The ability to sense the environment is heterogeneously distributed in cell populations

Introduction: Cellular responses to external stimuli vary substantially in cell populations. This raises a fundamental question: can signaling networks relay information about their environment to take appropriate action? Previous estimates of channel capacity are surprisingly low, close to 1 bit. This suggests that cells can barely detect the presence/absence of inputs in the environment.

Traditional estimates of channel capacity compute a single number for all cells in a population, implicitly assuming that individual cells have identical sensing abilities. Notably, cells typically exhibit heterogeneity in cell states. We argue that sensing fidelity is itself a cellular phenotype and hypothesize that it varies in cell populations. Unfortunately, however, there does not exist a conceptual framework to estimate the distribution of sensing abilities.

Using a new information theoretic framework, we show that using single cell signaling data and a stochastic model of signaling activity, we can infer a distribution over cell state dependent mutual information values. Moreover, we can identify biochemical states that make some cells better sensors than others.

We estimate the distribution of sensing abilities for two growth factor pathways. We show that while the traditional channel capacity estimate for both pathways is ∼1 bit, most individual cells are much better at sensing their environment. Importantly, we also verify our prediction that specific cell state variables dictate cells’ sensing abilities. We believe that our framework will be an important tool to explore variability in cellular sensing abilities and in identifying what makes some cells better and others worse in sensing their environment.

Materials and

Methods: Computational methods used in this work cannot be written out in plain text. The detailed methods can be found in the published manuscript: https://elifesciences.org/articles/87747.

The computational code used in this work can be found here: Scripts used to generate mutual information results as well the experimental data used to train our models can be obtained at https://github.com/adgoetz186/Cell_signalling_information. Script for maximum entropy inference of cellular parameters can be obtained at https://github.com/hodaakl/MaxEnt.

Results, Conclusions, and Discussions: Cell populations are characterized by heterogeneity in cell state variables that is responsible for important phenotypic differences and selective advantage. Therefore, it is reasonable to expect that cells’ ability to sense their environment depends on their state and is therefore variable across cells in a population. We developed a novel information-theoretic perspective that allowed us to quantify the distribution of single-cell sensing abilities using single-cell data and models of signaling networks. We also identified biochemical variable that make some cells better sensors than others. Importantly, we showed that individual cells in real cell populations were much better at sensing their environment compared to what is implied by the traditional estimate of channel capacities of signaling networks.

Acknowledgements (Optional):