CTCF-Binding Elements Mediate Accessibility of RAG Substrates During Chromatin Scanning

Abstract

RAG endonuclease initiates antibody heavy chain variable region exon assembly from V, D, and J segments within a chromosomal V(D)J recombination center (RC) by cleaving between paired gene segments and flanking recombination signal sequences (RSSs). The IGCR1 control region promotes DJ_H intermediate formation by isolating Ds, J_Hs, and RCs from upstream V_Hs in a chromatin loop anchored by CTCF-binding elements (CBEs). How V_Hs access the DJ_HRC for V_H to DJ_H rearrangement was unknown. We report that CBEs immediately downstream of frequently rearranged V_H-RSSs increase recombination potential of their associated V_H far beyond that provided by RSSs alone. This CBE activity becomes particularly striking upon IGCR1 inactivation, which allows RAG, likely via loop extrusion, to linearly scan chromatin far upstream. V_H-associated CBEs stabilize interactions of D-proximal V_Hs first encountered by the DJ_HRC during linear RAG scanning and thereby promote dominant rearrangement of these V_Hs by an unanticipated chromatin accessibility-enhancing CBE function.

Keywords: 3C-HTGTS; CBE orientation; CTCF-binding elements; HTGTS-Rep-seq; RAG chromatin scanning; V(D)J recombination; V(H) accessibility; V(H)81X; intergenic control region 1; loop extrusion.

Figure 1. V_H81X-CBE Greatly Enhances V_H81X Utilization in Primary Pro-B Cells

(A) Schematic of the murine Igh locus showing proximal V_Hs, Ds, J_Hs, C_H exons, and regulatory elements (not to scale). Red and blue bars represent members of the IGHV5 (V_H7183) and IGHV2 (V_HQ52) families, respectively. Teal blue triangles represent position and orientation of CTCF-binding elements (CBEs). Green arrow denotes position of the J_H4 coding end bait primer used to generate HTGTS-Rep-seq libraries. (B) Sequence of V_H81X-RSS (green) followed by WT (red) or scrambled (blue) V_H81X-CBE. (C) Relative V_H utilization ± SD in BM pro-B cells from WT (top) or V_H81X-CBE^scr/scr (bottom) mice. (D) Average utilization frequencies (left axis) and % usage (right axis) of indicated proximal V_H segments ± SD. For analysis, each library was normalized to 10,000 VDJ_H junctions. p values were calculated using unpaired, two-tailed Student’s t test, ns indicates p > 0.05, *p ≤ 0.05, **p ≤ 0.01 and ***p ≤ 0.001. See also Figures S1 and S2 and Tables S1, S3, and S4.

Figure 2. V_H81X-CBE Enhances V_H81X Utilization in DJ_H Rearranged v-Abl Pro-B Lines

(A) Schematic representation of the two murine Igh alleles in DJ_H rearranged v-Abl pro-B cell line (not to scale). One allele (top) harbors a non-productive VDJ_H rearrangement involving a distal V_HJ558 (V_H1-2P) that deletes the proximal V_H domain and is inert for V(D)J recombination. The other allele (bottom) harbors a D_HFL16.1 to J_H4 rearrangement (DJ_H allele) that actively undergoes V_H to DJ_H recombination upon RAG induction via G1 arrest. This D_HFL16.1J_H4 line served as the parent WT line and was used for all subsequent genetic manipulations. (B) Top line shows the sequence of WT V_H81X-CBE (red) while the bottom line shows V_H81X-CBE deletion (blue dashed line). (C) Average utilization frequencies (left axis) or % usage (right axis) ± SD of indicated proximal V_Hs in WT and V_H81X-CBE^del v-Abl pro-B lines; libraries were normalized to 3,500 VDJ_H junctions. As the WT line used for this experiment was the parent of all subsequent V_H-CBE mutant lines, we generated WT repeats at several points over the course of these experiments and used the average data, which were highly reproducible, for this and subsequent panels showing comparisons of mutants with WT controls (see STAR Methods for details). (D) Schematic of the 101-kb intergenic deletion extending from 302 bp downstream of V_H81X-CBE to 400 bp upstream of the D_HFL16.1J_H4 RC in the WT D_HFL16.1J_H4 v-Abl line and its V_H81X-CBE^del derivative. (E) Average utilization frequencies (left axis) or % usage (right axis) ± SD of indicated proximal V_Hs in Intergenic^del and Intergenic^del V_H81X-CBE^del v-Abl lines; libraries were normalized to 100,000 VDJ_H junctions. (F) Sequence of WT (red) and V_H81X-CBE inversion mutation (blue). (G) Average utilization frequencies (left axis) or % usage (right axis) ± SD of the indicated proximal V_Hs in D_HFL16.1J_H4 WT and V_H81X-CBE^inv v-Abl lines; libraries were normalized to 3,500 VDJ_H junctions. Statistical analyses were performed as in Figure 1. See also Figure S3 and Tables S1, S2, S3, and S4.

