MazF activation promotes translational heterogeneity of the grcA mRNA in Escherichia coli populations

Abstract

Bacteria adapt to adverse environmental conditions by altering gene expression patterns. Recently, a novel stress adaptation mechanism has been described that allows Escherichia coli to alter gene expression at the post-transcriptional level. The key player in this regulatory pathway is the endoribonuclease MazF, the toxin component of the toxin-antitoxin module mazEF that is triggered by various stressful conditions. In general, MazF degrades the majority of transcripts by cleaving at ACA sites, which results in the retardation of bacterial growth. Furthermore, MazF can process a small subset of mRNAs and render them leaderless by removing their ribosome binding site. MazF concomitantly modifies ribosomes, making them selective for the translation of leaderless mRNAs. In this study, we employed fluorescent reporter-systems to investigate mazEF expression during stressful conditions, and to infer consequences of the mRNA processing mediated by MazF on gene expression at the single-cell level. Our results suggest that mazEF transcription is maintained at low levels in single cells encountering adverse conditions, such as antibiotic stress or amino acid starvation. Moreover, using the grcA mRNA as a model for MazF-mediated mRNA processing, we found that MazF activation promotes heterogeneity in the grcA reporter expression, resulting in a subpopulation of cells with increased levels of GrcA reporter protein.

Keywords: Flow cytometry; Fluorescent reporter; Gene expression; Phenotypic heterogeneity; Toxin-antitoxin system; mazEF module.

Figure 1. Strong and weak repression of the mazEF transcription.

(A) A scheme of the mazEFG operon (Keseler et al., 2013). The transcriptional reporter P_maz-gfp includes intergenic region between the relA and mazE genes together with promoters P_maz1 (P₁) and P_maz2 (P₂), initially named P₂ and P₃, respectively. The promoters are located 13 nt apart, and the upstream promoter is 10 times more active than the downstream promoter (Marianovsky et al., 2001). Details of the reporter construction are given in File S1. (B) We used the reporter P_maz-gfp to quantify weak repression of the mazEF transcription by comparing the GFP signal in the ΔmazF strain (black histogram, strain NN229) and the isogenic wild-type strain (green histogram; strain NN227), measured in the exponential phase. The GFP fluorescence level of the wild-type strain NN227 was comparable to the fluorescence of the reporterless MG1655 strain (grey shaded histogram), thus corresponding to bacterial autofluorescence (see File S1 for Source Data).

Figure 2. Transcriptional de-repression and activation of the mazEF module, and constitutive gene expression under different conditions.

(A–D) GFP fluorescence of strain NN230 harboring the P_maz-gfp reporter is depicted in green; background GFP fluorescence of strain TB205 is depicted in grey. We measured the GFP signal in the cells gated on the size parameters and mCherry signal, to exclude filaments, dead cells and other debris that can influence the calculations. Analysis of the expression of the P_maz-gfp reporter and the constitutive mCherry reporter upon amino acid starvation induced with serine hydroxamate, as well as upon antibiotic treatments—namely ampicillin, chloramphenicol, nalidixic acid, spectinomycin—are reported in Table 1 for ‘harsh’ conditions and in Table 2 for ‘mild’ conditions. (E–H) mCherry fluorescence of strain NN230 harboring the λP_R-mCherry reporter is depicted in red; background mCherry fluorescence of strain MG1655 is depicted in grey. To infer population heterogeneity, we analyzed squared coefficient of variation (SCV; defined as squared standard deviation divided by the squared mean) in the mCherry fluorescence of the entire populations, without gating. (I) We first gated populations based on the size parameters, to minimize influence of cell size on the level of GFP fluorescence. Then, we analyzed the constitutive mCherry fluorescence in the fraction of cells without (GFP−) or with (GFP+) GFP signal of strain NN230, recovered after mild Spec-treatment. The same GFP threshold was applied in all replicates gated on approximately 20,000 cells by using FlowJo, and then two fractions per replicate, GFP- and GFP+ fraction, were analyzed in R. There is no correlation between GFP signal and mCherry signal in the analyzed replicates. For replicate 1: Spearman’s rho = 0.104, p = 5.7E − 50, N = 20,440 events; for replicate 2: rho = 0.088, p = 2.0E − 36, N = 20,546 events; for replicate 3: rho = 0.072, p = 2.5E − 25, N = 20,542 events.

