The Slowdown of Growth Rate Controls the Single-Cell Distribution of Biofilm Matrix Production via an SinI-SinR-SlrR Network

Abstract

In Bacillus subtilis, master regulator Spo0A controls several cell-differentiation pathways. Under moderate starvation, phosphorylated Spo0A (Spo0A~P) induces biofilm formation by indirectly activating genes controlling matrix production in a subpopulation of cells via an SinI-SinR-SlrR network. Under severe starvation, Spo0A~P induces sporulation by directly and indirectly regulating sporulation gene expression. However, what determines the heterogeneity of individual cell fates is not fully understood. In particular, it is still unclear why, despite being controlled by a single master regulator, biofilm matrix production and sporulation seem mutually exclusive on a single-cell level. In this work, with mathematical modeling, we showed that the fluctuations in the growth rate and the intrinsic noise amplified by the bistability in the SinI-SinR-SlrR network could explain the single-cell distribution of matrix production. Moreover, we predicted an incoherent feed-forward loop; the decrease in the cellular growth rate first activates matrix production by increasing in Spo0A phosphorylation level but then represses it via changing the relative concentrations of SinR and SlrR. Experimental data provide evidence to support model predictions. In particular, we demonstrate how the degree to which matrix production and sporulation appear mutually exclusive is affected by genetic perturbations. IMPORTANCE The mechanisms of cell-fate decisions are fundamental to our understanding of multicellular organisms and bacterial communities. However, even for the best-studied model systems we still lack a complete picture of how phenotypic heterogeneity of genetically identical cells is controlled. Here, using B. subtilis as a model system, we employ a combination of mathematical modeling and experiments to explain the population-level dynamics and single-cell level heterogeneity of matrix gene expression. The results demonstrate how the two cell fates, biofilm matrix production and sporulation, can appear mutually exclusive without explicitly inhibiting one another. Such a mechanism could be used in a wide range of other biological systems.

Keywords: biofilms; biosystems; gene expression; stochasticity.

FIG 1

Bistable expression of tapA is controlled by the Spo0A~P level via the SinI-SinR-SlrR network. (A) The schematic of the SinI-SinR-SlrR network used in our model (see Materials and Methods for details). (B) Deterministic model of the network predicts the existence of two steady states, namely, high (red) and low (blue), of the TapA steady-state concentration. At low Spo0A~P concentrations (unshaded region), only a low steady state exists (monostability). At high Spo0A~P concentrations (shaded in gray), two stable steady states coexist in the bistable region. Unstable steady state separating the two is show by dashed line. (C) Stochastic simulation of tapA expression as a function of time was performed using the fixed Spo0A~P concentrations at low (0.05 μM, blue line, value from the monostable region in B) or at high values (1 μM red line, value from the bistable region in B). The x and y axis indicate time (h), and TapA reporter levels are shown in the number of molecule per cell(#/cell).

FIG 2

Heterogeneity of tapA expression is controlled by growth rate and the SinI-SinR-SlrR network. (A) The dynamics and fluctuation (shaded area shows ± σ region) of growth rate (y axis) over time (x axis) were predicted by a Moser-type model (34). (B) Model-predicted Spo0A~P concertation (y axis) as a function of the growth rate (x axis) in the WT, ΔkinC, and ΔkinA strains. (C) Predicted Spo0A~P dynamics and fluctuations as a function of time (shaded area shows ± σ region) in the WT, ΔkinC, and ΔkinA strains. (D) Prediction of single-cell distribution of tapA expression at T4, T6, T8, and T12 (i.e., 4 h, 6 h, 8 h, and 12 h, respectively) in the WT, ΔkinC, and ΔkinA strains using the results of C as an input to the stochastic model of the SinI-SinR-SlrR network. Population mean levels of tapA expression (number of TapA molecules/cell) are indicated in each panel. Note that the maximum value for the first bin is indicated in the broken y axis with the same scale for the remaining bins used in each panel.

