Learn more: PMC Disclaimer | PMC Copyright Notice
Signal integration by DegS and RseB governs the σE-mediated envelope stress response in Escherichia coli
Associated Data
Abstract
In Escherichia coli, the σE transcription factor monitors and maintains outer membrane (OM) integrity by activating genes required for assembly of its two key components, outer membrane proteins (OMPs) and lipopolysaccharide (LPS) and by transcribing small RNAs to down-regulate excess unassembled OMPs. σE activity is governed by the rate of degradation of its membrane-spanning anti-σ factor, RseA. Importantly, the DegS protease can initiate RseA cleavage only when activated by binding to unassembled OMPs. The prevalent paradigm has been that the σE response is controlled by the amount of activated DegS. Here we demonstrate that inactivation of a second negative regulator, the periplasmic protein RseB, is also required for σE induction in vivo. Moreover, OMPs, previously known only to activate DegS, also generate a signal to antagonize RseB inhibition. This signal may be lipid related, as RseB is structurally similar to proteins that bind lipids. We propose that the use of an AND gate enables σE to sense and integrate multivariate signals from the envelope.
The outer membrane (OM) of Gram-negative bacteria is their first line of defense against harsh environments. Both key components of the OM, outer membrane proteins (OMPs) and lipopolysaccharide (LPS), have complex assembly and insertion pathways (1–4). In Escherichia coli, OM homeostasis is primarily maintained by a transcription factor, σE, whose regulon includes machinery for assembly and insertion of OMPs into the OM, small RNAs that down-regulate OMP expression, and components controlling LPS synthesis and transport (5, 6).
An elaborate regulatory apparatus monitors envelope status to control σE activity. Under steady-state growth, σE is predominantly bound to RseA, σE’s inner membrane (IM)-spanning anti-σ factor (Fig. 1A) (7–9). Accumulation of unassembled OMPs in the periplasm activates DegS, a trimeric IM-anchored serine protease, to initiate RseA degradation. Free DegS lacks significant catalytic activity because the oxyanion hole of the active site is inappropriately positioned (Fig. 1A) (10). However, binding of the YxF tripeptide at the C terminus of unassembled OMPs to the DegS PDZ domain stabilizes the active protease conformation, allowing it to cleave the periplasmic domain of RseA (Fig. 1B) (11–13), thereby activating the protease cascade that degrades RseA (14–18). This “activation” signal is an excellent indicator of OMP assembly problems, as the OMP C termini are inaccessible in properly assembled OM-localized OMPs (19). Once released from its inhibitory interactions with RseA, σE binds RNA polymerase and activates transcription. Thus, the intracellular concentration of active σE primarily reflects the rate of RseA degradation, enabling fine temporal control over σE activity (16).
Components controlling σE activity. (A) σE is sequestered at the IM by binding to the membrane-spanning anti-σ factor, RseA. The second negative regulator of σE, the periplasmic protein RseB, binds to RseA. Unliganded DegS lacks significant catalytic activity. (B) σE activity is controlled by an AND gate, requiring both DegS activation by unassembled OMPs and RseB inactivation by an alternate cellular signal.
σE also has a second negative regulator, the periplasmic protein RseB, which binds to RseA (Fig. 1A) (8, 20). There is conflicting data about the role of RseB. Strains lacking RseB display only twofold increase in σE activity, suggesting that RseB plays a minor role in the response (8), but in vitro, RseB completely inhibits RseA degradation by OMP-activated DegS (21). Moreover in vivo, OMP peptides are sufficient for σE induction, although were thought only to activate DegS, suggesting that antagonizing RseB inhibition was not critical to the response. Our work clarifies this issue. We show that σE activity is controlled by an AND gate, requiring both DegS activation and RseB inactivation for robust response (Fig. 1B). Consistent with this finding, RseB sets the sensitivity of the σE response to OMP signal. Additionally, we find that OMPs, previously known only to activate DegS, also antagonize RseB by inducing another cellular signal. Because RseB is structurally similar to proteins that bind lipids (22, 23), it suggests that this signal is lipid related. We propose that the σE pathway employs DegS and RseB to sense and integrate multivariate signals from the envelope.
