High-throughput and single-cell imaging of NF-kappaB oscillations using monoclonal cell lines

Abstract

Background: The nuclear factor-kappaB (NF-kappaB) family of transcription factors plays a role in a wide range of cellular processes including the immune response and cellular growth. In addition, deregulation of the NF-kappaB system has been associated with a number of disease states, including cancer. Therefore, insight into the regulation of NF-kappaB activation has crucial medical relevance, holding promise for novel drug target discovery. Transcription of NF-kappaB-induced genes is regulated by differential dynamics of single NF-kappaB subunits, but only a few methods are currently being applied to study dynamics. In particular, while oscillations of NF-kappaB activation have been observed in response to the cytokine tumor necrosis factor alpha (TNFalpha), little is known about the occurrence of oscillations in response to bacterial infections.

Results: To quantitatively assess NF-kappaB dynamics we generated human and murine monoclonal cell lines that stably express the NF-kappaB subunit p65 fused to GFP. Furthermore, a high-throughput assay based on automated microscopy coupled to image analysis to quantify p65-nuclear translocation was established. Using this assay, we demonstrate a stimulus- and cell line-specific temporal control of p65 translocation, revealing, for the first time, oscillations of p65 translocation in response to bacterial infection. Oscillations were detected at the single-cell level using real-time microscopy as well as at the population level using high-throughput image analysis. In addition, mathematical modeling of NF-kappaB dynamics during bacterial infections predicted masking of oscillations on the population level in asynchronous activations, which was experimentally confirmed.

Conclusions: Taken together, this simple and cost effective assay constitutes an integrated approach to infer the dynamics of NF-kappaB kinetics in single cells and cell populations. Using a single system, novel factors modulating NF-kappaB can be identified and analyzed, providing new possibilities for a wide range of applications from therapeutic discovery and understanding of disease to host-pathogen interactions.

Figure 1

p65-GFP reporter cells A549 SIB01, AGS SIB02 and L929 SIB01. Cells were seeded in 96-well plates and either left untreated (non-activated) or activated with TNFα (10 ng/ml) for 30 min. Cells were fixed and pictures were taken with a Leica DMR Microscope. Representative cells are shown. Scale bar = 10 μm.

Figure 2

The NF-κB readout: p65-GFP reporter cells A549 SIB01 and automated microscopy. a) Workflow of automated microscopy and picture analysis. Cells are fixed and stained with Hoechst 33342. Pictures are then acquired with the Scan^R system and nuclear (blue) and cytoplasmic (red) areas are defined in the Scan^R software. One activated (top) and one non-activated cell (bottom) is depicted. b) Translocation assay using Scan^R Analysis. Cells were seeded on 96-well plates, activated with TNFα (10 ng/ml), fixed, stained with Hoechst 33342 and then analyzed with automated microscopy. Scatter plots as depicted in the analysis software are shown. Cells are gated for circularity and size (Region R01), intensity of GFP and standard deviation of GFP intensity (Region R02). Cells in regions R01 and R02 are classified as active or inactive according to nuclear and cytoplasmic GFP intensity (Region R03 or R04). Cells with nuclear p65-GFP are also in region R03, whereas cells with mainly cytoplasmic p65-GFP are also in gate R04.

Figure 3

Inducer-specific activation profiles of cell lines A549 SIB01, AGS SIB02 and L929 SIB01. The cell lines were activated with the indicated inducer for times between 0 and 6 h: A549 SIB01 was activated for 0-345 min, AGS SIB02 and L929 SIB01 for 0-270 min, each at intervals of 15 min. Cells were fixed, stained with Hoechst 33342 and then analyzed with automated microscopy and image analysis software. For each cell line, single cells were analyzed and the mean percentage of cells with nuclear p65-GFP per well was calculated as described in Figure 2. Mean percentages from duplicate experiments are shown as bars. Results are representative of at least three independent experiments. Standard deviations are not shown for graphical reasons. One data point is missing in AGS H. pylori MOI 10 due to technical issues.

Figure 4

H. pylori induces damped oscillations of p65 nuclear translocations. a) p65-GFP expressing AGS SIB02 cells were infected with H. pylori, stained with Syto 61 and analyzed by confocal live-cell microscopy. The upper panel shows a single bacterium attaching to a cell shortly after acquisition begins (arrow indicates position of bacterium). Graph shows average intensity of GFP in a representative nuclear region of this cell. Scale bar = 10 μm. b) Graph shows average intensities of GFP in other cells from the same experiment, to which one or more bacteria attached at different time points. c) Alignment of normalized average intensities of GFP within representative nuclear regions of the nine single cells shown in (a) and (b). Mean GFP intensity of these nine cells is shown as a black line. While the first peak is remarkably similar, the peak interval and the second peak differ between cells. d) Peak interval of oscillating cells ranges from 40 and 140 minutes; intervals between 80-100 minutes were most frequent. Cells were treated as in (a) and 18 oscillating cells from four separate experiments were analyzed (see Additional file 8 for details).

Figure 5

H. pylori-induced oscillations of p65 nuclear translocation at the population level are dependent on MOI in simulations and experiments. a) Mathematical model for simulating p65 oscillations: Stimulus induces degradation of IκBα (1), IκBα competitively inhibits binding of p65 to DNA (2) and transcription and translation is simplified by an activation delay (3). b) Mathematical model predicts oscillations of nuclear p65 (green) and protein levels of IκBα (red) upon stimulation (grey) at the single-cell level. c) Simulation of population-level p65 oscillations after H. pylori infections at an MOI 100 and 1. Oscillations of p65 nuclear translocation within 500 individual cells were simulated (grey lines) and percentages of cells with nuclear p65 (blue line) were determined by a threshold set for nuclear p65. d) Experimental confirmation of predicted population-level oscillations of p65 (at MOI 100 and 1). AGS SIB02 cells were infected with H. pylori at the indicated MOI, fixed, analyzed by automated microscopy and then the percentage of cells with nuclear p65-GFP was determined, as described in Figure 2. Error bars = SD of experiment performed in triplicates. Results are representative of three independent experiments.