OUP user menu

Evidence for extracellular control of RpoS proteolysis in Escherichia coli

Anne-Marie Holland, Philip N. Rather
DOI: http://dx.doi.org/10.1111/j.1574-6968.2008.01255.x 50-59 First published online: 1 September 2008


The RpoS σ factor is required for the transition of Escherichia coli into stationary phase, as well as adaptation to environmental stresses and nutrient depletion. In this study, we report that under nutrient poor conditions, RpoS protein accumulation in E. coli was strongly enhanced by a secreted factor. Expression of a single copy RpoS′-′LacZ translational fusion was activated 12-fold by the signal, but a single copy rpoS-lacZ transcriptional fusion was only activated 1.6-fold. The extracellular signal activated the RpoS′-′LacZ translational fusion in dsrA, rprA or dsrA/rprA mutant backgrounds, but did not activate in an hfq mutant background. A RpoS379′-′LacZ translational fusion, missing the region of RpoS required for the RssB (SprE)/ClpXP-dependent proteolysis, was not activated by the extracellular signal. Furthermore, in a rssB(sprE)Tn10 background, the presence of extracellular signal did not significantly activate expression above the already elevated levels. Western and Northern blot analysis demonstrated that the extracellular signal significantly increased the levels of RpoS protein, but not mRNA. The extracellular signal did not bind to reversed-phase C-18 columns, was dialyzable, and stable to pH 2, pH 12 and heat. However, protease treatment drastically reduced signal activity. Extracellular signal activity was absent in an hldD (rfaD) mutant, but was present in cell lysates, suggesting that signal was unable to be exported in an hldD mutant.

  • RpoS
  • cell–cell signaling
  • Escherichia coli


Bacteria produce and secrete a variety of extracellular signals. These signals build up in a cell density-dependent manner and when a critical concentration is reached, gene expression can be coordinated in the bacterial population. This cell-to-cell signaling is also termed quorum sensing and comprehensive reviews on this subject have been compiled (Lazazzera, 2000; Fuqua et al., 2001; Miller & Bassler, 2001; Whitehead et al., 2001; Waters & Bassler, 2005). In gram-negative bacteria, two types of signaling molecules are commonly used. The first are the N-acyl homoserine lactones (AHLs) that are synthesized by an autoinducer synthase, encoded by members of the LuxI family (More et al., 1996; Fuqua et al., 2001). The second signaling molecule is autoinducer-2 (AI-2) that requires LuxS for production (Surette & Bassler, 1998; Schauder et al., 2001; Xavier & Bassler, 2003). AI-2 is a furanone that mediates both intra- and interspecies communication (Chen et al., 2002; Miller et al., 2004). Other signals produced by Escherichia coli that may mediate cell-to-cell signaling include AI-3, indole, and an unidentified signal that depends on CysE for production (Wang et al., 2001; Sperandio et al., 2003; Sturgill et al., 2004). Recently, a pentapeptide signal (NNKNN) was identified in E. coli as a signaling molecule (Kolodkin-Gal et al., 2007).