Results
Intertwined Roles of DegS and RseB in σE Induction.
A variant of DegS lacking its PDZ domain (DegSΔPDZ) cleaves RseA in an OMP-independent fashion in vitro, but cells harboring this truncated enzyme display only ∼1.5-fold induction of σE in vivo (Fig. 2A) (11, 12), raising the possibility that DegS activation per se may not be sufficient for full σE induction. The logical candidate for an additional regulator is RseB, which binds tightly to RseA and impedes DegS cleavage of RseA in vitro. Indeed, removal of RseB in a DegSΔPDZ strain led to synergistic σE induction (Fig. 2A). Although ΔrseB and DegSΔPDZ single mutants each activated σE only ∼1.5- to 2-fold, the double mutant ΔrseB DegSΔPDZ strain showed ∼12-fold activation. Additionally, amino acid substitutions in DegS that decrease the free-energy gap between its active and inactive conformations and increase basal cleavage activity in vitro (12, 13, 24) are also synergistic with ΔrseB (Fig. 2B). K243D, which breaks the inhibitory interactions between the PDZ and protease domains of DegS, and H198P, which stabilizes the functional conformation of the oxyanion hole, each induces σE only a few fold, but in combination with ΔrseB showed ∼15- to 20-fold σE induction. Thus, both DegS activation and RseB removal are required for robust activation of σE.
Synergistic effect of DegS activation and RseB removal on σE activity. ΔdegS and ΔrseBΔdegS strains carried plasmids for (A) WT DegS or DegSΔPDZ or (B) WT DegS, K243D DegS, or H198P DegS. Cells were grown to OD600 ∼0.1 and σE activity was measured by monitoring β-gal expression from a chromosomal σE-dependent lacZ reporter. Data were normalized to the ΔdegS strain expressing WT DegS from the plasmid and represent averages (±SD) of three independent experiments.
RseB Determines the Threshold of OMP Signal Required for σE Activation.
RseB modulates the σE induction of active DegS variants, leading us to consider whether it might also modulate σE induction when DegS is activated by OMPs. We compared σE induction in wild-type (WT) and ΔrseB cells using a library of inducers, which consisted of periplasmic cytochrome b562 fused to the C-terminal 50 amino acids of OmpC (Cyt-OmpC50) and variants in which the penultimate residue of the OmpC C terminus (YQF) was changed to correspond to each YxF motif identified in E. coli OMPs (11).
We expressed each YxF fusion protein from the Ptrc promoter in the presence of 1 mM of IPTG and measured σE activity. In WT cells, σE activation correlated with binding affinity of the YxF motif for the PDZ domain of DegS (Fig. 3A), consistent with the fact that this parameter governs DegS activation in vitro (12). In contrast, in ΔrseB cells, all YxF peptides displayed similar activation of σE and were significantly better inducers than in the WT background (Fig. 3B). Therefore, we tested whether the absence of RseB lowered the threshold for OMP activation by assaying σE induction from basal expression of fusion proteins (no IPTG; basal expression from the Ptrc promoter). Under these conditions, graded σE activation was observed in ΔrseB cells, but no induction was observed in WT cells (Fig. 3C). These results are consistent with the idea that RseB modulates σE activation by increasing the OMP peptide threshold.
