Absolute quantification of gene expression in individual bacterial cells using two-photon fluctuation microscopy Matthew L. Ferguson a , Dominique Le Coq b , Matthieu Jules b , Stéphane Aymerich b , Nathalie Declerck a , Catherine A. Royer a,⇑ a Centre de Biochimie Structurale, INSERM U554, CNRS UMR 5048, Université Montpellier 1 and 2, F-34090 Montpellier, France b INRA, UMR 1319 Micalis, Domaine de Vilvert, F-78350 Jouy-en-Josas, France article info Article history: Received 14 June 2011 Received in revised form 8 August 2011 Accepted 10 August 2011 Available online 22 August 2011 Keywords: Promoter activity Fluctuation microscopy Biological noise abstract Quantification of promoter activity or protein expression in gene regulatory networks is generally achieved via measurement of fluorescent protein (FP) intensity, which is related to the true FP concen- tration by an unknown scaling factor, thereby limiting analysis and interpretation. Here, using approaches originally developed for eukaryotic cells, we show that two-photon (2p) fluorescence fluctu- ation microscopy, specifically scanning number and brightness (sN&B) analysis, can be applied to deter- mine the absolute concentrations of diffusing FPs in live bacterial cells. First, we demonstrate the validity of the approach, despite the small size of the bacteria, using the central pixels and spatial averaging. We established the lower detection limit at or below 75 nM (3 molecules of FP/vol ex ) and the upper detec- tion limit at approximately 10 lM, which can be extended using intensity measurements. We found that the uncertainty inherent in our measurements (<5%) was smaller than the high cell–cell variations observed for stochastic leakage from FP fusions of the lac promoter in the repressed state or the 10 to 25% variation observed on induction. This demonstrates that a reliable and absolute measure of transcrip- tional noise can be made using our approach, which should make it particularly appropriate for the inves- tigation of stochasticity in gene expression networks. Ó 2011 Elsevier Inc. All rights reserved. Quantification of gene expression at the single cell level is key to understanding, predicting, and eventually modulating growth, development, and adaptation of cell populations and organisms [1,2]. Cell-to-cell variations in the abundance of gene products re- flect the stochastic noise inherent in the biochemical processes of gene expression and regulation networks as well as random fluctu- ations in cellular components and physiological or environmental factors. Investigations of this type typically use flow cytometry or wide field or confocal microscopy to measure the intensity of emis- sion of fluorescent proteins (FPs) 1 expressed from promoter or gene fusions or of single messenger RNA (mRNA) molecules by in situ hybridization with fluorescent oligonucleotides (see, e.g., Refs. [3– 5]). In these approaches, the measured fluorescence signal is consid- ered to be proportional to the intracellular concentration of fluores- cent molecules but does not yield their concentration directly. The scaling factor that relates the fluorescence intensities to the actual protein copy number is usually not known or is estimated indirectly using in vitro or in vivo calibration methods, thereby limiting the ex- tent to which the data can be interpreted. In addition, differences in intrinsic brightness are not considered. Furthermore, the detection limits of these techniques are often restricted by the relatively high autofluorescence of the bacterial cytoplasm or because of photoble- aching or long-range contaminating fluorescence emanating from bright cells. In the case of immobile molecules [6,7] or (alternatively) calibration from immobile particles [8], particle counting and noise measurements are possible down to one molecule per cell. However, the requirement for immobility is not always desirable because it relies on either highly localized protein fusions or fixed cells. In the current work, we tested the capacity of fluorescence fluc- tuation methods to provide an absolute and quantitative measure of gene expression in live bacterial cells. We imaged and analyzed individual cells of the model gram-positive bacterium Bacillus subtilis, producing fluorescent reporter proteins under the control of inducible promoters. We first demonstrate that fluorescence cor- relation spectroscopy (FCS) [9,10] acquired at single points is not ideally suited to our current application due to photobleaching of the FP molecules associated with their confinement inside the small bacterial cells. Successful implementation of fluctuation analysis in this context required the use of two-photon (2p) excitation coupled to an image scanning fluctuation approach termed scanning 0003-2697/$ - see front matter Ó 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.ab.2011.08.017 ⇑ Corresponding author. E-mail addresses: catherine.royer@cbs.cnrs.fr, royer@cbs.cnrs.fr (C.A. Royer). 1 Abbreviations used: FP, fluorescent protein; mRNA, messenger RNA; FCS, fluores- cence correlation spectroscopy; 2p, two-photon; sN&B, scanning number and brightness; GFP, green fluorescent protein; IPTG, isopropyl b-D-1-thiogalactopyrano- side; CFP, cyan fluorescent protein; 3D, three-dimensional; PSF, point spread function; cpspm, counts/s/molecule; mEGFP, monomeric enhanced GFP. Analytical Biochemistry 419 (2011) 250–259 Contents lists available at SciVerse ScienceDirect Analytical Biochemistry journal homepage: www.elsevier.com/locate/yabio