Fig 1.
Examples of initial networks and within-host activity over time.
A parasite (grey) is shown infecting the Slow Evolution network (A), where it self-replicates, activates the detector, downregulates signaling protein 3, and is destroyed by the effector. D = detector, S = signaling protein, E = effector, P = Parasite. Indigo nodes and arrows show connections fixed via pleiotropic action. Example infection where the infected host employs a mixed immune response (B), shown by the presence of immune effectors before the infection begins and by the increase in effector abundance following infection.
Fig 2.
The evolution of immunological dynamics depends on lifetime infection risk and pleiotropic constraint within the signaling network.
A) Normalized probability density function of the proportion of host immune responses that are induced by parasites. The x-axis shows the percent of the response that is induced by parasites, with the left-hand side being 0% of response induced, to 100% induced responses on the right. The y-axis corresponds to the relative likelihood of finding an immune response in the specified population that is X% induced. Non-pleiotropic networks are represented in green and pleiotropic networks in blue. All plots on the same row have the same chance of infection (10%, 50%, or 90% descending), and plots in the same column compare the non-pleiotropic network against the indicated pleiotropic constraint. Presented in each plot is the Pearson correlation coefficient calculated between the non-pleiotropic and pleiotropic networks. B) Heatmap of the magnitude of maximum immune response attained during infection vs proportion of immune response that is induced by parasites. The y-axis shows the peak of immune effector abundance achieved during infection range [0,1]. The x-axis shows the proportion of the peak response that was generated following infection range [0,1] Darker colors indicate more individuals expressing the magnitude and response combination.
Fig 3.
Analysis of network robustness to silenced signaling proteins under different implementations of pleiotropy.
The y-axis shows the mean absolute difference in effector levels between intact networks and single signaling protein knockout networks. Effector levels were measured during infection by a parasite that could not manipulate host immune signaling. Blue dots correspond to pleiotropic signaling protein knock outs, green dots correspond to knocking out of a single non-pleiotropic protein. All plots on the same row have the same chance of infection (10%, 50%, or 90% descending), and plots in the same column have the same implementation of pleiotropy. There are no pleiotropic nodes in non-pleiotropic networks (leftmost column), so nodes were just chosen at random twice. The networks used in this analysis were the most common network at the end of the simulation from which they originated. Asterisks denote significant differences between pleiotropic and non-pleiotropic knock outs.
Fig 4.
The immune response of the winning population of competitive scenarios was almost always more inducible than the immune response mounted by the losing population.
Immune response probability density function of pleiotropic winners (blue) and the last non-pleiotropic network (green) in the simulation. All plots on the same row have the same chance of infection (10%, 50%, or 90% descending), and plots in the same column have the same implementation of pleiotropy. The x-axis shows the percent of the response that is induced by parasites, with the left-hand side being 0% of response induced, to 100% induced responses on the right. The y-axis corresponds to the relative likelihood of finding an immune response in the specified population that is X% induced. 10 of 12 scenarios show pleiotropic winners being more inducible in their immune responses (as determined by right shifted density peaks) than their non-pleiotropic competitors. Each plot shows the results for competition after 250 generations of evolution.