RseB determines the threshold of OMP signal required for σE activation. (A) σE activity of WT strain expressing high levels of YxF fusion proteins. WT strain carried plasmids for cytochrome b562 (Cyt), a fusion of this protein to the C-terminal 50 residues of OmpC [Cyt-OmpC50(YQF); YQF], or its variants with the C-terminal motifs YYF, YFF, YRF, YKF, YNF, YTF, YSF, and YAF. Cells were grown to OD600 ∼0.1 and induced with 1 mM of IPTG. Samples were collected at four time points during exponential phase and assayed for β-gal activity. Data were normalized to σE activity of the WT strain expressing cytochrome b562 and represent averages (±SD) of three independent experiments. Kact represents the concentration of YxF peptide required for half-maximal activation of DegS cleavage of RseA in vitro (12); ND, not determined. (B) σE activity of ΔrseB strain expressing high levels of YxF fusion proteins. ΔrseB cells carrying plasmids for cytochrome b562 or YxF OMP fusion proteins were grown to OD600 ∼0.1, induced with 1 mM of IPTG, and assayed as described in A. Data were normalized to the WT strain expressing cytochrome b562 and represent averages (±SD) of three independent experiments. (C) σE activity of WT and ΔrseB strains under basal-level expression of YxF fusion proteins (no IPTG). The uninduced sample at OD600 ∼0.1 from A and B was used to measure σE activity from basal fusion protein expression. Data were normalized to the WT strain expressing cytochrome b562 and represent averages (±SD) of three independent experiments. (D) Catalytically active DegS is required for σE activation in the absence of RseB. ΔrseB cells carried plasmids for cytochrome b562 (Cyt) or the YQF-fusion protein, and ΔrseBΔdegS cells carried either Cyt or YQF-fusion protein and plasmid for constitutive expression of either WT DegS or DegS-S201A. Cells were grown to OD600 ∼0.1 and induced with 0.02 mM of IPTG. At an OD600 of 0.25–0.3, samples were harvested for β-gal assays. Data were normalized to the ΔrseB strain expressing cytochrome b562 and represent averages (±SD) of three independent experiments.
We explored alternative explanations for the effect of RseB on σE activation. We tested whether differential σE activation in WT and ΔrseB cells resulted from differential expression of the fusion proteins using SDS/PAGE of periplasmic extracts to examine the proteins in the high-expression regime, and quantitative RT-PCR to examine transcription of fusion proteins in the low-expression regime. Both studies showed that fusion proteins and their transcripts were present at equivalent levels in WT and ΔrseB cells, thus ruling out differential fusion protein expression as the cause of differential σE activation (Fig. S1 A and B). Moreover, the increase in σE activity in ΔrseB cells was not due to cleavage by another protease. Using a ΔrseBΔdegS strain expressing either WT DegS or catalytically inactive DegS (DegS-S201A), we demonstrate that active DegS is required for σE induction upon expression of the YQF fusion (Fig. 3D). These findings together with data that ΔrseB cells expressing low concentrations of YxF fusions exhibit σE activation roughly proportional to the YxF affinities for DegS (Fig. 3C) strongly support a model in which RseB modulates the threshold for OMP-peptide activation of DegS.
Sequence Elements Upstream of the C-Terminal YxF Motif Affect RseB Inhibition in Vivo.
The three C-terminal residues of OMP peptides are sufficient for activating DegS in vitro; additional residues N-terminal to these generally inhibit and sometimes enhance activation (13). We found that significantly more C-terminal OMP residues were necessary for induction in vivo. Using a C-terminal deletion series of fusion proteins containing 10, 20, 30, 40, and 50 C-terminal residues, we found that fusions ending with 20 (OmpC20) or more OmpC residues activated σE similarly, whereas the fusion ending with 10 OmpC residues (OmpC10) displayed weak activity (Fig. 4 A and B). Because σE induction in vivo involves both DegS activation and antagonizing RseB, we tested whether these additional OMP amino acids might be used for the latter process. Indeed, OmpC10 activated σE ∼75% as well as OmpC20 in ΔrseB cells using basal-expression conditions to ensure that σE activity is in the linear range (Fig. 4C). Increasing OmpC10 affinity for DegS by changing its terminal tripeptide to YYF did not significantly activate WT cells but did show full activation in ΔrseB cells, indicating that a process other than DegS activation limits the ability of OmpC10 to induce σE in WT cells (compare Fig. 4 B and C). These results demonstrate that an element located between 10 and 20 residues is required to antagonize RseB inhibition.
