Figure 2. Accessibility of adaptive mutations from each starting point.

(A) Additional screenings of NDM1 and VIM2 round one libraries. Activities of the top 30 variants isolated from additional screenings of three independently generated mutagenized libraries and normalized on the PMH fitness of NDM1-WT and VIM2-WT, respectively; presented as scatterplots. Mutants are labeled when variants possess a mutation previously encountered in the original directed evolution or if this mutation is located at the same position. Asterisks indicate that the variant has other mutations in addition to the labeled one. Mutations and activities of all improved variants are listed in Supplementary file 1H. The activities of all 2160 variants in the round one additional screenings are presented in Figure 2—figure supplement 1. (B) Epistasis analysis of mutations. Mutations occurring in the NDM1 and VIM2 trajectories were introduced into their respective wild-type sequence and the resulting PMH fitness was compared to the original PMH fitness change observed during the trajectory. (C) Fold change in PMH fitness induced by mutation W93G in NDM1-WT and VIM2-WT, respectively. (D) Fold change in PMH fitness induced by mutation of Trp93 in VIM2 to various hydrophobic residues. A full description of the lysate activities is presented in Supplementary file 1G. Errors bars represent the propagated standard deviation from triplicate measurements.

Figure 2—figure supplement 1. Screening of additional libraries.

Three additional mutagenized libraries were generated by performing error-prone PCR and cloning to the expression vector from each wild-type gene. The three libraries were transformed into E. coli BL21 (DE3), and pre-screened with 4 μg/ml ampicillin on agar plates. (A, B) 720 variants from each library (in total 2160 variants) were picked from the agar plates, and the cell lysate activity of variants was measured. The activity of variants is normalized to the wild-type enzyme. (C, D) Top 30 variants from each library were picked and re-assayed with triplicate. Mutations of positive variants are presented in Supplementary file 1G.

Figure 2—figure supplement 2. Differential effects of mutations (W93G and V72A) on six metallo-β-lactamases.

Fold change in PMH and β-lactamase activity of variants. (A) W93G and (B) V72A variants compared to the variants containing a Trp93 and Val72. Activity levels of purified enzymes were measured at single enzyme (5 μM for PMH and 1 nM for β-lactamase activity) and substrate (500 μM for PMH and 100 μM for β-lactamase activity) concentrations. The wild-type NDM1 contains an Ala at position 72; the activity presented is the ratio of the wild-type to the A72V variant. The wild-type EBL1 contains a Pro at position 72; the activity presented is the ratio of the P72A to the P72V variant. Errors bars represent the standard deviation calculated from three measurements. Changes in melting temperature (T_m) of (C) W93G and (D) V72A. Black squares denote the wild-type enzyme, and grey circles denote the mutant. Arrows indicate the change of melting temperature by the mutation. The melting temperature was calculated from the midpoint of the thermal denaturation curve of purified proteins. Asterisks indicate that the fold change in activity and change in T_m could not be determined, because one of the variants was not soluble. The catalytic activities and T_m of each mutant are listed in Supplementary file 1H.