacts was significantly elevated in the ZOI (dashed red box) and randomly perturbed outside of this region in Rag1^−/−E_VH1^−/− pro-B cells (Fig. 8d, e). In contrast, Eμ anchored interactions displayed no discernable regional contact pattern in Rag1^−/−E_VH1^−/− pro-B cells (Fig. 8d, e). Thus, in the absence of E_VH1, increased chromatin interactions are inversely correlated with reduced V_H gene usage within the ZOI. We propose that E_VH1 underpins an architectural structure that promotes ZOI V_H gene usage in part by coordinating and constraining long-range chromatin interactions.

Discussion

We report a major role for E_VHs and TEs in configuring Igh locus topology and in V(D)J recombination. E_VH1, the V_H14-2 Pr, and F.6 CBE are regulators of V- > DJ recombination potential in Abl-t pro-B lines. Deletion of E_VH1 in mice led to regionally reduced V_H gene usage in over a third of the V_H domain, and the V_H genes closer to E_VH1 were more significantly impacted. Likewise, expression of germline transcripts from several V_H genes in Abl-t E_VH1 KO lines and a set of Igh index genes within the ZOI on Rag2^−/−E_VH1^−/− pro-B cells were downregulated demonstrating a concordance of regional transcription regulation with V(D)J recombinational potential. E_VH1 is a modulator of locus conformation in both Abl-t pro-B lines and in mice. Deletion of E_VH1 led to locus decompaction of subTAD A in Abl-t pro-B lines and perturbed locus architecture in the E_VH1 ZOI in mice, identifying a link between D_H proximal V-DJ recombination, Igh gene transcription, and locus conformation. Deletion of E_VH1 also led to locus-wide hyper-compaction in both Abl-t pro-B lines and in mice.

There are several possible explanations to account for the observations of locus de-contraction around E_VH1 despite hypercontraction overall. Our Hi-C data sets indicate that chromatin folds within subTAD C and involving E_VH2-E_VH4 may form independently of subTAD A-B that is structured by IGCR1- E_VH1 or IGCR1-E_VH1-E_VH2 interactions with E_VH2 as the locus nexus. This view might imply that upon E_VH1 deletion the Igh locus should become decompacted overall since E_VH1 anchored chromatin contacts with E_VH2 are lost. However, there are two factors that may mitigate this outcome. First, we note compensatory chromatin contacts that occur in E_VH1 deleted loci such as new Eμ-E_VH2 contacts in Rag1^−/−E_VH1^−/− pro-B cells that might change locus conformation but would still contribute to overall locus compaction. Second, cohesin-mediated loop extrusion is also an important driver of chromatin loop formation^19,22,44 and enhancers can block the extrusion progress¹⁹. E_VH1 loss could lead to more processive loop extrusion that in turn generates greater locus compaction. A direct assessment of loop extrusion frequency in the Igh locus will be required to directly test this notion.

Our discovery of at least four E_VHs located among the intermediate and distal V_Hs was based on their epigenetic profile. Many enhancers regulate transcription with their cognate Prs via spatial proximity created by long-range interactions^52,53, a feature displayed by E_VH1. However, Igh genes may be regulated by more than one enhancer as intergenic antisense J558 transcript expression is Eμ independent⁵⁴ and we propose that this role may be filled by one or several E_VHs. Accordingly, V_H expression at the D_H distal end of the locus was upregulated upon E_VH1 deletion suggesting the presence of another regional enhancer such as E_VH2, 3 or 4. Detection of an E_VH interactome centered on E_VH1 was based on interactions with Eμ in the RC and with E_VH2-4 in the more distal regions of the locus and was confirmed in single-cell imaging studies. Competition between Eμ and the E_VHs is implied by the preferential formation of Eμ-E_VH2 contacts in Rag2^−/−E_VH1^−/− pro-B cells. Notably, elevated (>2 fold) usage of two V_H genes within the E_VH1 ZOI in E_VH1^−/− pro-B cells implies that rearrangement of some V_H genes within the same loop domain is differentially regulated by as yet undefined mechanisms.