The σ factor σ S (σ38 or σS) encoded by the rpoS gene is responsible for the activation of genes that function in general stress resistance and stationary phase functions (McCann et al., 1991; Hengge-Aronis, 1996, 2002). The regulation of RpoS expression is extremely complex and regulation is mediated at transcription, translation, and protein stability (Schellhorn & Stones, 1992; Gentry et al., 1993; Hengge-Aronis et al., 1993, 2002; Lange & Hengge-Aronis, 1994; Muffler et al., 1996a, b, c; Pratt & Silhavy, 1996; Schweder et al., 1996; Zhou & Gottesman, 1998; Repoila & Gottesman, 2001; Joloba et al., 2004; Hirsch & Elliott, 2005). Translation of rpoS mRNA can be affected by an increase in cell density, a decrease in temperature, high osmolarity or a shift to low pH (Lange & Hengge-Aronis, 1994; Sledjeski et al., 1996; Majdalani et al., 2001, 2002; Repoila & Gottesman, 2001; Hengge-Aronis, 2002; Repoila et al., 2003; Hirsch & Elliott, 2005). The rpoS mRNA forms a secondary structure at the 5′-leader region that folds to occlude the ribosome-binding site (RBS), inhibiting translation of the gene. Two small regulatory RNAs (sRNA), DsrA and RprA have been shown to relieve this inhibition through complementary base pairing with the 5′-leader segment freeing the RBS and therefore initiating translation during low temperature and cell surface stress, respectively (Muffler et al., 1996a, b, c; Sledjeski et al., 1996, 2001; Majdalani et al., 1998, 2001, 2002; Repoila & Gottesman, 2001; Repoila et al., 2003). OxyS, another sRNA also binds the rpoS mRNA 5′ leader, but functions to inhibit translation under oxidative stress conditions (Zhang et al., 1998, 2002). Activity of these sRNAs is dependent on a RNA-binding chaperone protein Hfq. The Hfq molecule is made up of two hexameric rings and facilitates binding of sRNAs to their mRNA targets (Muffler et al., 1996b; Sledjeski et al., 2001; Mφller et al., 2002; Geissmann & Touati, 2004). RpoS protein turnover is mediated by the response regulator, RssB or SprE, that targets degradation of RpoS by ClpXP protease in an ATP-dependent manner (Muffler et al., 1996a; Pratt & Silhavy, 1996; Bouche et al., 1998; Zhou & Gottesman, 1998; Becker et al., 1999; Klauck, 2001; Zhou et al., 2001; Mika & Hengge, 2005).

Previous studies have found evidence for extracellular control of rpoS transcription (Mulvey et al., 1990; Schellhorn & Stones, 1992; Sitnikov et al., 1996), but to our knowledge, regulation at the posttranscriptional level by extracellular signaling has not been reported. In this study, we present data indicating that RpoS stability in E. coli is increased by an extracellular signal. The stimulation of RpoS protein accumulation is likely via a mechanism that inhibits the RssB (SprE)-ClpXP proteolytic pathway.

Materials and methods

Bacterial strains and growth conditions

Bacterial strains used in this study included R090 λRZ5 RpoS379∷LacZ hybrid), R091 λRZ5 RpoS742∷LacZ hybrid), R0200 λRZ5 rpoS742lacZ transcriptional fusion), DDS1626 λGN272 (RpoS750∷LacZ hybrid). Luria–Bertani broth (LB) was prepared from 10 g tryptone, 5 g yeast extract and 5 g NaCl L−1 diluted to a 0.5 × concentration with sterile water for use as a growth medium.

Preparation of conditioned media (CM)

CM was prepared from 30 mL of 0.5 × LB inoculated with 20 μL of an overnight culture of strain DDS1626 or MG1655 in a 250-mL flask. Both strains produced equivalent amounts of signal. The flask was incubated with shaking (250 r.p.m.) at 37 °C and grown to OD600 nm=0.6 unless otherwise indicated. A cell-free supernatant solution was obtained by centrifuging at 6000 g for 10 min, adjusted the pH to pH 7.5 and filter sterilizing with 0.22-μm pore filters (Nalgene). This is referred to as CM throughout this study. CM was supplemented with 20 × TY, a tryptone yeast extract broth, to a final concentration of 0.25 × before use in each experiment. CM was stored at −80 °C until required for use.

β-Galactosidase assays

For each assay, 3 mL of either 0.5 × LB supplemented with 0.25 × TY or CM supplemented with 0.25 × TY was inoculated with 3 μL of an appropriate overnight culture (1/1000 dilution) and grown to OD600 nm=0.3 at 37 °C with shaking (250 r.p.m.). Cells were harvested and β-galactosidase assays were performed by the method of Miller (1977).

Treatment of CM with pH, heat, protease and dialysis

For treating with pH extremes, CM prepared from DDS1626 cells grown to an OD600 nm of 0.6 was adjusted to pH 12.0 and 2.0 by adding 5 M NaOH and 1.2 M HCl, respectively, in a dropwise manner. CM was left for 30 min at each pH and then adjusted to pH 7.5 before use in β-galactosidase assays. For control CM, equivalent drops of NaOH and HCl used to adjust pH to 12.0 and 2.0 were added to CM resulting in a pH 7.5. This media was then filter sterilized and used as control untreated CM for β-galactosidase assay. For heat treatment, CM was placed in a 95 °C water bath for 20 min, cooled to room temperature, brought to pH 7.5 and filter sterilized. For dialysis, 10 mL of CM was dialyzed against 0.5 × LB in 2-L beakers using dialysis membranes with a molecular weight cutoff of 10 kD (Spectrum) for 4 h. The filtration procedure was done at room temperature according to the manufacturer's instructions. Dialysate from each membrane along with 10 mL untreated CM were brought to pH 7.5 and filter sterilized.

