Figure 1.
A large-scale screen to determine the E. coli genes required for phage lambda infection.
(A) Schematic of the screen procedure. All of the strains in the “Keio Collection” were infected with phage lambda. Strains that exhibited significantly fewer or smaller plaques than the K-12 WT control were considered “hits”. (B) The results of the screen, grouped by functional category. The colored boxes indicate the annotation status of the gene and whether it had been previously associated with lambda phage infection (see legend).
Figure 2.
Schematics of E. coli pathways and networks involved in lambda infection.
The genes found in this screen are highlighted in yellow. (A) The lamB gene and several genes governing its transcriptional regulation. (B) Biosynthesis of the LPS inner core. Several genes shown here were also identified in a screen for T7 phage infectivity (highlighted in green). (C) Several entry points to central carbon metabolism, with corresponding transcriptional regulation. The asterisk indicates that glk was not among the original 57 “hits” found in the screen but is highlighted here because only one small plaque was found in the assay.
Figure 3.
E. coli growth curves under normal conditions and incubated with lambda phage.
All four replicates are shown with the average displayed as a bolded line. These growth curves have been normalized by growth rate and maximum growth capacity to facilitate comparison between strains (see Materials and Methods). (A) A typical set of K-12 WT uninfected (gray) and infected (blue) growth curves. (B) A sample of infected knockout strain growth curves to demonstrate the variability between strains, with respect to the response to infection. (C) The ΔlamB, ΔmalI, and ΔmalT strain growth curves, which appear in the same pathway (see Figure 2A), also exhibit very similar infected growth dynamics.
Figure 4.
Clustered E. coli growth curves under infection conditions for all of the reduced infectivity strains.
Each row of the heatmap is the derivative of an averaged, normalized, and smoothed time course for a single strain (see Materials and Methods). A dendrogram indicating the relative distance between row is shown at left. At right and proximal to the clustergram, the corresponding gene names are listed, followed by the cluster number for significant groups. Also shown are symbols indicating genes that fall into either the lamB regulation pathway (#, see Figure 2A), the LPS inner core biosynthesis pathway (*, see Figure 2B), central carbon metabolism and regulation (§, see Figure 2C), and tRNA thiolation (+, see Figure 9). At the far right are averaged time courses for each of the significant clusters.
Figure 5.
The effect of increased phage concentration on infection dynamics.
All of the strains in Cluster 2 and 3 (except ΔcyaA and Δcrr) were tested at three MOIs; the results for the ΔihfA strain are shown in (A) and the rest are found in Figure S2. The bold line indicates the average value of all four measurements. (B) The difference in normalized E. coli concentration between the high and low MOI tests at eight normalized time units is plotted for each strain (The ΔnusB strain time course progressed to only 7 normalized time units due to slow growth but exhibited little difference between MOI tests at that time). (C) For the LPS inner core biosynthesis strains, the high MOI infection growth curves (uninfected in gray and infected in blue) are shown in order of pathway occurrence (top to bottom) to highlight how the dynamics change depending on pathway position. MOI results for strains highlighted with an asterisk were not tested at the higher MOIs and therefore the low MOI results are shown. (D) The time (peak time) and absorbance (peak height) where clearance begins to be detected for each of the time courses shown in (C) is plotted relative to K-12 WT.
Figure 6.
Numerical simulations of E. coli growth during phage lambda infection.
(A) Schematic of the mathematical model, where the boxes represent the amounts of uninfected ([E]) and lysogenically-infected ([E*]) E. coli as well as phage ([l]). The arrows indicate the effects of one variable on the others, and are labeled with the relevant parameters. A detailed description of the model is given in Materials and Methods. (B) Simulated infection time courses where the model parameters were varied (inset) to produce trajectories that closely resembled the data shown in Figure 3B. (C) Clustergram of simulated time course derivatives for a variety of parameter combinations. 125 simulations were generated using the model (combinations of five possible values for each key parameter, see Figure S3 and Figure S4). Of these, the derivatives of all of the simulation time courses with low infectivity were clustered (middle), using the same technique as with the experimental data shown in Figure 4. Additionally, six of the simulations that exhibited no infectivity, and four of the simulations that exhibited high infectivity are shown at top and bottom as representative examples. Similar to Figure 4, a dendrogram is shown at left, and cluster indicators and growth curves representing the average behavior of each cluster are shown at right. At right and proximal to the clustergram are columns indicating the relative values of model parameters f, b and ki, as well as the product of ki and b, adjusted for display on the same color mapping by standard gamma correction of value shown.
Figure 7.
The use of phage production time courses to further discriminate between strains with similar E. coli infection curves.
(A) Certain combinations of model parameters produce indistinguishable simulated infection curve phenotypes. Some simulated time courses are shown. Inset quantifies the eventual deviation, as cosine of angle included between simulated time course vectors, with the variation of each parameter away from zero while other two parameters at constant ki = 1, b = 28, f = 0.75. (B) A phase plane-like diagram shows how the simulations shown in (A), although identical in terms of E. coli growth and lysis, are often different in terms of phage production. The regions in the diagram are labeled by the particular parameter combinations that simulate phage production in the region. The dashed line separating the gray and teal regions indicates a soft threshold where we considered dynamics significantly deviating from uninfected curves. (C) Experimentally measured phage production time courses for several of the Cluster 3 strains, shown as fold change from infection phage concentration.
Figure 8.
Determining the function of yneJ.
Single cell analysis of E. coli infection in the presence of GFP expressing lambda phage was performed to assess yneJ's effect on infection rate and the lytic-lysogenic decision. Growth of the infected K-12 WT and ΔyneJ strains were observed at 60× magnification and assessed for GFP expression. (A) Images of K-12 WT and ΔyneJ cells infected with GFP expressing phage. A recently lysed cell can be seen in the bottom left corner of the K-12 WT image. (B) Table summary of data obtained from single cell analysis shows reduced infectivity and no significant change in fraction lytic for ΔyneJ. (C) Bar plot showing the results of quantitative real-time RT-PCR of lamB mRNA, for ΔyneJ as well as several cell lines deficient in known lamB transcriptional regulators. The error bars indicate the standard deviation. The asterisk indicates a fold-change between the K-12 WT and the strain of interest with a p-value ≤0.001.
Figure 9.
Pathways involved in sulfur metabolism and the thiolation of tRNA nucleosides affect lambda replication both positively and negatively.
Several genes identified in this study fall within tRNA thiolation pathways, as shown schematically here. A star on the secondary structure drawing at right indicates the tRNA modification location. Infection dynamics are displayed (Absorbance (600 nm) vs. Time (hours). The scaling is equivalent for all plots) for knockout strains, demonstrating consistent dynamics within pathways. The pathway that includes ΔtusA, ΔtusBCD, ΔtusE, and ΔmnmA shows dynamics consistent with decreased infectivity (yellow), while the iron-sulfur dependent pathway that goes through IscU shows dynamics consistent with increased infectivity (blue). ΔiscS has a very slow growth rate and is therefore not shown here.