OMP peptide relieves RseB inhibition of RseA. (A) C-terminal 20 amino acids of OmpC are required for σE activation in WT cells. WT strain carried plasmids for either cytochrome b562 (Cyt) or a fusion of this protein to the C-terminal 50 (C50), 40 (C40), 30 (C30), 20 (C20), and 10 (C10) residues of OmpC. Cells were grown to OD600 ∼0.1 and induced with 1 mM of IPTG. Samples were collected at four time points during exponential phase and assayed for β-gal activity. Data were normalized to the σE activity of WT strain expressing cytochrome b562 and represent averages (±SD) of two independent experiments. (B) Increasing affinity of the C-terminal 10 residues of OmpC for DegS does not promote significant activation in WT cells. WT cells carried plasmids expressing cytochrome b562, a fusion of cytochrome b562 to the C-terminal 20 or 10 residues of OmpC (C20 YQF, C10 YQF) or its variant with YYF C-terminal motif (C20 YYF, C10 YYF). Strains were grown and sampled as in A. The background β-gal activity of cytochrome b562 was subtracted from the β-gal activity of fusion proteins and the data are presented as percentage of activity of C20 YQF. Data represent averages (±SD) of three independent experiments. (C) C-terminal 10 amino acids of OmpC are sufficient for σE activation in ΔrseB cells. ΔrseB cells carrying plasmids mentioned in B were grown to OD600 ∼0.1. β-gal activity was measured under basal protein expression. The background β-gal activity of cytochrome b562 was subtracted from the β-gal activity of fusion proteins and the data are presented as percentage of activity of C20 YQF. Data represent averages (±SD) of three independent experiments.
We asked whether upstream OMP sequences vary in their ability to counter RseB inhibition. Our test system compared σE induction by “natural inducers” composed of the authentic 50 C-terminal residues of several OMPs with that of an OmpC50 fusion protein having the same YxF motif as its “natural inducer” counterpart. Four natural inducers (OmpX, MipA, OmpG, and FadL) were better than their respective OmpC controls at inducing σE (Fig. 5A). Importantly, these same natural inducers performed the same as or worse than their OmpC counterparts in ΔrseB cells, where induction is dependent solely on DegS activation (Fig. 5B). We conclude that OMPs vary in their ability to antagonize RseB inhibition, just as they vary in their ability to affect DegS activation.
Residues upstream of the C-terminal YxF motif encode RseB modulatory sequences. (A) Comparison of σE activation strengths of natural OMPs with OmpC carrying matched YxF C terminus in WT cells. WT strain carried plasmids expressing cytochrome b562 (Cyt), a fusion of this protein to the C-terminal 50 residues of OmpC [Cyt-OmpC50(YQF); YQF] or its variants with the C-terminal motifs YYF, YFF, YRF, YKF, YNF, YTF, YSF, YAF, and a fusion of cytochrome b562 to ∼50 C-terminal residues of YcbB, YshA, OmpX, MipA, Tsx, OmpT, OmpG, or FadL. Cells were grown to OD600 ∼0.1 and induced with 1 mM of IPTG. Samples were collected at four time points during exponential phase and assayed for β-gal activity. Data were normalized to σE activity of WT strain expressing cytochrome b562. Data shown are the averages (±SD) of three independent experiments. (B) Comparison of σE activation strengths of natural OMPs with OmpC carrying matched YxF C terminus in ΔrseB cells. ΔrseB cells carried plasmids expressing cytochrome b562 (Cyt), a fusion of this protein to the C-terminal 50 residues of OmpC [Cyt-OmpC50(YQF); YQF] or its variants with C-terminal motifs YRF, YKF, YSF, YAF, and a fusion of cytochrome b562 to ∼50 C-terminal residues of OmpX, MipA, OmpG, or FadL. Cells were grown to OD600 ∼0.1 and induced with 0.003 mM of IPTG. Samples were collected at four time points during exponential phase and assayed for β-gal activity. Data were normalized to σE activity of ΔrseB strain expressing cytochrome b562. Data shown are the averages (±SD) of three independent experiments. (Inset) To show clear difference in the σE activation strengths of OmpG and YSF fusion proteins in ΔrseB strain, β-gal activity was measured following protein expression with 0.02 mM of IPTG.