Partial purification of the signal

CM was prepared from a 200-mL culture of E. coli MG1655 grown to an OD600 nm of 1.1. Cell-free supernatants were prepared by centrifuging cells at 6000 g and filter sterilizing the supernatant using a 0.22-μm filter unit. The resulting supernatant was lyophilized and resuspended at a 50 × concentration. The material was loaded into a C-18 SepPak column pre-equilibrated and prewashed with 100% and 0.5% methanol, respectively. The active signal did not bind to the column and was present in the flow through. This active material from the flow through was then separated using gel-filtration on a Sephadex G-10 column by eluting with 0.5 × M9 salts media. 2.5 mL fractions were collected and stored at −80 °C until use. To assay each fraction, 0.5 mL was added to 2.5 mL of 0.5 × LB pH 7.5 and DDS1626 was added at a 1 : 1000 dilution. Cells were grown to an OD600 nm of 0.3 and assayed for β-galactosidase activity.

Western blot analysis

RpoS protein levels were assessed for both wild type strains DDS1626 and MG1655 and a DDS1626 hfqkan mutant. Three microliters of overnight culture of each strain was used to inoculate 3 mL of 0.5 × LB or 3 mL of CM, and grown to an OD600 nm of 0.6 at 37 °C. Cells were harvested and resuspended in Laemelli sample buffer and heated to 95 °C for 10 min. Samples were run on 10% Tris-HCl gel (BioRad) and total protein was electrotransferred onto nitrocellulose and probed using a 1 : 2500 dilution of a mouse antibody to σS (Neoclone). This was followed by incubation with horseradish peroxidase (HRP)-conjugated anti-mouse antibodies (Amersham Biosciences) used at a 1 : 26 000 dilution. Detection was performed using ECL Western Blotting Detection Reagents according to the manufacturer's instructions (Amersham Biosciences).

Northern blot analysis

rpoS mRNA levels were assayed using Northern blot analysis. Cell samples were collected at the same time as those for the Western blot analysis described above. RNA was extracted using the MasterPure RNA Purification Kit (Epicentre). RNA was separated on formaldehyde agarose (1.2%) gels, and blotted onto a nitrocellulose membrane. A digoxigenin-labeled rpoS probe was synthesized using PCR and used as a probe for the Northern blot analysis.

PCR-mediated allelic replacement

The dsrA, rprA and hfq genes were inactivated using the technique described by Datsenko & Wanner (2000). The following primers were used for PCR-mediated disruptions, hfq gene: 5′-CCACGAAGGCGC GTGCTCTTCCACCACCGAGTGGGACGGCACATATGAATATCCTCCTTTAG-3′ and 5′-AACGGAAGCCAACAATCTGCGAGTTCGCTTCAGGCAGGATTT GTGTAGGCTGGAGCTGCTTC-3′, rprA gene: 5′-TCGACGCAAAAGTCCGTATG CCTAGTATTAGCTCACGGTTGTGTAGGCTGGAGCTGCTT-3′ and 5′-CGAAGCGGAAAAATGTTTTTTTTTGCCCATCGTGGGAGACATATGAATATCCTCCTTA-3′, and dsrA:5′-TTTCTTGTCAGCGAAAAAAATT GCGGATAAGGTGATGAACTGTGTAGGCTGGAGCTGCTT-3′ and 5′-GCGTCTCTGAAGTGAATCGTTGAATGCACAATAAAAAAATCATATGAATATCCTCCTTA-3′. DNA products were generated from pKD3 or pKD4 by 30 cycles of PCR amplification (94 °C denaturation, 55 °C annealing and 72 °C extension for 1 min) and used to transform DDS1626/pKD46 by electroporation. Transformants were selected on kanamycin 20 μg mL−1 or chloramphenicol 25 μg mL−1. Disruptions were confirmed using PCR.

