Use of a Fluorescence-Based Assay To Measure Escherichia coli Membrane Potential Changes in High Throughput

Abstract

Bacterial membrane potential is difficult to measure using classical electrophysiology techniques due to the small cell size and the presence of the peptidoglycan cell wall. Instead, chemical probes are often used to study membrane potential changes under conditions of interest. Many of these probes are fluorescent molecules that accumulate in a charge-dependent manner, and the resulting fluorescence change can be analyzed via flow cytometry or using a fluorescence microplate reader. Although this technique works well in many Gram-positive bacteria, it generates fairly low signal-to-noise ratios in Gram-negative bacteria due to dye exclusion by the outer membrane. We detail an optimized workflow that uses the membrane potential probe, 3,3'-diethyloxacarbocyanine iodide [DiOC₂(3)], to measure Escherichia coli membrane potential changes in high throughput and describe the assay conditions that generate significant signal-to-noise ratios to detect membrane potential changes using a fluorescence microplate reader. A valinomycin calibration curve demonstrates this approach can robustly report membrane potentials over at least an ∼144-mV range with an accuracy of ∼12 mV. As a proof of concept, we used this approach to characterize the effects of some commercially available small molecules known to elicit membrane potential changes in other systems, increasing the repertoire of compounds known to perturb E. coli membrane energetics. One compound, the eukaryotic Ca²⁺ channel blocker amlodipine, was found to alter E. coli membrane potential and decrease the MIC of kanamycin, further supporting the value of this screening approach. This detailed methodology permits studying E. coli membrane potential changes quickly and reliably at the population level.

Keywords: CCCP; DiOC2(3); E. coli; Gram-negative; amlodipine; antimycin; barium chloride; high throughput; membrane potential; valinomycin.

FIG 1

Effect of EDTA on DiOC₂(3)-loaded E. coli. (A) The fluorescence spectra (excitation wavelength, 450 nm) of DiOC₂(3)-loaded E. coli without (left) and with (right) EDTA treatment. The red dotted line indicates cells treated with 5 μM CCCP, whereas black is DMSO only. The arrow indicates the emission (Em.) wavelength at which subsequent experimental data were collected. AU, arbitrary units. (B) DiOC₂(3) signal, normalized to the mean, at 670-nm emission from three biological replicates with and without 5 μM CCCP. n = 3. Error bars represent standard errors of the means (SEM). (C) Fluorescence microscopy images showing dye uptake in cells not subjected to (top) and subjected to (bottom) a 5-min EDTA treatment. The scale bar represents 3 μm.

FIG 2

Effects of EDTA and valinomycin on E. coli. Viable cell counts from E. coli without EDTA treatment or with a 5-min EDTA treatment plated on LB agar with 135 mM K⁺ in the presence of 0, 5, or 20 μM valinomycin. The % DMSO concentration remained constant under all three conditions. Error bars represent SEM. n = 4 to 7. Unpaired t tests, with Welch’s correction for unequal variances, were used for data analysis (**, P < 0.01).

FIG 3

Response of E. coli to valinomycin in various KCl concentrations. Fluorescence intensities of cells treated with valinomycin were normalized to fluorescence of valinomycin-treated cells in 0 KCl buffer (A) and to the corresponding buffers without valinomycin (1% DMSO only) (B). Error bars represent SEM. Three biological replicates were performed. (C) Assay dynamic range was determined by calculating the change in membrane potential upon valinomycin addition for each extracellular K⁺ concentration relative to 1 mM K⁺ (see Materials and Methods).

FIG 4

DiOC₂(3)-loaded E. coli exhibits increased depolarization in response to increasing CCCP concentrations. Dose-response curves of the membrane potential response to CCCP indicate an IC₅₀ of 0.10 ± 0.03 μM. Error bars represent SEM.

FIG 5

Effects of amlodipine, antimycin, barium chloride, and sodium azide on E. coli membrane potential. Dose-response curves are shown for each compound tested; data were normalized to the fluorescence of cells treated with vehicle only (DMSO, water, or ethanol). The IC₅₀ for each compound is shown. Not determined (ND) indicates that a dose-response curve was unable to be fit to the data. Three biological replicates were performed to determine membrane potential response to each compound; data points represent averages from these experiments. Error bars represent SEM.