Our studies provide evidence that architectural “stripes”, formed by cohesin-mediated loop extrusion⁴⁴, initiate from both the 3’ and 5’ ends of the Igh locus in pro-B cells. Stripes form when one subunit of cohesin stalls near a strong CTCF loop anchor and the second subunit progressively extrudes the chromatin loop to form a multiplicity of contacts and that ultimately brings the far end of the locus into spatial proximity with the stall site and forms a loop domain⁴⁴. These loop domains could provide a platform to “reel-in” V_Hs toward the RC from various positions across the locus. E_VH1 is a stripe origin that extends in the 3’ direction and ultimately intersects with IGCR1, creating a loop domain encompassing the proximal and intermediate V_Hs and promoting spatial proximity to the RC. A RAG scanning event initiating within the IGCR1-E_VH1 loop domain would create V_H gene proximity with the RC. Loss of the IGCR1- E_VH1 chromatin loop might preclude efficient RAG scanning through this region of the locus.

E_VH1 affects regional transcription and V_H to DJ_H rearrangement as well as overall locus conformation and loop extrusion. We favor the view that the participation of E_VH1 in structuring the Igh locus through chromatin looping and loop extrusion is a key to determining its regional influence on V-DJ rearrangements. However, it is difficult to exclusively assign any one aspect of E_VH1 function to its influence on V-DJ recombination, as some features, such as transcription may be necessary but not sufficient.

Recent studies identified the E88 enhancer as a structural organizer of the Igk locus, and, like E_VH1, its deletion negatively impacts rearrangement of nearby V_κ genes⁵⁵. However, E88 is distinct from E_VH1 as it affected transcription only within the region in which V_κ gene rearrangement is impacted⁵⁵, whereas E_VH1 influences GLT expression throughout the V_H locus and is a component of an enhancer hub. This difference may also be related to the structural roles E_VH1 plays within the Igh locus as an architectural stripe origin and a modulator of locus compaction.

Deletion of E_VH1 results in a reduction of V_H11-2 rearrangement, the V gene most predominantly observed in B1a cells from C57Bl/6 mice⁴¹. These V_H-D-J rearrangements, with similar CDR3 sequences, are greatly expanded in B1 cells presumably by their cognate antigen PtC, and accordingly, the E_VH1 deletion results in a decrease in splenic B1a cells which bind PtC. Why do we therefore not see a decrease in B1a B cells from the PerC? One explanation could be that E_VH1 does not have as much influence in fetal vs adult life, and most B1 cells in the PerC arise from fetal precursors⁵⁶. Alternatively, strong pressure for expansion of PtC-binding B1 cells in the PerC might lead to expansion of the residual V_H11-2 precursors with the relevant CDR3 sequence. In contrast, the splenic B1a repertoire is more diverse and thus presumably has less selection pressure for V_H11.2⁺ B1a cells. Regardless, these data show a biological effect of E_VH1 deletion on a protective common antibody response and that perturbation of the preselected IgH repertoire can be reflected in the periphery.

Methods

Mice, cell lines and cell culture

C57BL/6 (WT) or Rag deficient mice on the C57BL/6 background were purchased from Jackson Laboratories and E_VH1^−/− mice were constructed using genomic editing and then maintained in colonies at the University of Illinois College of Medicine, or Scripps Research. All procedures involving mice were approved by the Institutional Animal Care Committee of the University of Illinois College of Medicine, and the Scripps Research Institute, in accordance with protocols approved by the UIC and Scripps Research Institute Institutional Animal Care and Use Committees. Mice were housed in sterile static microisolator cages on autoclaved corncob bedding with water bottles. Food was irradiated (Envigo 7912), water was autoclaved and both were provided ad libitum. The standard photoperiod is 14 hours of light and 10 hours of darkness for mouse rooms. Mice receive autoclaved nesting material to enrich their environments. Cage bedding is changed in either a biosafety cabinet or a HEPA-filtered animal transfer station at least weekly. Housing density and cage size are consistent with the recommendations of the Guide for the Care and Use of Laboratory Animals. The ambient temperature and humidity of the rodent housing rooms are consistent with the recommendations of the Guide for the Care and Use of Laboratory Animals.