We examined the DegS activation parameters of the OmpX20 and OmpC20 fusion proteins. As OmpX20 activated WT cells twofold better than OmpC20, but was equivalent to OmpC20 in activating in ΔrseB cells, their differential activation of σE in vivo is attributed to differential antagonism of RseB (Fig. 6 A and B). In an in vitro assay measuring DegS cleavage of the periplasmic domain of RseA, both OmpC20 and OmpX20 peptides supported roughly equal rates of DegS cleavage at peptide concentrations below 5 μM, with the OmpC20 peptide showing higher activity at saturating concentrations (Fig. 6C). These results are consistent with our hypothesis that the induction potential of OmpX20 in vivo reflects stronger antagonism of RseB, as OmpX is not better than OmpC at activating DegS in vitro. Importantly, OmpX20 did not antagonize RseB repression of RseA cleavage in vitro (Fig. 6D). Therefore, relief of RseB inhibition is not mediated via direct binding by OmpX20 to RseB, and likely requires additional regulatory components not present in our biochemical assays (see below and Discussion).
Comparison of σE activation, DegS activation, and RseB inactivation strengths of OmpC20 and OmpX20. (A) σE activation of WT cells by OmpC20 and OmpX20. Cells carrying plasmids for cytochrome b562 (Cyt), a fusion of this protein to the C-terminal 20 residues of OmpC (OmpC20) or OmpX (OmpX20) were grown to OD600 ∼0.1 and induced with 1 mM of IPTG. Samples were collected at four time points during exponential phase and assayed for β-gal activity. Data were normalized to σE activity of WT strain expressing cytochrome b562. Data shown are the averages (±SD) of three independent experiments. (B) σE activation of ΔrseB cells by OmpC20 and OmpX20. Cells carrying plasmids mentioned in A were grown to OD600 ∼0.1. β-gal activity was measured under basal protein expression. Data were normalized to σE activity of ΔrseB strain expressing cytochrome b562. Data shown are the averages (±SD) of three independent experiments. (C) Rates of cleavage in vitro of 35S-RseA121-261 (100 μM) by DegS (0.5 μM trimer) were determined in the presence of different concentrations of the OmpX20 or OmpC20 peptides. Data were fit to the Hill equation. For the OmpX20 peptide, half-maximal stimulation was observed at a peptide concentration of 3.6 μM, and the maximal degradation rate constant was 2,350 M-1·s−1. For the OmpC20 peptide, half-maximal stimulation was achieved at a peptide concentration of 10 μM, and the maximal degradation rate constant was 6,860 M-1·s−1. (D) Kinetics of the cleavage in vitro of 35S-RseA121-261 (31 μM) by DegS (0.84 μM trimer) was assayed using the OmpX20 or OmpC20 peptides (6.5 μM) with RseB present (33 μM) or absent. The lines are linear fits.
OMPs Generate Two σE Induction Signals in Vivo.
If OMP accumulation in vivo activates DegS and also induces a second signal that antagonizes RseB, then OMP accumulation might activate σE in a DegS variant unable to bind OMP peptides (DegSΔPDZ). We previously found that overexpression of OmpC50 did not alter DegSΔPDZ activity in vivo (11). Here, we tested whether OmpX50, a strong antagonizer of RseB, activated σE in a DegSΔPDZ strain. Indeed, in DegSΔPDZ, overexpression of OmpX50 induced σE activity twofold, whereas overexpression of OmpC50 did not alter σE activity (Fig. 7). This result is consistent with the idea that some OMPs induce a second signal in vivo that antagonizes RseB inhibition.
OmpX fusion protein induces σE in a DegSΔPDZ strain. Strains carrying DegSΔPDZ on the chromosome as the sole copy of DegS carried plasmids expressing cytochrome b562 (Cyt) and a fusion of this protein to the C-terminal 50 residues of OmpC (OmpC50) or OmpX (OmpX50). Cells were grown to OD600 ∼0.1 and induced with 1 mM of IPTG. Samples were collected at four time points during exponential phase and assayed for β-gal activity. Data were normalized to σE activity of strain expressing cytochrome b562 and represent averages (±SD) of three independent experiments.
Discussion
We have dissected the intertwined roles of DegS and RseB in control of the σE response. Our results discriminate between two models for allosteric activation of DegS and show that DegS and RseB serve as joint gatekeepers for the response. Perturbing either gatekeeper leads to minor induction of the σE response but joint perturbation leads to complete derepression. Two events are thus required for maximal σE response, (i) OMP activation of DegS and (ii) relief of RseB inhibition by an envelope signal. We argue that the use of an AND gate enables the cell to simultaneously sense and integrate multivariate envelope signals.
