AUTHORS
Pablo Domizi, PhD, Cameron Baker, MS2, Tanxin Liu, MSc3, Milton Merchant4, Astraea Jager5, Yuxuan Liu, PhD6, Adam J de Smith, PhD7, Carol Fries Simpson, MD8, Kara L. Davis, D.O.9
AFFILIATIONS
1 Stanford University, Stanford, California, United States
2 University of Rochester, Rochester, New York, United States
3 University of Southern California, Los Angeles, California, United States
4 Stanford University, Stanford, California, United States
5 Stanford University, Stanford, California, United States
6 Stanford University, Stanford, California, United States
7 University of Southern California, Los Angeles, California, United States
8 University of Rochester, Rochester, New York, United States
9 Stanford University, Stanford, California, United States
ABSTRACT
Background: RAG1 and RAG2 form the recombinase complex essential for V(D)J recombination during normal B-cell development. Their expression is tightly coordinated and restricted to developmental stages undergoing immunoglobulin heavy- and light-chain rearrangement. RAG1 recognizes recombination signal sequences (RSSs) and catalyzes DNA cleavage, whereas RAG2 binds histone H3 lysine 4 trimethylation (H3K4me3), promoting recombinase activity at transcriptionally active chromatin. Consequently, excess RAG2 relative to RAG1 may increase the potential for off-target recombinase activity at non-immunoglobulin loci, contributing to DNA double-strand breaks and genomic instability. Consistent with this mechanism, aberrant RAG-mediated recombination is a well-established driver of oncogenic alterations in B-cell acute lymphoblastic leukemia (B-ALL). However, whether disruption of RAG1/RAG2 stoichiometry has prognostic significance in B-ALL or defines a distinct leukemic cell state remains unknown.
Methods: To define physiologic RAG1/RAG2 expression, we analyzed publicly available single-cell RNA-sequencing datasets from human fetal (n=8) and adult (n=45) bone marrow. Leukemic RAG1/RAG2 expression was evaluated in two independent single-cell B-ALL cohorts comprising an internal University of Rochester Medical Center cohort (n=5) and the Pediatric Cancer Atlas (n=88). Cells were classified according to their RAG1/RAG2 expression ratio as RAG1-biased, RAG1-prone, concordant, RAG2-prone, or RAG2-biased. Differential gene expression and pathway enrichment analyses were performed using pseudobulk EdgeR/Limma models adjusted for patient identity and molecular subtype. Within the MP2PRT cohort, whole-genome sequencing data (n=1,493) were analyzed to identify off-target RAG-mediated structural variants (SVs), defined as non-Ig/TCR rearrangements harboring recombination signal sequence (RSS) motifs at breakpoint junctions. Clinical significance was evaluated in two independent pediatric B-ALL bulk RNA-seq cohorts (MP2PRT, n=1,451; TARGET, n=282). Patients were stratified into RAG2-biased, concordant, or RAG1-biased groups using the 25th and 75th percentiles of the RAG1/RAG2 ratio, and event-free survival (EFS) was assessed using subtype-adjusted Cox proportional hazards models.
Results: During normal B-cell development, RAG1 expression is higher across most developmental stages, whereas RAG2 predominance was largely restricted to the large pre-B stage. In contrast, leukemic B cells exhibited expanded populations of RAG2-biased cells spanning both large and small pre-B developmental states. Compared with RAG1-dominant cells, RAG2-biased and RAG2-prone leukemic cells were enriched for hematopoietic stem and progenitor cell transcriptional programs with reciprocal depletion of mature B-cell identity signatures, consistent with reduced B-lineage commitment and increased developmental plasticity. Consistent with this phenotype, decreasing RAG1/RAG2 ratios were associated with an increased burden of off-target RAG-mediated SVs, linking RAG2 predominance to heightened genomic instability. Clinically, patients who relapsed were more than twice as likely to exhibit RAG2-biased disease than non-relapsing patients (OR 2.24, 95% CI 1.73–2.90, P=3.98×10⁻¹⁰). Consistent with these findings, a low RAG1/RAG2 ratio (<Q25) independently predicted inferior event-free survival (EFS) in both the MP2PRT and TARGET cohorts (adjusted HR>1, P<0.05). Furthermore, high-risk molecular subtypes exhibited significantly lower RAG1/RAG2 ratios than favorable-risk subtypes.
Conclusions: An imbalanced RAG1/RAG2 expression ratio identifies a reproducible high-risk subgroup of pediatric B-ALL across two independent clinical cohorts. Single-cell transcriptomic analyses demonstrate that RAG2-dominant leukemic cells exhibit immature, progenitor-like programs consistent with increased developmental plasticity, while whole-genome sequencing links RAG2 predominance to increased off-target RAG-mediated genomic instability. Together, these findings establish the RAG1/RAG2 ratio as a robust prognostic biomarker and implicate disrupted recombinase stoichiometry as a potential driver of aggressive leukemia biology and adverse clinical outcomes.
DISCLOSURES
Pablo Domizi, PhD: No relevant conflicts of interest to declare.