Comparison and calibration of different reporters for quantitative analysis of gene expression

Abstract

Absolute levels of gene expression in bacteria are observed to vary over as much as six orders of magnitude. Thermodynamic models have been proposed as a tool to describe the expression levels of a given transcriptional circuit. In this context, it is essential to understand both the limitations and linear range of the different methods for measuring gene expression and to determine to what extent measurements from different reporters can be directly compared with one aim being the stringent testing of theoretical descriptions of gene expression. In this article, we compare two protein reporters by measuring both the absolute level of expression and fold-change in expression using the fluorescent protein EYFP and the enzymatic reporter β-galactosidase. We determine their dynamic and linear range and show that they are interchangeable for measuring mean levels of expression over four orders of magnitude. By calibrating these reporters such that they can be interpreted in terms of absolute molecular counts, we establish limits for their applicability: autofluorescence on the lower end of expression for EYFP (at ∼10 molecules per cell) and interference with cellular growth on the high end for β-galactosidase (at ∼20,000 molecules per cell). These qualities make the reporters complementary and necessary when trying to experimentally verify the predictions from the theoretical models.

Figure 1

Gene expression levels in E. coli. (Red) The estimated absolute expression level of several bacterial and viral promoters obtained from the literature are shown (see the Supporting Material and Table S4 for the corresponding references and assumptions made to determine the level of expression). For comparison, the results from two recent cell censuses of E. coli are also shown as histograms of the number of proteins (7,69). Note that the range of expression spans greater than six orders of magnitude for a given set of measurements illustrating the wide dynamical range associated with bacterial promoters. The discrepancy between the two cell censuses of E. coli is further explored in Fig. S9.

Figure 2

Fold-change of different regulatory motifs. The states and weights from the thermodynamic models are shown for the case of (A) simple repression by Lac repressor and (B) simple activation by CRP. The corresponding fold-change in gene expression as a function of transcription factor concentration predicted by the model is shown (C) for repression and (D) for activation. The fold-change values span over several orders of magnitude. Refer to Bintu et al. (12) for a derivation of the respective formulas and their parameters which are characteristic of bacterial promoters. The data for Lac repressor in panel C has been taken from Oehler et al. (70).

Figure 3

Absolute in vivo fluorescence calibration. (A) Representative fluorescence snapshots of a single molecule seen as a diffraction limited spot and (B) their corresponding fluorescence traces for a single bleaching event of the EYFP-Tet repressor fusion bound to the genomic DNA. The size of a pixel corresponds to ∼143 nm. A total of 200 frames with a 250-ms interval were taken for all traces. (C and D) Snapshots and fluorescent traces for multiple bleaching events of the EYFP-Tet repressor fusion. (Red lines) Least-squares fit to a single or multiple step function. (E) Distribution of fluorescence of bleaching steps for the in vivo sample.

Figure 4

Relation between the mean cell fluorescence and β-galactosidase activity. The fluorescence per cell is plotted against the β-galactosidase activity. Each point corresponds to the same construct bearing either EYFP or lacZ as a reporter in the same strain background and at the same concentration of IPTG. (Blue line) Linear fit fixing the intercept to zero with a slope of (9.6 ± 0.7) × 10⁻⁵ fluorescence units/MU or an estimated 0.1 YFP molecules/LacZ monomer. (Gray-shaded area) Range of YFP where the fluorescence signal is comparable to the cell autofluorescence (see discussion in the main text and Fig. S12). (Red-shaded area) Range where our assay can detect LacZ expression affecting cell growth (refer to the main text and to Table S1). (Blue line) The expression values of several natural promoters, some of which are also shown in Fig. 1, are plotted.

Figure 5

Fold-change in gene expression measured by LacZ and EYFP. The fold-change of a construct bearing a single Lac repressor binding site (Oid, O1, and O2) in the lacI+ and lacI++ backgrounds is compared when lacZ and EYFP are used as reporters. The line has a slope of one. The point in the plot displaying the lowest fold-change corresponds to fluorescence levels that are near the detection limit. This results in the very large error bar shown.