Functional requirement of terminal inverted repeats for efficient ProtoRAG activity reveals the early evolution of...

The discovery of ProtoRAG in amphioxus indicated that vertebrate RAG recombinases originated from an ancient transposon. However, the sequences of ProtoRAG terminal inverted repeats (TIRs) were obviously dissimilar to the consensus sequence of mouse 12/23RSS and recombination mediated by ProtoRAG or RAG made them incompatible with each other. Thus, it is difficult to determine whether or how 12/23RSS persisted in the vertebrate RAG system that evolved from the TIRs of ancient RAG transposons. Here, we found that the activity of ProtoRAG is highly dependent on its asymmetric 5'TIR and 3'TIR, which are composed of conserved TR1 and TR5 elements and a partially conserved TRsp element of 27/31 bp to separate them. Similar to the requirements for the recombination signal sequences (RSSs) of RAG recombinase, the first CAC in TR1, the three dinucleotides in TR5 and the specific length of the partially conserved TRsp are important for the efficient recombination activity of ProtoRAG. In addition, the homologous sequences flanking the signal sequences facilitate ProtoRAG- but not RAG-mediated recombination. In addition to the diverged TIRs, two differentiated functional domains in BbRAG1L were defined to coordinate with the divergence between TIRs and RSSs. One of these is the CTT* domain, which facilitates the specific TIR recognition of the BbRAGL complex, and the other is NBD*, which is responsible for DNA binding and the protein stabilization of the BbRAGL complex. Thus, our findings reveal that the functional requirement for ProtoRAG TIRs is similar to that for RSS in RAG-mediated recombination, which not only supports the common origin of ProtoRAG TIRs and RSSs from the asymmetric TIRs of ancient RAG transposons, but also reveals the development of RAG and RAG-like machineries during chordate evolution.