High‐Throughput Characterization of VDJ Recombination Signal Sequences

Abstract

Developing B- and T-cells produce a diverse antigen receptor repertoire by rearranging antigen receptor genes during VDJ recombination. Recombination activating proteins, RAG1 and RAG2, catalyze VDJ recombination by cleaving DNA between recombination signal sequences (RSS) and antigen receptor gene segments, and the double-strand breaks are repaired by nonhomologous end-joining machinery. The RSS is defined as containing a consensus heptamer and a nonamer sequence separated by a less conserved spacer sequence. However, many RSSs deviate from the consensus sequence. Therefore, RAG1/2 must be promiscuous to facilitate recombination of poorly conserved RSSs, but it must also be precise to avoid off-target cryptic RSSs. Recent cryo-EM and X-ray crystallographic studies on RAG1/2 showed the RSS heptamer undergoes a dramatic structural transition in the RAG1/2 active site. Before cleavage, the heptamer region untwists by 180°, so RAG1/2 sidechains contacting the major groove switch to contacting the minor groove and vice-versa. RAG1/2 also makes few base-specific contacts prior to nicking. Therefore, we hypothesized that sequence-specific structural properties of the RSS heptamer plays a role in RAG1/2 specificity by facilitating these structural transitions in the RAG1/2 active site. To test this hypothesis and better characterize the DNA sequence specificity of RAG1/2, we modified an episomal-based VDJ recombination assay to quantify RAG1/2 activity in an unbiased approach using a large sequence-diverse set of RSSs. While the consensus heptamer sequence was highly preferred, RAG1/2 showed similarly high activity on many different sequences. Notably, purine/pyrimidine (R/Y) content of RSS heptamers was a strong predictor of RAG1/2 activity, preferring alternating R/Y sequence motifs at nucleotide positions 5-7 of the RSS heptamer. RAG1/2 also favored A+T base-pairs in the first two positions of the RSS spacer. These data show RAG1/2 is highly active on many DNA sequences deviating from the consensus RSS, and R/Y content of RSS heptamers is a good predictor of RAG activity. Molecular dynamics simulations of RSSs using the parmbsc1 force field showed that DNA sequences preferred by RAG1/2 had unique twist distributions, particularly between nucleotide positions 5-7. In conclusion, these data support our hypothesis that RAG1/2 can recognize many different DNA sequences, and sequence-specific structural features of alternating R/Y nucleotides promote RAG1/2 cleavage. This mechanism of promiscuously selecting sequences for VDJ recombination yields a diverse antigen receptor repertoire, but it also highlights the constant threat of off-target VDJ recombination events. Further investigation of RAG1/2 specificity will help elucidate the genetic instructions guiding VDJ recombination to antigen receptor gene segments.