Fig 1.
The evolutionary history, genotype, and phenotype of 1B4.
(A) The evolutionary lineage from 1A0 (SBW25) to 1B4 consists of nine genotypes (rows). Each consecutive genotype contains an additional mutation (columns: gene and nucleotide change). Grey bars indicate the mutations present in each genotype; grey circles mark the first occurrence. (B) Colonies grown on King’s Medium B (KB) agar for 48 h. 1B4 and 1A0-carB* give dimorphic colonies: opaque (Op) and translucent (Tr; 1B4) or smooth (Sm; 1A0-carB*). Bar ~0.5 cm. Saturation and exposure of images was altered. (C) Light microscope images of cells counterstained with India ink. 1B4 and 1A0-carB* have Cap- and Cap+ cells (white and black arrows). Bar ~4 μm. Saturation and exposure of images was altered. (D) Proportion of Cap+ cells in evolutionary series and engineered genotypes. Introduction of the c2020t carB mutation to ancestors increases capsulation (1A0-carB*, 1A4-carB*). Removal of the mutation from 1B4 (1B4-carBwt) reduces capsulation. Bars are means of five replicates, and error bars denote +/- 1 SE. Two sample t tests for the indicated pairs of genotypes revealed significant differences in capsulation levels (p < 0.001). (E) A 1B4-ΔwcaJ-wzb colony grown on KB medium. Bar ~0.5 cm. (F) A 1B4-wcaJ-lacZ colony grown on medium containing X-gal. Light microscopy was used to check blue sectors contained a greater proportion of Cap+ cells than elsewhere. Bar ~1 cm. Image contrast and exposure was altered.
Fig 2.
Intracellular metabolic pathways downstream of carB.
The pyrimidine biosynthetic pathway is a molecular link between the mutant carB genotype and the CAP capsule phenotype. The switch-causing carB mutation reduces concentrations of intermediates in the pyrimidine biosynthesis pathway (shown in black), exposing a decision point at which uridine triphosphate (UTP) is used either by PyrG for nucleotide biosynthesis (leading to the Cap- phenotype, components in red), or by GalU for polymer biosynthesis (generating Cap+ phenotype, components in blue). Metabolic pathways related to UTP biosynthesis are shown in grey. High-flux reactions recycling nucleotides are shown by thick arrows, and nucleotide biosynthetic pathways are shown by thin arrows. Figure based on amino acid homology with established Escherichia coli pathways [14].
Fig 3.
The SBW25 CAP biosynthetic gene cluster.
(A) Transposon insertions in SBW25 genes pflu3658–78 (filled triangles = precise insertion site identified, and unfilled = precise site unknown; numbers relate to S1 Table strain names). See S1 Fig for reactions catalysed by enzymes disrupted by transposon insertion. (B) Comparison of the SBW25 and E. coli K-12 colanic acid biosynthetic loci. The SBW25 locus is predicted to contain five genes involved in CAP precursor biosynthesis (yellow), wcaJ (red), four glycosyl transferases (purple), two acetyl transferases (blue), five genes involved in polymerisation and transport (green), and four genes of unknown function (white). Homologous gene products (connected by grey lines) were identified with Basic Local Alignment Search Tool protein (BLASTP, a tool used to search a query protein against a protein database).
Fig 4.
Deleterious mutations in carB and ndk affect capsulation and intracellular concentrations of pyrimidine intermediates.
The ndk gene was deleted from SBW25 and 1B4 using genetic engineering techniques, giving SBW25-Δndk and 1B4-Δndk, and the effects on capsule switching and pyrimidine pathway intermediates were determined. Bars are means of five replicates, and error bars show +/- 1 SE. (A) Capsulation levels in evolved and engineered genotypes. Deletion of ndk caused an increase in capsulation in SBW25 (i.e., in the absence of a carB mutation; Wilcoxon rank-sum test p < 0.01) and 1B4 (i.e., in the presence of c2020t carB; Welch two-sample t test; p < 0.001). No significant difference in capsulation levels was found between 1B4-Δndk and the ndk transposon mutant, JG176ΔCre (two-sample t test, p > 0.5). (B) Uridine monophosphate (UMP), UDP, and UTP levels were measured in the indicated genotypes (see S7 Table and S8 Table for independent measurements and measurement of additional metabolites and genotypes). The c2020t carB mutation significantly reduced concentrations of UDP and UTP (two-sample t tests for 1A4 versus 1B4, p < 0.05). Similarly, deletion of ndk significantly reduced concentrations of UMP, UDP, and UTP (two-sample t tests for SBW25 versus SBW25-Δndk, p < 0.05). Coexistence of c2020t carB and the ndk deletion resulted in an increase in UMP, UDP, and UTP (two-sample t tests or Wilcoxon rank-sum tests for 1B4 versus 1B4-Δndk and/or JG176ΔCre, p < 0.05). No significant differences in UMP, UDP, or UTP concentrations were found between 1B4-Δndk and the ndk transposon mutant, JG176ΔCre (two-sample t tests or Wilcoxon rank-sum tests, p > 0.05).
