Systems-level analyses of osmoregulation in Saccharomyces cerevisiae

Developing a predictive dynamic model of a biological system often requires that the system be extensively characterized genetically and biochemically. But, relatively few systems are sufficiently well characterized to be amenable to quantitative modeling. Here I present two studies in which my coworkers and I combine time-lapse microscopy of living single cells with tools from the engineering disciplines to model an endogenous stress-response system while exploiting few of the previously known system details. Our strategies are very general and highlight the promise of studying other biological systems in an analogous manner. We investigate the frequency dependence of the osmotic-shock response in Saccharomyces cerevisiae , which is mediated largely by the MAP kinase Hog1. The activity of Hog1 correlates with its enrichment in the nucleus, and we monitor its localization while simultaneously applying salt pulses spanning a range of frequencies. Using linear systems theory and our frequency-response data alone, we derive a quantitative model of the system capable of predicting the Hog1 response to an arbitrary input. We further use system-identification techniques to recast our model into biologically interpretable equations, which correspond very highly with the known network structure. Our analysis suggests that the reactions dominating the stress response occur on a timescale shorter than that required for gene expression, even though minor stress elicits a transcriptional response. We find that gene expression plays a role in facilitating the response to future shocks. We next explore how perfect adaptation is achieved in the system. The yeast osmoregulation system is a closed feedback loop, and extensive theoretical work from control engineering shows that only a special type of negative feedback, termed "integral feedback", can permit perfect adaptation. We determine the network location of the integrating reaction(s) responsible for this paramount system feature by utilizing small-molecule inhibitors, a range of salt inputs (e.g., steps and ramps), and theoretical arguments. We conclude that there is only one effective integrator in the system; it requires Hog1 kinase activity, and it regulates glycerol synthesis but not leakage.