Modern experimental methods help reveal more and more details of signalling networks in terms of the molecules employed and their interactions. However, the complexity of these networks often makes it difficult to understand how specific messages are passed on. One example is calcium signalling where a response can be caused by many different agonists and, in turn, can regulate diverse cellular processes downstream. An important question is thus how cells can reliably distinguish between different calcium-mediated signals in the light of inevitable random fluctuations in molecule numbers. We employ stochastic models of calcium oscillations and calcium-sensitive proteins, such as calmodulin, PKC-alpha, glycogen phosphorylase, calcineurin and calcium/calmodulin kinase 2. These models are simulated under a variety of cellular conditions, e.g. stimulus strength, type of calcium dynamics etc. Then, the information-theoretic measure transfer entropy is estimated on the simulation outputs to quantify the amount of information that is transferred under different conditions. We find that the different types of protein regulation show specific maxima of information transfer for different calcium signals and kinetic parameters. This means that proteins can be tuned to specific calcium signals and that signalling to these proteins is effectively turned on or off depending on the shape of the signal. Our method allows us to compare calcium signalling to different target proteins in terms of their information processing capabilities under different cellular conditions and molecular fluctuations. Our method is general enough to also be applied to other signalling pathways.
