|Home | About | Journals | Submit | Contact Us | Français|
Nuclear processes such as transcription, DNA replication, and recombination are dynamically regulated by chromatin structure. Transcription is known to be regulated by chromatin-associated proteins containing conserved protein domains that specifically recognize distinct covalent post-translational modifications on histones. However, it has been unclear whether similar mechanisms are involved in mammalian DNA recombination. Here, we show that RAG2 – an essential component of the RAG1/2 V(D)J recombinase, that mediates antigen receptor gene assembly1 – contains a plant homeodomain (PHD) finger that specifically recognizes histone H3 trimethylated at lysine 4 (H3K4me3). The high-resolution crystal structure of the RAG2 PHD finger bound to H3K4me3 reveals the molecular basis of H3K4me3-recognition by RAG2. Mutations that abrogate RAG2's recognition of H3K4me3 severely impair V(D)J recombination in vivo. Reducing the level of H3K4me3 similarly leads to a decrease in V(D)J recombination in vivo. Notably, a conserved tryptophan residue (W453) that constitutes a key structural component of the K4me3-binding surface and is essential for RAG2's recognition of H3K4me3 is mutated in patients with immunodeficiency syndromes. Together our results identify a novel function for histone methylation in mammalian DNA recombination. Furthermore, our results provide the first evidence suggesting that disrupting the read-out of histone modifications can cause an inherited human disease.
Many studies suggest that V(D)J recombination is regulated by modulating the chromatin structure of antigen receptor loci during lymphoid development2. Several studies have analyzed the pattern of histone modifications present at the immunoglobulin heavy chain locus during B cell development3-5. However, mechanisms linking histone modifications to the function of the RAG recombinase have remained elusive.
Since RAG2 contains a noncanonical plant homeodomain (PHD) finger6,7 – a module that can mediate interactions with chromatin8-10 – we asked whether a polypeptide encompassing the RAG2 PHD finger (RAG2PHD: aa 414-527) can recognize modified histone proteins. In an in vitro screen of peptide microarrays containing ~70 distinct modified histone peptides, we found that RAG2PHD specifically binds to histone H3 trimethylated at lysine 4 (H3K4me3) (Fig. 1a; Fig. S1; data not shown). The specificity of this interaction was confirmed by peptide pull-down assays (Fig. 1b; Fig. S2; Fig. S3). RAG2 has a C-terminal extension of 40 aa that is essential for phosphoinositide (PtdInsP)-binding7 (aa 488-527), but this region is dispensable for H3K4me3-binding as the minimal PHD finger alone (aa 414-487) is sufficient for H3K4me3-recognition (Fig. 1c). In addition, the acidic hinge region of RAG2 (aa 388-412), previously implicated in histone-binding11, is dispensable for recognition of H3K4me3 (Fig. 1d). Moreover, mutations in the acidic hinge region, which had previously been shown to interfere with histone binding11, had no effect on the H3K4me3 interaction (Fig. 1d). Using calf thymus histone protein pull-down assays, we confirmed that the interaction between RAG2PHD and H3K4me3 occurs in the context of full-length histone proteins (Fig. 1e). Further, RAG2PHD bound to native mononucleosomes purified from HeLa cells, but not mononucleosomes reconstituted from bacterially-expressed recombinant histones, indicating that the RAG2PHD-nucleosome interaction is dependent upon post-translational histone modifications (Fig. 1f). Although full-length RAG2 binds H3K4me3 peptides, core RAG2 (aa 1-387) – the minimal portion of RAG2 required for V(D)J cleavage in vitro – which lacks the PHD finger, does not (Fig. 1g). Consistent with the pull-down results, tryptophan fluorescence measurements using RAG2PHD determined dissociation constants (Kd) of ~4 μM for H3K4me3, ~60 μM for H3K4me2, ~120 μM for H3K4me1, and ~500 μM for H3K4me0 (Fig. S5). Thus, the RAG2 PHD finger is a chromatin-binding module that recognizes H3K4me3.
