The noise in the crowd: How gene interactions influence the evolution of cell-to-cell variation

These interact in a network known as a gene network. However, the process of gene expression is subject to chance and the expression of individual genes in the network is unpredictable. How do genes in gene networks evolve to deal with this inherent noise while maintaining the function of the gene network? This question is being addressed by the Molecular Systems Evolution research group at the Max Planck Institute for Evolutionary Biology.

Cells as stochastic machines
The typical textbook depiction of genes organised in well-defined networks gives the impression that the cell functions like a finely tuned, programmed machine, but this is far from being the case. Studies in the field of single-cell biology have shown that gene expression is inherently a noisy process in which cells with identical genetic backgrounds can express genes in very different ways, resulting in a form of cellular individuality. The cell-to-cell variability in the amount of gene products is referred to as "expression noise", and it has been shown that this noise propagates from one gene to another within the network and may be amplified.

High-throughput single-cell genomics studies have also shown that genes vary greatly in the amount of noise they exhibit: Some are expressed with high accuracy, while others are much less predictable. This large variation in noise within the genome suggests that expression noise is shaped by natural selection, but how selection acts on genes in networks is largely unknown.

In silico evolution of gene networks
A new study by Nataša Puzović, Tanvi Madaan and Julien Dutheil from the Max Planck Institute for Evolutionary Biology in Plön investigates the evolution of expression noise in gene regulatory networks using a computational approach. They conducted an in silico evolution experiment in which they generated thousands of populations of model gene regulatory networks and simulated their evolution over several generations.

The researchers found that the evolved amount of noise of a gene is strongly correlated with its position in the network. Central, strongly networked genes that regulate other genes develop a strongly deterministic gene expression, while peripheral genes at the end of the regulatory chain are rather unpredictable.
The model of gene networks allows complex effects of the network structure to be deciphered, despite necessary approximations. The authors show that global network characteristics influence the average noise in the network, suggesting that the entire network topology should not be ignored when analysing expression noise.

A domino effect
The study described here shows that selection at the network level to perform a particular cellular function leads to differential selection pressure on the expression noise of individual genes and that this effect is modulated by the structure of the network itself. He proposes noise propagation as the underlying mechanism for the observed variability of expression noise in the genomes of organisms. This can be understood as an example of the domino effect: If a central gene is noisy, all other connected genes are affected and the entire network collapses. In contrast, a domino falling at the end of the chain has very little impact. Consequently, the burden of reducing expression noise at the network level is greater for genes that control other genes.

This study shows that taking into account selection at multiple levels of organisation is necessary to understand the evolution of life forms that consist of many interacting components. It also shows that natural selection acts not only on the intermediate level of expression, which has been the focus of molecular biology for half a century, but also on its variance and heterogeneity, a dimension that we are only beginning to fully decipher with the advent of single-cell omics.

Understanding how genes work
A fundamental goal of biology is to find out how genes work together to create a functioning organism. It is important to understand how changes in these genes can lead to disease or, conversely, to disease resistance. This is the only way to pave the way for new treatment approaches and methods. The individual genes cannot be analysed in isolation; rather, it is important to consider and understand their interaction as a system. Similarly, we must not ignore the fact that such systems are the result of millions of years of evolution. Computer modelling, such as that used in the study described here, which integrates our knowledge of both molecular biology and evolutionary processes, is the key to achieving this goal.

Press release of the "idw - Informationsdienst Wissenschaft" from 24 April 2023