Role of RAG1 autoubiquitination in V(D)J recombination

The variable domains of Ig and T-cell receptor genes in vertebrates are assembled from gene fragments by the V(D)J recombination process. The RAG1–RAG2 recombinase (RAG1/2) initiates this recombination by cutting DNA at the borders of recombination signal sequences (RSS) and their neighboring gene segments. The RAG1 protein is also known to contain a ubiquitin E3 ligase activity, located in an N-terminal region that is not strictly required for the basic recombination reaction but helps to regulate recombination. The isolated E3 ligase domain was earlier shown to ubiquitinate one site in a neighboring RAG1 sequence. Here we show that autoubiquitination of full-length RAG1 at this specific residue (K233) results in a large increase of DNA cleavage by RAG1/2. A mutational block of the ubiquitination site abolishes this effect and inhibits recombination of a test substrate in mouse cells. Thus, ubiquitination of RAG1, which can be promoted by RAG1’s own ubiquitin ligase activity, plays a significant role in governing the level of V(D)J recombination activity.V(D)J recombination plays a central role in the production of antigen receptors by recombining V, D, and J gene segments from their genomic clusters to give rise to the highly varied populations of immunoglobulins and T-cell receptors (1). Recombination starts with the introduction of double-strand breaks by the RAG1/RAG2 protein complex at a pair of recombination signal sequences (RSS) (2, 3), distinguished by the length of the spacer DNA separating their conserved heptamer and nonamer elements. Recombination requires one RSS with a 12-base pair spacer and another with a 23-base pair spacer. Each pair of breaks is then processed by the nonhomologous DNA end-joining group of proteins to produce a junction of two segments of coding sequence (a coding joint) and a junction of the two RSSs (a signal joint) (4). The purified RAG1/2 protein complex displays the correct specificity for pairs of RSSs (5, 6), and has thus been used as a model for the initiation of V(D)J recombination. Until recently, the RAG proteins used for these studies have generally been minimal “core” regions of RAG1 and RAG2 (amino acids 384–1,008 of 1,040 in mouse RAG1 and 1–387 of 527 in RAG2), which are sufficient for specific binding and cleavage activity in a purified cell-free system. Ectopic expression of these truncated proteins supports V(D)J recombination in suitable cell lines, although with differences from the full-length proteins that will be discussed here.A complex composed of core RAG1 and RAG2 is more active than its full-length counterpart in cleavage of extrachromosomal substrates in a hamster cell line, but overall recombination is reported to be lower (7), indicating a defect in the stages of recombination subsequent to DNA cleavage. Similarly, mice or pre-B cells missing the RAG2 C-terminal noncore region are defective in the V to DJ recombination step of Ig heavy chain joining, although the earlier D to J joining step is normal (8). The mice also display an increased prevalence of lymphomas (9). A plant homeo domain (PHD) within the RAG2 C terminus is known to bind to chromatin, and specifically to histone 3 trimethylated on lysine 4 (H3K4me3), which is presumably an important step in directing RAG1/2 to loci bearing this “activating” modification (10). The lack of this domain may largely explain the defective functions of the RAG2 core protein. Similarly, although core RAG1 can support D to J rearrangement at the Ig heavy chain locus in RAG1^−/− pro-B cells, the level is reduced compared with that of full-length RAG1 (FLRAG1) (11), and deletions of certain smaller regions within the RAG1 N terminus have even greater effects (11). Some naturally occurring truncations of the RAG1 N terminus lead to human immunodeficiency (12). The functions of the parts of RAG1 and RAG2 outside of the catalytically essential cores have been reviewed (13). There is also evidence that the RAG1 and RAG2 C termini interact: DNA cleavage by RAG1/2 combinations containing both regions was greatly reduced but was restored upon addition of an H3K4me3-containing peptide (14). Relief of this autoinhibition may synergize with the chromatin-binding effect of the PHD domain to target recombination to the appropriate loci.The significant modulation of recombination in cells, and/or of DNA cleavage in vitro, by these “dispensable” regions of both RAG1 and RAG2 is further modified by covalent modifications of the proteins, which affect their stability or activity. RAG2 becomes phosphorylated at a specific site in its C terminus (T490) at the G1/S stage of the cell cycle, and is then ubiquitinated by the Skp2-SCF ubiquitin ligase, a central regulator of cell cycle progression, leading to its degradation in S phase (15, 16). Phosphorylation of RAG1 at residue S528 by the AMP-dependent protein kinase has also been described (17), in this case leading to increased activity of RAG1/2 both for cell-free DNA cleavage and for recombination in cells.The N terminus of RAG1 contains a Zn-binding motif (amino acids 264–389) that includes a C₃HC₄ RING (really interesting new gene) finger motif closely associated with an adjacent C₂H₂ Zn finger. This domain was shown to have ubiquitin ligase (E3) activity (18, 19), a common feature of RING finger domains, when combined with ubiquitin, the ubiquitin-activating (E1) enzyme, and an appropriate ubiquitin-conjugating (E2) enzyme. A naturally occurring human mutation in this RING finger motif (C328Y) was found to cause the primary immunodeficiency disease Omenn’s syndrome (20). A study of the equivalent mutation in mouse RAG1 (C325Y) showed that it greatly reduced recombination of an extrachromosomal plasmid, as did mutation of the neighboring residue (P326G) (21). Other RING finger residues critical for ubiquitin ligase activity appeared to contribute to robust recombination of extrachromosomal substrates (22). In biochemical experiments carried out with an N-terminal fragment of RAG1 (residues 218–389), the principal site of autoubiquitination was found to be a residue neighboring the RING finger, K233; mutation of this residue (K233M) essentially abolished autoubiquitination of the fragment (18).In this article, we assess the site or sites and extent of autoubiquitination of RAG1, the consequences of this modification for RAG1/RAG2 activity in a cell-free system and in cells, and the functional relationship between this modification and the histone-recognizing PHD domain of RAG2. We prepare FLRAG1 in complex with either full-length RAG2 (FLRAG2) or core RAG2 and find that FLRAG1 undergoes autoubiquitination specifically at K233. The ubiquitination of RAG1 protein enhances coupled cleavage by the RAG1/RAG2 complex of a 12/23 RSS pair by about fivefold. RAG1 autoubiquitination also ap-pears to be important for supporting V(D)J recombination in cells.