| Abstract | | Background: A large fraction of camelid (camels and llamas) antibodies is composed of heavy chain-only homodimers, able to recognise antigens with their variable domain. Events in somatic assembly and maturation of antibodies such as hypermutations and rearrangement of variable loops (CDRs - complementary determining regions) and selection among a wide range of framework variants are generally considered to be random processes. Methods: An original algorithmic approach (Global Sequence Signature-GSS) was developed, able to take into account multiple functional and/or local sequence properties to detect scattered evolutionary constraints into sequences. Results: Using the GSS approach, we show that the length of the main hypervariable loop (CDR3) is linked to the nature of 19 surrounding residues on the scaffold. Surprisingly, the relation between CDR3 size and scaffold residues strongly depends on the considered species, illustrating either significant differences in selection mechanisms or functional constraints during antibody maturation. Conclusions: Combined with the statistical coupling analysis (SCA) approach at the level of scaffold residues, this study has unravelled a robust interaction network on antibody structure surrounding the CDR3 loop. General significance: In addition to the general applicability of the GSS algorithm, which can bring together functional and sequence data to locate hot spots of constrained evolution, the relationship between CDR3 and scaffold discussed here should be taken into account in protein engineering when designing antibody libraries.
|