Figure 3. grcA reporters constructed to infer MazF-mediated mRNA processing.

(A) The grcA_wt reporter comprises 278 nt upstream of the start codon of grcA, as well as the first 15 codons fused in frame to EmgfpΔACA, which is a gfp variant devoid of ACA sites in the open reading frame (Sauert, 2015; Oron-Gottesman et al., 2016). The grcA_ATA reporter is a modified grcA_wt reporter harboring two nucleotide substitutions (C to T) at the positions previously determined to be processed by MazF (Vesper et al., 2011). (B) GFP and mCherry signals do not significantly differ between the grcA_wt and the grcA_ATA reporter measured in the exponential phase. Strains NN207 (depicted in blue) and NN214 (depicted in orange) harbor the grcA_wt and the grcA_ATA reporter, respectively, together with the constitutive λP_R-mCherry reporter. (C) The differences are neither significant for the variation in GFP and mCherry fluorescence. Variation is defined as the squared coefficient of variation, SCV. We analyzed N = 5 independent replicates.

Figure 4. Effect of the mild MazF stress on bacterial growth and cell length.

(A) Exponentially growing cultures of strain NN204 were divided into two flasks: 0.1% Ara was added to the first flask to ectopically induce mazF expression from the chromosomally integrated P_BAD-mazF system, and the second flask served as a control. The maximum bacterial growth rate was measured as the slope of ln-transformed OD₆₀₀ data measured at t₁ = 3 h and t₂ = 4.5 h, and the final OD₆₀₀ was measured at t = 8.5 h. Here depicted are average values of N = 3 independent replicates; error bars present standard deviation. (B) An exponentially growing culture harboring the λP_R-mCherry reporter and the chromosomally integrated system P_BAD-mazF (strain NN207) was split in two flasks; 0.2% Ara was added to one flask, and the second flask served as the uninduced control. The samples were taken and cells were fixed after 3 h and 22 h. mCherry images were acquired with an epifluorescence microscope to infer variation in the cell length, indicated as mean ± standard deviation.

Figure 5. Effect of ectopic mazF expression on translation of the grcA_wt and grcA_ATA reporters.

Fluorescence was measured at different time points: during exponential phase before mazF expression (t0), and 2 h (t1), 5.5 h (t2), and 21 h (t3) without or with addition of 0.1% Ara to ectopically induce mazF expression. In addition, we analyzed cultures that were washed 3 h after arabinose induction and recovered for 1.5 or 17 h in fresh media, respectively. (A) For analysis of the mean level of GFP fluorescence, the mean of each replicate was divided by the mean GFP fluorescence of the respective culture in the exponential phase, and referred to as ‘normalized mean’. The normalized mean level of log₁₀-transformed GFP fluorescence was not significantly different between strain NN206 harboring the grcA_wt reporter (blue) and strain NN212 harboring the grcA_ATA reporter (orange). (B) Variation in the GFP fluorescence (SCV) was significantly higher in the grcA_wt reporter than in the grcA_ATA reporter in the overnight recovery phase after 3 h of stress caused by ectopic mazF expression (time point t3), and throughout the growth curve without added Ara. Significant p-values of t-tests are depicted (*p < 0.05, **p < 0.01). All p-values are listed in File S1. (C) The GFP fluorescence distribution is skewed towards higher values for grcA_wt reporter in the overnight recovery phase after 3 h of ectopic mazF expression (blue). The histogram of the grcA_ATA reporter is depicted in orange, and the reporterless strain NN204 is depicted in grey as the background level of GFP fluorescence, i.e., bacterial autofluorescence. The average value of skewness (kurtosis) is 0.023 (26.50) for the grcA_wt reporter and −0.227 (42.37) for the grcA_ATA reporter, measured in N = 3 replicates. The differences between the reporters are significant (p = 0.003 for skewness, p = 0.00098 for kurtosis), and all skewness and kurtosis values are reported in File S1. Here, the histograms of one replicate per strain are plotted. (D) For comparison, the GFP fluorescence distributions are not significantly different between grcA_wt and grcA_ATA reporters after 2 h of ectopic mazF expression.