Bone marrow (BM) was collected from humerus, tibia and femur bones of WT and E_VH1^−/− mice. Rag deficient CD19 + pro-B cells were isolated from BM using anti-CD19 coupled magnetic beads (Miltenyi) and cultured in the presence of IL7 (1% vol/vol supernatant of a J558L cell line stably expressing IL7) for 4 days. The Abelson-MuLV transformed (Abl-t) pro-B cell line, 445.3 (Rag1−/−) on the C57Bl/6 background was kindly provided by Dr. B. Sleckman (University of Alabama at Birmingham)⁵⁷. A newly derived subclone, 445.3.11 from the Abl-t 445.3 line were cultured in RPMI 1640 (Cellgro), 10% (v/v) FBS, 4 mM glutamine (Gibco), 1 mM sodium pyruvate (Gibco), 1X nonessential amino acid (Gibco), 5000 units/ml Penicillin and 5000 mg/ml Streptomycin (Gibco), 50 mM β-mercaptoethanol (Sigma) and maintained at approximately 5x10e5 cells/ml. Splenic T cells were enriched using Mouse T Cell Enrichment Columns (MTCC-5; R&D Systems) and cultured at a density of 5 × 10⁵ to 1 × 10⁶, stimulated in RPMI 1640 and glutamine (4 mM) and penicillin-streptomycin supplemented with FCS (10% v/v), and activated with Con A (5 ng/ml; 15324505; MP Biomedicals).

Flow cytometry and FACS

FACS isolation of bone marrow derived pro-B cells: Total bone marrow (BM) was collected from humerus, tibia and femur bones from groups of at least 3–10 mice. CD19⁺ cells were isolated from BM using CD19-conjugated MACS beads and were preincubated with Fc Block for 5 minutes and then stained with specific antibodies. Isolation of pro-B cells for VDJ-seq: Mice: WT (n = 3), E_VH1^−/− (n = 2). CD19⁺ cells were stained with stained with Abs to B220-APC-Cy7, CD19-FITC, CD93-PE-Cy7, CD2-PE, CD43-BV421 and IgM Fab-AF647 and were sorted on a BD FACSAria II using BD FACS Diva 6.1.3 software. Sorted pro-B cells (CD19⁺CD93⁺IgM^-CD2^-CD43⁺) were used to isolate gDNA. Antibodies were used at 1 μl/1 × 10⁶ cells in 1 ml up to 30 × 10⁶ cells except for IgM (0.5 μl/1 × 10⁶ cells). Isolation of pro-B cells for qPCR and repertoire analysis: CD19⁺ cells were stained with antibodies to CD19-PerCP (dil 1/40), B220-APC-Cy7 (dil 3/100), CD93-PE-Cy7 (AA4.1; dil 1/50), CD2-PE (dil 1/50), CD43-APC (S7; dil 1/100) and IgM-e450 (dil 1/50) and were sorted on a BD FACSAria Fusion 5-18 using BD FACS Diva 8.0.1 software. FACS purified pro-B cells (CD19⁺B220⁺CD93⁺ IgM^-CD2^-CD43⁺) were used to isolate gDNA for qPCR and repertoire analysis shown in Fig. 4f, h. Flow cytometry analyses were performed using live pro-B cells (5 × 10⁻⁵) washed in PBS plus 2% FCS, and stained with antibodies (CD19-PerCp (dil 1/40), B220-APC Cy7 (dil 3/100), CD93 PE-Cy7 (dil 1/50), CD43-APC (dil 1/100), CD2-PE (dil 1/2500), IgM-e450 (dil 1/100)) by gating for Fixable Viability Stain 510 (FVS510; dil 1/5000) (BD Biosciences) and/or forward- and side scatter on an Attune NxT Acoustic Focusing Cytometer lasers B, R, V, Y using NxT software version 2.7.0 (ThermoFisher). Assessment of splenic marginal zone (MZ) and follicular (Fo) B cells from mice was performed by flow cytometry. Splenocytes were isolated and the live cells were enumerated by using the viability stain FVS-510 (dil 1/5000), and the antibodies B220-APC Cy7 (dil 3/100), CD21-FITC (dil 1/100), CD23-biotin (dil 1/50) and streptavidin-APC (dil 1/500). Live splenic B1 B cells were analyzed by treating with viability stain FVS-510 (dil 1/5000), and with antibodies to CD19-PerCp (dil 1/40), B220-APC Cy7 (dil 3/100), CD23-FITC (dil 1/50), anti-IgM-e450 (dil 1/100), IgD AF700 (dil 1/50) or with CD19-PerCp (dil 1/40), B220-APC Cy7 (dil 3/100), CD43-APC (dil 1/100), IgM-e450 (dil 1/100), IgD AF700 (dil 1/50) and PtC liposomes (FITC) (dil 1/2000) and analyzed on a Attune NxT Acoustic Focusing Cytometer with gating as described^58,59 with FlowJo 10.8.1 or Summit 4.3 software. All antibodies and reagents are listed in (Supplementary Table 1).