Figure 3. V_H81X-CBE Promotes Interactions of Its Flanking V_H with the DJ_HRC

(A) Schematic representation of the 3C-HTGTS method for studying chromosomal looping interactions of a bait region of interest with the rest of Igh locus (see text and STAR Methods for details). (B) Schematic of the NlaIII restriction fragment (indicated by a blue asterisk) and the relative positions of the biotinylated (cayenne arrow) and nested (blue arrow) PCR primers used for 3C-HTGTS from V_H81X bait in (C). (C) Top panel: schematic representation of chromosome interactions of V_H81X-CBE containing NlaIII fragment with other Igh locales. Bottom two panels: 3C-HTGTS profiles of Rag2^−/− derivatives of control, V_H81X-CBE^del, and V_H81X-CBE^inv D_HFL16.1J_H4 v-Abl lines using V_H81X-CBE locale as bait (blue asterisk). Owing to a D_HFL16.1 to J_H4 rearrangement in the lines, the region spanning IGCR1, DJ_H substrate and iEm appears as a broad interaction peak. As v-Abl lines lack locus contraction, we detected few substantial interactions with the upstream Igh locus beyond the most proximal V_Hs (see legend of Figure S3). Two independent datasets are shown from libraries normalized to 105,638 total junctions. See also Table S4.

Figure 4. V(D)J Recombination of V_H2-2 Is Critically Dependent on Its Flanking CBE

(A) Sequence of WT V_H2-2-CBE (red) and its scrambled mutation (blue). (B) Average utilization frequencies (left axis) or % usage (right axis) ± SD of indicated proximal V_Hs in WT and V_H2-2-CBE^scr v-Abl lines. Each library was normalized to 3,500 VDJ_H junctions. Statistical analyses were performed as in Figure 1. See also Figure S4A and Table S1. (C) Illustration of NlaIII restriction fragment (blue asterisk) and relative positions of biotinylated (cayenne arrow) and nested (blue arrow) primers used for 3C-HTGTS analyses in (D). Due to repetitive sequences in the restriction fragment that harbors V_H2-2-CBE, the downstream flanking restriction fragment was used as bait. (D) Representative 3C-HTGTS interaction profiles of V_H2-2 locale (blue asterisk) in Rag2^−/− control and V_H2-2-CBE^scr v-Abl lines, plotted from libraries normalized to 84,578 total junctions. See Figure S4B for an independent repeat. See also Tables S3 and S4.

Figure 5. V_H81X-CBE Is Required for Dominant V_H81X Usage in the Absence of IGCR1

(A) Schematic of 4.1 kb IGCR1 deletion. (B) Average utilization frequencies (left axis) or % usage (right axis) ± SD of proximal V_Hs in IGCR1^del and IGCR1^del V_H81X-CBE^del v-Abl lines. Each library was normalized to 100,000 VDJ_H junctions. Statistical analyses were performed as in Figure 1. See also Figures S5A and S5B and Tables S1 and S2. (C) Representative 3C-HTGTS interaction profiles of V_H81X bait (blue asterisk) in Rag2^−/− control, IGCR1^del, and IGCR1^del V_H81X-CBE^del D_HFL16.1J_H4 v-Abl lines performed using the strategy shown in Figure 3B, plotted from libraries normalized to 106,700 total junctions. Bottom panel shows a zoom-in of the region extending from upstream of IGCR1 to downstream of Cd exons. Pink rectangles marked with “D” indicate the IGCR1 region deleted in the IGCR1^del and IGCR1^del V_H81X-CBE^del lines. See Figure S5C for an additional repeat. (D) Representative 3C-HTGTS interaction profiles of iEm bait (blue asterisk) in Rag2^−/− v-Abl DJ_H lines of the indicated genotypes following NlaIII digest using the strategy shown in Figure S5D. Each library was normalized to 273,547 total junctions. Bottom: zoom-in of the proximal V_H region. See Figure S5D for an independent repeat and Figure S6 for related repeats. See also Figure S7 and Tables S3 and S4.

