Transcription fluctuation effects on biochemical oscillations
Biochemical oscillation systems often consist of negative feedback loops with repressive transcription regulation. Such systems have distinctive characteristics in comparison with ordinary chemical systems: i) the numbers of molecules involved are small, ii) there are typically only a couple of genes in a cell with a finite regulation time scale. Due to the fluctuations caused by these features, the system behavior can be quite different from the one obtained by rate equations, because the rate equations ignore molecular fluctuations and thus are exact only in the infinite molecular number limit. The molecular fluctuations on a free-running circadian system have been studied by Gonze et al. (2002) by introducing a scale parameter $Ω$ for the system size. They consider, however, only the first effect, assuming that the gene process is fast enough for the second effect to be ignored, but this has not been examined systematically yet. In this work, we study fluctuation effects due to the finite gene regulation time by introducing a new scale parameter $τ$, which we take as the unbinding time of a nuclear protein from the gene. We focus on the case where the fluctuations due to small molecular numbers can be ignored. In simulations on the same system studied by Gonze et al., we find the system is sensitive to the fluctuation in the transcription regulation; the period of oscillation fluctuates about 30 min even when the regulation time scale $τ$ is around 30 s, that is even smaller than 1/1000 of its circadian period. We also demonstrate that the distribution width for the oscillation period and the amplitude scales with $\sqrtτ$, and the correlation time of the oscillation scales with $1/τ$ in the small $τ$ regime. The relative fluctuations for the period are about half of that for the amplitude, namely, the periodicity is more stable than the amplitude.