Reversible adhesion by type IV pili leads to formation of permanent localized clusters

Abstract

The formation of long-lived, multicellular clusters is a fundamental step in the physiopathology of many disease-causing bacteria. Experiments on abiotic surfaces suggest that bacterial colonization, including initial cluster formation, requires (1) irreversible adhesion, (2) cell proliferation, and (3) a phenotypic transition. However, here we show that on infection of a polarized MDCK epithelium, Pseudomonas aeruginosa (PA) forms long-lived - i.e., permanent - bacterial clusters without requiring irreversible adhesion, cell proliferation, or a phenotypic transition. By combining experiments and a mathematical model, we reveal that the cluster formation process is mediated by type IV pili (T4P). Furthermore, we unveil how T4P quantitatively operate during adhesion, finding that it is a stochastic process that involves an activation time, requires the retraction of pili, and results in reversible attachment. We explain how such reversible attachment process leads to the formation of permanent bacterial clusters and quantify the cluster growth dynamics.

Keywords: Cell biology; Mathematical biosciences; Microbiology.

Graphical abstract

Figure 1

Morphology of formed PA aggregates (A, B and F) Transwell grown MDCK monolayers were infected with PA, incubated for 1 h and fixed. (A) Top view and orthogonal section (upper and lower panel respectively) showing a confocal micrograph of a monolayer with several extruded apoptotic cells with adhered bacteria. After infection with PA-GFP (green), samples were labeled with Annexin V-Alexa 647 (blue), fixed, permeabilized and stained with phalloidin-rhodamine for F-actin (red). (B and F) Scanning electron micrographs. (B) Bacterial aggregate. Arrows indicate apoptotic host cell material. (C–E) Time-lapse confocal microscopy images.Annexin V: blue. (C) The monolayer was infected with PA-GFP (green). Upper panel: Equatorial plane of an apoptotic cell. The angle between the longitudinal axis of bacteria and the tangent of the cell surface was measured as indicated. The 93% of the angles fell between 45 and 135° (lower panel) showing that bacteria attach by the pole. (D) CheA was used as a reporter of the flagellar pole (left panels show a fixed bacterium expressing CheA-GFP (green) and stained with an anti-PA antibody that labels the flagellum (yellow)). Right panel: time-lapse image of the equatorial plane of an apoptotic cell 15 min after infection. Bacteria attach by the pole opposite the flagellum. (E) Micrographs show the equatorial plane of the same apoptotic cell at the beginning (upper panel) and 15 min after infection (lower panel) with PA-GFP (green). Bacteria adhered to zones of more intense annexin V labeling. (F) Bacteria attach to zones of the surface with vesiculated morphology.

Figure 2

Formation of aggregates on apoptotic cells extruded from a monolayer (A,C and E) Time-lapse confocal imaging of PA-GFP (green) adhering on apoptotic cells (blue). (A) 3D reconstructions of successive z-stacks (C) Snapshots of the equatorial plane of the cell. (B and D) Growth curves of four different experimental aggregates. (B) The number of bacteria (aggregate size, denoted by $n$) was obtained from the entire z stack. (D) The number of bacteria found on the equatorial plane. (E) Snapshots of the equatorial plane in an experiment where initially the monolayer was inoculated with PA-GFP (green) and after 30 min PA-mCherry (red) was added.

Figure 3

Growth dynamics of the aggregate (A) Semi-log plot of the cumulative distribution of bacterial dwelling times on the cell membrane. Circles correspond to WT data, squares to $\Delta$pilAdata, and diamonds to the $\Delta$pilT data, whereas the solid, dashed, and dotted-dashed curves to Equation2 applied to WT, $\Delta$pilA, and $\Delta$pilTdata, respectively. The inset displays the distributions for short dwelling times, in the range $\left[\mathrm{0,70}\right]$. (B) Scheme of the three-states model, see (Equation 1a), (Equation 1b), (Equation 1c). For $r_{12}=0$, the dynamics reduces to a 2-state model, with only states 0 and 1. (C) Temporal evolution of the probability $P\left(n,t\right)$ (color coded) of finding that at time $t$ the aggregate size is $n$; see Equation5. (D) Comparison of the exact (red) and approximate (blue) solution of $P\left(n,t\right)$, evaluated at various times $t$. (E) Aggregate size $n$ versus time. Circles correspond to the growth of an experimental aggregate, while the red and blue curve correspond to exact and approximate solution of $P\left(n,t\right)$, respectively. Schemes (F) and (G) illustrates that the vacant-occupied dynamics of a small cell membrane area does not convey information about the arrow of time, (F), whereas from the temporal evolution of the aggregate is possible to identify it (G). (H) The increase in entropy $H$, Equation6, puts in evidence the arrow of time and the irreversible character of the growth of the aggregate.