Quantitative RT-PCR and ChIP

RNA extraction was performed from 2–3 × 10⁶ cells using TRIzol (Life Technologies) according to manufacturer instructions. RNA samples were further treated with the DNase I kit (Invitrogen) to remove contaminating DNA per the manufacturer’s instruction. First-strand cDNA synthesis was done with RNA (1–4 µg) and the SuperScript III Reverse Transcriptase kit (Invitrogen) according to the manufacturer’s instructions. Quantitative (q) real-time RT-PCR was performed using primers (Supplementary Table 2)⁴⁰ and Fast SYBR Green PCR Mix (Applied Biosystems) and Real-Time PCR System viiA7 (Applied Biosystems) as described⁶⁰ except that primers for 18 S rRNA were used⁶¹ to normalize samples. Semi-quantitative RT-PCR assays for V_HJ558 transcripts were carried out using Platinum Taq DNA polymerase (Invitrogen) (25 μl) at 94 °C, 30 s for 1x; 94 °C 30 s, 60 °C 30 s, 72 °C 30 s, for 29x, 31x and 33x and 7 µl were analyzed by gel electrophoresis. The 18 S loading control was assessed by qRT-PCR. Semi-quantitative RT-PCR assays were carried out for the V_H81X and V_H2-5 genes using the same conditions as for V_HJ558 except that the PCR products were harvested at 32 cycles and for 18 S rRNA at 28 cycles. All results represent the average of at least three independent experiments from 3–5 mice for each genotype or 2 biallelic identical 445.3.11 KO subclones. Each sample was assayed in triplicate and SEMs were calculated. ChIP assays for CTCF binding were performed with anti-CTCF antisera as described previously (Supplementary Table 1)⁶².

CRISPR-Cas9 mediated genomic editing of cell lines and mice

The optimal Guide (g)RNA sequence closest to the genomic targets were identified using a combination of the http://crispr.mit.edu design tool and BLAST. GRNAs were cloned into the pX330 vector at the BbsI site (Supplementary Table 3)⁶³. GRNA efficiency was tested in HEK293T cells using previously described protocol⁶⁴. The Abl-t 445.3.11 cell line was co-transfected with gRNAs plasmid constructs (1 μg/each) together with a pmax GFP plasmid (1 μg) (Lonza) using the Amaxa Cell Line Nucleofector Kit (Lonza) (Y001 program) and a Nucleofector II (model AAD-1001N) according to the manufacturer’s instructions. Cells were allowed to recover for 48 h and then GFP⁺ cells were purified by FACS (MoFlo Astrios) supported by Summit software (version 4.3) (Beckman Coulter, Indianapolis, IN). Cells were submitted to limiting dilution, subclones were expanded for 10–14 days and gDNA was harvested using the DNA spooling method. Subclones were screened for insertion/deletions (indels) by PCR using Platinum Taq DNA polymerase (Invitrogen) and target specific primers at 94  °C, 30 s for 1 cycle; 94 °C 30 s, 58–66 °C 30 s, 72 ^oC 30–60 s for 35 cycles; 72 ^oC 5 min for 1 cycle (Supplementary Table 4). PCR products were examined by Sanger sequencing to authenticate CRISPR/Cas9 indels.

