Fig. 1. The types of reactions catalyzed by complexes of RAG-1 and RAG-2 proteins. (a) RAG proteins recognize a pair of RSSs (depicted here as open and closed triangles for a 12 bp and a 23 bp spacer RSS, respectively) and catalyse two DSBs by a two-step reaction. Firstly, following synapsis of the signals (which for simplicity is not depicted here), single-stranded nicks are generated in a coordinated fashion at the border between the coding regions (shaded boxes) and the heptamers of the RSSs. The 3′OH groups generated at the nicks attack the phosphodiester bond of the complementary DNA strands and cleave the DNA by a direct transesterification reaction. This creates hairpinned coding-ends (CEs) and blunt signal-ends (SEs). (b) CEs and SEs held in a synaptic complex can be ligated to hybrid joints by RAG proteins. For this, RAGs catalyze the transesterification of the 3′OH groups at the SE to the phosphodiester bonds of the opposite CE hairpin, resulting in a ‘cross-over’ ligation. (c) SEs bound by RAG proteins are able to transpose into DNA by attacking staggered phosphodiester bonds on both strands of a target site with their 3′OH groups. The phosphodiester bonds involved in the transesterifications with the signal ends are 5 bp apart, which results in 5 nucleotides of single-stranded, complementary regions flanking the transposed DNA (as indicated). If the transposition occurred in vivo, the single-stranded regions would be filled in by DNA polymerase, leading to 5-bp duplications at the integration site.

Fig. 2. Schematic representation of V(D)J recombination. A pair of RSSs is recognized and synapsed by a complex of RAG-1 and RAG-2 proteins, and DSBs are generated in a coordinated fashion at the heptamer side of the RSSs (open and black triangles) and the coding regions (shaded boxes). The coding ends (CEs) are processed by nicking at, or close to, the hairpin tip, followed by the optional addition of nucleotides by TdT, which leads to the generation of P- and N-region diversity (as indicated; see also Fig. 3). Blunt signal-ends are usually ligated without further modification to generate an SJ. In contrast, CEs may be subjected to trimming of the ends, alignment via microhomologies and fill-in of single-stranded gaps, before ligation into a CJ can occur.

Fig. 3. Possible scenarios for coding-end processing. (a) Hairpinned (HP) coding ends can be nicked 5′ of the tip, at the tip or 3′ thereof (indicated by encircled 1, 2 and 3, respectively), which may be catalyzed by a complex of RAG-1 and RAG-2 proteins (or possibly by a complex of Nbs1–Mre11–Rad50). Nicking up- or down-stream of the tip results in the generation (by DNA polymerase) of short P-region sequences. Coding end hairpin opening can therefore result in 5′ overhangs, blunt ends or 3′ overhangs that may further be modified by TdT-catalyzed N-region addition (as indicated), DNA-polymerase fill-in (as indicated), or nuclease digestion (not shown). (b) Processing of coding ends often results in noncompatible ends that cannot be directly ligated. In these cases, annealing via short 1–3 bp microhomologies close to the coding ends will most probably occur, resulting in 5′ and 3′ flap structures (as indicated). 3′ flap structures might be a substrate for the 3′ flap endonuclease activity of the RAG proteins (R) or the 3′→5′ exonuclease activity of Mre11 (M). 5′ flaps could be removed by the 5′ flap endonuclease FEN-1 (F), and single-stranded gaps can be filled-in by DNA polymerases (Pol). Single-stranded nicks are then a substrate for a complex of XRCC4 and DNA ligase IV (X4–L4), catalyzing the ligation into a CJ.