Sequence divergence of coiled coils—structural rods, myosin filament packing, and the extraordinary conservation of cohesins

Abstract

The amino acid sequences of the long, anti-parallel coiled coils of the cohesin subunits SMC1 and SMC3 are almost totally conserved in mammals. To understand this exceptional conservation more broadly, we analyzed amino acid sequence variation for several groups of coiled-coil proteins. Some long coiled coils, including giantin, NuMA, and Ndc80p/Nuf2p diverge ∼20% from humans to rodents, suggesting they function as spacer rods, whose sequence divergence is constrained only by the need to maintain the coiled-coil structure. Other coiled coils such as skeletal muscle myosin, intermediate filaments, and the lamins diverge only 1–3%. We suggest that this sequence divergence is constrained by the extensive packing contacts over the entire surface of the coiled-coil. The coiled coils of SMC5/6 and SMC2/4 (condensin) are slightly more constrained than the presumed spacer rods, diverging 10–15%. Conversely, the coiled coils of SMC1/3 (cohesin) diverge only 0.0–1.0%. This extreme constraint suggests that the entire surface of the coiled coil is intimately involved in the mechanism of sister chromatid cohesion. Direct binding of the coiled coils to chromatin, or perhaps the need to avoid such binding, are two possible mechanisms. Finally, analysis of the heptad repeat shows that the a and d positions are more constrained in spacer rods, and the bcefg positions more constrained in skeletal muscle myosin.