Directed evolution of an aspartate aminotransferase with new substrate specificities

Abstract

The substrate specificity of aspartate aminotransferase was successfully modified by directed molecular evolution using a combination of DNA shuffling and selection in an auxotrophic Escherichia coli strain. After five rounds of selection, one of the evolved mutants showed a 10(5)-fold increase in the catalytic efficiency (kcat/Km) for beta-branched amino and 2-oxo acids and a 30-fold decrease in that for the native substrates compared with the wild-type enzyme. The mutant had 13 amino acid substitutions, 6 of which contributed 80-90% to the total effect. Five of these six substitutions were conserved among the five mutants that showed the highest activity for beta-branched substrates. Interestingly, only one of the six functionally important residues is located within a distance of direct interaction with the substrate, supporting the idea that rational design of the substrate specificity of an enzyme is very difficult. The present results show that directed molecular evolution is a powerful technique for enzyme redesign if an adequate selection system is applied.

Figure 1

Sequence analysis of the evolved AspATs that have high activity for 2-oxovaline. (A) Lines show the entire coding region of the aspC gene. Longer bars indicate the nucleotide positions of missense mutations, which caused amino acid substitutions, and shorter bars indicate those of silent mutations. Asterisks are the missense mutations conserved in all five mutant AspATs. (B) Amino acid substitutions. Residues were numbered according to the sequence of cytosolic AspAT from pig as described (27).

Figure 2

Stabilization of the activation free energy for various amino acid substrates by AV5A-7, relative to the wild-type AspAT. ΔΔG_T^‡ = RTln{(k_cat/K_m)_AV5A-7/(k_cat/K_m)_WT}. Substrate amino acids are shown by three-letter abbreviations. Positive bars indicate that the catalytic efficiency of AV5A-7 is higher than that of the wild-type AspAT, and negative bars indicate vice versa. Exact values could not be determined for valine and isoleucine; therefore, ΔΔG_T^‡ values for their 2-oxo acids are shown for these amino acids (asterisks).

Figure 3

Stereo representation of the structure of the wild-type E. coli AspAT complexed with 2-methyl-l-aspartate (16). The side chains of the six functionally important residues mutated in this study are shown by full bonds (Asn³⁴, Ile³⁷, Ser¹³⁹, Asn¹⁴², Asn²⁹⁷, and Val³⁸⁷). The coenzyme pyridoxal 5′-phosphate, the substrate analog, and the side chains of Trp¹⁴⁰ and Arg³⁸⁶ are also shown (open bonds). Asn²⁹⁷ belongs to the other subunit of the dimer (asterisk).