An epigenetic switch governing daughter cell separation in Bacillus subtilis

Abstract

Growing cells of Bacillus subtilis are a bistable mixture of individual motile cells in which genes for daughter cell separation and motility are ON, and chains of sessile cells in which these genes are OFF. How this ON/OFF switch is controlled has been mysterious. Here we report that a complex of the SinR and SlrR proteins binds to and represses genes involved in cell separation and motility. We also report that SinR and SlrR constitute a double-negative feedback loop in which SinR represses the gene for SlrR (slrR), and, by binding to (titrating) SinR, SlrR prevents SinR from repressing slrR. Thus, SlrR indirectly derepresses its own gene, creating a self-reinforcing loop. Finally, we show that, once activated, the loop remains locked in a high SlrR state in which cell separation and motility genes are OFF for extended periods of time. SinR and SlrR constitute an epigenetic switch for controlling genes involved in cell separation and motility.

Figure 1.

The SinR SlrR switch. (A) Genetic circuit controlling matrix production and cell chaining. SinI is an anti-repressor that inhibits the SinR repressor, thereby derepressing the matrix operons (yqxM and eps) and the gene for SlrR. The resulting synthesis of SlrR has two consequences. It leads to the formation of the SlrR•SinR heterocomplex, which represses autolysin and motility genes. Synthesis of SlrR also leads to inhibition of SinR (through titration of SinR), thereby further derepressing slrR as well as the matrix operons. (B) SinR and SlrR constitute a double-negative loop that operates at the level of protein–protein interaction (SlrR–SinR) and gene transcription (SinR slrR). The loop is self-reinforcing in that inhibition (titration) of SinR by SlrR derepresses slrR, resulting in the accumulation of SlrR, which in turn further inhibits (titrates) SinR. The SinR SlrR switch has an SlrR LOW state (left loop) in which autolysin and motility genes are ON, and an SlrR HIGH state (right loop) in which these genes are OFF. The switch can be driven into the SlrR HIGH state either stochastically by noise during growth or deterministically during biofilm formation (see the text). Transcriptional regulation is indicated in red, and protein–protein interactions are indicated in blue.

Figure 2.

SlrR controls cell chaining during biofilm formation. (A) Cells from a colony of wild-type strain (WT) grown on solid biofilm medium exhibited extensive chaining, whereas ΔslrR mutant cells (strain YC131) were impaired in chaining. (B) Induction of slrR from an IPTG-inducible promoter (P_hy-slrR; strain YC280) caused extensive cell chaining in cells grown on solid biofilm medium. (The more wrinkled appearance of the ΔslrR mutant in B than in A is due to leakiness of P_hy-slrR.) (C,D) Neither ΔtasA (strain YC283; C) nor ΔepsH (strain YC282; D) was able to block cell chaining induced by overproduction of SlrR.

Figure 3.

SlrR represses autolysin genes. (A, B) Expression of P_lytA-lacZ (A) or P_lytF-lacZ (B) in wild-type cells and in cells mutant for slrR, sigD, or both slrR and sigD. Cells were grown in shaking culture in biofilm medium. (C) Fluorescence microscopy of wild-type cells harboring a P_slrR-gfp fusion (YC175) grown in shaking culture in biofilm medium. The left panel shows cells stained in red with the membrane dye FM4-64, the middle panel shows fluorescence (green) from GFP, and the right panel is an overlay. (D) Expression of P_lytF-lacZ in a strain (YC600) that contains an IPTG-inducible copy of slrR grown on solid (top) or in liquid (bottom) biofilm medium in the absence or presence of an inducer.

Figure 4.

Chaining phenotype of mutants grown in shaking culture in biofilm medium. (A, top row) A ΔsinI mutant (RL3853) was impaired in chaining, overexpression of sinI (YC227; a functional gfp-sinI translational fusion was used) promoted extensive cell chaining, and a ΔslrR mutation reversed the chaining phenotype caused by overexpression of sinI (YC589). (Bottom row) A ΔsinR mutant (RL3856) showed little chaining (the strain contained an ΔepsH mutation to prevent aggregation), a ΔsinR mutation reversed the chaining phenotype caused by overexpression of sinI (YC228), a ΔsigD mutant showed extensive cell chaining (RL4169), and the chaining phenotype was not reversed by ΔslrR (YC205). (B,C) Induction of slrR (from P_hy-slrR) promoted chaining in shaking culture (strain YC280; B) but not in a strain that with a ΔsinR mutation (strain YC284; C). (D,E) Shown are point mutants of sinR that promoted chaining (D shows strains YC606, YC607, and YC608, respectively, from top to bottom panels). (E) Chaining caused by the point mutations was reversed by the presence of a ΔslrR mutation (strains YC620, YC621, and YC622, respectively, from top to bottom).