To construct E_VH1^−/− mice, Cas9 mRNA and sgRNAs E_VH1 g1 and E_VH1 g2 were injected into pronuclei of mouse zygotes at the Scripps Research Institute Mouse Genetics Core facility (Supplementary Table 3)⁶⁴. The E_VH1^−/− mouse contains a 515 bp deletion (chr12:114182511-114183025, mm10) spanning E_VH1.

V(D)J recombination assays in Abl-t pro-B lines

Abl-t 445.3.11 Rag1^−/− cells were transfected with pMSCV-IRES-Bsr-Rag1 plasmid (a gift from Dr. D. Schatz, Yale University) using an Amaxa Cell Line Nucleofector Kit (Lonza) (Y001 program) and a Nucleofector II (Lonza, model AAD-1001N) according to the manufacturer’s instructions. Stable transfectants were selected in 20 µg/ml blasticidin for 10–14 days and subsequently maintained in 10 µg/ml blasticidin. Cells (1 × 10⁷) were treated with STI-571 (2.5 µM) for 48 hours and gDNA and RNA were isolated. RNA was prepared from cells (9 × 10⁶) using TRIzol (Life Technologies) following the manufacturer’s instructions and gDNA was prepared from cells (1 × 10⁶) by spooling. D_H- > J_H rearrangement was assayed using gDNA (50 ng) with DFL16.1 F and JH1R primers and the Fast SYBR green master mix (Applied Biosystems) with qPCR of the Mb1 gene as the loading control. V- > DJ recombination was assayed using cDNA (2 µl) in Fast SYBR Green PCR Mix (10 µl reaction volume) in a viiA7 system (Applied Biosystems) with 18 S RNA as a loading control. Primers are listed in Supplementary Tables 2 and 5.

DNA FISH and nuclear volume analyses

DNA probes for fluorescent in situ hybridization (FISH) were prepared from locus-specific BACs or short probes. All genomic coordinates are chr12 (mm10). The BACs probes were RI (113813346-113989276) (BAC 373N4), RII (114705813-114886430) (BAC 70F21), RIII (115539177-115773011) (BAC 368C22 or 230L2)¹⁷ and H14 (112907511 – 113142759) located ∼44 kb 3′ of HS7 of the 3′RR (BAC RP23-201H14)⁶⁵. Short probes (4.2–7.5 kb) were 3’Eα (113221845-113226083), Eµ (113425893–113430719), Site I.3 (113987323–113994353), E_VH1 (114184075–114191532) and E_VH2 (114600831–114606206) were generated by PCR using gDNA with primers listed in (Supplementary Table 6). Probes were labeled by nick translation in the presence of Alexa Fluor 555 (red) and 488 (green) and Alexa Fluor 647 (blue). Labeled probes were hybridized with fixed cells and FISH was performed using a Nikon W1 dual CAM spinning disk confocal microscope (Center for Advanced Microscopy and Nikon Imaging Center, Northwestern University). Serial optical sections (n = 30–40; each slice is 0.1 mm thick) spaced by 0.2 mm were acquired. The data sets were deconvoluted using Imaris (version 9.0) software and optical sections were merged to produce 3D images. Spatial distances between probes were measured as previously described^17,27. Purified pro-B cells from at least two mice or two slides from cell lines of different genotypes were investigated. P values of statistical significance were calculated using Mann–Whitney U-test, Wilcoxon rank sum test and Chi square test (Supplementary Table 7).