Activation of DegS in Vivo.
There are two competing models for the role of the PDZ domain in allosteric activation of DegS. In the “peptide-activation” model, the PDZ domain positions a bound OMP peptide, allowing its penultimate side chain to contact the protease domain and alter the conformation or dynamics of the active site to increase activity of DegS (10, 25). Conversely, in the “inhibition-relief” model, the unliganded PDZ domain stabilizes the inactive conformation of the protease domain, and OMP-peptide binding relieves these inhibitory interactions (11–13, 24). Our in vivo results provide two lines of evidence that support the inhibition-relief model. First, weakening the PDZ/protease interface increases rather than decreases DegS activity. K243D eliminates a salt bridge between the PDZ and protease domains (13) and has threefold higher basal σE activity and four- to fivefold higher peptide-stimulated σE activity compared with WT DegS (Fig. 2B and Fig. S2A). Importantly, K243D does not increase activity in the context of DegSΔPDZ (Fig. S2B), as expected if K243 exerts its effect solely by altering the PDZ/protease interface. Mutations such as H198P, which stabilize the active conformation of DegS (24, 26), provide a second line of evidence that the PDZ domain is inhibitory. H198P increases σE activity both in strains containing DegSΔPDZ, which does not bind OMP peptides, and WT DegS (Fig. 2B and Fig. S2 A and B). In summary, our results indicate that the behavior of DegS in vivo is consistent with the predictions of the inhibition-relief model for allosteric activation of DegS.
Relief of RseB Inhibition Requires a Second Signal.
The role of OMP peptides in activating DegS has been well characterized (12, 13, 24). Here we show that OMP peptides also antagonize RseB inhibition. OMP amino acids between 10 and 20 residues from the C terminus are uniquely required for σE induction in the presence of RseB (Fig. 4). OMP peptides differ in their ability to antagonize RseB (Figs. 5 and 6 A and B), just as they differ in their ability to activate DegS. Moreover, the OmpX peptide, identified as more proficient at RseB antagonism, can induce σE even in a DegS variant that lacks the OMP docking domain (DegSΔPDZ), whereas the OmpC peptide, which is less proficient at RseB antagonism, cannot (Fig. 7). Taken together, these results argue that OMP peptides play a dual role in inducing the σE response, both activating DegS and relieving RseB inhibition.
How might OMPs generate a signal that antagonizes RseB in vivo? Our results argue strongly that they generate this signal indirectly, as OmpX is not able to antagonize RseB inhibition in an in vitro cleavage system (Fig. 6D). Because RseB is structurally similar to the LolA, LolB, and LppX lipid-binding proteins (22, 23), the RseB signal is most likely a lipophilic compound such as a periplasmic lipid, lipoprotein, or LPS. Intriguingly, there is a connection between OMP and LPS assembly. The β-barrel assembly machinery (BAM complex) is necessary to insert both OMPs and LptD, the critical player in LPS assembly, into the OM (27–31). Accumulation of OMPs in the periplasm could therefore titrate the BAM machinery away from LptD. Decreased levels of LptD would reduce LPS insertion at the OM, resulting in its accumulation in the periplasm, which could signal the σE pathway via RseB. This model is consistent with the previous data that changes in LPS structure induce σE response (32, 33). We are currently testing whether defects in LPS synthesis/transport activate σE in an RseB-dependent manner.
Output Regime of the σE Pathway.
To provide an optimal response to inducing signals, signal-transduction pathways must sense signals over a wide concentration range, but at the same time discriminate between “stressed” and “unstressed” conditions. The E. coli σE pathway achieves this distinction by using an AND gate to trigger the system. A robust response requires both activating DegS and antagonizing RseB inhibition. The synergistic nature of these two events is shown by differential σE induction by activated DegS variants with and without RseB. When RseB is present, these variants show only minor σE activation, but in the absence of RseB, each of these variants shows high σE induction, consistent with their demonstrated ability to cleave RseA in an in vitro assay (Fig. 2).
