Local and global consequences of flow on bacterial quorum sensing

Abstract

Bacteria use a chemical communication process called quorum sensing (QS) to control collective behaviours such as pathogenesis and biofilm formation(1,2). QS relies on the production, release and group-wide detection of signal molecules called autoinducers. To date, studies of bacterial pathogenesis in well-mixed cultures have revealed virulence factors and the regulatory circuits controlling them, including the overarching role of QS(3). Although flow is ubiquitous to nearly all living systems(4), much less explored is how QS influences pathogenic traits in scenarios that mimic host environments, for example, under fluid flow and in complex geometries. Previous studies(5-7) have shown that sufficiently strong flow represses QS. Nonetheless, it is not known how QS functions under constant or intermittent flow, how it varies within biofilms or as a function of position along a confined flow, or how surface topography (grooves, crevices, pores) influence QS-mediated communication. We explore these questions using two common pathogens, Staphylococcus aureus and Vibrio cholerae. We identify conditions where flow represses QS and other conditions where QS is activated despite flow, including characterizing geometric and topographic features that influence the QS response. Our studies highlight that, under flow, genetically identical cells do not exhibit phenotypic uniformity with respect to QS in space and time, leading to complex patterns of pathogenesis and colonization. Understanding the ramifications of spatially and temporally non-uniform QS responses in realistic environments will be crucial for successful deployment of synthetic pro- and anti-QS strategies.

Figure 1. Fluid flows locally repress QS by advecting signaling molecules

a, (i) Growth curves obtained by measuring mKate fluorescence intensity from the S. aureus sarA P1-mKate reporter under different flow conditions; flow rates of 0–10 µL/min. Relative cell number denotes the number of cells divided by the initial cell number. Cell counts were obtained as dividing the summed mKate fluorescence intensity by the experimentally determined average fluorescence signal of individual cells. (ii) Normalized QS output is the GFP fluorescence intensity from the QS-controlled agr P3-gfpmut2 reporter divided by the cell counts as a function of time for different flow conditions. We imaged the surface area of 400 µm × 500 µm, which is one quarter of the area of the channel. Data points indicate means and error bars denote standard deviations with n = 4 independent replicates. b, Fluorescence images of S. aureus under no flow and flow conditions 6 h after inoculation. Left panels (i) and (iv) show the constitutive sarA P1-mKate reporter. Middle panels (ii) and (v) show the QS-controlled agr P3-gfpmut2 reporter. Right two panels (iii) and (vi) show the merged images for the left and middle panels. Top panels (i), (ii), and (iii); no flow, bottom panels (iv), (v), and (vi); flow rate of 1 µL/min. The images are based on n = 4 independent replicates.

Figure 2. Forming a thick biofilm promotes QS under flow

a, A three-dimensional rendering of biofilms (127 µm × 127 µm × 50 µm) of S. aureus grown in a microfluidic channel (Q_fluid = 0.1 µl/min). The figure is the merged image (see Fig. 1 for details); red indicates QS-off cells, and yellow indicates QS-on cells. The images are based on n = 4 independent replicates. b, Central merged images show single optical sections of the xy plane, 10 µm above the surface-biofilm interface, with z-projections shown at right (xz plane) and below (yz plane). The same biofilm region is shown (i) under flow for 13 h after inoculation and (ii) following 13 h of flow and 3 h of no flow. The images are based on n = 4 independent replicates. c, Merged images for the base of biofilms grown under different flow rates, (i) 0.1 µL/min and (ii) 10 µL/min, were taken 13 h after inoculation. The images are based on n = 4 independent replicates. d, Normalized QS output was measured as a function of cell number in a volume of 5000 µm³ (100 µm² area × 50 µm height).

Figure 3. Flows assist long-distance QS by enhancing autoinducer accumulation

a, Merged images (see Fig. 1 for details) at different downstream locations along a microfluidic channel (Q_fluid = 0.5 µl/min): (i) 50 mm, (ii) 240 mm, (iii) 260 mm, (iv) 280 mm and (v) 300 mm 14 h after the start of the experiment. The images are based on n = 3 independent replicates. b, Spatio-temporal evolution of the normalized QS output (Q_fluid = 0.5 µl/min). Data indicate means of normalized QS output for triplicate data. c, Model prediction for the autoinducer concentration profile as a function of space and time (see Supplementary Equations).

Figure 4. QS is activated inside crevices

a, Flow networks with crevices or pores in (i) the small intestine of mice (image courtesy of Dr. Anisa Ismail), (ii) tooth cavities (image courtesy of Dr. Wonhee Lee), (iii) corrugated industrial pipes, and (iv) cracks in rocks. Scale bars: 120 µm, 10 mm, 2 cm, and 5 cm, respectively. b, Image of fluorescent 1 µm diameter beads flowing into a corrugated microfluidic channel under a flow rate of 1 µL/min. The images based on n = 4 independent replicates. c, Merged images of S. aureus in a complex topography. As in Fig. 1, red shows QS-off cells, and yellow shows QS-on cells. The images are based on n = 6 independent replicates. d, Normalized QS output inside of crevices in the absence of antagonist (black) and when 1 µM antagonist (AIP-II) was continuously flowed in beginning 8 h after inoculation (red). Data points indicate means and error bars denote standard deviations with n = 4 independent experiments.