Nuclear volumes of the IL7 expanded Rag2^−/− pro B cells were analyzed by staining with Cell Tracker Red CMTPX fluorescent dye (Life Technologies, C34552) and DAPI and then fixed with para-formaldehyde (4%) as previously described²². Cells were imaged using a Nikon W1 dual CAM spinning disk confocal microscope (Center for Advanced Microscopy and Nikon Imaging Center, Northwestern University). Serial optical sections (n = 30–40) spaced by 0.2 μm were acquired. Nuclear volumes were determined using Imaris 9.2.1 software for a total of 120 nuclei and represented as mean ± SEM (210.4 µm³ ± 5.5).

V(D)J-seq

Total BM cells were harvested from long bones of 4–5 mice that were 6 to 10 weeks old. CD19 + cells were isolated using anti-CD19 conjugated MACS beads (Miltenyi, Auburn CA) and then pre-incubated with CD16/32 Fc Block for 5 min and stained with antibodies for CD19-BB515, CD93-PE/Cy7, IgM Fab-AF647, CD2-PE, CD43-BV421 (Supplementary Table 1). Genomic DNA (gDNA) was isolated from purified pro-B cells (DNeasy, Qiagen) and the V(D)J-seq protocol performed while omitting the negative depletion step and using custom J_H primers (Supplementary Table 8)^39,66. GDNA was sonicated to a range of 500 bp to 1000 bp using a Covaris S2 (Covaris). Library barcoding was carried out using NEBNext Multiplex Oligos for Illumina (E7600S). Samples were paired-end 2 × 300 sequenced on an Illumina MiSeq System (San Diego, CA) at the NGS Core, Scripps Research.

Paired end sequencing data was aligned to assembled V, D, and J exons using MiXCR⁶⁷, with parameters: allowChimeras false, minSumScore 120.0, maxHists 5, relativeMinVFR3CDR3Score 0.7. V/D/J segment reference libraries were from MiXCR built in Mus Musculus library. Next, unique CDR3s were assembled using MiXCR assemble, with the parameters: minimalClonalSequenceLength 12, badQualityThreshold 20, maxBadPointsPercent 0.7, qualityAggregationType Max. The frequency of V_H gene usage was then counted for 1) single V_H genes, and for 2) V_H gene families using a customized python script. V_H gene usage data determined using MiXCR software have been deposited in the GEO database (accession number: GSE203484) (Supplementary Table 10).

Statistics

P values were calculated by using two-tailed Student’s t test, two-tailed Chi square test, two-tailed Mann-Whitney U test or two tailed Mann-Whitney Wilcoxon test as indicated. Box plots represent the distribution of spatial distances with medians or means as indicated (middle line), 25th and 75th percentile (box), 10th and 90th percentile (whiskers), and outliers (single points). Box and whisker plots were generated using GraphPad or RStudio ggplot2.

3C library construction and analysis

3C chromatin was prepared from CD19⁺ IL7 expanded Rag2^−/− pro-B cells, the Abl-t 445.3.11 line and ConA activated splenic T cells. Optimized 3C library construction and assays for the Igh locus using Hind III were performed as described^62,68. Briefly, Rag2^−/− pro-B cell, the Abelson transformed (Abl-t) pro-B cell line 445.3.11 and ConA activated T cells were crosslinked using 1% formaldehyde and template concentration using the Mb1 primers was determined⁶⁰. Quantitative PCR (qPCR) in combination with 5’FAM and 3’BHQ1 modified probes (IDT) was used to detect of 3C products and primers were designed using Primer Express software (ABI) (Supplementary Table 9). Primer and probe optimization were carried out according to the manufacturer’s recommendations, (http://www3.appliedbiosystems.com/cms/groups/mcb_support/documents/generaldocuments/cms_042996.pdf). A template in which all possible 3C ligation products are present in equimolar concentration was used to control for differences in amplification efficiency between primer sets (Supplementary Table 9). The data were normalized using the interaction frequency between two fragments within the non-expressed Chr5 gene desert to facilitate sample-to-sample comparisons (Supplementary Table 9). The relative crosslinking frequency between two Igh restriction fragments was calculated: X_Igh = [S_Igh/S_GD] Cell Type/[S_Igh/S_GD] Control mix. S_Igh is the signal obtained using primer pairs for two different Igh restriction fragments and S_GD is the signal obtained with primer pairs for the GD locus fragments. The crosslinking frequency for the GD fragments was set to 1 to allow sample-to-sample comparisons. Data are represented as mean ± SEM. A complete laboratory protocol for 3C is available upon request. For all qPCR 3C reactions, 100 ng of chromatin was used.