To understand the molecular basis of the interaction between the RAG2 PHD finger and H3K4me3, we determined the crystal structure of the RAG2PHD-H3K4me3 complex at 1.15 Å resolution (Fig. 2; Fig. S6; Table S1). In three previously published PHD-H3K4me3 structures12-14, the H3 peptide extends straight through the peptide-binding groove. However, in the RAG2PHD-H3K4me3 structure, the H3 peptide is kinked by ~90° at Q5. The K4me3 sidechain is recognized by cation-π interactions within an aromatic channel delimited by Y415 on the left, M443 on the back, and W453 on the right (Fig. 2a). Despite not revealing any primary sequence conservation with the K4me3-binding residues of ING2, BPTF and YNG1 (Fig. S7), RAG2 forms a similar K4me3-binding pocket (Fig. 2c). Interestingly, RAG2PHD lacks the canonical “aromatic cage” often employed to bind trimethylated lysine residues15 (Figs. 2b and 2c). Instead of being closed on both sides, the back, and the top (as observed with other PHD fingers12-14) the RAG2PHD K4me3-binding surface is open on the top, and resembles an “aromatic channel” rather than an “aromatic cage” (Figs. 2b and 2c). This “channel” conformation may provide a mechanism to modulate histone binding16. Aside from the K4me3 and Q5 residues, the remaining side chains of H3 form no specific interactions with RAG2PHD. Finally, unlike other H3K4me3-binding PHD fingers9,12-14,17, RAG2PHD does not have acidic residues (Asp or Glu) positioned to electrostatically interact with H3R2 (Fig. 2c).
The three residues in RAG2 that form the aromatic channel critical for trimethyl-lysine recognition are completely conserved through evolution (Fig. S8). As expected, mutating any one of these residues (Y415A, M443A, W453A/R) abrogated H3K4me3-binding by RAG2PHD (Fig. 3a) and by full-length RAG2 (Fig. 3b). Since the W453R mutation has been implicated in the pathogenesis of Omenn's Syndrome18 – a rare severe combined immunodeficiency (SCID)19 – and the molecular mechanism linking this mutation to Omenn's Syndrome remains unknown7, we further characterized W453R's role in histone binding. Consistent with the critical role of this residue in forming the recognition surface for H3K4me3, introducing this mutation into the RAG2 PHD finger (RAG2PHD-W453R) abolished the ability of RAG2PHD to bind either full-length histone H3 (Fig. 3c) or intact nucleosomes (Fig. 3d). Thus, the interaction of RAG2 with histone proteins in vitro is dependent upon H3K4me3-binding. We note that RAG2PHD-W453R is properly folded, as indicated by a comparison of the 1H-15N HSQC spectra of RAG2PHD and RAG2PHD-W453R (Fig. S9). In addition, the Y415A and M443A substitutions had no effect on PtdInsP-binding by RAG2PHD (Fig. S10), and W453R-associated recombination defects were previously demonstrated to be independent of the PtdInsP-binding activity of RAG27. Thus, the identification of multiple different point mutations that selectively abrogate RAG2's recognition of H3K4me3 allowed us to study the functional significance of this interaction.
Before analyzing the effect of H3 methylation on V(D)J recombination in vivo, we first wanted to confirm that RAG2Y415A, RAG2M443A, and RAG2W453R disrupt H3K4me3-binding without affecting protein folding or the inherent catalytic properties of RAG2. We tested the ability of recombinant wild-type and mutant RAG2 proteins to catalyze V(D)J cleavage in vitro on a naked DNA substrate. In these assays, the recombinant proteins, which all expressed and purified equally well (Fig. 3e), were incubated with core RAG1 (aa 384−1008), and a DNA substrate. All three H3K4me3-binding mutants catalyzed V(D)J cleavage at wild-type levels (Fig. 3f). Therefore, the H3K4me3-binding mutants – RAG2Y415A, RAG2M443A, and RAG2W453R – are properly folded and catalytically active and thus, can be utilized for in vivo functional analyses.
Next, to address whether the recognition of methylated H3 by RAG2PHD plays a role in regulating RAG2 function in vivo, we performed extrachromosomal V(D)J recombination assays with the H3K4me3-binding mutants. Fibroblast cell lines were transfected with an exogenous recombination substrate that becomes partially chromatinized when introduced into cells and has nucleosomes positioned over the recombination signal sequences20, along with full-length RAG1, and either full-length wild-type RAG2, RAG2Y415A, RAG2M443A, or RAG2W453R. Strikingly, despite being expressed at comparable levels to wild-type RAG2 (Fig. 4A-panel a), all three H3K4me3-binding mutants exhibited a profound decrease (>90%) in recombination activity (Fig. 4A-panel b). Thus, the H3K4me3-recognition activity of RAG2 is required for V(D)J recombination in vivo.
