Trehalose Is Not Relevant for In Vivo Activity of ς^S-Containing RNA Polymerase in Escherichia coli

Abstract

The ς^S- and ς⁷⁰-associated forms of RNA polymerase core enzyme (E) of Escherichia coli have very similar promoter recognition specificities in vitro. Nevertheless, the in vivo expression of many stress response genes is strongly dependent on ς^S. Based on in vitro assays, it has recently been proposed that the disaccharide trehalose specifically stimulates the formation and activity of Eς^S and thereby contributes to promoter selectivity (S. Kusano and A. Ishihama, J. Bacteriol. 179:3649–3654, 1997). However, we demonstrate here that a trehalose-free otsA mutant exhibits growth phase-related and osmotic induction of various ς^S-dependent genes which is indistinguishable from that of an otherwise isogenic wild-type strain and that stationary-phase cells do not accumulate trehalose (even though the trehalose-synthesizing enzymes are induced). We conclude that in vivo trehalose does not play a role in the expression of ς^S-dependent genes and therefore also not in sigma factor selectivity at the promoters of these genes.

Many stress-responsive genes in Escherichia coli require the rpoS-encoded ς^S subunit of RNA polymerase for expression. While exponentially growing cells contain very little ς^S, entry into stationary phase or exposure to hyperosmolarity, low or high temperature, or acidic pH results in a rapid induction of ς^S (for recent reviews on the function and regulation of ς^S, see references 6 and 7). Nevertheless, the “housekeeping” ς⁷⁰ still remains the most abundant sigma factor, with ς^S reaching a cellular content of approximately 30% of that of ς⁷⁰ in stationary phase (10). In contrast to alternative sigma factors, ς^S has an in vitro promoter recognition specificity that strongly overlaps with that of ς⁷⁰. Under standard in vitro transcription conditions, promoter recognition by the RNA polymerase core enzyme (E) associated with either of the two sigma factors can be observed for many genes (20, 23, 24). In vivo, however, especially the strongly stress-responsive genes are clearly dependent on Eς^S for expression. This discrepancy has been attributed to the artificial conditions of the standard in vitro transcription assay, since variations in the concentrations of different salts, in the sigma factor-to-core enzyme ratio, and in the superhelicity of the DNA templates have been found to selectively improve in vitro recognition by Eς^S of promoters that in vivo are ς^S dependent (2, 13). Yet the observed effects, even when additive, are small, and/or the required conditions are not present in vivo at the time of induction of ς^S-dependent genes (e.g., most ς^S-dependent genes are induced during the transition into stationary phase, whereas a significant reduction in negative DNA supercoiling is found only in late stationary phase [13]).

Recently, the disaccharide trehalose has also been implicated in the formation and the activity of the Eς^S holoenzyme (14). Trehalose synthesis is itself under the control of ς^S (8, 12), and accumulation of trehalose has been demonstrated in hyperosmotically stressed cells of E. coli, where it serves as an osmoprotectant (3, 17). Trehalose is known as a general stress protectant in many organisms, and its protein- and membrane-protective properties are well documented (1). The hypothesis that trehalose is also important for the selective activation of transcription by Eς^S was entirely based on in vitro experiments (14). For the present report, we have tested this hypothesis in vivo.

Effect of an otsA mutation on the expression of ς^S-dependent genes during entry into stationary phase.

Trehalose synthesis requires two enzymes, trehalose-6-phosphate synthase (encoded by the otsA gene) and trehalose-6-phosphate phosphatase (otsB) (4). otsB and otsA constitute an operon that exhibits ς^S-dependent activation in response to high levels of osmolarity or during entry into stationary phase (8, 12). If the above-mentioned hypothesis is correct, a trehalose-free otsA mutant should be impaired in the stationary-phase induction of ς^S-dependent genes. This was tested by using lacZ and phoA fusions that represent different classes of ς^S-regulated genes. The expression of all these genes is strongly ς^S dependent in vivo, but whereas osmY and bolA are negatively regulated by cyclic AMP (cAMP)-cAMP receptor protein (CRP) (15, 16), csiD requires cAMP-CRP for activation (18, 25). While osmY, bolA, and osmB exhibit strong osmotic as well as stationary-phase induction (11, 15, 16, 26), csiD expression is activated exclusively upon carbon starvation (18, 25). Figure 1 demonstrates that an otsA::Tn10 null mutation did not alter stationary-phase induction of any of these genes, i.e., the ability to synthesize trehalose is irrelevant for their in vivo expression.

FIG. 1.

Stationary-phase induction of ς^S-dependent genes is not affected in trehalose-free strains. The following MC4100-derived strains carrying reporter lacZ or phoA fusions (solid symbols) and their otsA::Tn10 derivatives (open symbols) were tested: RO151 [carrying csi-5(osmY)::lacZ(λplacMu55)] (9), RH95 (carrying λMAV103::bolAp1::lacZYA) (16), DW12 [carrying csi-12(csiD)::lacZ(λplacMu15)] (25), and LB78 (carrying osmB411::TnphoA) (5). Cells were grown in Luria-Bertani medium (19), and optical densities at 578 nm (OD₅₇₈; circles) and specific LacZ and PhoA activities (triangles) were determined as described previously (references 19 and 5, respectively).