FIG 3

The behavior of the SinI-SinR-SlrR network is determined by both the Spo0A~P level and the growth rate. (A) Changes in the predicted ratio of the total concentrations of SinR to SlrR as a function of growth rate in the absence of transcriptional regulation. The inset illustrates degradation rates of SlrR and SinR and their gene positions on the chromosome. (B) Bifurcation between mono- (clear) and bistability (shaded) of tapA expression state controlled by Spo0A~P level (y axis) and growth rate (x axis) via the SinI-SinR-SlrR network. (C) High (red) and low (blue) steady states of TapA concentration as a function of growth rate at the fixed 2 μM Spo0A~P condition as indicated with the dashed line in B. The shaded region shows a bistable region in which a high TapA-expressing state is possible (red). Decrease of the growth rate outside the bistable region lead to switch into deactivated tapA expression state (blue line). (D) The proposed feed-forward network showing how growth rate regulates biofilm matrix production via the SinI-SinR-SlrR network.

FIG 4

Repression of tapA expression caused by a slowdown of growth rate. (A) Predicted changes in Spo0A~P concentration as a function of growth rate in the WT, ΔkinA, ΔkinC, and Δsda strains superimposed on Fig. 3B displaying bifurcation (dashed line) between mono- (clear) and bistability (shaded) of tapA expression state controlled via the SinI-SinR-SlrR network. (B) Stochastic simulation of tapA expression dynamics in the WT, ΔkinA, ΔkinC, and Δsda strains. The x and y axis indicate time (h) and population-averaged (mean of n = 1,000 simulations) tapA expression levels (number of molecules/cell), respectively. (C) Experimentally measured tapA expression at a population level in the WT, ΔkinA, ΔkinC, and Δsda strains. Culture samples were collected at the indicated times (x axis) after the start of incubation and assayed for β-galactosidase activity from PtapA-lacZ (Miller units [MU], y axis). The mean activities of three independent experiments are shown with standard deviations as error bars.

FIG 5

The dynamics of matrix gene expression under different growth dynamics. (A) Stochastic simulation of growth rate (y axis) as a function of time (x axis) under normal (black) and slow (red) growth conditions. When different growth dynamics depicted in A were used as an input in our stochastic simulations, the model predicted stochastic simulation of tapA expression (mean number of TapA molecule per cell, y axis) as a function of time (x axis) under normal (black) and slow (red) growth conditions (B). (C, D) Experimentally measured expression of tapA (C) and epsA (D) at a population level in the WT cells grown in normal (black) and nitrogen-reduced (reduced to 1/10 of the normal level, red) MSgg media. Culture samples were collected at the indicated times (x axis) after the start of incubation and assayed for β-galactosidase activity from PtapA-lacZ (Miller units [MU], y axis). The mean activities of three independent experiments are shown with standard deviations.

FIG 6

Acceleration of Spo0A~P dynamics decrease mutual exclusiveness of sporulation and matrix production. (A) Stochastic simulation of tapA expression in sporulating cells (spo; slow growth); nonsporulating cells (non-spo; fast growth); and all cells in the WT, Δsda, and ΔkinC strains. (B) Fluorescent images of the WT, Δsda, and ΔkinC cells harboring both the GFP-LCN (unstable GFP) reporter under the control of tapA promoter (PtapA-gfp-lcn) and the mCherry reporter under the control of the spoIIQ promoter (PspoIIQ-mCherry). Cells were cultured in MSgg medium and processed for imaging at 8 h after the start of culture. Cells displaying both GFP (tapA expression) and mCherry (spoIIQ expression for sporulation) are indicated with arrows. Scale bar: 5 μm. (C) The experimentally measured fraction of cells expressing tapA in sporulating (spo) cells; nonsporulating (non-spo) cells; and all cells in WT, Δsda, and ΔkinC strains. Error bars indicate standard deviation for n = 9 images taken from 3 independent cultures for each strain.