Natural and Synthetic Suppressor Mutations Defy Stability-Activity Tradeoffs

Abstract

Thermodynamic stability represents one important constraint on protein evolution, but the molecular basis for how mutations that change stability impact fitness remains unclear. Here, we demonstrate that a prevalent global suppressor mutation in TEM β-lactamase, M182T, increases fitness by reducing proteolysis in vivo. We also show that a synthetic mutation, M182S, can act as a global suppressor and suggest that its absence from natural populations is due to genetic inaccessibility rather than fundamental differences in the protein's stability or activity.

Figure 1

Overlay of TEM-1 (PDB 1BTL, shown in gray) and the extended spectrum variant TEM-52 (PDB 1HTZ, shown in blue). Position 238 lines the active site and is separated from position 182 by ∼18 Å (shown in yellow spheres).

Figure 2

Denaturant melts of TEM variants as detected by (A) CD and (B) intrinsic fluorescence. At least three states are present at equilibrium as indicated by CD, but only two are detectable by fluorescence. Because the intermediate and unfolded states have the same fluorescence signal, these melts highlight the N-to-I transition. All substitutions studied impact only this transition, which means that they impact the native state stability but not stability of the intermediate. Wild-type data appear in gray, closed circles, G238S in black, open circles, G238S/M182T in blue, open squares, and G238S/M182S in red, open triangles.

Figure 3

Immunoblots of E. coli expressing TEM variants. TEM levels in the periplasm and insoluble fraction were quantified using immunoblotting. Whole cell extracts were prepared by boiling in sample buffer, and periplasmic and insoluble fractions were prepared as described in the Materials and Methods. Purified wild-type TEM was loaded as a standard at two dilutions (Std. 1 and Std. 2). G238S is much less abundant than the wild type, and the presence of M182T or M182S increased abundance in all contexts.

Figure 4

Correlations between abundance and stability in the (A) native expression and (B) overexpression strains. Soluble, periplasmic TEM abundance, as quantified by immunoblotting, has a modest correlation with stability when expression levels are native-like, but no correlation when overexpressed in a strain lacking Lon and OmpT proteases.

Figure 5

Pulse proteolysis of wild-type TEM (black circles) overlaid with the fit to intrinsic fluorescence denaturation data (gray line). Samples are equilibrated in varying concentrations of urea and treated with thermolysin for a short pulse designed to leave the native state intact. Close agreement with intrinsic fluorescence, which tracks the N-to-I transition, demonstrates that I has similar proteolytic susceptibility to U.

Figure 6

Correlations between in vitro activities and MICs for (A) BP and (B) CFX. k_cat and k_cat/K_m correlate with MICs for BP and CFX, respectively, suggesting that an enzyme’s intrinsic activity for its substrate is predictive of its fitness.