Figure 5.

Amino acid substitutions in SinR that promote chaining are clustered in two domains. Two residues (V26 and A27) are located in the DNA-binding domain, whereas the other five (D55, E67, E79, D84, and A85) are clustered in the multimerization domain that is responsible for SinR–SinR interactions.

Figure 6.

SlrR and SinR form a complex that binds to the promoter for the lytABC operon but not the control region for slrR and eps. (A) Shown is a pull-down experiment in which both His₆-SinR and SlrR were detected in the eluate from Ni-NTA beads using antibodies that cross-reacted with both proteins. In a control without His₆-SinR, SlrR was detected only in the flow-through fraction (unbound). (B-E) The top panels show EMSAs using a radiolabeled DNA probe that contained the promoter for the lytABC operon (P_lytA), purified GST-SlrR fusion protein, and SinR. In B and C, increasing SinR (B) or GST-SlrR (C) was added at 0, 62.5, 125, 250, 500, and 1000 nM. In D, a fixed amount of SinR (500 nM) was mixed with increasing GST-SlrR at 0, 62.5, 125, 250, 500, and 1000 nM. In E, a fixed amount of GST-SlrR (63 nM) was mixed with increasing SinR at 0, 62.5, 125, 250, 500, and 1000 nM. The bottom panels in B–D are control experiments in which the proteins were added at the same concentrations as in the top panels except that the radiolabeled DNA probe contained the promoter for an unrelated ywbH gene. (F) DNase I footprinting experiment to map the binding site for heteromeric SlrR•SinR in the promoter region for the lytABC operon. In the middle two lanes, 1 μM GST-SlrR alone was incubated with a DNA probe, whereas in the right two lanes, 1 μM both GST-SlrR and SinR were added. The indicated region of ∼47 bp was protected from DNase I in the presence of both GST-SlrR and SinR. Bands that were protected from DNase I or were hypersensitive to the nuclease are indicated by vertical bars and asterisks, respectively. Within the protected region, two SinR operator-like sequences (indicated as SinR box) and two identical repeats are labeled. Also labeled are the −35 and −10 regions and the +1 start site for the σ^D-dependent promoter. (G) EMSAs using a radiolabeled DNA probe for the intergenic control region for slrR and eps (P_slrR/eps), and purified GST-SlrR fusion protein and SinR. In the left panel, increasing amounts of GST-SlrR were added (0, 31.2, 62.5, 125, 250, and 500 nM). In the middle panel, increasing amounts of SinR were added (0, 15.6, 31.2, 62.5, 125, and 250 nM). In the right panel, a fixed amount of SinR (125 nM) was mixed with increasing amounts of GST-SlrR (0, 31.2, 62.5, 125, 250, 500 nM).

Figure 7.

The SinR SlrR slrR switch controls cell chaining during exponential-phase growth. Shown are a domesticated strain (PY79; A) and an undomesticated strain (3610; B) and derivatives of each mutant for sinI, sinR, or slrR. The strains contained a P_hag-gfp fusion to visualize σ^D-directed transcription (green). The cells were grown to mid-exponential phase in LB medium and stained with the membrane dye FM4-64 (red). PY79 and 3610 showed a mixture of single cells that were ON for σ^D, and chains of cells that were OFF for σ^D. All of the mutants were impaired in chaining and were ON for σ^D.

Figure 8.

The feedback loop exhibits hysteresis. (A, top panel) Cells of an undomesticated (swrA⁺) strain bearing an IPTG-inducible copy of slrR (YC281) were largely in the form of single cells when grown in LB to mid-exponential phase in the absence of an inducer. (Bottom panel) A derivative of the above strain that was mutant for slrR (YC280) contained few chains under the same conditions. (B) After treatment with 100 μM IPTG for 60 min to induce SlrR production, the proportion of chains markedly increased for both strains. (C–E) The cells were washed to remove IPTG, and samples were withdrawn and diluted fourfold into fresh LB medium at hourly intervals. At 120 and 180 min after suspension, cells of YC281 (top panels in D,E) were still largely in the form of chains, whereas cells of YC280 (bottom panels in D,E) had largely reverted back to single cells. (F, top panel) After further rounds of growth and suspension in fresh LB medium, the YC281 cells were still largely in the form of chains. (G,H) Quantitative results showing the ratio of chains versus single cells for YC281 (G) and YC280 (H) cells before and after induction shown in A–F.