Multiplexing information flow through dynamic signalling systems

Abstract

We consider how a signalling system can act as an information hub by multiplexing information arising from multiple signals. We formally define multiplexing, mathematically characterise which systems can multiplex and how well they can do it. While the results of this paper are theoretical, to motivate the idea of multiplexing, we provide experimental evidence that tentatively suggests that the NF-κB transcription factor can multiplex information about changes in multiple signals. We believe that our theoretical results may resolve the apparent paradox of how a system like NF-κB that regulates cell fate and inflammatory signalling in response to diverse stimuli can appear to have the low information carrying capacity suggested by recent studies on scalar signals. In carrying out our study, we introduce new methods for the analysis of large, nonlinear stochastic dynamic models, and develop computational algorithms that facilitate the calculation of fundamental constructs of information theory such as Kullback-Leibler divergences and sensitivity matrices, and link these methods to a new theory about multiplexing information. We show that many current models such as those of the NF-κB system cannot multiplex effectively and provide models that overcome this limitation using post-transcriptional modifications.

Ligand induced receptor multimerization achieves the specificity enhancement of kinetic proofreading without associated costs

“…Receptor signaling mechanisms have often been viewed as serving a single purpose, such as specificity enhancement by KPR, but we have demonstrated (Section II C) that both KPR and multimeric receptors are multifunctional with overlapping yet non-redundant signal processing capacities. Additionally, both receptor mechanisms are likely capable of producing multiple outputs from different receptor bound states, enabling combinatorial signal processing [69,80,81]. We expect that a deeper theoretical understanding of the range of signal processing functions that can be implemented by a given signaling mechanism will be useful in elucidating different roles of pleiotropic receptors and other receptor signaling mechanisms.…”

“…Intrinsic stochasticity of the reaction kinetics in both the KPR and multimer models leads to variance in the receptor output, which can significantly limit the accuracy of downstream signal processing [1,11,31,36,57,59,69,72,79]. Such noise affects the precision of ligand identification and classification [10,36,69], accuracy of decision making with many possible responses (sometimes referred to as functional plasticity) [3,5,43], and the identification and disentanglement of mixtures of ligands presented to the receptor [69,[80][81][82][83]. KPR is known to be particularly vulnerable to noise because the specificity enhancement is accompanied by signal attenuation [36,59,70].…”

Determinants of Ligand Specificity and Functional Plasticity in Type I Interferon Signaling

“…One commonly considered mechanism is to differentiate transient from sustained receptor activation (Figure 1C). Time course discrimination has been proposed in ERK and NF-kB signaling as well as in responses to metabolic and environmental stressors (57)(58)(59)(60).…”

“…A third mechanism for absolute discrimination is combinatorial encoding, where different ligands induce different compositions of receptor outputs (60,61). This mechanism appears to be most useful for decomposition of mixtures of ligands, a signal processing problem which is not treated here.…”