Base Flipping in V(D)J Recombination: Insights into the Mechanism of Hairpin Formation,the 12/23 Rule,and the Coordination of Double-Strand Breaks

Tn5 transposase cleaves the transposon end using a hairpin intermediate on the transposon end. This involves a flipped base that is stacked against a tryptophan residue in the protein. However, many other members of the cut-and-paste transposase family, including the RAG1 protein, produce a hairpin on the flanking DNA. We have investigated the reversed polarity of the reaction for RAG recombination. Although the RAG proteins appear to employ a base-flipping mechanism using aromatic residues, the putatively flipped base is not at the expected location and does not appear to stack against any of the said aromatic residues. We propose an alternative model in which a flipped base is accommodated in a nonspecific pocket or cleft within the recombinase. This is consistent with the location of the flipped base at position −1 in the coding flank, which can be occupied by purine or pyrimidine bases that would be difficult to stabilize using a single, highly specific, interaction. Finally, during this work we noticed that the putative base-flipping events on either side of the 12/23 recombination signal sequence paired complex are coupled to the nicking steps and serve to coordinate the double-strand breaks on either side of the complex.Antibody and T-cell receptor (TCR) diversity is generated by V(D)J recombination initiated by the RAG proteins, RAG1 and RAG2. The recombination signal sequences (RSSs), where recombination takes place, have a distinctive arrangement resembling transposon ends. The relationship between V(D)J recombination and transposition was established beyond doubt by the discovery of RAG-mediated transposition and by the identification of a triad of conserved active-site residues. This evidence placed RAG1 firmly within the family of transposases and retroviral integrases that have a characteristic DDE triad of amino acid residues that coordinate catalytic metal ions in the active site (1, 26, 30, 35, 39, 46). Later, the Transib family of transposons was identified as the likely ancestral group of RAG1 (33).In V(D)J recombination, the RAG proteins excise the DNA between a pair of RSSs. This fragment is the equivalent of an excised transposon, and it takes no further part in the canonical V(D)J recombination reaction. Instead, the variable regions of the genes encoding antibodies and TCR are created by the imprecise rejoining of the flanking DNA, referred to as the “coding flank.” A key feature of the cleaved coding flanks is that they have covalently closed hairpin ends. The asymmetric resolution of these hairpins contributes to the diversification of the coding sequences during rejoining. The hairpins themselves arise as a consequence of the molecular mechanism RAG-mediated RSS cleavage.The crystal structure for the catalytic core of the human immunodeficiency virus type 1 integrase protein revealed a structural fold shared in common with RNase H and the Holliday junction resolving enzyme RuvC (22). RNase H and RuvC monomers each perform a simple nicking reaction that requires a single phosphoryl transfer event. Cut-and-paste transposition, which requires at least three phosphoryl transfer steps at each transposon end, therefore presents a mechanistic challenge. One solution to this challenge was revealed by the discovery of the DNA-hairpin cleavage-intermediate in V(D)J recombination and Tn10 transposition (Fig. ​(Fig.1)1) (34, 57). However, it is interesting to note that the existence of this intermediate was first suggested by Coen and colleagues on the basis of the genomic scars produced by excision of the hAT family transposon Tam3 in Antirrhinum majus (14).Open in a separate windowFIG. 1.Hairpin-processing reactions of opposite polarity. Most prokaryotic and eukaryotic members of the DDE family have hairpin intermediates of opposite polarity. In this paper, we refer to the two strands of DNA as “first strand” or “second strand” depending on the order of cleavage. The first strand therefore corresponds to the transferred and nontransferred strands of the prokaryotic and eukaryotic elements, respectively. Scissile phosphates are in red. The transposon end and RSS are shown as gray triangles. (Left panel) In Tn5 and Tn10, the first step of the reaction is a nick on the bottom (first) strand that exposes the 3′-OH at the end of the transposon. The second strand is cleaved by a direct transesterification reaction, which generates a “proximal-hairpin” intermediate on the transposon end (5, 34). Resolution by a nick at the tip of the hairpin yields a blunt transposon end. Distortion of the DNA helix can be