To directly test the importance of H3K4 methylation and the interaction between RAG2 and H3K4me3 for V(D)J recombination in vivo, we used two different methods to reduce endogenous H3K4 methylation levels, and then repeated the extrachromosomal V(D)J recombination assays. First, H3K4 methylation levels were reduced (Fig. 4B-panel a) by knocking down expression of the common histone H3 lysine 4 methyltransferase component WDR515,21 with shRNA. Consistent with the decreased recombination observed with the H3K4me3-binding mutants, we observed a marked reduction in the recombination activity of wildtype RAG2 (~60%) in cells carrying the WDR5 shRNA (Fig. 4B-panels b, d). Significantly, V(D)J recombination by RAG2W453R – which does not recognize H3K4me3 – was unaffected by the presence of WDR5 shRNA (Fig. 4B-panels c, d), indicating that the reduction observed for wildtype RAG2 is specifically due to reduced H3K4me3-binding. In a second independent method, we reduced H3K4me3 levels by transiently expressing the H3K4me3 demethylase SMCX22. As with WDR5 shRNA, we observed that decreases in H3K4me3 levels (Fig. 4C-panel a) resulted in a significant reduction in recombination (~45%) in cells expressing SMCX (Fig. 4C-panels b, d). V(D)J recombination by RAG2W453R was unaffected by SMCX expression (Fig. 4C-panels c, d). Thus, reducing either the levels of H3K4 methylation or the ability of RAG2 to bind H3K4me3, impairs V(D)J recombination.
Since H3K4me3 is observed at actively rearranging gene segments23 (Fig. S12), we assayed the ability of the H3K4me3-binding mutants to perform chromosomal V(D)J recombination at the endogenous murine IgH locus. RAG2-/- Pro-B cells were transduced with lentiviruses encoding either RAG2, RAG2Y415A, RAG2M443A, or RAG2W453R and the extent of DH-to-JH recombination in the transduced cell populations was measured using a standard semi-quantitative PCR strategy. Despite being expressed at comparable levels to wild-type RAG2 (Fig. 4D-panel a), all three H3K4me3-binding mutants exhibited a dramatic reduction in DH-to-JH recombination (Fig. 4D-panel b). Thus, we conclude that recognition of H3K4me3 by RAG2PHD is crucial for V(D)J recombination at the endogenous immunoglobulin locus.
Our findings provide the first direct molecular link between histone methylation and mammalian DNA recombination. Prior to this study, H3K4me3 had only been demonstrated to function in the regulation of gene expression15. Our results, demonstrating a novel function for this mark in V(D)J recombination, highlight that chromatin structure can be coupled to diverse nuclear processes via protein modules that recognize modified histones. We have also provided the first evidence that disrupting the read-out of histone modifications can cause an inherited human disease. Omenn's Syndrome has been observed in patients carrying a W453R mutation in RAG218. Since RAG2W453R exhibits wildtype enzymatic activity in vitro7 (Fig. 3f), it has been unclear how this mutation causes immunodeficiency. Here, we have shown that the W453R mutation disrupts the read-out of H3K4me3, thereby providing a molecular explanation of how RAG2W453R causes Omenn's Syndrome. While the W453R, M443A, and Y415A point mutations – which selectively disrupt H3K4me3-recognition – severely disrupt V(D)J recombination, complete deletion of the RAG2 C-terminus (including the PHD finger) only partially compromises V(D)J recombination activity24-27 (Fig S11). We propose that the RAG2-H3K4me3 interaction serves two functions: (1) increases the stable association of RAG1/2 complexes at target sites and (2) might relieve an inhibitory activity present in the C-terminal portion of RAG27 (Fig S13). Taken together with other studies that have demonstrated a link between the writing of histone modifications and human disease15,28,29, our results highlight the fundamental role chromatin plays in human health. We postulate that in the coming years other naturally occurring mutations that cause inherited human diseases will be linked to disruption of lysine methylation signaling pathways.
