Transcription-coupled eviction of histones H2A/H2B governs V(D)J recombination

Abstract

Initiation of V(D)J recombination critically relies on the formation of an accessible chromatin structure at recombination signal sequences (RSSs) but how this accessibility is generated is poorly understood. Immunoglobulin light-chain loci normally undergo recombination in pre-B cells. We show here that equipping (earlier) pro-B cells with the increased pre-B-cell levels of just one transcription factor, IRF4, triggers the entire cascade of events leading to premature light-chain recombination. We then used this finding to dissect the critical events that generate RSS accessibility and show that the chromatin modifications previously associated with recombination are insufficient. Instead, we establish that non-coding transcription triggers IgL RSS accessibility and find that the accessibility is transient. Transcription transiently evicts H2A/H2B dimers, releasing 35-40 bp of nucleosomal DNA, and we demonstrate that H2A/H2B loss can explain the RSS accessibility observed in vivo. We therefore propose that the transcription-mediated eviction of H2A/H2B dimers is an important mechanism that makes RSSs accessible for the initiation of recombination.

Figure 1

Increased IRF4 levels in pro-B cells from three transgenic lines. (A) Diagram of the λ5/V_pre-B construct. (B) Left: IRF4 protein levels in purified primary pro-B and pre-B cells (NTG BM). Right: IRF4 levels in cultured pro-B cells from the PIP3, PIP4 and PIP2 lines compared to cultured pro-B cells from NTG. Protein loading was normalised to β-tubulin.

Figure 2

Increased IRF4 levels in pro-B cells cause premature light-chain recombination. (A, B) Upper: Schematics of the Igκ and Igλ loci. Arrows indicate the locations of primers used in the PCR assay, filled triangles represent RSSs; Eλ_2–4/Eλ_3–1 indicate the Igλ recombination enhancers; κ3′E, the Igκ 3′ enhancer; κEi, the Igκ intronic enhancer and V, J and C indicate variable, joining and constant regions, respectively. The probe to detect Igκ recombination is shown as a black bar. (A, B) Lower: Southern blots showing the level of (A) Igκ and (B) Igλ recombination products in cultured pro-B cells from the PIP3, PIP4 and PIP2 transgenic lines and NTG and pro-B cells and pre-B cells that had been purified directly from the bone marrow of NTG (NTG BM). Duplicate samples were analysed for each transgenic line. The fold increase is compared to cultured pro-B cells from NTG. The amounts of DNA were normalised to a region of the β-globin gene. LC is the loading control, DNA was added in three-fold increments. (C) Graph showing the relative IRF4, Igλ and Igκ recombination levels in the transgenic lines. The recombination data are an average of three experiments; s.d. is shown. Source data for this figure is available on the online supplementary information page.

Figure 3

Increased IRF4 levels in pro-B cells cause premature non-coding transcription of the Igκ and Igλ loci. (A, B) Upper: schematic of the Igκ and Igλ loci. Labels are as for Figure 2. Lower: Southern blots showing the level of non-coding transcripts at (A) the Igκ locus through the κJC and Vκ₀₂ regions, and (B) at the Igλ locus through the Vλ_1/2 and Jλ₁ gene segments. For the V genes, a probe was used that detects transcripts from both Vλ₁ and Vλ₂. Transcripts were measured in pro-B cells from transgenic mice, NTG and from pre-B cells from NTG. The fold increase shown is compared to pro-B cells from NTG. The amounts of cDNA were normalised to Hprt transcripts; duplicate samples are shown from each transgenic line. The asterisks indicate where lanes have been removed from a continuous gel. Source data for this figure is available on the online supplementary information page.

Figure 4

Increased IRF4 levels in pro-B cells stimulate H3K4me3 at the Igκ and Igλ loci. (A) Schematic of the Igλ and Igκ loci. Labels are as for Figure 2. (B) H3K4me3 levels at the Igλ (left) and the Igκ (right) loci in primary pro-B cells and pre-B cells from NTG and pro-B cells from PIP3 transgenic mice. The levels of H3K4me3 are shown as a percentage of the positive control, the CD19 promoter; the negative control (AP) is the α-amylase promoter. Int III, Int IV and Intgene are intergenic regions. The s.d. from three independent experiments is shown.