French press lysis of cells

Thirty milliliters of 0.5 × LB was inoculated with 20 μL of overnight cultures of PR1 wild-type, and MJ11 hldDcat (Joloba et al., 2004) and grown to an OD600 nm of 0.6 at 37 °C. Cultures were centrifuged at 6000 g for 10 min at 4 °C and the resulting supernatant was adjusted to pH 7.5 and saved at −80 °C until needed for use as CM. The corresponding cell pellets were immediately washed in 10 mL 0.5 × LB and resuspended in 10 mL 0.5 × LB. Each cell suspension was passed twice through a French Press (Thermo Spectronic) at 900 PSI and centrifuged at 7000 g for 10 min at 4 °C. The cell lysate was brought to 10 mL with 0.5 × LB, which corresponded to a 3 × concentrate. The cell lysates were adjusted to pH 7.5, filter sterilized and stored at −80 °C.


Regulation of a RpoS′-′LacZ translational fusion by nutrient conditions

In the course of studies examining regulation of a RpoS′-′LacZ translational fusion in strain DDS1626 in which RpoS sequences to amino acid 250 are fused to lacZ (Sledjeski et al., 1996, 2001), we found that expression was highly dependent on the relative media strength. When grown in regular LB media (1 ×), the expression of β-galactosidase from an RpoS′-′LacZ fusion in strain DDS1626 was measured at 78±1.3 Miller units in cells at mid-log phase (OD600 nm=0.3). However, when grown in 0.5 × (i.e. diluted 50% with sterile water) and 0.25 × LB, RpoS′-′LacZ expression dropped significantly and was measured at 5±1.3 and 11±1.8 Miller units, respectively. When cells were grown in media that was further diluted to 0.125 × LB, the expression of RpoS′-′LacZ was strongly induced with 210±8.5 Miller units. Under these latter conditions, cells grew very slowly and the high levels of RpoS′-′LacZ expression in 0.125 × media was likely due to carbon starvation.

Since RpoS′-′LacZ expression was high in normal LB (1 ×), we hypothesized that previous studies to identify extracellular signals that regulate RpoS expression (Hengge-Aronis, 2002) may have failed because the already elevated expression would mask further induction by potential cell–cell signaling pathways. Therefore, we reinvestigated the role of cell-to-cell signaling in RpoS regulation using 0.5 × LB where expression was low.

Activation of RpoS expression by an extracellular signal

When exposed to CM prepared from cells grown in 0.5 × LB and harvested at various densities, the expression of β-galactosidase from a RpoS′-′LacZ translational fusion in strain DDS1626 was slightly activated by CM prepared from cells at low density (OD600 nm=0.1 and 0.2), but sharply higher in CM prepared from cells at an OD600 nm of 0.4, 0.6 and 0.8, with 8.9, 12.4 and 11.6-fold activation, respectively (Fig. 1). The signal activity then decreased at higher densities, with 5.3-fold activation of RpoS′-′LacZ in CM prepared from cells at OD600 nm=1.2 (Fig. 1). Therefore, signal activity appears to increase with cell density up to late-log phase, and then decrease in early stationary phase. The growth rate of DDS1626 cells in each CM was similar to that of cells grown in LB, although occasionally a longer lag period was observed in CM.

Figure 1

Activation of RpoS′-′LacZ by an extracellular signal. CM was harvested at the indicated densities from Escherichia coli DDS1626 RpoS750∷LacZ grown in 0.5 × LB. To restore nutrients, a concentrated solution of tryptone/yeast extract was added to each preparation of CM at a final concentration of 0.25 × relative to normal LB. Control 0.5 × LB received the same supplement and all media was used at pH 7.5. DDS1626 RpoS750∷LacZ was inoculated from a low-density culture into each of the above media (3 mL) at a 1 : 1000 dilution. Cells were grown with shaking at 37°C and cells were harvested for β-galactosidase assays at an OD600 nm of 0.3. The reported values represent the average of quadruplicate samples from two independent experiments.