Biotinylated peptides were synthesized at Stanford or Yale Protein and Nucleic Acid facilities. Peptide microarray experiments were performed essentially as described17. Biotinylated peptide pull-down assays, calf thymus histone binding assays, and mononucleosome gel shift assays were performed as described9. Tryptophan fluorescence assays were performed essentially as described13. NMR structure determination was performed as described7. The structure determination is described in Methods. In vitro V(D)J cleavage assays were performed as described31. The extrachromsomal V(D)J recombination assays and the endogenous V(D)J recombination assay are described in Methods. Information about antibodies is available in Methods.
We would like to thank Karim-Jean Armache and Ji-Joon Song for the generous gift of recombinant mononucleosomes; Robert Kingston and members of the Kingston lab for helpful discussions; and Nelson Lau, Ji-Joon Song, and Martin Gellert for critical reading of this manuscript. This work was supported by NIH grants to M.A.O., O.G., D.I., and T.G.K. as well as a Korea Research Foundation grant to S.H. O.G. is a recipient of a Burroughs Wellcome Career Award in Biomedical Sciences and a Kimmel Scholar Award. S.R.-M. has been the recipient of a fellowship from the Human Frontier Science Program. A.J.K. is funded by Stanford University through a Genentech Foundation Predoctoral Fellowship. A.G.W.M. is a Howard Hughes Medical Institute Predoctoral Fellow.
Antibodies used in the study: Histone H3 (Abcam); Flag M5 (Sigma); Flag M2 (Sigma); GST (Santa Cruz); donkey anti-rabbit IgG HRP-linked F(ab')2 fragment (GE Healthcare); sheep anti-mouse IgG HRP-linked whole Ab (GE Healthcare); and Alexa Fluor 647 chicken anti-rabbit IgG (Invitrogen). GST-RAG2PHD (aa 414-487), GST-RAG2PHD (aa 414-527), and FLAG-tagged RAG2 were described previously7. Point mutations were generated by site-directed mutagenesis PCR using Pfu turbo DNA polymerase (Stratagene). The GST-PHD finger of hING2 (aa 200-281) was described previously9.
Peptide microarray experiments were performed as described previously17. Briefly, biotinylated histone peptides were printed in 6 replicates onto a streptavidin-coated slide (ArrayIt) using a VersArray Compact Microarrayer (Bio-Rad). After a short blocking incubation with biotin (Sigma), the slides were incubated with GST-fused RAG2PHD in peptide binding buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 0.1% Nonidet P-40, 20% fetal bovine serum) overnight at 4 °C with gentle agitation. After washing with the same buffer, slides were probed first with anti-GST antibody and then fluorescein-conjugated secondary antibody and visualized with GenePix 4000 scanner (Molecular Devices).
Biotinylated peptide pull-down assays were performed as described previously9. Briefly, 1μg of biotinylated peptides were incubated with 1 μg of GST-PHD fingers in peptide binding buffer (50 mM Tris-HCl, pH 7.5, 300 mM NaCl, 0.1% Nonidet P-40) overnight at 4°C. After 1h incubation with streptavidin beads (Amersham), complexes were washed 3 times with the binding buffer, and the bound proteins were subjected to either Western or Coomassie analysis.
Calf thymus histone binding assays were performed as described previously9. Briefly, 10 μg of GST-RAG2PHD was incubated with 25 μg of calf thymus histones (Worthington) in binding buffer (50mM Tris-Hcl 7.5, 1M NaCl, 1% Nonidet P-40) overnight at 4°C. After being incubated with glutathione beads for 1h, complexes were washed 3 times with binding buffer, and bound proteins were subjected to Western analysis.
Mononucleosome gel shift assays were performed as described previously9. Briefly, 1.5 μg of mononucleosomes isolated from HeLa cells or assembled from recombinant histones32 were incubated with GST-RAG2PHD in binding buffer (20mM HEPES 7.9, 80mM KCl, 0.1mM, ZnCl2, 0.1% EDTA and 10% glycerol) at 30°C for 30 minutes. The reaction was then subjected to 5% TBE native gel electrophoresis and visualized with ethidium-bromide staining.
