Jeet Guram
Dr. Ely
Biology 303 – Genetics
8 November 2007
The Significance of Epistasis in Protein Evolution
Intragenic interactions in which the effects of one gene or set of genes influence the effects of another are referred to as epistasis. This phenomenon is of particular relevance to the field of evolutionary genetics. For example, a specific mutation may affect an individual’s fitness in unforeseen ways if the gene on which the mutation acts influences other genes.
Shimon Bershtein and colleagues explored the relationship between epistasis and mutational effects in a study involving TEM-1 β-lactamase, an enzyme conferring resistance to the antibiotic ampicillain.1 In their study, the TEM-1 β-lactamase gene was cloned into a plasmid and transformed into E. coli host cells. The cells were placed in one of three growth media: one population was exposed to no ampicillin (control group), another was exposed to low levels of ampicillin (12.5 µg ml-1 – the “12.5 group”), and the third was exposed to high levels (250 µg ml-1 – the “250 group”).1 After allowing for growth in the three separate media, the plasmid DNA was extracted from the host cells. Random in vitro mutagenesis was then run on the DNA, with an average of two mutations per gene per round of mutagenesis. The resulting mutant DNA was ligated into empty vectors and re-transformed into host cells for another round of growth. The process was repeated, with DNA being extracted only from surviving bacteria, for a total of ten rounds of mutagenesis. Each group was separately mutated.
The researchers measured the effects of the accumulating deleterious mutations on the fitness of the TEM-1 β-lactamase enzyme. Fitness was calculated by determining the fraction of mutants that conferred ampicillin resistance after each successive round of mutagenesis; cells containing mutants not conferring resistance died. The number of cumulative mutations was also measured. Two mathematical models were established to quantify the effects of the mutations on fitness. The first model was designed under the premise that the individual mutations did not interact through epistasis, and that their effects on fitness were merely additive. The second model assumed that the mutations did interact, so fitness would decline more rapidly as mutations accumulated.
Once the rounds of mutagenesis were completed and data had been gathered, the researchers found that the fitness plots correlated much more closely with the second mathematical model. For example, during the first five rounds of mutagenesis, an average of 3.6 non-synonymous mutations had accumulated in each surviving member of the 12.5 group.1 During the remaining five rounds, only 1.9 mutations were added in surviving mutants.1 Thus, the tolerance to these deleterious mutations decreased sharply as mutations were accumulated.
Bershtein’s team drew a second relevant conclusion. The 12.5 group initially tolerated far more mutations than did the 250 group. This result was expected since the 250 group was growing in significantly harsher conditions. After two rounds of mutagenesis, the 12.5 group had accumulated a total of 52 mutations while the 250 group had accumulated only 24.1 However, by the end of the experiment, both groups had accumulated approximately the same number of mutations—171 and 170, respectively—even though one group had been exposed to 20 times as much of the antibiotic.1 The team concluded that the enzyme’s fitness remained relatively unaffected under a “threshold” number of mutations before dropping essentially to zero at this value. The existence of such a “threshold” inherently supports the role of epistasis in modulating the effects of mutations; later mutations interacted with earlier ones to create a complex, joint effect.
By showing that individual mutations interact with each other so significantly, the study strongly suggested that epistasis plays a role in protein evolution. Evolution acts on favorable phenotypes; inheritable phenotypes that increase individual fitness rise in frequency over time within a population.3 The genotypes encoding such phenotypes also increase in incidence.
Jesse Bloom and colleagues analyzed a possible instance of epistasis influencing protein evolution.2 Bloom’s group asked if extra protein stability increased the ability of proteins to evolve useful new functions. Evolution does not select for extra protein stability; once a protein reaches the minimum stability level necessary, there is no selective pressure for additional stability since this addition would offer no immediate advantage. However, mutations conferring additional stability could make the protein more receptive to the conformational changes demanded by mutations conferring new functions.
The researchers first ran simulations using electronic modelhypothetical lattice proteins. These model proteins are extremely simplified but useful for studying protein evolution. Two model proteins were created; both could bind a certain ligand equally well. The first model protein had a Gibbs Free Energy of -0.5.2 The second protein was more stable, with a Free Energy of -1.5.2 Researchers then identified a series of mutations that would enable the proteins to bind to new ligands. These mutations were applied to the genetic sequences of both proteins, and the results of these applications are presented in Figure 1 below.2 The bars show the distribution of the stabilities of the mutant proteins, while the circles indicate the stabilities of mutants with newly evolved functions.
The mutations conferring additional functions did, in fact, tend to have a destabilizing effect, so the second protein was more robust to helpful mutations. The team thereby proposed a potential role of epistasis in protein evolution. A mutation conferring additional stability could potentially be involved in the evolution of a new function, even though the actual function is brought about by a second mutation influencing vastly different genes than those determining stability.
Bloom’s group then ran a second model, this one on actual physical proteins. Two laboratory-evolved variants of the cytochrome P450 superfamily of heteroproteins were used: P450 21B3 and P450 5H6. Both proteins catalyze subterminal hydroxylation of medium and long-chain fatty acids. The 5H6 derivative, however, is substantially more stable than the other. As with the model lattice proteins, the researchers determined a series of mutations that would confer additional abilities to the proteins. Thousands of different mutations (each designed to enable the protein to act on one of five new ligands) were applied to the two derivatives. Again, the stabilized protein much more readily tolerated the necessary mutations, as is shown in figure 2 below.2 The counts above the bars in the graph indicate the number of improved mutants over the total number of different mutants created for each target ligand.
