Sign epistasis caused by hierarchy within signalling cascades

Manjunatha Kogenaru,Sander J. Tans,Philippe Nghe

doi:10.1038/s41467-018-03644-8

Abstract

Sign epistasis is a central evolutionary constraint, but its causal factors remain difficult to predict. Here we use the notion of parameterised optima to explain epistasis within a signalling cascade, and test these predictions in Escherichia coli. We show that sign epistasis arises from the benefit of tuning phenotypic parameters of cascade genes with respect to each other, rather than from their complex and incompletely known genetic bases. Specifically, sign epistasis requires only that the optimal phenotypic parameters of one gene depend on the phenotypic parameters of another, independent of other details, such as activating or repressing nature, position within the cascade, intra-genic pleiotropy or genotype. Mutational effects change sign more readily in downstream genes, indicating that optimising downstream genes is more constrained. The findings show that sign epistasis results from the inherent upstream-downstream hierarchy between signalling cascade genes, and can be addressed without exhaustive genotypic mapping.

Highlights

Sign epistasis is a central evolutionary constraint, but its causal factors remain difficult to predict
The approach provides a single criterion for the presence of sign epistasis within the full phenotype space: it exists when the optimal value of a phenotypic parameter of one gene depends on a phenotypic parameter that corresponds to another gene, and is not affected by the same mutations
In order to analyse the origin of the observed sign epistasis, we developed a theoretical approach inspired by geometric fitness models[9,10]

Summary

Introduction

Sign epistasis is a central evolutionary constraint, but its causal factors remain difficult to predict. A number of random mutants of tetR and lacI transcription factors are paired in all combinations to produce a library of cascades, whose ability to transduce signals is quantified by measured input–output relations These data indicate a significant number of sign epistatic interactions between tetR and lacI, with the corresponding sign change occurring predominantly for mutations in the downstream lacI. The approach provides a single criterion for the presence of sign epistasis within the full phenotype space: it exists when the optimal value of a phenotypic parameter of one gene depends on a phenotypic parameter that corresponds to another gene, and is not affected by the same mutations This methodology is general and may be applied to any pair of traits that are genetically independent, but contribute in concert to organismal functions or fitness

Objectives

Methods

Results

Conclusion