FITNESS LANDSCAPE OF A BIOPOLYMER PARTICIPATING IN A MULTISTEP REACTION

A biopolymer such as an enzyme participates in a multi-step reaction. When we take such a biopolymer as a target of evolutionary molecular e ngineering, the problem of frustration or interference among partial-f itness landscapes for elementary reaction steps arises. We examined th e effect of the frustration on a fitness landscape for the whole react ion. From the results of a previous paper we assumed that each partial -fitness landscape is of the Mt, Fuji-type, based on the mutational ad ditivity. Computer simulation for two-step or three-step reactions sho wed that the fitness landscape for a whole reaction was of the rough M t. Fuji-type, in which local optima were located near the top. However , there were fitter double-point mutants around each local optimum. Th is conclusion was supported by the result of the mathematical analysis of a simplified landscape for an L-step reaction: even if there are l ocal optima on the fitness landscape, several fitter sequences exist a t the Hamming distance of L from every local optimum in almost all cas es. Fitness distributions around the global optimum also showed that t he fitness landscape can be approximated to a Mt. Fuji-type landscape. (C) 1998 Academic Press Limited.