Figure 5

Non-coding transcription increases accessibility at RSSs in pre-B cells. (A) Upper: schematics of the Jλ₁ (left) and the Jκ₃ (right) gene segments and RSSs. The grey triangle represents the RSS and black and grey boxes represent the heptamer and nonamer, respectively. The restriction sites used to probe accessibility are shown. Arrows show the positions of the PCR primers. Lower: accessibility of the Jλ₁ and Jκ₃ gene segments in primary pro-B and pre-B cells from NTG and pro-B cells from the PIP3 transgenic line. Error bars show the s.d. from at least three experiments. (B) Histone H4 acetylation levels at the Igλ locus (left) and the Igκ locus (right) in primary pro-B and pre-B cells from NTG and pro-B cells from PIP3 transgenic mice. The levels of acetylation are shown as a percentage of the positive control, the CD19 promoter. The s.d. from three independent experiments is shown.

Figure 6

Eviction of an H2A/H2B dimer makes RSSs accessible for RAG cutting. Schematic (upper panel), RAG cutting (middle panel) and Sal I cutting (lower panel) of (A) free DNA, hexasome and nucleosome where the heptamer and the RSS spacer are not protected by histones in the hexasome, (B) free DNA and hexasome where the heptamer of the RSS is not protected by histones and (C) reconstitutes where the nucleosome and hexasome protect the RSS. The solid lines in the schematic show the region protected by the hexasome; the dotted lines show the additional region protected by the nucleosome (Supplementary Figure S7). In (C), the amount of DNA in the nucleosome and hexasome preparations is identical. The grey triangle represents the RSS and the heptamer and nonamer of the RSS are represented by black and grey boxes, respectively.

Figure 7

RSSs are transiently and stochastically made accessible via eviction of H2A/H2B dimers. (A) Accessibility at the RSSs is transient. Nuclei were isolated from pre-B cells and accessibility at the Jλ₁ Sty I site measured at the times shown following addition of α-amanitin. The data are the average of two experiments. (B) Transcription-dependent eviction of H2B at the Jκ1 RSS. The levels of H2B and H3 were compared in primary pre-B cells at the Jκ1 RSS in the presence (black bars) or absence (grey bars) of α-amanitin (left). Changes in H2B in the presence and absence of α-amanitin at a region not known to be transcribed (Intgene III) are shown on the right. Values at Jκ1 were normalised to those at a non-transcribed region; the values at the Int III region are averaged values using normalised amounts of DNA. The upstream primer is in a region that is deleted upon recombination; thus the changes are measured in pre-B cells that have not yet undergone recombination. The data are an average of five experiments; error bars show s.d. (C) Eviction of H2B correlates with the presence of RNA polymerase II. The presence of RNA polymerase II as determined by chromatin immunoprecipitation at the Jκ1 RSS and at the non-transcribed Intgene III region in the presence and absence of α-amanitin. Values were normalised to a non-transcribed region (Int V); the average of three experiments with s.d. is shown. (D) Eviction of H2B correlates with increased RAG1 binding. Binding was examined at the Jκ1 RSS, the Intgene III region and the Gapdh gene as an example of a transcribed region. Numbers were normalised to Gapdh or, in the case of Gapdh, by using equivalent amounts of DNA. The data are the average of three experiments; s.e. is shown. (E) Nucleosome mapping at the Jκ region and adjacent to κ3′E. Nuclei were isolated from 103/BCL-2 cells that had been grown at 39.7 °C to induce IgL recombination and digested with increasing amounts of micrococcal nuclease. The isolated DNA was digested with the restriction enzymes shown and the probes used in indirect end-labelling are indicated (black bars). The region adjacent to κ3′E (middle panel) is shown as a control since promoters and enhancers are thought to promote positioning of adjacent nucleosomes, generating a characteristic nucleosome ladder. A more smeared pattern is apparent at the Jκ RSSs (right).

Figure 8

Differences in STAT5-mediated repression of Igλ and Igκ recombination. (A) Chromatin immunoprecipitation of IRF4 at the Igλ (E3-1) and Igκ (k3’E) recombination enhancers in pro-B cells and pre-B cells from NTG and pro-B cells from the PIP3 transgenic mice. The data are an average of at least three experiments; s.d. is shown. (B) STAT5 does not bind to the Eλ_3–1 enhancer (E3–1) but is enriched at the Irf4 promoter in pro-B cells. (C) Binding of STAT5 to the Irf4 promoter decreases at the pro-B/pre-B transition.