Wisdom of the crowds: A suggested polygenic plan for small-RNA-mediated regulation in bacteria

Abstract

The omnigenic/polygenic theory, which states that complex traits are not shaped by single/few genes, but by situation-specific large networks, offers an explanation for a major enigma in microbiology: deletion of specific small RNAs (sRNAs) playing key roles in various aspects of bacterial physiology, including virulence and antibiotic resistance, results in surprisingly subtle phenotypes. A recent study uncovered polar accumulation of most sRNAs upon osmotic stress, the majority not known to be involved in the applied stress. Here we show that cells deleted for a handful of pole-enriched sRNAs exhibit fitness defect in several stress conditions, as opposed to single, double, or triple sRNA-knockouts, implying that regulation by sRNA relies on sets of genes. Moreover, analysis of RNA-seq data of Escherichia coli and Salmonella typhimurium exposed to antibiotics and/or infection-relevant conditions reveals the involvement of multiple sRNAs in all cases, in line with the existence of a polygenic plan for sRNA-mediated regulation.

Keywords: Complex system biology; Microbiology; Molecular biology.

Graphical abstract

Figure 1

The level of numerous sRNAs changes in response to various stress conditions and antibiotic treatments (A and B) Monitoring temporal expression of osmotic-stress-related (A) and nonrelated (B) sRNAs during osmotic stress (20% sucrose) of MG1655 E. coli cells. Expression levels in stressed vs. nonstressed were determined by qPCR, and the fold changes are given. The bars show the SEM between three biological repeats. (C) E. coli sRNAs induced under stress conditions. Increase in E. coli sRNA levels upon high osmolarity, iron limitation, and nutrient deprivation (stationary phase) compared with the levels in non-stressed cells. See Venn diagram in Figure S2. See also S. typhimurium sRNAs induced under stress conditions in Figure S3. (D) E. coli sRNAs induced upon antibiotic treatments. sRNA expression levels in RB001 E. coli cells (isolated from a patient with urinary tract infection) treated with 10 different antibiotics (4x MIC; see Table S4 for MIC values), which represent 6 groups based on their targets (indicated at the bottom) for 10, 30, 60, and 120 min compared with the sRNA levels at time 0. RNA-seq results were analyzed using DESeq2. The heatmap shows the Log₂ fold changes. Az: azithromycin, Cam: chloramphenicol, Doxy: doxycycline, Gent: gentamicin, Cip: ciprofloxacin, Novo: novobiocin, Phl: phleomycin, Nitro: nitrofurantoin, Tmp-Smx: sulfamethoxazole-trimethoprim, CTX: ceftriaxone.

Figure 2

sRNAs act together to affect bacterial fitness during various stresses (A) Top 20 sRNAs enriched at the poles. The NER (normalized enrichment ratio) index for each sRNA was calculated by dividing the polar fold enrichment after stress by the polar fold enrichment before the stress. sRNAs that were chosen for the sRNA-multi-KO strain are designated by an asterisk. (B) Growth at high osmolarity of wild-type E. coli MG1655 (WT) strain compared with the sRNA-multi-KO strain deleted for five sRNAs, chosen unbiasedly as described in the text (Δ5-1). Cells were grown to mid-logarithmic phase and stressed by media replacement with LB supplemented with 20% sucrose. OD results were normalized to time 0, and data are represented as mean ± SEM from 3 independent experiments. (C) Fitness analysis of WT cells (defined as 1), of cells deleted for each of the five sRNAs individually, or of cells with different sRNAs knockout combinations (Δ2, Δ3, Δ4, and Δ5-1), as well as of the Δ5-1 cells overexpressing the five sRNAs deleted in this strain (complementation, Δ5-1 + pOE), during osmotic shock. See Figures S5B and S5C for overexpression validation. Fitness was calculated as explained in the text, and data are represented as mean ± SEM from 3 independent experiments. Statistical analysis for the differences between mutants and WT was performed using t test. The calculated p value is ∗∗∗∗<0.0001. (D) Growth at high osmolarity of wild-type E. coli MG1655 compared with a strain deleted for five osmotic stress-related sRNAs KO (Δ5-2). Experimental details are as in (B). Data are represented as mean ± SEM from 3 independent experiments. (E) Fitness analysis of the sRNA-multi-KO strain not stressed (mock) and subjected to osmotic shock, oxidative stress, or iron depletion, compared with WT (growth rate of WT in LB is defined as 1). Fitness was calculated as explained in the text based on the growth curves shown in Figure 2B (mock), Figure S5D (osmotic stress), Figure S5E (oxidative stress), and Figure S5F (iron limitation). Statistical analysis was conducted as in (C). The calculated p values are ∗ 0.0108 and ∗∗ 0.0023 for osmotic shock and oxidative stress, respectively. The bars show the SEM between three biological repeats. (F) Growth during oxidative stress for WT, ΔoxyS, or sRNA-multi-KO strains. Cells were grown to mid-logarithmic phase and stressed by adding hydrogen peroxide to a final concentration of 1 mM. OD results were normalized to time 0, and data are represented as mean ± SEM from 3 independent experiments. (G) Comparative swimming ability of WT (defined as 1), sRNA-multi-KO (Δ5), and single sRNA KO strains. Hfq deletion strain was used as a control for severe swimming ability defect. Statistical analysis was conducted as in Figure 2C. The calculated p value is ∗∗∗∗<0.0001. The bars show the SEM between at least three biological repeats (n = 3, for Δ5-1 n = 6).

Figure 3

A scheme describing putative aspects of the polygenic and spatial plans for sRNA-mediated regulation A combined mechanism relying on stress-induced expression of a collection of core and peripheral sRNA genes (left panel) and on accumulation of the sRNAs and their chaperon Hfq at the cell pole (right panel) is suggested to underlie regulation by sRNAs. The identity and the number of core and peripheral sRNA genes that are expressed are expected to vary depending on the stress, the transcription plan, the sRBPs involved, and the level of transcriptional plasticity (see discussion). Illustration was created with BioRender.com.