Do stationary-phase cells accumulate trehalose?

The hypothesis that trehalose is involved in the selective activation of ς^S-dependent genes during entry into stationary phase (14) relies on trehalose being synthesized under these conditions. Whereas stationary-phase activation of otsA and otsB has been demonstrated (8), several authors have mentioned (as unpublished results) that the activation of the genes did not automatically correlate with an accumulation of trehalose itself (8, 22). Here, we demonstrate that stationary-phase cells do not accumulate trehalose, at least not in amounts comparable to those found after osmotic upshift (Fig. 2). This was invariably observed no matter whether cells were grown in rich or minimal glucose medium or whether early- or late-stationary-phase cells were tested. By contrast and as expected, trehalose accumulation was observed for osmotically stressed cells. As previously reported (3, 17), the intracellular trehalose concentration under these conditions is in the 100 to 200 mM range (with the limit of detection by thin-layer chromatography being at least fivefold lower). But even osmotic upshift did not elicit trehalose production when applied to stationary-phase cells (Fig. 2). Carbon-starved cells may be unable to accumulate trehalose, perhaps because of a reduction in the cellular content of the precursor UDP-glucose (21). Moreover, this result indicates that stationary-phase cells, which are highly osmotolerant, are not dependent on high internal concentrations of trehalose in order to cope with high levels of osmolarity. These data do not exclude the possibility that smaller amounts of trehalose (i.e., below the limit of detection by thin-layer chromatography) are synthesized in stationary-phase cells, as suggested by the fact that the otsBA operon is clearly stationary phase induced. At such lower concentrations, trehalose may have more specific functions or targets, as suggested by the finding that the otsA mutant is partially impaired in stationary-phase thermotolerance (8).

FIG. 2.

Stationary-phase cells do not accumulate trehalose. Strain MC4100 was grown in Luria-Bertani medium (lanes 1 to 4) or M9 minimal medium (19) containing 0.1% glucose (lanes 5 to 8). Exponentially growing cells (lanes 1, 2, 5, and 6) and stationary-phase cells (lanes 3, 4, 7, and 8) were assayed for trehalose accumulation as follows. First, 0.3 M NaCl was added to aliquots of exponentially growing cells (at an optical density at 578 nm [OD₅₇₈] of 0.3) and of stationary-phase cells (2 h after the onset of stationary phase). Incubation of NaCl-treated (lanes 2, 4, 6, and 8) and NaCl-free (lanes 1, 3, 5, and 7) cultures was continued for 60 min (exponential phase) and 90 min (stationary phase). Cells (corresponding to 10 ml of a culture at an OD₅₇₈ of 0.5) were harvested, the pellet was resuspended in 15 μl of 15 mM trichloroacetic acid and centrifuged after a 10-min incubation on ice, and the entire supernatants were separated overnight on a thin-layer chromatography plate (Merck) with a butanol-ethanol-water mixture (5:3:2). As a standard, pure trehalose (10 μl of a 10 mM solution) was used (lane 9). The plate was soaked in 20% H₂SO₄, dried, and incubated at 180°C for 15 min.

Does trehalose play a role in ς^S-dependent gene expression in hyperosmotically stressed cells?

Upon osmotic upshift, the intracellular trehalose concentration increases to considerable levels (3, 17). Since these seem to be the only conditions under which the in vivo content of trehalose is at least of the same order of magnitude as that required to observe in vitro effects on holoenzyme formation and the activity of ς^S (14), we tested whether osmotic induction of ς^S-dependent genes in vivo is affected by the otsA::Tn10 mutation (Fig. 3). Again, this was not the case for osmY, bolA, and osmB (csiD is not osmotically inducible at all).

FIG. 3.

Osmotic induction of ς^S-dependent genes is not affected in trehalose-free strains. Strains RO151, LB78, and RH95 carrying reporter gene fusions in osmY, osmB, and bolA, respectively (see legend to Fig. 1), and their otsA::Tn10 derivatives were grown in M9 medium with 0.4% glycerol. After growth for more than three generations, mid-log-phase cultures (of an optical density at 578 nm [OD₅₇₈] of 0.3) were divided into two aliquots each, one of which was supplemented with 0.3 M NaCl. OD₅₇₈ values (circles) and specific LacZ and PhoA activities (triangles) were determined in NaCl-free (solid symbols) and NaCl-containing (open symbols) cultures.

Conclusions.

While trehalose in high concentrations (0.5 to 1 M) measurably affects the in vitro formation and activity of the Eς^S holoenzyme form of RNA polymerase (14), we conclude from the data presented here that trehalose is not relevant for ς^S-dependent gene expression in vivo. This is so during entry into stationary phase, when trehalose does not measurably accumulate at all, as well as in hyperosmotically stressed exponential-phase cells, where trehalose can accumulate to substantial concentrations. Consequently, trehalose also cannot contribute to in vivo sigma factor selectivity at ς^S-regulated promoters.