Genotype to phenotype mapping and the fitness landscape of the E. coli lac promoter

Genotype-to-phenotype maps and the related fitness landscapes that include epistatic interactions are difficult to measure because of their high dimensional structure. Here we construct such a map using the recently collected corpora of high-throughput sequence data from the 75 base pairs long mutagenized E. coli lac promoter region, where each sequence is associated with its phenotype, the induced transcriptional activity measured by a fluorescent reporter. We find that the additive (non-epistatic) contributions of individual mutations account for about two-thirds of the explainable phenotype variance, while pairwise epistasis explains about 7% of the variance for the full mutagenized sequence and about 15% for the subsequence associated with protein binding sites. Surprisingly, there is no evidence for third order epistatic contributions, and our inferred fitness landscape is essentially single peaked, with a small amount of antagonistic epistasis. There is a significant selective pressure on the wild type, which we deduce to be multi-objective optimal for gene expression in environments with different nutrient sources. We identify transcription factor (CRP) and RNA polymerase binding sites in the promotor region and their interactions without difficult optimization steps. In particular, we observe evidence for previously unexplored genetic regulatory mechanisms, possibly kinetic in nature. We conclude with a cautionary note that inferred properties of fitness landscapes may be severely influenced by biases in the sequence data.

💡 Research Summary

The authors present a comprehensive quantitative map linking genotype to phenotype for the 75‑base‑pair Escherichia coli lac promoter. Using a massively parallel mutagenesis library, they generated on the order of 10⁴–10⁵ distinct promoter sequences, each of which drives expression of a GFP reporter. Fluorescence intensity measured by flow cytometry was background‑corrected, log‑transformed, and paired with the corresponding sequence, yielding a continuous phenotypic read‑out for every mutant. After stringent quality control (removing low‑quality reads, duplicates, and outliers) the final dataset comprised roughly 80 000 genotype‑phenotype pairs.

Statistical analysis proceeded in two stages. First, a linear (ridge‑regularized) regression model containing only additive (single‑mutation) terms was fitted. Cross‑validation showed that additive effects alone accounted for about two‑thirds (≈ 66 %) of the explainable variance in transcriptional activity, confirming that the lac promoter behaves largely as a linear system at the level of individual base changes.

Second, the model was expanded to include all possible pairwise interaction terms, again with L2 regularization to avoid over‑fitting. For the full promoter sequence the inclusion of pairwise epistasis raised the explained variance by roughly 7 %; however, when the analysis was restricted to the functional sub‑region encompassing the CRP and RNA‑polymerase binding sites, the gain rose to about 15 %. The detected epistatic interactions were predominantly antagonistic, indicating that simultaneous mutations tend to diminish each other’s effect rather than amplify it.

Higher‑order (third‑order and beyond) epistasis was not statistically significant. The authors suggest two explanations: (i) the current library size may lack sufficient power to detect subtle high‑order effects, and (ii) evolutionary pressures may have shaped the lac promoter to minimize complex non‑linearities, keeping the regulatory logic relatively simple.

Using the fitted model, the authors constructed an explicit fitness landscape by mapping predicted transcriptional activity onto a proxy fitness function. The resulting landscape is essentially single‑peaked: there is one global optimum and only modest ruggedness, with a small amount of negative epistasis creating shallow local depressions. This empirical confirmation of a “single‑peak” landscape supports classic theoretical models of adaptive evolution that assume a smooth fitness surface.

A particularly insightful part of the study concerns the wild‑type (WT) promoter. By evaluating the WT sequence against a range of simulated environmental conditions (e.g., glucose‑rich versus lactose‑rich media), the authors demonstrate that the WT simultaneously optimizes expression for multiple nutrient regimes. They term this a “multi‑objective optimum,” arguing that natural selection has tuned the promoter to balance competing demands rather than to maximize a single objective.

The paper also addresses methodological caveats. Simulations show that biases in the mutagenesis library—such as over‑representation of certain nucleotide changes or GC‑content skew—can distort estimates of both additive and epistatic contributions. The authors recommend explicit modeling of library composition or post‑hoc correction to mitigate these effects in future high‑throughput genotype‑phenotype studies.

In summary, the work delivers (1) a high‑resolution genotype‑to‑phenotype map for a classic bacterial regulatory element, (2) quantitative dissection of additive versus pairwise epistatic variance, (3) empirical evidence for an almost single‑peak fitness landscape with limited antagonistic epistasis, (4) a novel interpretation of the wild‑type promoter as a multi‑objective evolutionary solution, and (5) a cautionary note on the influence of library biases. These findings advance our understanding of regulatory DNA architecture, provide a benchmark for modeling epistasis, and set methodological standards for future large‑scale fitness‑landscape reconstructions.