Mechanism of increased RpoS expression by cell-to-cell signaling

Since stimulation of transcription or translation could account for the increased expression of the RpoS′-′LacZ fusion by the extracellular signal, we examined the individual contributions of transcription and translation to the increased expression. For this purpose, we used lacZ transcriptional and translational fusions that contained the rpoS promoter and coding region extending to position 742 (amino acid 247) (Muffler et al., 1996c). In strain R0200, this region of rpoS was fused in a single copy to lacZ as a transcriptional fusion and in strain R091, this region of rpoS was fused to lacZ as a translational fusion. As seen previously with the RpoS′-′LacZ translational fusion in DDS1626, the translational RpoS′-′LacZ fusion in R091 was activated 5.4-fold by CM (Fig. 2a). However, the rpoS-lacZ transcriptional fusion in R0200 was only activated 1.7-fold (Fig. 2a).

Figure 2

Effects of extracellular signal on RpoS mRNA and protein accumulation. CM prepared from cells at an OD600 nm of 0.6 and control LB was prepared as described in Fig. 1. (a) Strains R0200 rpoS742lacZ, R091 RpoS742′-′lacZ hybrid and R090 RpoS379′-′LacZ hybrid were inoculated at a 1 : 1000 dilution into each medium. Cells were grown with shaking at 37°C and cells were harvested for β-galactosidase assays at an OD600 nm of 0.3. The reported values represent the average of quadruplicate samples from two independent experiments. In (b) and (c), cells of DDS1626, DDS1626 hfqkan or MG1655 wild-type were grown in 0.5 × LB or CM prepared as described in Fig. 1 from cells at an OD600 nm of 0.6 and duplicate sets of samples were harvested for RNA and total protein. The Northern blot analysis in (b) was carried out with a PCR generated RpoS probe using the ORFmer primer set (Sigma-Genosys). The Western blot analysis was carried out using a RpoS-specific antibody (Neoclone) as described in Materials and methods.

To verify these results in an independent manner, we directly examined rpoS mRNA accumulation and RpoS protein accumulation in cells exposed to the extracellular signal. In two different E. coli backgrounds, DDS1626 and MG1655, the accumulation of rpoS mRNA was slightly stimulated by the extracellular signal in CM (Fig. 2b). However, the levels of RpoS protein were markedly increased in cells exposed to CM (Fig. 2c). These results confirm the lacZ fusion results and indicate the extracellular signal increases RpoS expression in a posttranscriptional manner.

Role of genes required for RpoS translation in activation by an extracellular signal

The results in Fig. 2 suggested that the extracellular signal increased RpoS expression in a posttranscriptional manner. The role of the small RNAs, DsrA and RprA, and the RNA-binding protein Hfq in this activation were investigated. An hfqkan mutation in DDS1626 abolished the ability of CM to activate the RpoS′-′LacZ fusion (Fig. 3). In addition, this result was independently confirmed using Western blot analysis of the native RpoS protein, where in a hfqkan background, there was no increase in RpoS levels upon exposure to CM (Fig. 2c). Next, we tested DDS1626 RpoS′-′LacZ derivatives with either dsrAcat or rprAKm mutations. The basal levels of RpoS′-′LacZ expression was reduced by either mutation, however, the extent of activation by CM in either mutant background was actually higher than in wild type (Fig. 3). A double dsrAcat, rprAkm mutant was constructed and although the basal levels of RpoS expression were further reduced in the double mutant, compared to each single mutant, the extent of activation by CM in the double rsrA/dsrA mutant was 10-fold (Fig. 3). This indicated that the loss of both small RNAs still allowed for activation of RpoS expression by the extracellular signal.

Figure 3

Role of Hfq, RprA, DsrA and RssB(SprE) in activation of RpoS′-′LacZ by an extracellular signal. DDS1626 RpoS750∷LacZ derivatives containing the indicated mutations were assayed in control LB and CM prepared from DDS1626 cells grown in 0.5 × LB and harvested at an OD600 nm of 0.6. Cells were harvested and assayed as described in Fig. 1. The reported values represent the fold-activation of RpoS′-′LacZ in each strain when grown in CM, relative to the expression when grown in control 0.5 × media. Values represent the average of quadruplicate samples from two independent experiments.