Hi-C library construction and analyses

Genome-wide in situ Hi-C libraries were constructed from Rag1^−/−, Rag1^−/−E_VH1^−/− pro-B cells expanded in IL7 for 4–5 days using Arima Hi-C kits (Arima Genomics, San Diego, CA) as recommended by the manufacturer. In situ Hi-C was performed using two biological replicates that yielded a minimum of 1.3 billion read pairs and 0.72 billion from pro-B cells of each genotype (Supplementary Table 11) (GEO Accession No. GSE201357). Published in situ Hi-C data sets for mouse embryonic fibroblasts (MEFs) were constructed with Arima Hi-C kits (GEO Accession No. GSE113339)⁴³ and data was handled in parallel with the pro-B cell data. Hi-C data have been deposited in the GEO database (accession number GSE201357).

In situ Hi-C data was processed using the Juicer pipeline (v.1.5), CPU version⁶⁹. The pipeline was set up with BWA (0.7.15–r1140)⁷⁰ to map each read end separately to the mm10 reference genome (GRCm38). Duplicate and near-duplicate reads, as well as reads that map to the same fragment were removed. Among the remaining reads, those with mapping quality score (MAPQ) < 30 are retained. Arima-HiC-specific restriction sites (mm10_GATC_GANTC.txt) were obtained from ftp://ftp-arimagenomics.sdsc. edu/pub/JUICER CUTSITE FILES. For individual replicates, raw reads from each replicate were mapped separately to the reference genome, and then filtered by the MAPQ score. For merged replicates of each sample type, valid read pairs from both replicates were merged, mapped to the reference genome, and then filtered by the MAPQ score. Hi-C contact matrices were extracted from.hic files using the Dump command provided by a Java-based program in Juicer tools⁶⁹. Hi-C maps were then normalized with the Knight and Ruiz (KR) matrix balancing method⁷¹. Heatmap resolution of all individual replicates was computed using the Juicer script (calculate_map_resolution.sh) and found to be >10 kb. The reproducibility of Hi-C data was computed using the Stratum-adjusted correlation coefficient (SCC) on Chr12 using HiCRep at different resolutions⁷². The replicates were merged and displayed at 10Kb resolution.

Extraction of virtual 4C interaction matrices: The hic files were used to generate virtual 4C viewpoints from dumped matrices generated in Juicebox. KR normalized observed read matrices were extracted at 10 kb resolution. The biological replicates had stratum adjusted correlation coefficient (SCC)⁷² greater than 0.9 and were merged. The interaction profile of virtual 4C were plotted by running a rolling window of 30 kb with a 10 kb slide.

