! SUMMARY Errors made by the RAG1/RAG2 endonuclease during V(D)J recombination can lead to genome instability and the development of leukemia and lymphoma. One important cause of such instability is improper action of RAG at "cryptic" recombination signal sequences (RSSs) that are present abundantly in the genome. Our prior work provided mechanistic and structural insights into RAG protein-DNA complexes and revealed that RAG1 and RAG2 bind to numerous sites in the genome outside of the antigen receptor loci, virtually all of which are active promoters or enhancers. We further demonstrated that such "off-target" RAG binding occurs through two modes, one driven by a histone code "reader" function of RAG2 (largely at promoters) and the other dependent on regulatory portions of RAG1 (largely at enhancers). The thousands of off-target RAG binding sites and the millions of cryptic RSSs in the genome raise fundamental questions about how the genome is protected from devastating instability caused by RAG. The central objective of our proposed experiments is to determine the rules that govern RAG off-target activity by understanding the interactions that dictate RAG localization in the genome and the mechanisms that determine which cryptic RSSs are cleaved by RAG and which are spared. We will use complementary biochemical, biophysical, genetic, and genomic approaches to achieve the following aims: Aim 1. Determine the rules governing the selection of cryptic RSS targets by RAG. We and others have identified intriguing sequence and topological features of the cryptic RSSs cleaved by RAG, but it is unknown to what extent, or how, these features dictate RAG activity. We will systematically determine how cryptic RSS orientation, location, and sequence influence RAG off-target activity and in the process, test a provocative new hypothesis that RAG acquires its targets in part through linear "tracking" along DNA. Aim 2. Determine the domains and interactions directing RAG1 to active enhancers. Much off-target cutting by RAG occurs in enhancers, but the molecular interactions that dictate localization of RAG to these regions are not known. We will identify the portions of RAG1 and the RAG1-binding partners that specify enhancer binding and test the hypothesis that RAG1, like RAG2, is a reader of the histone code. Aim 3. Determine how retargeting of RAG alters cRSS selection and off-target cleavage patterns. We will use RAG mutants and a novel, inducible in vivo targeting system to direct RAG to new sites in the genome and determine the spectrum of new cleavage sites that arise. We will also reconstitute RAG- mediated cleavage at "strong" cryptic RSSs normally ignored by RAG to uncover the epigenetic and chromatin architectural parameters that specify RAG off-target activity. Together, our proposed studies have a dual significance, both for basic mechanisms of RAG function and for the causes of cancer.