Extracellular control of RpoS proteolysis

The inability of extracellular signal to activate RpoS expression in a hfq mutant could indicate that the extracellular signal works by a pathway that increases RpoS translation. Alternatively, since there is essentially little or no RpoS expression in a hfq mutant, either in the presence or absence of extracellular signal, the positive effect of signal may be masked if it stimulated RpoS expression at a point downstream of translation. Moreover, β-galactosidase expression from the RpoS742′-′LacZ fusion also reflects proteolysis of RpoS by the ClpXP protease because the ‘turnover’ sequence within RpoS is centered at amino acid 173 (Becker et al., 1999). Cleavage of the RpoS portion of RpoS-LacZ hybrid protein facilitates rapid degradation of the entire hybrid protein, resulting in loss of β-galactosidase activity (Muffler et al., 1996c). The RpoS′-′LacZ fusion in DDS1626 extends to amino acid 250 of RpoS and also contains this turnover region for proteolysis. To test if alterations in RpoS proteolysis contributed to the increased expression of RpoS′-′LacZ upon exposure to CM, we first utilized the RpoS379′-′LacZ hybrid fusion in R090, which encodes a hybrid protein composed of RpoS sequences up to amino acid 126 and is missing the ClpXP cleavage site. Therefore, this fusion only reports transcriptional and translational changes and because it is missing the ClpXP cleavage site, the basal level of expression of this fusion is higher than RpoS742′-′LacZ. The expression of β-galactosidase from RpoS379′-′LacZ was only stimulated 1.8-fold by CM (Fig. 2) and this stimulation could be explained by the minor effect of CM on rpoS transcription. This suggested that the cell-to-cell signaling pathway primarily acted in a manner that inhibited RpoS proteolysis and required amino acids beyond residue 126 for this effect.

RpoS proteolysis is mediated by the response regulator RssB (also called SprE) that acts to deliver RpoS to the ClpXP protease (Muffler et al., 1996a; Pratt & Silhavy, 1996; Becker et al., 1999; Klauck et al., 2001; Zhou et al., 2001). If the cell-to-cell signaling pathway acted to inhibit RpoS proteolysis, we would expect to see no difference in RpoS levels upon exposure to CM in a rssB (sprE) mutant. However, if the extracellular signaling pathway acted independently of inhibiting RpoS proteolysis, then in a rssB mutant, the presence of extracellular signal should result in the same degree of activation as in wild-type cells.

As expected, a rssBTn10 mutation in the DDS1626 background increased the levels of β-galactosidase from the RpoS′-′LacZ fusion 10.4-fold above wild type when grown in LB media (data not shown). However, the addition of CM only stimulated this β-galactosidase activity by twofold in the rssBTn10 background (Fig. 3). This twofold increase could be attributed to the minor effect of extracellular signal on transcription (see Fig. 2a, b).

Partial characterization of the activating signal

To rule out the possibility that the activation of RpoS′-′LacZ by CM resulted from variables other than the presence of extracellular signal, such as oxygen depletion or growth dependent depletion of an inhibitor of expression, the signal activity was partially purified to demonstrate that adding back partially purified signal activated RpoS expression. The extracellular signal did not efficiently bind to a reversed-phase C-18 Sep Pak column. However, a large percentage of the LB components were retained on the C-18 column based on the fact that the flow-through material was colorless. The active, flow-through material from the C-18 column was then further purified using Sephadex G-10 size exclusion chromatography. Analysis of the activity in the resulting fractions demonstrated a peak of activity at fractions 14 and 15, a position well after the void volume at fraction 10/11 (Fig. 4a).

Figure 4

Characterization of the extracellular signal. In (a), partially purified CM was fractionated on a Sephadex G-10 column as described in the Materials and methods. Individual fractions were tested for activation of RpoS′-′LacZ by growing cells of DDS1626 in each fraction and harvesting cells at an OD600 nm of 0.3 for β-galactosidase assays. A representative experiment is shown. This experiment was repeated three additional times with similar results. In (b), CM was subjected to the indicated treatments and assayed for activity using DDS1626. The relative activity of CM after treatment was determined by the dividing the activation values for the treated CM by the untreated control CM.