Generation of Hi-C difference maps: Experimentally measured Hi-C contact matrices of individual replicate and merged samples were quantile normalized against the Hi-C contact matrices of the uniformly sampled random ensemble of the corresponding cell type. Normalizing each sample against the target distribution of uniformly sampled random ensemble will remove the between-replicate biases in frequencies due to differences in sequencing depth⁷³. Hi-C difference maps were assembled by quantile normalization of Hi-C contact frequencies using a null distribution generated using previously described methods^74,75. The source code for the null model chromatin folding by fractal Monte Carlo is available via git repository at https://bitbucket.org/aperezrathke/chr-folder and step by step procedure to generate the null distribution is explained https://bitbucket.org/aperezrathke/chr-folder under “Null distribution” subsection. To generate the target distribution of the corresponding cell type, we followed the approach of reference⁷⁴. Briefly, we generated a random ensemble of 300,000 polymer chains that are sampled uniformly from all geometrically feasible polymer chains confined in the nuclear volume. Each polymer chain consists of 2000 spherical monomer beads. Individual monomer beads have a diameter of ~51 nm and represent ~10 kb of DNA, which is the resolution of the Hi-C data. To model the effects of confinement and volume exclusion, we constrained each self-avoiding polymer chain to reside within a spherical nuclear volume that is appropriate for the cell type. For pro-B cells, the nuclear volume is taken as 213 µm³; this was scaled to ~779 nm³ to accommodate a chromatin segment of 20 MB, which preserves a constant base pair density relative to the entire genome. We generated a random ensemble of 300,000 polymer chains of length 20 MB of 2000 monomers, all residing in the scaled volume. For the Igh locus, we considered the region of chr12:113,215,000–chr12:116,045,000 of 3.31 MB length. We took each chromatin polymer chain of the 20 Mb segment and randomly selected 331 consecutive monomeric units to match the size of the Igh locus. The resulting random ensemble for the Igh locus has 300,000 polymer chains, each consisting of 331 spherical monomer beads. For MEF, the nuclear volume is taken as 881 µm^3[ 76 this was scaled to ~534 nm³ for the Igh locus of 3.31 MB. We then directly generated a random ensemble of 300,000 polymer chains of length 3.31 MB of 331 monomers. The contact frequency matrix of randomly sampled chromatin chains was obtained by counting the frequency of interacting monomer pairs. A pair of chromatin monomers are considered as interacting, if their Euclidean distance is ≤80 nm⁷⁷. Contact matrices of each pro-B cell KO sample were quantile normalized against the contact matrices of the uniformly sampled random ensemble of the corresponding cell type. These quantile normalized frequencies of each sample are then compared with each other, from which differential plots are obtained to identify differences between the two biological samples. Specifically, quantile normalized frequencies of MEF are subtracted from quantile normalized frequencies of each pro-B cell KO sample.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Source data

Source data ^{(8.5MB, xlsx)}

Author contributions

A.L.K. and A.J.F. conceptualized the research. K.H.B., S.P., X.Q., E.K., J.X., J.F.C., and R.W. designed and performed experiments. S.N., H.F., X.L., and J.L. performed data analyses and computation., N.B. provided resources. A.L.K., K.H.B., and S.P. wrote the manuscript. A.L.K., A.J.F., S.P., and S.N. edited the manuscript., A.L.K. and A.J.F. supervised the study, and all authors contributed to and approved the manuscript.

Peer review

Peer review information

Nature Communications thanks David Schatz, and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.

Data availability

The data that support this study are available from the corresponding author upon reasonable request. The Hi-C and VDJ-seq data sets generated in this study are available in the NCBI Gene Expression Omnibus (GEO) repository under accession numbers GSE203484, GSE201357. Public ChIP data sets used in this study were: GSM1635413, GSM1635411, GSM546523, GSM1038263, GSM2255552, GSM546527, GSM539537, GSM987808, GSM4350110, GSM4805384, GSM1296534, GSM1156665, GSM1156667, GSM1006370 and the mouse reference genomes: mm10 (GRCm38) and mm9 (GSCv37). DNA FISH distance measurements, qPCR, and qRT-PCR are provided in the Supplementary/Source Data file. All raw FACS data is available upon request to the corresponding author and will be provided within four weeks from the request data. Source data are provided with this paper.

Code availability

Source code for the null model chromatin folding by fractal Monte Carlo is available via the git repository at https://bitbucket.org/aperezrathke/chr-folder and step by step procedure to generate the null distribution is explained https://bitbucket.org/aperezrathke/chr-folder under “Null distribution” subsection.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-023-36414-2.