Figure 6. Restoration of a CBE Converts V_H5-1 into the Most Highly Rearranging V_H

(A) Schematic showing the sequence of V_H5-1-RSS and its downstream non-functional, “vestigial” CBE. The yellow shaded box highlights the CpG island that is methylated in normal pro-B cells. Bottom sequence shows the four nucleotides mutated (highlighted in blue) to eliminate the CpG island and restore consensus CBE sequence. Two additional nucleotides were mutated just downstream of the CBE to generate a BglII site for screening. (B) Average utilization frequencies (left axis) or % usage (right axis) ± SD of the indicated proximal V_Hs in WT and V_H5-1-CBE^ins v-Abl lines. Each library was normalized to 3,500 VDJ_H junctions. Statistical analyses were performed as in Figure 1. See also Figure S4C and Table S1. (C) Illustration of the MseI restriction fragment (blue asterisk) and the relative positions of biotinylated (cayenne arrow) and nested (blue arrow) primers used for 3C-HTGTS analyses in (D). (D) Representative 3C-HTGTS interaction profiles of the V_H5-1 locale (blue asterisk) in Rag2^−/− control and V_H5-1-CBE^ins v-Abl lines, plotted from libraries normalized to 37,856 total junctions. See Figure S4D for an independent repeat. See also Figure S7 and Tables S3 and S4.

Figure 7. Model for RAG Chromatin Scanning via Loop Extrusion

Figure shows a working model for potential roles of V_H-associated CBEs during RAG scanning over chromatin. Numerous variations of the model are conceivable. (A) From its location in the initiating RC, RAG linearly scans cohesin-mediated extrusion loops proceeding through Ds, to allow their utilization; but is largely impeded further upstream by the IGCR1 anchor. After formation of a DJ_HRC, residual lower level scanning of upstream sequences beyond the IGCR1 impediment allows the most proximal V_H-CBEs to mediate direct association with the DJ_HRC enhancing utilization of their associated V_H. V_Hs further upstream likely access the DJ_HRC by diffusion with proximal CBEs also enhancing DJ_HRC interactions and flanking V_H utilization. (B) In the absence of IGCR1, loop extrusion progresses upstream allowing RAG to scan the most proximal V_Hs where associated CBEs promote DJ_HRC interaction, accessibility, and dominant over-utilization in V(D)J joins. Utilization is most robust for proximal V_H81X, which provides the first V_H-CBE encountered during linear scanning. V_H5-1 is bypassed due to lack of a CBE. Scanning can sometimes bypass V_H81X-CBE and continues to the first few upstream V_Hs, with their CBEs similarly promoting utilization. (C) If both IGCR1 and the V_H81X-CBE are mutated, loop-extrusion continues unabated to the V_H2-2-CBE and to progressively lesser extents to immediately upstream V_H-CBEs. (D–F) CBEs not directly flanking distal V_Hs theoretically also may augment V_H utilization. (D) A distal V_H locus CBE associates strongly with chromatin or associated factors (e.g., CTCF/cohesin) at the DJ_HRC. (E) Cohesin rings load near this DJ_HRC-associated distal V_H locus CBE and initiate loop extrusion. (F) Loop-extrusion allows RAG to scan downstream (or upstream, not illustrated) V_Hs lacking directly associated CBEs from the DJ_HRC where the active/transcribed chromatin in which they lie facilitates access for V(D)J recombination (see text for more details and relevant references. See also Figure S1.