The effects of high pH (12.0), low pH (2.0) and extreme heat (95 °C) on signal activity were tested by exposing CM to the above conditions for 30 min for pH treatment and 20 min for heat treatment. The pH-treated CM was then adjusted back to pH 7.5 by adding either NaOH or HCl, respectively. In Fig. 4b, the effect of each treatment on activity is shown. The activity in CM was stable to pH 2 and heating at 95 °C for 20 min, with 84% and 80% activity, respectively, remaining after each treatment (Fig. 4). Exposure to pH 12 resulted in 64% remaining activity, indicating the signal was partially sensitive to alkaline conditions. Consistent with the gel-filtration analysis, the activating signal was small in size and only 15% activity remained after dialysis using membranes with a molecular weight cutoff of 10 kDa (Fig. 4). Finally, protease treatment of CM resulted in a 79% reduction in activity. The partially purified signal obtained after gel filtration was also subjected to the above treatments. The signal was stable to pH 2, pH 12 and heat and pronase treatment resulted in a 80% reduction in activity (data not shown).

Role of previously characterized E. coli cell-to-cell signaling pathways in RpoS regulation

Previously characterized E. coli signals that participate in cell-to-cell signaling include: AI-2 and AI-3, dependent on LuxS for production (Surette & Bassler, 1998; Schauder et al., 2001; Sperandio et al., 2003); indole, dependent on TnaA (Wang et al., 2001), and an uncharacterized signal that requires CysE for production (Sturgill et al., 2004). CM prepared from an E. coli tnaA, cysE double mutant or a luxS mutant activated the RpoS′-′LacZ fusion in a manner that was similar to wild type (data not shown).

In a previous study, a mutation in hldD (formerly rfaD) was shown to result in a sixfold increase in expression of the RpoS′-′LacZ translational fusion in DDS1626 (Joloba et al., 2004). The hldD gene encodes an ADP-l-glycerol-d-mannoheptose-6-epimerase involved in the synthesis of the inner lipopolysaccharide core. To determine whether this increased RpoS expression was the result of enhanced extracellular signal activity, we prepared CM from E. coli MJ11 hldDcat and the isogenic wild-type parent PR1. Unexpectedly, the CM from MJ11 exhibited a significant decrease in signal activity, with activation values approximately sixfold lower than wild-type PR1, 2.6- vs. 16-fold, respectively (Fig. 5). To reconcile the increased expression of RpoS′-′LacZ with the decreased extracellular signal activity, we hypothesized that the extracellular signal was being produced in MJ1, but failed to get properly exported due to the disruption in lipopolysaccharide synthesis and outer membrane structure caused by the hldDcat mutation. The resulting increase in the levels of intracellular signal could then lead to higher levels of RpoS expression. To test this hypothesis, cells of PR1 wild-type and MJ11 hldDcat were lysed using a French Press and the cell lysate from each strain was tested for signal activity. The cell lysate from MJ11 hldDcat activated the RpoS fusion 7.7-fold, which was significantly higher than the cell lysate from wild-type PR1, which activated the RpoS fusion 4.1-fold. Therefore, despite the lack of extracellular signal activity in MJ11 hldDcat, the levels of intracellular signal activity were significantly higher than wild type.

Figure 5

Effect of an hldD mutation on extracellular signal activity. CM or cell lysates were prepared as described in the Materials and methods from Escherichia coli PR1 wild-type or MJ11 hldDcat at an OD600 nm of 0.6. Each medium and control LB was assayed with DDS1626 RpoS750∷LacZ as described in Fig. 1. The reported values represent the fold-activation of RpoS′-′LacZ under each condition, relative to the values when grown in control 0.5 × LB.


In this study, we report that RpoS expression in E. coli is stimulated by a secreted extracellular signal. A key parameter in our ability to detect this extracellular activating signal was the use of nutrient poor growth conditions. In 0.5 × LB, the basal level of expression from a RpoS′-′LacZ fusion was c. 10-fold lower than in normal LB. This lower level of expression allowed us to detect the activating signal in CM prepared from cells grown in 0.5 × LB. Although activation of RpoS expression can still be observed with CM prepared from cells grown in 1 × LB, the peak activation was only two- to threefold (data not shown). One variable contributing to the reduced basal expression in 0.5 × LB appears to be the lower NaCl concentration (data not shown). This raises the obvious question regarding whether the E. coli signal is simply acting like NaCl and raising the osmolarity of the media. Although this is a possible mechanism, there are no precedents for this type of cell-to-cell signaling. In addition, signal activity was lost in a hldD mutant and also by protease treatment, indicating that this was unlikely due to an osmotic effect.

