A statistical modelling reveals bi-directional chromatin scanning by RAG in the human TCR-α locus

Rearrangements of germline variable (V), diversity (D), and joining (J) coding gene segments generate a diverse T cell receptor (TCR) repertoire. Two different molecular mechanisms govern a stochastic nature of gene choices by the recombination activating gene (RAG) protein complex. One is diffusion access in which a chromatin loop containing a coding gene bounces back and forth until the loop encounters a distant gene. The other is a linear chromatin scan, which extrudes chromatin loops to assemble a gene pair in a distance-dependent manner. However, the extent to which those two mechanisms underlie V(D)J recombination in human TCRs still remains unclear. Applying statistical modelling to TCR sequence data, we infer the stochastic process of gene choices during this recombination. The inferred usage of V and J gene segments vary from gene to gene. This pattern of gene usage is consistent across individuals, suggesting a fundamental bias in the gene choice. Our modelling shows the dependency of VJ pairing on their linear distance in the TCR- locus; proximal-proximal and distal-distal VJ pairs are more preferred than proximal-distal or distal-proximal gene pairs. These results support bi-directional RAG scanning into proximal and distal sides of the TCR- V gene locus.