Figure 2
The results of Bloom’s study bring up some interesting questions. Bloom may have been able to create a laboratory model of epistatic mutations, but is there any case of a protein actually having evolved a new function via epistatic stabilization mutations that were neutral with respect to natural selection? If so, does the role of epistasis make evolutionary trajectories dependent on chance events?
A remarkable study by Eric Ortlund and colleagues shed light on these issues.5 Ortlund’s team began by noticing similarities in two hormone receptors present in all jawed vertebrates: glucocorticoid receptors (GR) and mineralocorticoid receptors (MR). GRs bind exclusively to the adrenal steroid cortisol and modulate blood glucose levels. MRs bind mainly to aldosterone and modulate salt and electrolyte homeostasis, but they also have a weak affinity for cortisol.
By analyzing an exhaustive database of genetic sequences of extant GRs and MRs, the researchers were able to “resurrect” the ancestral parent from which these related proteins descended, the “ancestral corticoid receptor” or AncCR. In determining the genetic sequence of AncCR, researchers applied the parsimony principle and determined the maximum likelihood phylogenetic sequence of this protein. AncCR was then expressed in vivo, and its functions were analyzed. The ancestral protein was found to have binding affinities similar to those of MRs. The researchers concluded that GR’s cortisol specificity must be evolutionarily derived.
In determining how this specificity evolved, Ortlund’s team pinpointed the specific genetic mutations that occurred in the protein sequence around the time when the switch to GR function occurred—between 420 and 440 million years ago.5 The team was able to determine 37 local mutations that occurred during this period.5 Two of the 37 had already been established as increasing cortisol specificity, so the researchers applied this pair of mutations (mutations group “X”) to AncGR1—an ancestor of the GR that was a descendent of AncCR but still lacked cortisol specificity.5 The mutations were introduced into the AncGR1 genetic sequence, and the resulting protein was expressed in vivo. As expected, increased affinity of the receptor for cortisol was observed.
Three other mutations were suspected to play a role in establishing specificity, since these five mutations (these three plus the two in X) are collectively conserved in all modern strains of GRs descending from AncGR1. This other group of three mutations (mutations group “Y”) was introduced next. Surprisingly, these additional mutations rendered the protein completely nonfunctional. This result led the team to hypothesize that epistatic mutations that conferred additional stability to the protein were introduced along with X and Y.
The identity of these subtle epistatic mutations was determined by analyzing the protein structure. Since these mutations did not directly affect the affinity of the protein for cortisol, they were not included in the 37 previously identified mutations.5 The team determined the most likely sequence of stabilizing or “permissive” mutations (mutations group “Z”). When applied by itself to AncGR1, Z did not have any significant impact on hormone binding sensitivity. When applied in conjunction with X and Y, however, the resulting protein had a cortisol-specific sensitivity. Figure 3 below shows various combinations of mutations applied to AncGR1 and their effects on hormone sensitivity.5 95% confidence intervals are included in parenthesis. From these results, the team concluded that mutation sequences X, Y, and Z acted in tandem in the evolution of the GR.
Figure 3
By identifying the ancestral predecessor of two proteins and charting the sequence along which this ancestor evolved, Ortlund’s team was able to prove that epistatic structural mutations have played a role in the evolution of protein functions. This process seems to contradict conventional evolutionary thought, since the epistatic mutations that were so central to the process conferred no individual increase in fitness.
So is evolution dependent on chance? As Ortlund explains, “If there are few potentially permissive substitutions, and these are nearly neutral, then whether they will occur is largely a matter of chance. If the historical ‘tape of life’ could be played again, the required permissive changes might not happen, and a ridge leading to a new function could become an evolutionary road not taken.”5
References
1.Bershtein, Shimon, Michal Segal, Roy Bekerman, Nobuhiko Tokuriki, and Dan S. Tawfik. "Robustness-Epistasis Link Shapes the Fitness Landsape of a Randomly Drifting Protein." Nature 444 (2006): 929-932. 7 Nov. 2007 <
2. Bloom, Jesse D., Sy T. Labthavikul, Christopher R. Otey, and Frances H. Arnold. "Protein Stability Promotes Evolvability." Proceedings of the National Academy of Sciences of the United States of America 103.15 (2006): 5869-5874. 7 Nov. 2007 <
3. Campbell, Neil A., and Jane B. Reece. Biology. 7th ed. San Francisco, CA: Pearson Education, Inc., 2005.
4. Klug, William S., Michael R. Cummings, and Charlotte A. Spencer. Concepts of Genetics. 8th ed. Upper Saddle River, NJ: Pearson Education, Inc., 2006.
5. Ortlund, Eric A., Jamie T. Bridgham, Matthew R. Redinbo, and Joseph W. Thornton. "Crystal Structure of an Ancient Protein: Evolution by Conformational Epistasis." Science 317 (2007): 1544-1548. 7 Nov. 2007 <