This dual requirement for σE induction is also evident in a physiological setting. Whereas high-level expression of a majority of YxF peptides failed to strongly activate σE in WT cells, even weak peptides could achieve significant σE induction in the absence of RseB (Fig. 3 A and B). Thus, removal of RseB magnifies the effects of OMP activation of DegS. RseB can be viewed as a “noise-filtering gatekeeper,” which suppresses the contribution of noise in the signaling process and improves the information quality of the stress signal. This effect of RseB could be explained if RseB and active DegS (stabilized by OMP peptide) compete for binding to RseA. In WT cells, a larger number of active DegS molecules and therefore higher peptide amounts would be required for RseA cleavage compared with ΔrseB cells.
Use of an AND gate enables the cell to integrate multiple signals from the OM before committing to significant activation of σE. Whereas the DegS gatekeeper responds to an unfolded OMP signal, RseB is likely to respond to a lipid signal. Because the σE regulon encodes the machinery for assembly and insertion of both OM proteins and lipids, this signal integration would permit σE to finely tune production of OM biogenesis machinery to the stresses at hand.
Materials and Methods
Media and Antibiotics.
Luria–Bertani (LB) broth was prepared as described (34). When required, LB was supplemented with 100 μg/mL of ampicillin (Ap) or 20 μg/mL of chloramphenicol (Cm). IPTG was added to induce the expression of cytochrome b562-OMP-protein fusions from Ptrc promoter.
Strains and Plasmids.
Strains and plasmids used are listed in Table S1, and details of their construction are available upon request. The ΔrseB strain carrying plasmids for strong σE-activating OMP peptides and the ΔrseBΔdegS strain carrying plasmids for constitutively active DegS variants were unstable as glycerol stocks and lost σE activity over time. Plasmids were freshly transformed to construct these strains.
β-Galactosidase Assays.
σE activity was measured by monitoring β-galactosidase (β-gal) expression from a chromosomal σE-dependent lacZ reporter gene in Φλ(rpoHP3::lacZ), as described (35–37). Cells were diluted from overnight cultures to OD600 ∼0.01 in LB and grown at 30 °C. For single point assays, the sample was harvested at an OD600 listed in the figure legends, and σE activity was calculated as Miller units. For differential plots, several samples were collected between 0.1 and 0.4 OD600, and β-gal activity/0.5 mL cells was plotted against OD600. The slope of these plots is the differential rate of β-gal synthesis and was used as the measure of σE activity. All assays were performed at least twice and usually three times to ensure reproducibility. For β-gal assays involving OMP fusions, the proteins were overexpressed with 1 mM of IPTG in WT cells and with low induction in ΔrseB cells, enabling quantifiable σE activity measurements within the linear range of each strain.
Biochemical Assays.
DegS variants (residues 27–355) with an N-terminal His6 tag and RseA (residues 121–216) with a C-terminal His6 tag were expressed and purified as described (12). RseB (residues 24–318) with a C-terminal Leu-Glu-His6 tag was expressed in E. coli strain X90(DE3) and purified by Ni++-NTA chromatography. The active dimeric fraction of RseB was separated from the inactive oligomeric fraction by gel-filtration chromatography and stored in 50 mM of NaHPO4 (pH 8.0), 200 mM of NaCl, and 10% glycerol (21). The OmpX20 and OmpC20 peptides were synthesized by the Massachusetts Institute of Technology Biopolymers Laboratory and purified by reverse-phase HPLC; their molecular weights were confirmed by MALDI-TOF mass spectrometry. DegS cleavage of 35S-labeled RseA121-216 was performed as described and quantified by scintillation counting of cleavage products soluble in cold trichloroacetic acid (12).
Acknowledgments
We thank Dr. Hana El-Samad and members of the C.A.G. laboratory for helpful suggestions. This work was supported by National Institutes of Health Grants GM-036278-27 (to C.A.G.) and AI-16982-31 (to R.T.S.).
Footnotes
The authors declare no conflict of interest.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1019277108/-/DCSupplemental.