Regulation of V(D)J recombination by nucleosome positioning at recombination signal sequences

Abstract

A key component in the regulation of V(D)J recombination is control of the accessibility of RAG proteins to recombination signal sequences (RSS). Nucleosomes are known to inhibit this accessibility. We show here that the signal sequence itself represses accessibility by causing nucleosome positioning over the RSS. This positioning is mediated, in vitro and in vivo, by the conserved nonamer of the RSS. Consistent with this strong positioning, nucleosomes at RSSs are resistant to remodelling by nucleosome sliding. In vivo we find that consensus RSSs are preferentially protected, whereas those that lack a consensus nonamer, including some cryptic RSSs, fail to position nucleosomes. Decreased protection of these non-consensus RSSs correlates with their increased use in recombination assays. We therefore suggest that nucleosome positioning by RSSs provides a previously unanticipated level of protection and regulation of V(D)J recombination.

Fig. 1. Mapping of nucleosomes on 12- and 23-bp spacer RSS fragments. (A) Restriction enzyme map of the 284-bp Acc65I–NotI fragment from plasmid pFM210. The triangle represents a functional 12-bp spacer RSS consisting of a nonamer (N), a 12-bp spacer (12) and a hexamer (H). (B) Native polyacrylamide gel electrophoresis of mononucleosomes reconstituted on the radiolabelled pFM210 Acc65I–NotI fragment. The nucleosomes are labelled ‘MIDDLE’ or ‘END’ according to their translational position. ‘FREE’ indicates DNA that was not reconstituted into nucleosomes. ‘HEXAMER’ indicates reconstitutes that lack one H2A/B heterodimer as determined by protection of only 105 bp following micrococcal nuclease digestion. Nucleosome positions, according to the mapping in (C), are shown in the schematic drawing to the right. (C) Restriction enzyme digestions of the 147-bp fragments obtained following micrococcal nuclease digestion of the ‘END’ or ‘MIDDLE’ reconstituted nucleosomes. (D) Native polyacrylamide gel electrophoresis of mononucleosomes reconstituted on the radiolabelled Jκ1C fragment carrying a 23-bp spacer RSS. A summary of the nucleosome position mapping is given on the right.

Fig. 2. Stimulation of V(D)J cleavage and nucleosome sliding by NURF. (A) Free DNA (Free) or the middle nucleosome complex (MIDDLE) carrying the 12-bp spacer RSS (Figure 1B) or 23-bp spacer RSS (Figure 1D) were incubated with recombinant RAG1 and RAG2 in the absence or presence of NURF. V(D)J cleavage products (161 and 123 bp) were detected using native gel electrophoresis. Addition of NURF to free DNA (lane 3) shows NURF has no effect on RAG cutting of this template. Addition of HMGB1 was required for efficient cleavage at the 23 RSS. The cleavage products for the 23 RSS DNA fragment are 162 and 123 bp. (B) Nucleosomes reconstituted into the middle position of DNA fragments carrying the 12- or 23-bp spacer RSS were incubated with NURF in the presence or absence of ATP and analysed by native gel electrophoresis. The mobilities of nucleosomes in the centre or at the ends of the fragment are shown in cartoons to the left.

Fig. 3. The position of nucleosomes is influenced by the position of the RSS. (A) Schematic representation of the relocation of the 12-bp spacer RSS from the centre of the Acc65I–NotI fragment in plasmid pFM210 to the left end in plasmid pMAB5, or the right end in MAB20. The RAG cutting site is indicated by a boxed number. (B–D) Mapping of nucleosome positions on the pMAB4, pMAB5, pMAB20 Acc65I–NotI fragments, respectively. Restriction enzyme analysis of the 147-bp fragments obtained following micrococcal nuclease digestion of ‘MIDDLE’ nucleosome complexes is on the left. A summary of the nucleosome positions is shown on the schematic drawing to the right.

Fig. 4. Mutation of the conserved nonamer affects nucleosome positioning and mobility. (A) Alignment of the pFM210 12-bp spacer RSS and the Jκ1C 23-bp spacer RSS with the sea urchin 5S rDNA nucleosome positioning sequence and the artificial nucleosome positioning sequence TG (Shrader and Crothers, 1989). The nonamer (N) and hexamer (H) sequences are underlined. (B) Nucleosome mapping on the Acc65I–NotI fragment from pMAB7 containing a mutated nonamer sequence within the 12-bp spacer RSS was performed as described above using ‘MIDDLE’ nucleosome complex. The major position is shown schematically to the right. (C) Nucleosome sliding assay. Nucleosomes reconstituted in the MIDDLE position of the pMAB5 Acc65I–NotI or the pMAB7 Acc65I–NotI fragment were incubated with NURF and analysed by native gel electrophoresis.