The fluorescence spectra were recorded at 25°C on Fluoromax3 spectrofluorometer. The samples of 10 μM RAG2PHD containing progressively increasing concentrations (up to 2 mM) of histone H3 peptides (aa 1-12) were excited at 295 nm. Emission spectra were recorded between 305 and 405 nm with a 0.5 nm step size and a 1 s integration time and averaged over 3 scans. The Kd's were determined by a nonlinear least-squares analysis using the equation: ΔI=(ΔImax*L)/(KD+L), where L is concentration of the histone peptide, ΔI is observed change of signal intensity, and ΔImax is the difference in signal intensity of the free and bound states of the RAG2 PHD finger. The Kd value was averaged over two experiments for the H3K4me3 binding and over three separate experiments for the binding of H3K4me0, H3K4me1 and H3K4me2 peptides.
The RAG2 PHD finger (residues 414-487) was prepared as described16. Crystals of its complex with H3K4me3 were obtained at 3 mg/ml protein concentration and 1:1.5 molar ratio of protein to peptide with a precipitant solution of 26 % PEG 3350 and 0.18 M potassium thiocyanate. X-ray diffraction data were collected from a single crystal of RAG2-PHD - H3K4me3 complex on a Mar225 CCD detector at ID-22 1 Å beamline in Advanced Photon Source (APS) at -160°C. The crystal belongs to the C2 space group with 2 RAG2-PHD-H3K4me3 complexes per asymmetric unit (Table S1). Ramachandran statistics: 93.3% allowed region, 6.7% additionally allowed region. The data set was processed and scaled at 1.15 Å using HKL200033. Crystallographic phases were obtained by molecular replacement with PHASER34, using as a model the 2.4 Å resolution structure of RAG2-PHD-H3K4me3 determined previously35. The model was traced using COOT36 and refined using CNS37 and SHELX38. Individual anisotropic displacement parameters were refined for all atoms, and hydrogens were added in the late stages of refinement.
FLAG-tagged RAG2 and derivatives were used for in vitro V(D)J cleavage assays as described previously39. Briefly, V(D)J cleavage assays were initiated by the addition of R1 (~80 ng) and R2 (~10 ng) proteins to a 10-μl reaction mixture containing 0.25 pmol of 32P-labeled, uncleaved 12-RSS substrate (VDJ100/101) in 60 mM KGlu, 1 mM MnCl2, 25 mM Hepes (pH 7.5), and 2mM DTT. After a 2-hr incubation at 30°C, the reactions were stopped by addition of 95% formamide loading dye. The samples were denatured by heating at 95°C for 5 minutes and reaction products were visualized by autoradiography of samples separated by denaturing electrophoresis.
Extrachromosomal V(D)J recombination assays were performed as described previously40 in either Br3neo40 or HT10809 human fibroblastoid cells, and the plasmid pGG49 was used as the reporter41. Full-length RAG1 was transiently expressed using the pcDNA6-myc-hisA vector (Invitrogen). RAG2 and derivatives were transiently expressed using the p3xFLAG-CMV vector (Sigma). Expression of all proteins was confirmed by Western analysis. For the WDR5 shRNA experiments, HT1080 cells were first stably transfected with either a WDR5 shRNA vector or a control vector that lacks the shRNA insert. These cells were subsequently transfected with RAG1, RAG2, and a recombination substrate. Activity was normalized to that of wild-type RAG2 in the absence of WDR5, which was defined as 100%. For the SMCX experiments, Br3neo cells were simultaneously co-transfected with RAG1, RAG2, a recombination substrate, and either an SMCX mammalian expression vector, or a control vector that lacks the SMCX insert. Activity was normalized to that of wild-type RAG2 in the absence of SMCX, which was defined as 100%.
Note added in proof: While this work was under review, another study also reported that the RAG2 PHD finger binds to methylated H3K430.
Atomic coordinates and structure factors of the RAG2PHD-H3K4me3 peptide complex have been deposited in the Protein Data Bank with the accession code of 2v89. Reprints and permissions information is available at npg.nature.com/reprintsandpermissions.
Supplementary Information is linked to the online version of the paper at www.nature.com/nature.