Evolution of the RAG1-RAG2 locus: both proteins came from the same transposon

Abstract

The RAG1 and RAG2 proteins are essential subunits of the V(D)J recombinase that is required for the generation of the enormous variability of antibodies and T-cell receptors in jawed vertebrates. It was demonstrated previously that the 600-aa catalytic core of RAG1 evolved from the transposase of the Transib superfamily transposons. However, although homologs of RAG1 and RAG2 genes are adjacent in the purple sea urchin genome, a transposon encoding both proteins so far has not been reported. Here we describe such transposons in the genomes of green sea urchin, a starfish and an oyster. Comparison of the domain architectures of the RAG1 homologs in these transposons, denoted TransibSU, and other Transib superfamily transposases provides for reconstruction of the structure of the hypothetical TransibVDJ transposon that gave rise to the VDJ recombinases at the onset of vertebrate evolution some 500 million years ago.

Figure 1

TransibSU transposons in sea urchins and starfish. A: The RAG1-RAG2-like locus in the purple sea urchin genome. DECR (GenBank: XP_793296) and RHPN (GenBank: XP_785878) are neighbor genes flanking the RAG1-RAG2-like locus. Numbers indicate positions of SPRAG1L, SPRAG2L and their flanking genes in Scaffold70 (GenBank: NW_003577031). Scaffold616 (GenBank: NW_003577577) encodes the elongator complex protein 2 gene (Ecp2, conserved in all metazoans). Numbers indicate the beginning and the end of the coding exons (black rectangles) and position of the second intron (between the exons 2 and 3). B: In the green sea urchin L. variegates genome, a unit coding for both LVRAG1L and LVRAG2L is inserted into the second intron of Ecp2, this gene is orthologous to the purple sea urchin Ecp2. Both genomes contain a single copy of the Ecp2 gene. The green sea urchin LVRAG1L/LVRAG2L unit is assembled from two scaffolds. Scaffold positions corresponding to loci coding for Ecp2, LVRAG1L and LVRAG2L are shown above and the green (Scaffold19893) and black (Scaffold71) lines, respectively. The N-terminal part of LVRAG1L is lost (1-272-aa of SPRAG1), the core and C-terminal part is almost intact (it corresponds to pos. 437–978 of SPRAG1; disrupted by one stop codon). The vertical red line indicates the boundary between the LVRAG2L parts (aa positions 1–184 and 185–467), encoded by the two scaffolds due to incomplete assembly. C: The TransibSU transposons in the Bat star genome (Patiria minata, sea star or starfish). One copy of the TransibSU-1_PM transposon can be assembled from three contigs [GenBank: AKZP01053170-AKZP01054172]. The RAG2L protein in this copy was lost due to deletion of a region between the FL1 and FL2 parts (dark blue and brown arrows). The 94% identical copies of FL1 and FL2 constitute the termini of a 3734-bp contig [GenBank: AKZP01064647]. The central part of this contig encodes the PMRAG2L protein. This contig represents a part of the second copy of the TransibSU-1_PM transposon. A copy of the second TransibSU-2_PM transposon is present at the 3’ and 5’ termini of the AKZP01110315 (green line) and AKZP01110316 (blue line) contigs, which are assembled into one scaffold. This transposon is flanked by the CCAGG target site duplication (yellow ovals). Due to the sequencing problems, the internal part of this transposon is not complete. D: Termini of TransibSU, Transibs and V(D)J recombination signal sequences are shown. E: Commonly accepted phylogeny of species colonized by the TransibSU and TransibVDJ transposons. Magenta stars denote the presence of TransibSU transposons and the red stars denote the RAG1-RAG2 locus. The magenta oval indicates an unknown RAG1-RAG2-enocding transposon reported recently as a polymorphic insertion in a lancelet genome [24].

Figure 2

Phylogeny of Transib transposase core sequences. The Transib superfamily is composed of four clades: Transib, TransibSU, Chapaev and Chapaev3. The vertebrate RAG1 proteins cluster with the TransibSU clade. Species name abbreviations: HM - Hydra magnipapillata, NV - Nematostella vectensis, SP - Strongylocentrotus purpuratus, BF - Branchiostoma floridae, PM - Petromyzon marinus, CE - Caenorhabditis elegans, XL – Xenopus laevis, HS - Homo sapiens, FR - Takifugu rubripes, CL - Carcharhinus leucas, GG - Gallus gallus, AA - Aedes aegypti, AG - Anopheles gambiae, DP - Drosophila pseudoobscura, HR - Helobdella robusta, AC - Anolis carolinensis, ACa - Aplysia californica, Transib1-Transib5 are from Drosophila melanogaster. PMRAG1L is encoded by the Bat star TransibSU-1_PM transposon, SPRAG1L, ECRAG1L, AFRAG1L are TransibSU transposases from Strongylocentrotus purpuratus, Evechinus chloroticus sea urchin (encoded by the GenBank: GAPB01003278.1 transcribed RNA sequence) and Asterias forbesi starfish (GenBank: GAUS01036390.1 transcribed RNA sequence), respectively. In addition to Echinoderms, the TransibSU clade includes also a transposon from Pacific oyster (see Additional file 4; the oyster transposase core is ~40% identical to its homologues in Echinodermata). The tree was obtained using the PhyML: RtREV model, discrete gamma model with 10 categories and estimated gamma shape parameter, best of NNIs and SPRs tree topology search. Statistical support is indicated by aLRT-SH values above 95% [27]. Domain structure of RAG1 and transposases encoded by known members of the Transib superfamily was used to infer the domain structure of the transposase encoded by the TransibVDJ transposon, the hypothetical direct ancestor of the vertebrate RAG1-RAG2 locus. The losses of the CHAPA, RING and C2H2 domains in different clades that were inferred under the parsimony principle are indicated by “L:”.