Feedback noise-suppression -- limits, trade-offs and ways to avoid them

28 October 2005

Johan Paulsson
Department of Systems Biology
Harvard Medical School


Life in single cells is dictated by chance: reactions that involve small numbers of molecules generate spontaneous fluctuations that then enslave all dependent processes. Harmful fluctuations can supposedly be suppressed by negative feedback control, increasing synthesis at low concentrations and vice versa. However, designing efficient feedback is difficult in the presence of any lags or delays, especially if there are multiple types of noise present. Based on generalized theory for homeostatic control, I will discuss how feedback deals with a combination of low-copy fluctuations, noisy signaling, and environmental enslavement. If there are lags in the loop, one type of fluctuation is then typically suppressed at the expense of amplifying another. Sharp delays, as may result from the elongation phase of replication, transcription or translation, further imply a time window in which noise accumulation cannot be prevented, regardless of feedback loops. This fundamentally limits what can be controlled, i.e., imposes a strictly non-zero variance that increases with the relative length of the delay. Finally I will discuss how cells use more exotic strategies to avoid the trade-offs or circumvent the delays, sometimes using processes that are non-Markovian in the regulated species. The talk will contain both general mathematical results and biological mechanisms.

All results are unpublished and in progress, but for somewhat related material, see:


J Paulsson, "Models of stochastic gene expression", Phys. Life Rev. 2:157-75A 2005.

J Paulsson, "Summing up the noise in genetic networks", Nature 427:415-18A 2004. PubMed.

J Paulsson, M Ehrenberg, "Noise in a minimal regulatory network: plasmid copy number control", Quart. Rev. of Biophys. 34:1-59Z 2001.

current theory lunch schedule