Directed Evolution of Aminoglycoside Phosphotransferase (3′) Type IIIa Variants That Inactivate Amikacin but Impose Significant Fitness Costs

The rules that govern adaptive protein evolution remain incompletely understood. Aminoglycoside aminotransferase (3′) type IIIa (hereafter abbreviated APH(3′)-IIIa) is a good model enzyme because it inactivates kanamycin efficiently; it recognizes other aminoglycoside antibiotics, including amikacin, but not nearly as well. Here we direct the evolution of APH(3′)-IIIa variants with increased activity against amikacin. After four rounds of random mutation and selection in Escherichia coli, the minimum inhibitory concentration of amikacin rose from 18 micrograms/mL (wild-type enzyme) to over 1200 micrograms/mL (clone 4.1). The artificially evolved 4.1 APH(3′)-IIIa variant exhibited 19-fold greater catalytic efficiency (k (cat)/K (M)) than did the wild-type enzyme in reactions with amikacin. E. coli expressing the evolved 4.1 APH(3′)-IIIa also exhibited a four-fold decrease in fitness (as measured by counting colony forming units in liquid cultures with the same optical density) compared with isogenic cells expressing the wild-type protein under non-selective conditions. We speculate that these fitness costs, in combination with the prevalence of other amikacin-modifying enzymes, hinder the evolution of APH(3′)-IIIa in clinical settings.