Protein stability promotes evolvability

Abstract

The biophysical properties that enable proteins to so readily evolve to perform diverse biochemical tasks are largely unknown. Here, we show that a protein's capacity to evolve is enhanced by the mutational robustness conferred by extra stability. We use simulations with model lattice proteins to demonstrate how extra stability increases evolvability by allowing a protein to accept a wider range of beneficial mutations while still folding to its native structure. We confirm this view experimentally by mutating marginally stable and thermostable variants of cytochrome P450 BM3. Mutants of the stabilized parent were more likely to exhibit new or improved functions. Only the stabilized P450 parent could tolerate the highly destabilizing mutations needed to confer novel activities such as hydroxylating the antiinflammatory drug naproxen. Our work establishes a crucial link between protein stability and evolution. We show that we can exploit this link to discover protein functions, and we suggest how natural evolution might do the same.

Fig. 1.

Increased stability enhances evolvability of a model lattice protein. (A) The original model protein (right side) that had been evolved to bind to a rigid ligand (left side, in bold). (B) Mutants of a stabilized model protein were more likely than mutants of the original model protein to show improved binding to the four new ligands shown below the bars. The bars show the number of mutants of 1,500 screened that bound the new ligand with at least twice the affinity of the parent. (C) The stabilized model protein was more evolvable because more of its destabilized but improved mutants satisfied the minimal stability cutoff. The bars show the distribution of stabilities among all mutants in the libraries, and the circles show the stabilities of the improved mutants.

Fig. 2.

Distribution of mutations in the two P450 error-prone PCR libraries. (A) The distribution of mutations among 20 randomly chosen 21B3 mutants and 21 randomly chosen 5H6 mutants. The distributions are statistically indistinguishable (P = 0.84). (B) The distribution of mutations among all 41 sequenced mutants is consistent with the theoretical prediction for an error-prone PCR library (lines) (49, 50) (P = 0.11). (C) The mutations are uniformly distributed along the gene (P = 0.66). The lines show the cumulative fraction of mutations that occur at or before that position in the gene. All P values are from Kolmogorov–Smirnov tests (53) and represent the probability that the samples or theoretical curves would differ by at least this much if they were generated by the same underlying distribution.

Fig. 3.

Increased stability enhances evolvability of the P450 BM3 heme domain. (A) The stable 5H6 protein yielded more mutants with new or improved activity than the marginally stable 21B3 protein. The counts above the bars give the number of improved mutants of the total number of mutants screened. (B) Some of the improved 5H6 mutants were greatly destabilized relative to the parent protein, whereas the stabilities of the improved 21B3 mutants clustered around those of the parent protein (circles show T₅₀ values for improved mutants).

Fig. 4.

The functionally innovative but destabilizing L75R mutation can only be tolerated by the stabilized parent. (A) Leucine 75 is positioned close to the substrate in the hydrophobic P450 BM3 substrate-binding pocket (30). Mutating L75 to a positively charged arginine conferred naproxen activity on the stabilized 5H6 parent but disrupted the proper folding of the marginally stable 21B3 parent. (B) The antiinflammatory drug naproxen, which contains a negatively charged carboxylic acid group.