An overlay of the DIC and fluorescence images of E. coli cells expressing the fluorescent protein Venus, tethered to the membrane protein Tsr. Single Tsr–Venus fusion molecules (yellow spots) can be detected when they anchor to the inner membrane of the cell.

Most of what we assume to be true about gene expression is based on genetic and biochemical studies on total pools of molecules and cells, and even single-cell measurements have so far lacked the sensitivity to allow observation of protein expression at the single-molecule level. Now Sunney Xie and colleagues describe two powerful techniques that can track single protein expression, even of low-copy number proteins.

The technique reported in Nature (440, 358–362; 2006) is based on the 'veteran' gene reporter b-galactosidase (b-gal) that is expressed from the lacZ gene. Although b-gal is a highly-sensitive probe, the fluorescent molecules it produces, following substrate hydrolysis, are not retained in the cell. The authors used closed microfluidic chambers to trap the fluorescent molecules excreted by the cells in the small volume of the chambers. In doing so, they were able to obtain real-time quantitative information on gene expression in live Escherichia coli cells with single molecule sensitivity. Furthermore, they showed that this technique was also applicable to budding yeast and mouse embryonic stem cells expressing b-gal from the GAL1 or ROSA promoters, respectively.

The second technique, reported in Science (311, 1600–1603; 2006), replaces the native lacZ gene with a fusion protein of a fluorescent tag (YFP–Venus) and Tsr (a membrane protein), so it can be used as a reporter for monitoring protein expression from the lac promoter. By tracking the disappearance of the fluorescence signal after photobleaching, the authors could show that each fluorescent peak corresponded to a single molecule.

In both studies, the authors concluded that protein molecules are produced in bursts randomly occurring over time, that the number of molecules per burst follows an exponential distribution, and that each burst results from a stochastically transcribed single mRNA. Furthermore, the burst size and frequency could be determined either by real-time quantitative monitoring of protein production or by measuring the steady-state distribution of the number of protein copies within a population of cells.

Xie and colleagues have developed two highly related methods that allow single-molecule sensitivity at a single-cell level. These techniques offer new possibilities for understanding gene expression and will allow genome-wide characterization of low-copy number proteins.