The nonamer code for RAG-mediated recombination in vivo

V(D)J recombination, essential for adaptive immunity, is initiated by RAG1/2, which recognizes recombination signal sequences (RSSs) flanking gene segments at antigen receptor loci. RSSs comprise conserved heptamer and nonamer motifs separated by a 12/23-base spacer. While the first three heptamer nucleotides (5′-CAC) are strictly conserved, RSS promiscuity enables RAG “off-target” activity at RSS-like sequences, termed “cryptic RSSs” (cRSSs), contributing to chromosomal deletions and lymphoid tumorigenesis. Notably, the nonamer exhibits substantial sequence variability across physiological RSSs. In addition, many cRSSs lack a discernible nonamer-like sequence, limiting the ability to predict RAG targets based on RSS consensus. Using a high-throughput approach, we characterized here the nonamer properties supporting RAG-mediated recombination. While the consensus nonamer (5′-ACAAAAACC) exhibited strong functionality, numerous sequences significantly different from it promoted a high recombination rate. These functional nonamers balance two, somewhat opposing, features: affinity for the RAG nonamer-binding domain—primarily via a shared purine–A/T–purine motif at positions 4–6—and nucleosome repulsion. Moreover, nonamers of genomic cRSSs mimic canonical nonamers mainly through nucleosome-repelling sequences. This study provides a model for both physiological RAG activity and its off-target effects, enhancing our understanding of immune repertoire formation and the genetic basis of lymphoid cancers.