E. faecalis populations growing in the presence of ampicillin and sulbactam undergo significant compositional changes that depend on drug concentration, initial population density, and composition.

(a) Diagram illustrating the experiment. A mixture of florescent ampicillin-sensitive and ampicillin-resistant E. faecalis bacteria, prepared at a defined density and ratio, is added to a 3.5 mm glass bottom dish containing 3 ml liquid growth media supplemented with drugs. After 24 hours of incubation at 37°C the biofilm is imaged directly in the glass bottom dish using confocal fluorescence microscopy to obtain a z-stack through the biofilm (25 cross-sectional slices of dimensions 80 µm x 80 µm with 2.5 µm of vertical separation). Z-stacks were obtained from five non-overlapping positions within the biofilm. The three slices depicted correspond to a biofilm with starting composition of 15% resistant and 85% sensitive bacteria treated with ampicillin at 1 µg/ml and sulbactam at 1 µg/ml. A sample of the planktonic population is extracted and mounted on a glass slide for imaging using confocal fluorescent microscopy. (b) Representative images (densest slice) of mixed ampicillin-resistant (green) and ampicillin-sensitive (magenta) biofilms after the incubation period. Initial bacterial density, initial resistant:sensitive ratio, as well as ampicillin and sulbactam concentrations were varied to examine effects on final population composition. The first row corresponds to untreated biofilms, the second to biofilms grown with ampicillin at 1 µg/ml, and the third to biofilms grown with both ampicillin and sulbactam at 1 µg/ml. All panels are shown at the same scale.

Biofilm and planktonic populations exhibit a non-monotonic dependence between final and initial resistant fractions after antibiotic treatment, with final composition of biofilm and planktonic populations being nearly identical.

Rows indicate different initial starting OD’s. (a) & (d) Final resistant fraction versus initial resistant fraction in the biofilm for the no drugs case (green), and after treatment at Amp. 1 µg/ml (blue), Amp. 2 µg/ml (pink), and Amp. 3 µg/ml (purple). (b) & (e) Final resistant fraction versus initial resistant fraction in the planktonic. (c) & (f) Final resistant fractions in the biofilm versus final resistant fractions in the planktonic for the no drugs case and all Amp. concentrations. Error bars throughout all panels represent standard error across the five non-overlapping regions of the biofilm and the three experimental replicas.

The combination of ampicillin and sulbactam increases the final resistant fraction in both biofilm and planktonic populations and leads to a nearly identical final composition.

Rows indicate different initial starting OD’s. (a) & (d) Final resistant fraction for sulbactam experiments versus final resistant fraction for no drug experiments for both biofilm (square) and planktonic (triangle). (b) & (e) Final resistant fraction for ampicillin and sulbactam experiments versus final resistant fraction of just ampicillin experiments (blue: Amp µg/ml-Sulb µg/ml, purple: Amp µg/ml-Sulb. 2 µg/ml). (c) & (f) Final resistant fractions in the biofilm versus final resistant fractions in the planktonic for the Sulb. 1 µg/ml case and at the two cases with drug combinations. Error bars throughout all panels represent variation across the five non-overlapping regions of the biofilm and the three experimental replicas.

Correlation analysis reveals that the biofilm does not exhibit spatial structure.

(a) Schematic of the correlation metric used to quantify structure: the ratio of the number of resistant bacteria within a distance r of a sensitive bacteria in the original data to that in the randomized data. Values equal to one indicate that the biofilm lacks spatial structure, as the relative positions are the same as in a completely random scenario. Values below one indicate a decrease in the number of resistant cells close to sensitive cells in relation to the random scenario. (b), (c) & (d) Correlation values for three different initial resistant fractions (10%, 25% & 50%) at an initial OD of 0.01. (e), (f) & (g) Correlation values for three different initial resistant fractions (10%, 25% & 50%) at an initial OD of 0.001. Dashed line at 1 µm indicates approximate cell length. Error bars represent variation across the five non-overlapping regions of the biofilm and the three experimental replicas.

(a) Representation of the model. Sensitive and resistant bacteria grow as they consume glucose, and die with rates that depend on the ampicillin concentration. Resistant bacteria degrade both ampicillin and sulbactam. (b) & (c) Comparison between experimental data (solid lines) and model prediction (dashed lines) for the final resistant fraction across all drug combinations at an initial OD of 0.01. (d) & (e) Comparison between experimental data and model predictions, including fitting, at an initial OD of 0.001. In panel (d), the no drugs and Amp. 1 µg/ml curves are fitted to extract model parameters. All other curves shown in the figure are model predictions.

Experimental data (solid lines) and model predictions (dashed lines) for the total number of sensitive bacteria within a biofilm z-stack at initial OD 0.001.

The data in the absence of drugs (green) is used to fit the growth parameters and , as well as the initial total biomass and glucose concentrations. Predictions for the lowest ampicillin concentration (blue) are generated using the parameter values obtained from the fitting procedure. Predictions at Amp. 2 µg/mL (pink) and Amp. 3 µg/mL (purple) are obtained by increasing the fitted initial ampicillin concentration twofold and threefold, respectively.