The RpoS-activating signal was stable to acid and heat, but moderately sensitive to alkaline conditions (pH 12.0 for 30 min). In addition, dialysis experiments suggested the signal was below 10 kDa in size. The G-10 gel-filtration studies indicate the signal partitions with the included volume, suggesting the signal is likely to be significantly smaller than 10 kDa in size. In addition, the signal is sensitive to proteases. Whether the signal is a small peptide remains to be determined. Recent studies have indicated that E. coli can produce peptide-signaling molecules (Kolodkin-Gal et al., 2007). However, there are a number of differences between the signal that activates RpoS and the peptide reported in the above study. These differences include the fact that the RpoS-activating signal is unable to bind C-18 columns and is relatively stable at pH extremes.

The production of the signal did not require any of the known loci that participate in extracellular signal production in E. coli. However, spent culture supernatants from a hldD (rfaD) mutant were unable to activate RpoS expression. This defect in signal activity was likely due to a failure to export the signal because cell lysates from a hldD mutant exhibited twofold higher levels of signal activity than wild-type cells. We propose that this export defect is due to the pleiotropic effects of the outer membrane perturbation. For example, heptoseless mutants exhibit; (1) sensitivity to detergents and hydrophobic antibiotics, (2) increased phosphatidylethanolamine in the outer membrane, (3) resistance to lipopolysaccharide-specific phages, and (4) a significant reduction in porin proteins (Nikaido & Vaara, 1985; Austin et al., 1990; Parker et al., 1992; Schnaitman & Klena, 1993). With such an extensive disruption in the structure of the outer membrane, it is likely that movement across the outer membrane in heptoseless mutants is deficient. For example, if a specific outer membrane protein was required for the export (or diffusion) of the activating signal to the outside of the cell, this protein may not be correctly localized or inserted in the hldD mutant. This would lead to accumulation of the signal inside the cell.

Although the activation of RpoS′-′LacZ by the extracellular signal was Hfq dependent, we do not propose that translational changes are involved. The expression of RpoS379′-′LacZ, which contains the native transcriptional and translational signals, but is lacking the turnover sequence for proteolysis is not significantly activated by the extracellular factor. We propose that the apparent requirement for Hfq is due to the fact that RpoS is completely off in a hfq mutant and any pathway that increases expression at a downstream point would be masked. Interestingly RpoS′-′LacZ expression was still activated in a rprA/dsrA double mutant, suggesting the possibility that additional small RNAs are required for translational activation. Additional small RNAs have been described and one or more may regulate RpoS translation (Wassarman et al., 2001; Zhang et al., 2003).

The mechanism by which the extracellular signal increases RpoS expression is likely via inhibiting the RssB(SprE)-ClpXP proteolytic pathway. In a rssB mutant, the levels of RpoS′-′LacZ were only stimulated twofold, an increase that can be attributed to the weak effect of CM on rpoS transcription. Signals that control RssB activity include acetyl phosphate, the IraP protein, the ArcB/ArcA two-component system, and nutrient (glucose, phosphate, nitrogen) starvation (Bouche et al., 1998; Mandel & Silhavy, 2005; Mika & Hengge, 2005; Bougdour et al., 2006). How the extracellular signal integrates into the pathway for RpoS degradation is currently unknown. It may increase the expression or activity of IraP, a protein that binds to RssB and inhibits activity (Bougdour et al., 2006). Alternatively, the signal may act in a pathway that interferes with phosphotransfer to RssB or stimulates a phosphatase and thereby reduces activity and subsequent RpoS proteolysis. The benefit derived from the use of a diffusible signal to increase RpoS accumulation in an E. coli population remains to be determined. The fact that signal activity peaks at mid-log phase and then decreases in stationary phase argues against a mechanism strictly related to increasing RpoS expression in stationary phase cells.


This study was supported by a grant from the National Science Foundation MCB0406047. We are grateful to R. Hengge for providing various lacZ fusions. In addition, we thank Katy Clemmer for help with β-galactosidase assays.


  • Editor: Reggie Lo


View Abstract