Genome-wide analysis of the general stress response network in Escherichia coli: sigmaS-dependent genes, promoters, and sigma factor selectivity

Abstract

The sigmaS (or RpoS) subunit of RNA polymerase is the master regulator of the general stress response in Escherichia coli. While nearly absent in rapidly growing cells, sigmaS is strongly induced during entry into stationary phase and/or many other stress conditions and is essential for the expression of multiple stress resistances. Genome-wide expression profiling data presented here indicate that up to 10% of the E. coli genes are under direct or indirect control of sigmaS and that sigmaS should be considered a second vegetative sigma factor with a major impact not only on stress tolerance but on the entire cell physiology under nonoptimal growth conditions. This large data set allowed us to unequivocally identify a sigmaS consensus promoter in silico. Moreover, our results suggest that sigmaS-dependent genes represent a regulatory network with complex internal control (as exemplified by the acid resistance genes). This network also exhibits extensive regulatory overlaps with other global regulons (e.g., the cyclic AMP receptor protein regulon). In addition, the global regulatory protein Lrp was found to affect sigmaS and/or sigma70 selectivity of many promoters. These observations indicate that certain modules of the sigmaS-dependent general stress response can be temporarily recruited by stress-specific regulons, which are controlled by other stress-responsive regulators that act together with sigma70 RNA polymerase. Thus, not only the expression of genes within a regulatory network but also the architecture of the network itself can be subject to regulation.

FIG. 1.

(A) Comparison of genome-wide gene expression in rpoS⁺ and rpoS::Tn10 strains (MC4100 and RH90, respectively) under three different growth and stress conditions. RNA was prepared during entry into stationary phase in LB medium (OD₅₇₈ = 4), 20 min after the addition of 0.3 M NaCl in minimal medium and 40 min after a shift to pH 5 in LB medium. Cy3- and Cy5-labeled cDNAs obtained from these RNA preparations were analyzed by whole-genome microarray analysis, and normalized intensities are visualized as scatter plots (MC4100 versus RH90). All data are the average of three independent identical experiments. (B) The numbers of σ^S-controlled genes (i.e., genes with an at least twofold difference in expression in rpoS⁺ and rpoS::Tn10 strains) identified under one, two, or all three conditions tested are shown as a Venn diagram.

FIG. 2.

Cluster of σ^S-controlled genes at around 79 min on the E. coli chromosome that includes several acid resistance genes (gad and hde genes). Genes identified as σ^S controlled under all three growth and stress conditions (core genes) are shown by black arrows, and σ^S-controlled genes identified only under one or two conditions are indicated by hatched arrows.

FIG. 3.

A common sequence pattern in the 200-bp regions upstream of σ^S-controlled core genes strongly resembles an extended −10 region previously discussed for σ^S-dependent promoters. Relative frequencies of nucleotides as identified by BioProspector (40) are shown in the table and correspond to positions −14 to −4 in putative σ^S-dependent promoters. A consensus sequence (Con) is given as well as a degenerate consensus (Deg), which also takes into account the second-most-frequent nucleotide if it occurs in more than 30% of the sequences identified (K stands for T or G, Y stands for T or C, and R stands for A or G). The consensus sequence is also shown as a sequence logo (14).

FIG. 4.

Expression and σ^S dependence of gadA::lacZ and gadB::lacZ fusions under different growth and stress conditions. Strains JK86 and JK87, which carry transcriptional single-copy lacZ fusions in gadA and gadB (black and hatched bars, respectively), as well as their rpoS::Tn10 derivatives (grey and white bars, respectively) were grown in LB medium. During log-phase growth, an aliquot of the culture was shifted to pH 5 (see Materials and Methods for details). Samples were taken during log phase (OD₅₇₈ = 0.4; pH 7), 40 min after shift to pH 5, and during entry into stationary phase (OD₅₇₈ = 4; pH 7). Specific β-galactosidase activities were measured (the data given represent the average of three independent experiments each).

FIG. 5.

The role of σ^S, GadX, and GadE in the expression of acid resistance genes under different stress conditions. (A) The gadE transcriptional start site was determined by primer extension experiments with RNA prepared from strain MC4100 carrying the translational gadE::lacZ fusion plasmid (see Materials and Methods for details). During log-phase growth in LB, an aliquot of the culture was shifted to pH 5. Samples for RNA preparation were taken during log phase (OD₅₇₈ = 0.4; pH 7), 40 min after shift to pH 5, and during entry into stationary phase (OD₅₇₈ = 4; pH 7). The reverse transcript and the transcriptional start site in the sequence are indicated by asterisks, and two putative −10 regions are indicated along the sequence. (B) Ex-pression of gadE was assayed by a transcriptional gadE::lacZ fusion present on a plasmid, because in single-copy constructs, measurable activities were extremely low. Translational single-copy fusions, which exhibit higher activities, yielded results similar to those obtained with the multicopy transcriptional fusions. Strain MC4100 and its rpoS and gadX mutant derivatives carrying these fusions were grown and sampled as described in the legend to Fig. 4, and specific β-galactosidase activities were measured. (C) Summarizing model. Solid arrows indicate regulatory influences relevant during entry into stationary phase, and dotted arrows indicate regulatory influences upon shift to or growth at acidic pH.

FIG. 6.

σ^S dependence of many genes is altered in the absence of the global regulator Lrp. Ratios of expression in rpoS⁺ and rpoS mutant strains were determined by microarray analysis of lrp⁺ and lrp::Tn10 backgrounds and are shown in a scatter plot. Genes with a >2-fold difference in this ratio (i.e., in σ^S dependence) in lrp⁺ and lrp mutant strains fall outside of the diagonal field marked by hatched lines.

FIG. 7.

Functional annotations of σ^S-controlled genes. Numbers shown were obtained for core genes. For gene names and further functional details, see Table 1 and the text.