Schematic representation of the system and of the two main experimental steps. The first step is P. aeruginosa PAO1 GFP culture and inoculation in the microchannel (PDMS on glass, 100 μm × 100 μm cross-section) using a pressure pump. The bacterial suspension is then left for 3 hours without flow to allow cells to adhere. The second step consists in flowing the culture medium at constant flow rate through the microchannel, while recording pressure fluctuations and imaging biofilm development via photonic microscopy. UVC radiation is used during the second step of the experiment to constrain the biofilm in a specific part of the channel.

Impact of nutrient limitation on the longitudinal distribution of biofilm. (a): Fluorescence intensity integrated for 72 hours for the flow rates Q = 0.02 μL/min, 0.2 μL/min, 2 μL/min and 20 μL/min with 1× concentrated brain heart infusion (BHI) culture medium, along with Q = 0.02 μL/min with either 0.2×BHI or 1×BHI supplemented with 8 g/L glucose. Nutrient limitation is observed only for Q = 0.02 μL/min and is strongly dependent upon the concentration of BHI components. (b): Simulations for Q = 0.02 μL/min for two values of ϕmax and the corresponding Damköhler numbers. The Damköhler numbers for the case with 0.2×BHI were simply obtained by multiplying those for the case 1×BHI by a factor 5. Each experimental curve in (a) and (b) is averaged over 3 replicates. In the model, we considered that the fluorescence signal is proportional to the product of biomass and concentration. We then integrated this signal in time and normalized it with the maximum value as .

Summary of various quantities in the experiments and models.

σ0 is the shear stress in empty channel. is the corresponding shear rate. D is an estimate diffusion coefficient for the limiting component. Pe is the Péclet number. ϕmax is the maximum volumic fraction of biofilm. Da is the Damköhler number. τreac is the reaction time. Pe/Da is the ratio of Péclet to Damköhler numbers. To calculate dimensionless numbers, we used D = 10−9 m2s−1 and L = 10 mm.

Values obtained from direct fitting of the experimental data

Temporal dynamics of growth and detachment for Q = 0.2 μL/min, 2 μL/min and 20 μL/min. (a): Summary of the two main stages of biofilm development. The figure shows the evolution of the hydraulic resistance (black solid line) in time, calculated from pressure measurements, and microscopy images corresponding to the different phases. It also shows changes in biofilm colonization extracted from either integrated fluorescence intensity (green squares) or from image segmentation (red solid line). (b), (c) and (d): Temporal dynamics of growth and detachment for the different flow rates. (e), (f) and (g): Wavelet scalograms corresponding to (b), (c) and (d).

Flow modifies the apparent doubling time of bacteria on surfaces. (a), (b), (c) and (d): Composite brightfield and GFP images of development stages, starting from single cells that form microcolonies and then evolve towards a biofilm. (e) and (f) show, respectively, the average number of cells on the surface as a function of time and the corresponding doubling time for flow rates (Q = 0.2 μL/min, 2 μL/min and 20 μL/min). For (f), statistical differences were examined by unpaired student test with Gaussian distribution of data and equal standard deviations. Error bars indicate standard error of mean (SEM) and symbols denote statistical significance (****: p-value <0.0001, ***: p-value = 0.0002, ns: p-value > 0.05). The doubling time was calculated by a linear fitting of the logarithm of the number of cells. The slope was used to estimate growth rate and doubling time. Cell count was calculated from image segmentation of four positions in two channels to generate 8 measurements by condition (n= 8) for (Q = 0.2 μL/min, 2 μL/min and 20 μL/min).

Spatio-temporal dynamics of sloughing in Stage II. (a) and (b): composite bright field and GFP image of a 40 hour biofilm during Stage II. Image before (top) and after (bottom) for (a) a minor sloughing event and (b) a major sloughing event. (c) and (d): kymographs showing the fluorescence intensity (averaged in the radial direction y) as a function of both the longitudinal direction (x) and time. Fluorescence intensity values were normalized by the maximum value. Plots on the right-hand side show the corresponding hydraulic resistance as a function of time.

2D simulations of flow around a biofilm (white) in a channel of 100 μm width and 500 μm length using COMSOL multiphysics. (a): Streamlines and magnitude of the velocity (m/s). (b): Shear rate in (s−1). (c): Pressure field in (Pa). The white arrows indicate the flow direction.

Evolution of the distribution and maximum of the volume fraction for the different flow rates. (a), (b), (c) and (d): Fractions of biofilm in the microchannel, either calculated from hydraulic resistance (black solid line) or estimated from integrated GFP intensity (green dotted line), for the different flow rates. (e): Distribution of biofilm fraction between 24 and 72 h for all flow rates, represented as whisker boxes. (f): Log-log plot of 1 − ϕmax as a function of the flow rate (Q = 0.2 μL/min, 2 μL/min, 20 μL/min and 200 μL/min). The red dotted line simply shows the slope for an evolution with the square root of the flow rate.

Fraction of biofilm obtained from image segmentation for the Δpsl (red line) and Δpel strains (green line) compared to the wild-type strain (blue line), for (Q = 0.2 μL/min, 2 μL/min).

Stage II detachment events can be described as a jump process. (a) Power spectrum for the different flow rates (Q = 0.2 μL/min, 2 μL/min and 20 μL/min). Slopes are indicative and were calculated in the interval between 0 and 0.25 Hz. (b) Probability density function of the time between two successive jump events, δt, for (Q = 0.2 μL/min, 2 μL/min and 20 μL/min). The histograms are calculated from experiments, while the solid lines are fitted Gamma distributions (δt)a−1/(baΓ(a)) exp (−δt/b). (c) Probability density function of the relative amplitude of jump events, , for (Q = 0.2 μL/min, 2 μL/min and 20 μL/min). The histograms are calculated from experiments, while the solid lines are fitted log-normal distributions . (d), (e) and (f): Stochastic simulations of the volume fraction as a function of time for (Q = 0.2 μL/min, 2 μL/min and 20 μL/min).