Gene expression noise results in protein number distributions ranging from long-tailed to Gaussian. We show how long-tailed distributions arise from a stochastic model of the constituent chemical reactions and suggest that, in conjunction with cooperative switches, they lead to more sensitive selection of a subpopulation of cells with high protein number than is possible with Gaussian distributions. Single-cell-tracking experiments are presented to validate some of the assumptions of the stochastic simulations. We also examine the effect of DNA looping on the shape of protein distributions. We further show that when switches are incorporated in the regulation of a gene via a feedback loop, the distributions can become bimodal. This might explain the bimodal distribution of certain morphogens during early embryogenesis.fluctuations ͉ genetic switches ͉ single cell T he inevitable noise in gene expression, manifested at the subcellular level as distributions in protein numbers, has been observed experimentally in both prokaryotes and eukaryotes (1-6). Recent studies have investigated how organisms tolerate this noise and the kinds of regulatory strategies they use to control or minimize it (7,8). One example where it has been suggested that noise is exploited for the benefit of the organism is bacterial chemotaxis (9). Analyses of noise in gene expression have highlighted the analogy with quantum many-body systems (10), and the authors of refs. 11-15 have explored the contribution of intrinsic and extrinsic sources, as well as the relative contribution of transcription and translation, focusing on the standard deviation of protein fluctuations. If the protein distributions were Gaussian, the mean, , and standard deviation, , would provide a complete description of the noise characteristics. However, recent experiments have revealed that protein distributions are often nonGaussian and also time-dependent, showing a crossover from long-tailed to Gaussian (3). It is important, therefore, to understand the origins and implications of the long-tailed nature of the protein distributions.We implement a stochastic chemical model of gene expression and show that it leads to distributions that fit the experimental observations, presented in this paper and in ref. 3, of protein distributions at different stages of bacterial growth. The predictions of the simulations were experimentally tested by single-cell-tracking experiments. We suggest that long-tailed protein distributions filtered by appropriate switches can lead to selection at the subcellular level. For example, a switch with a sharp threshold can be used to select a subpopulation of cells with a large concentration of a particular protein from a population of cells with a long-tailed distribution of that protein. By contrast, we show that symmetric Gaussian distributions are not as sensitive to switches.When long-tailed distributions are combined with switches via a positive feedback loop, we find that the system becomes bistable and can result in bimodal protein di...