Preventing E. coli Biofilm Formation with Antimicrobial Peptide-Functionalized Surface Coatings: Recognizing the Dependence on the Bacterial Binding Mode Using Live-Cell Microscopy

Abstract

Antimicrobial peptides (AMPs) can kill bacteria by destabilizing their membranes, yet translating these molecules' properties into a covalently attached antibacterial coating is challenging. Rational design efforts are obstructed by the fact that standard microbiology methods are ill-designed for the evaluation of coatings, disclosing few details about why grafted AMPs function or do not function. It is particularly difficult to distinguish the influence of the AMP's molecular structure from other factors controlling the total exposure, including which type of bonds are formed between bacteria and the coating and how persistent these contacts are. Here, we combine label-free live-cell microscopy, microfluidics, and automated image analysis to study the response of surface-bound Escherichia coli challenged by the same small AMP either in solution or grafted to the surface through click chemistry. Initially after binding, the grafted AMPs inhibited bacterial growth more efficiently than did AMPs in solution. Yet, after 1 h, E. coli on the coated surfaces increased their expression of type-1 fimbriae, leading to a change in their binding mode, which diminished the coating's impact. The wealth of information obtained from continuously monitoring the growth, shape, and movements of single bacterial cells allowed us to elucidate and quantify the different factors determining the antibacterial efficacy of the grafted AMPs. We expect this approach to aid the design of elaborate antibacterial material coatings working by specific and selective actions, not limited to contact-killing. This technology is needed to support health care and food production in the postantibiotic era.

Keywords: antibiotics resistance; antimicrobial peptides; biofilms; fimbriae; image analysis; live-cell microscopy; microfluidics; surface coatings.

Figure 1

Surface modification of microfluidic channels. (a) Chemical structure of AMC-109. (b) Chemical structure of AMC-25-04. (c) Cartoon illustrating how the microfluidic channel was set up. (d) Chemical structure of O-(propargyloxy)-N-(triethoxysilylpropyl)urethane. The blue broken line indicates the potential origin of fragment C₃H₈N⁺. (e) Fluorescence micrographs of silane-modified glass showing a glass surface after modification with Azide-fluor 488 in the presence of all reagents needed for click reaction (left panel) and a negative control where copper ions were omitted (right panel). The surfaces were gently scratched with a steel needle. (f) Part of the ToF-SIMS spectrum showing total intensity counts of a clean glass surface (top), a surface modified with silane (center), and a surface modified first with silane and then with AMC-25-04 through click reaction (bottom). The blue-shaded peak corresponds to the ionized fragment C₃H₈N⁺ (m/z ≈ 58.07, cf. panel [d]). (g) Chemical structure of AMC-25-04 where the ionized fragment with m/z ≈ 830.6 is indicated by a green broken line and the fragment C₂₁H₃₂N⁺ (m/z ≈ 298.8) corresponding to tri-tert-butyl tryptophan is indicated by a red broken line. (h) Part of the ToF-SIMS spectrum analogous to that of panel (f) highlighting the green-shaded peak corresponding to the fragment with m/z ≈ 830.6 from the AMC-25-04 molecule. (i) ToF-SIMS micrographs showing the distribution of the ionized fragment C₂₁H₃₂N⁺ over an area of 0.5 × 0.5 mm² (256 × 256 pixels) of surfaces modified with only silane (left) and silane + AMC-25-04 (right). The color scale shows the number of fragments detected for each pixel.

Figure 2

Method for the analysis of binding and growth. (a) Number of identified bacteria on the imaged surface from the beginning to the end of the experiment. The gray-shaded region indicates the period of bacteria injection. The red broken line indicates the best fit of eq 1 to the number data. (b) Selection of traces showing the formation of a microcolony plotted in the spatial (upper panel) and time (lower panel) domains, respectively. The mother–daughter relations are shown by a color code. (c) Plot showing all data points, color-coded according to the tracing procedure for which GRs were calculated. (d) Boxplot showing the distribution of mean GRs for the bacteria present divided into 20 min intervals. (e) Boxplot showing the distribution of the mean lengths of the bacteria for which GR was presented in (d). (f) Cartoon diagram and plot detailing the analysis of bacterial alignment, i.e., the angle θ between a bacterium’s major axis and the flow direction. The combined scatter and boxplot show the median angle, θ_M, versus the standard deviation (STD) of θ for each bacterium at an early (30 min, blue points) and a late (160 min, red points) time point of the experiment. (g) Cartoon diagram and plots detailing the analysis of bacteria’s wiggling movements around their median major axes. The scatter plot in the upper panel shows for a single example bacterium the instantaneous separations, l × sin(∂θ), between each position l along the bacterium’s major axis with length L and the median major axis M. The bar plot in the lower panel shows the distribution (standard deviations) of the instantaneous separations, l × sin(∂θ), for all positions l of all bacteria present early (30 min, blue bars) and late (160 min, red bars) in the experiment.