detected by permanganate sensitivity of the T−1 and T+2 residues on the second strand. The insert shows the crystal structure of the Tn5 transposon end, highlighting the flipped base at position +2 (19). Two tryptophan residues are also shown. One acts as a “wedge” or “probe” residue inserted into the DNA helix, while the other provides stacking interactions that stabilize the flipped base. The W323 probe residue resides within the catalytic core close to the DDE residue E326, whereas the W298 stacking residue is in the inserted subdomain (see text for further details). Base flipping takes place after the first nick and is probably maintained for all subsequent steps, including integration (3, 7). (Right panel) In V(D)J recombination and the hAT family of transposons, the polarity of the reaction is reversed. The first nick is on the top strand providing a 3′-OH group on the flanking DNA (53, 71, 77). Transesterification yields a “distal hairpin” intermediate on the flanking DNA that is processed by the host. The positions of relevant thymidine residues in our substrates are indicated.All DDE family transposases, including RAG1, cut the DNA to expose the 3′-OH at the end of the element (or RSS). However, the fate of the opposite strand and the order of strand cleavage events vary within the group (reviewed in references13, 18, and 55). Some enzymes, such as the retroviral integrases and the bacteriophage Mu transposase, nick and integrate the 3′-OH directly without second-strand cleavage. The cut-and-paste transposons, which cleave both strands of DNA, can be divided into two groups. With some notable exceptions such as the piggyBac element, most prokaryotic family members cleave the bottom strand of the recombination site first, whereas most eukaryotic members cleave the top strand first (8, 10, 20, 41, 47, 48, 77). For those family members with a hairpin mechanism, the inverted polarity of the first step dictates the reversal of all subsequent steps (Fig. ​(Fig.1).1). In consequence, most eukaryotic members of the family can achieve transposition with one less phosphoryl transfer reaction than the prokaryotic members, which are obliged to resolve the hairpin intermediate. The eukaryotic members can simply release the hairpin ends or, as in the case of RAG, hand them on to host factors for further processing (40).Insight into the hairpin mechanism was provided by a crystal structure for the Tn5 transpososome, in which the penultimate base on the second, nontransferred, strand was flipped from the helix and stacked against a tryptophan side chain in the protein (Fig. ​(Fig.1)1) (19). The flipped base seemed to provide the steric freedom that is presumed to be required for making and resolving the hairpin intermediate. Two groups searched for a residue in RAG1 that performs a function equivalent to the stacking tryptophan in the Tn5 transposase (27, 45). This work identified several candidate residues on the basis of their respective mechanistic defects and their rescue by modified DNA substrates.Here we have further assessed the candidate stacking residues using biochemical techniques previously used to study the dynamics of base flipping in Tn5 and Tn10 transposition (6, 7). We have identified a distortion at position −1 of the V(D)J coding flank DNA. It is introduced after the first nick at the RSS and is therefore reminiscent of the flipped base at the end of Tn5. The distortion is perfectly correlated with the ability of wild-type and mutant RAG-RSS complexes to perform the hairpin step of the reaction. We conclude that this base is probably equivalent to the flipped base in Tn5. However, none of the candidate aromatic residues seems to fulfill the function of the putative stacking tryptophan residue. We therefore propose a model in which base flipping in RAG recombination is significantly different from that in Tn5 transposition.Canonical V(D)J recombination occurs within a 12/23 RSS paired complex (24, 36, 60, 72, 73). This restriction is known as the 12/23 rule. More recently a further restriction, the so-called “beyond 12/23” (B12/23) rule has been proposed to explain the exclusion of direct Vβ-to-Jβ joining in the TCR β region, despite the presence of appropriately oriented pairs of 12 and 23 RSSs (4, 21, 31, 32).Little is known of the mechanisms that enforce the 12/23 rule or coordinate cleavage on either side of the complex. However, during this work, we observed that the coding flank distortion was coupled on either side of a 12/23 RSS paired complex: the distortion of a nicked coding flank is suppressed by an unnicked partner. We present a model and discuss the biological significance of this conformational coupling and its relevance to the B12/23 rule.