Abstract There is increasing evidence that some proteins fold during translation, i.e. cotranslationally, which implies that partial protein function, including interactions with other molecules, could potentially be unleashed early on during translation. Although little is known about cotranslational assembly mechanisms, for homomeric protein complexes, translation by the ribosome, folding and assembly, should be well-coordinated to avoid misassembly in the context of polysomes. We analysed 3D structures of homomers and identified a statistically significant trend conserved across evolution that supports this hypothesis: namely that homomeric contacts tend to be localized towards the C-terminus rather than N-terminus of homomeric polypeptide chains. To probe this in more detail, we expressed a GFP-based library of 611 homomeric E. coli genes, and analyzing their folding and assembly in vivo . Consistent with our hypothesis, interface residues tend to be located near the N-terminus in cotranslationally aggregating homomers. In order to dissect the mechanisms of folding and assembly under controlled conditions, we engineered a protein library with three variable components: (i) the position and type homomerization domain, (ii) the reporter domain and (iii) the linker length that connects the two. By analyzing the misassembly rates of these engineered constructs in vivo , in vitro and in silico , we confirmed our hypothesis that C-terminal homomerization is favorable to N-terminal homomerization. More generally, these results provide a set of spatiotemporal constraints within polypeptide chains that favor efficient assembly, with implications for protein evolution and design.

Cotranslational assembly imposes evolutionary constraints on homomeric proteins