Figure 3

Efficacy of AMPs delivered in growth media. In all plots, green denotes control experiments (N = 6), orange denotes experiments with added AMC-109 (N = 5), and red denotes experiments with added AMC-25-04 (N = 5). Statistical significance was tested by Student’s t-test where n.s. denotes not significant, *p < 0.05, **p < 0.01, ***p < 0.005, and ****p < 0.001. (a) The plot shows the average and standard error (SE) of the median GR values for each bin in the boxplots (cf. Figure 2d) of all individual experiments. (b) The bars show the average and SE of median GRs measured before (“initial GR”) and after (“final GR”) the transition time was reached in the individual experiment. (c) Bars show the average and SE of the transition times determined in each experiment. The transition times correspond to the inflection points of the Gompertz’s fits to the data in the boxplots (cf. Figure 2d). (d) The plot shows batch-culture growth curves for E. coli in the presence of different concentrations of AMC-109. (e) The plot shows batch-culture growth curves for E. coli in the presence of different concentrations of AMC-25-04. (f) The plot shows the average and SE of the median bacterial size values for each bin in the boxplots (cf. Figure 2e) of all individual experiments. (g) Bars show the average and SE of median sizes measured after the transition time was reached in the individual experiment. (h) Micrographs detail the appearance of E. coli toward the end of a control experiment (upper panel) and an experiment where AMC-109 was added (lower panel). Both bacteria and AMP nanoparticles (small round features) are visible in the lower micrograph.

Figure 4

Efficacy of AMP-coated surfaces. In all plots, green denotes control experiments (N = 6), blue denotes experiments with WT E. coli (N = 5), and violet denotes experiments with fimbriae-deficient E. coli-ΔFimA (N = 5). Statistical significance was tested by Student’s t-test where n.s. denotes not significant, *p < 0.05, **p < 0.01, ***p < 0.005, and ****p < 0.001. (a) The plot shows the average and SE of the median GR values for each bin in the boxplots (cf. Figure 2d) of all individual experiments. (b) The bars show the average and SE of median GRs measured before (“Initial GR”) and after (“Final GR”) the transition time was reached in the individual experiment. (c) The combined scatter and boxplot show the median angle, θ_M, versus the STD of θ for all bacteria at an early (30 min, light blue) and a late (160 min, dark blue) time point of the experiment (cf. Figure 2f). (d) Scatter plots in the upper panels show for a single example bacterium the instantaneous separations, l × sin(∂θ), between each position l along the bacterium’s major axis with total length L and the median major axis M. The bar plots in the lower panels show the distribution (standard deviations) of the instantaneous separations, l × sin(∂θ), for all positions l of all bacteria (c.f. Figure 2g). (e) The plot shows the average and SE of the median GR values for each bin in the boxplots (cf. Figure 2d) of all individual experiments. (f) The bars show the average and SE of median GRs measured before (“Initial GR”) and after (“Final GR”) the transition time was reached in the individual experiment. (g) The combined scatter and boxplot show the median angle, θ_M, versus the STD of θ for all bacteria at an early (30 min, pink) and a late (160 min, violet) time point of the experiment (cf. Figure 2f). (h) Scatter plots in the upper panels show for a single example bacterium the instantaneous separations, l × sin(∂θ), between each position l along the bacterium’s major axis L and the median major axis M. The bar plots in the lower panels show the distribution (standard deviations) of the instantaneous separations, l × sin(∂θ), for all positions l of all bacteria (cf. Figure 2g). (i) Bars show the average and SE of bacteria’s mean binding times, T_1/2, determined in the different experiments.