Fig 5.
Changes in pyrimidine biosynthetic pathway flux alter proportion of Cap+ cells.
(A) Addition of 2 mM uracil to 1B4 growth medium causes a reduction in the proportion of Cap+ cells (two-sample t test, p < 0.001); addition of 2 mM arginine has no effect (two-sample t test, p > 0.1). The addition of both uracil and arginine reduces capsulation levels below those seen with uracil alone, possibly as a result of complex regulatory interactions. Each point is the mean of five replicates, and error bars show +/- 1 SE. Asterisks indicate a significant difference between each genotype in the presence and absence of uracil. (B) Seven independently isolated switcher genotypes (see Table 1) show varying proportions of Cap+ cells. Addition of 2 mM uracil significantly reduces Cap+ levels in carB switchers (two-sample t tests, p < 0.05) but has no effect on Cap+ in the pyrH switcher (Re1_4, p > 0.1) or 1A4 (p > 0.1). Bars are means of five replicates, and error bars show +/- 1 SE. (C–D) The proportion of Cap+ cells when overexpressing carB, pyrH, ndk, galU, or pyrG in 1B4 (C) and SBW25 (D). “pSX” bars are means of 18 replicates (six replicates from each of three biological replicates), while others are means of nine or six replicates (three replicates from each of two or three biological replicates). Significance tests are for differences between the relevant genotype and empty pSX. (E) Light microscope image of 1B4 + pSX (left) and 1B4 + pSX-galU (right). Overexpression of galU results in cell chains. Bar ~5 μm. The saturation of the images was altered. (F) Fluorescence microscope image of 1B4 (left) and JG49 (right) cells stained with Fluorescent Brightener 28 (cellulose binding dye). Inactivation of galU gives short cells. Bar ~5 μm. The exposure of the second image was altered.
Table 1.
Mutations causing capsule switching.
Fig 6.
The growth-capsulation model for stochastic capsule switching.
(A) A schematic model for capsule switching. Central to the switch is the bifurcation of the UTP pool into nucleotide metabolism (via PyrG) and CAP synthesis (via GalU). When UXP (a signal molecule proportional to UTP) levels are high, cells commit resources to nucleotide metabolism, and expression of the GalU branch is basal. As UXP levels fall, the likelihood that CAP synthesis is activated increases because of the accumulation of a positive regulator sensitive to UXP levels. Once a cell commits to CAP synthesis, nucleotide metabolism slows, and continued CAP synthesis positively feeds back to maintain low UXP levels. To enable a return to Cap-, we model negative feedback via a repressor activated concomitantly with CAP synthesis. The length of time cells remain Cap+ depends on many factors that are each subject to molecular noise. This introduces stochasticity to the system; we model this using the Gillespie algorithm (see Results text for more detail). (B) A simulation of the model showing that low UXP production can induce switching through changes in CAP expression. Low UXP promotes CAP expression through increased availability of a presumed transcriptional activator of CAP biosynthetic genes. Cap+ persists until the CAP repressor, induced by CAP, is synthesized and outcompetes the activator. Bound repressor causes CAP mRNA levels to fall because of halted synthesis and degradation. Cap- persists until repressor concentration decreases and the DNA binding site is free. UXP is low, and the cycle repeats. See also S5 Fig, S1 Code, and S2 Code. (C) Same as B but with elevated UXP production. The positive regulator-repressor system is not utilized; thus, CAP expression remains basal.