Fig. 5. Effects of the nonamer sequence on the accessibility of the RSS in vivo. (A) Partial map of the plasmid recombination substrates, pMAB1 and pMAB9. The 12-bp spacer RSS (white arrowhead) and the 23-bp spacer RSS (black arrowhead) are indicated. The RSS sequences contain restriction enzyme sites for XhoI (12-bp spacer RSS) or ScaI (23-bp spacer RSS). In pMAB9 the sequence of the nonamer within the 12-bp spacer RSS is mutated. Following digestion with ScaI or XhoI in nuclei, DNA was digested with PvuI and NcoI to give the RSS parental band or with BamHI to give the parental band of the plasmid backbone. A map of the expected sizes for parental bands and digestion within the RSS is given above the plasmid diagram; that for accessibility of the plasmid backbone ScaI site is given below the plasmid diagram. The Southern probe is indicated. (B) Southern analysis of pMAB1 and pMAB9 after transfection into NIH 3T3 cells. Nuclei were prepared from transfected cells and subjected to restriction enzyme cleavage by either XhoI or ScaI as described in (A). Quantitation of the average accessibility from five experiments and the standard deviation is given below the blots. (C) Analysis of RSS accessibility at endogenous loci. Accessibility of Jκ1 23 RSS to ScaI is shown in NIH 3T3 cells and the lymphoid cell line, 63-12. This is compared to accessibility at two other ScaI sites that are not associated with an RSS: one within the kappa locus and one within the β-globin locus. Accessibility of the Vκ21c 12 RSS to StyI and the human TCRαJ34 12 RSS to BglII are compared with the accessibility of adjacent sites in NIH 3T3 and 63-12 cells and in HeLa cells, respectively. As a control, cutting at a MspI site, 25 bp within a highly positioned nucleosome at the c-fos promoter is compared with cutting at an adjacent site within the c-fos locus in A431 cells. All experiments were internally controlled: the cutting at the RSS was compared with the cutting at an adjacent site using DNA prepared from the same digested nuclei sample.

Fig. 6. Loss of nucleosome positioning at non-consensus RSSs is correlated with increased recombination. (A) Reconstitution of nucleosomes onto a 277-bp fragment where the nonamer has been substituted by the cryptic nonamer from TAL2. Free DNA and nucleosomes in the END and MIDDLE positions are indicated. The predominant complexes (middle-1, -2, -3) were excised and the nucleosome positions mapped. The various positions detected are shown diagrammatically to the left. Identical results were obtained using the LMO2 cryptic nonamer (data not shown). (B) Accessibility of the non- consensus nonamer RSSs in vivo. The TAL2 cryptic RSS was substituted for the 12 RSS in pMAB1 to generate pMAB15. Mutation of 1 bp within the linker of the cryptic RSS created an XhoI site. Accessibility of the cryptic RSS, the 23 RSS and the neutral ScaI site was analysed for the transfected, chromatinized plasmid templates as described above. Quantitation of the accessibility (an average of at least three experiments) is shown below the blot. Accessibility of the endogenous TAL2 locus was analysed in HeLa cells using a DraIII site within the heptamer of the RSS. This was compared with a neighbouring DraIII site that lies distant to the RSS and to the accessibility of a DraIII within a consensus 12 RSS at Vκ2–24. (C) RAG cutting at non-consensus and consensus RSS in vitro. Naked DNA fragments covering the TAL2 cryptic RSS and the VκL8 consensus RSS were incubated with purified RAG proteins for increasing times in vitro. The extent of RAG cutting was analysed by native gel electrophoresis and was quantitated using a PhosphorImager. (D) Recombination at non-consensus and consensus RSSs. Plasmids carrying either a consensus (pMAB1) or cryptic (pMAB15) RSS were used as substrates in a recombination assay. Reporter plasmids that have undergone recombination are both ampicillin and chloramphenicol resistant whereas those that have not are only ampicillin resistant. The number of colonies given is the sum of six experiments.