Despite several similarities in structure and kinetic behavior, the bacterial and vertebrate forms of the enzyme dihydrofolate reductase (DHFR) exhibit differential specificity for folate. In particular, avian DHFR is 400 times more specific for folate than the Escherichia coli reductase. We proposed to enhance the specificity of the E. coli reductase for folate by incorporating discrete elements of vertebrate secondary structure. Two vertebrate loop mutants, VLI and VLII containing 3-7 additional amino acid insertions, were constructed and characterized by using steady-state kinetics, spectrofluorimetric determination of ligand equilibrium dissociation constants, and circular dichroism spectroscopy. Remarkably, the VLI and VLII mutants are kinetically similar to wild-type E. coli reductase when dihydrofolate is the substrate, although VLII exhibits prolonged kinetic hysteresis. Moreover, the VLI dihydrofolate reductase is the first mutant form of E. coli DHFR to display enhanced specificity for folate [(kcat/Km)mutant/(kcat/Km)wt = 13]. A glycine-alanine loop (GAL) mutant was also constructed to test the design principles for the VLI mutant. In this mutant of the VLI reductase, all of the residues from positions 50 to 60, except the strictly conserved amino acids Leu-57 and Arg-60, were converted to either glycine or alanine. A detailed kinetic comparison of the GAL and wild-type reductases revealed that the mutations weaken the binding by both cofactor and substrate by up to 20-fold, but under saturating conditions the enzyme exhibits a kcat value nearly identical to that of the wild type. The rate of hydride transfer is reduced by a factor of 30, with a compensating increase in the dissociation rate for tetrahydrofolate. Although key stabilizing interactions have been sacrificed (it shows no activity toward folate), the maintenance of the correct register between key residues preserves the activity of the enzyme toward its natural substrate. Collectively, neither specific proximal point site mutations nor larger, more distal secondary structural substitutions are sufficient to confer a specificity for folate reduction that matches that observed with the avian enzyme. This is consistent with the hypothesis that the entire protein structure must contribute extensively to the enzyme's specificity.