RseB Binding to the Periplasmic Domain of RseA Modulates the RseA:ςE Interaction in the Cytoplasm and the Availability of ςE·RNA Polymerase*

The Escherichia coli ςEregulon has evolved to sense the presence of misfolded proteins in the bacterial envelope. Expression of periplasmic chaperones and folding catalysts is under the control of ςE RNA polymerase. The N-terminal domain of RseA sequesters ςE in the cytoplasmic membrane, preventing its association with core RNA polymerase. The C-terminal domain of RseA interacts with RseB, a periplasmic protein. The relative concentration of ςE:RseA:RseB is 2:5:1 and this ratio remains unaltered upon heat shock induction of the ςE regulon. Purification from crude cellular extracts yields cytoplasmic, soluble ςE RNA polymerase as well as membrane sequestered ςE·RseA and ςE·RseA·RseB. RseB binding to the C-terminal domain of RseA increases the affinity of RseA for ςE by 2- to 3-fold (K d 50–100 nm). RseB binds also to the misfolded aggregates of MalE31, a variant of maltose binding protein that forms inclusion bodies in the periplasm. We discuss a model whereby the RseB-RseA interaction represents a measure for misfolded polypeptides in the bacterial envelope, modulating the assembly of ςE RNA polymerase and the cellular heat shock response.

Folding and assembly of proteins in the bacterial envelope requires mechanisms and factors that are distinct from those of the cytoplasmic heat shock chaperones GroEL/GroES and DnaK/DnaJ/GrpE (1). Protein misfolding in the cytoplasm is sensed directly by the general heat shock transcription factor 32 (RpoH) (2). In contrast, two independent pathways have evolved to sense the presence of misfolded polypeptides in the periplasm and in the outer membrane of Escherichia coli (3). The CpxRA system comprises a two-component signal transduction system. It appears that the histidine kinase CpxA senses the presence of misfolded polypeptides in the periplasm, resulting in autophosphorylation and phosphotransfer to the aspartyl of the cytoplasmic CpxR response regulator. CpxR then activates the transcription of various genes that act on the bacterial envelope (3).
The second pathway requires E , 1 a sigma factor that asso-ciates with core RNA polymerase to generate the transcriptionally active ␣ 2 ␤␤Ј⅐ E (E⅐ E ) complex (4 -7). E RNA polymerase recognizes specific promoters and under heat shock conditions causes a 2-to 3-fold induction of transcription (5,6). Similar to 32 regulation, this modest increase in transcriptional activation is presumably due to the high level of basal expression of E -dependent promoters, because folding catalysts in the bacterial envelope are required at all times, even prior to the exposure to extreme stress. Consistent with this view is the observation that rpoE, the structural gene for E , is essential for the growth of E. coli under all conditions (8). The activity of E RNA polymerase is controlled by rseA and rseB (regulator of sigma E), two genes that are encoded within the E operon rpoErseABC (9,10). Deletion of rseA leads to a 9-fold increase in transcription of E -dependent promoters, whereas deletion of rseB results in a 2-fold stimulation (9,10).
RseA is inserted into the cytoplasmic membrane via a single transmembrane segment, positioning the N-terminal domain in the cytoplasm and the C-terminal domain in the bacterial periplasm. Co-immunoprecipitation and affinity chromatography experiments revealed the existence of interactions between E and RseA and between RseA and RseB. The N-terminal cytoplasmic domain of RseA alone is sufficient to bind E and to prevent transcription of E RNA polymerase (9,10), presumably by sequestering the sigma factor from RNA polymerase. Gross and co-workers (11) showed that, under conditions of extreme stress, i.e. when misfolded outer membrane proteins accumulate in the bacterial periplasm, RseA may be rapidly degraded, leading to transcriptional activation at E -dependent promoters. Proteolysis of RseA appears to require DegS (11). DegS is a putative periplasmic protease anchored to the inner membrane (12). Although these results provide an explanation for the regulation of E RNA polymerase by controlling the stability of the RseA anti-sigma factor, this model does not account for the formation of E ⅐RseA⅐RseB complexes.
We report here the intracellular concentration of E :RseA: RseB, revealing a ratio of 2:5:1. When examined under heat shock, this ratio appeared largely unaltered, and we observed only a modest increase in RseA degradation. Affinity purification experiments revealed the presence of E ⅐RNA polymerase, E ⅐RseA, and E ⅐RseA⅐RseB complexes. RseA binding to RseB increased the affinity of the anti-sigma factor for E . Furthermore, RseA⅐RseB binding sequesters E in the membrane, preventing the association of the sigma factor with core RNA polymerase. We discuss a model whereby the binding of RseB to the C-terminal domain of RseA modulates its affinity for E and the activity of E RNA polymerase.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Growth Conditions-Strains and plasmids used are listed in Table I. Sequences of primers used in this study can be obtained from the authors upon request. Cells were grown in Luria broth (LB) medium (13) at 30°C, using the appropriate antibiotic at the following concentrations: ampicillin (100 g/ml), spectinomycin (50 g/ ml), tetracycline (15 g/ml), and kanamycin (50 g/ml). Induction of RseA His6 and His6 E was accomplished by addition of arabinose (0.2%) to cultures of E. coli BC1⌬(rseArseB) harboring pBC3 or pDM2232. Mutations were transduced into various backgrounds using the P1 bacteriophage (13).
Cloning Procedures and Gene Replacement-Plasmid pBC3 encodes RseA His6 , a variant of RseA with a six-histidine tag appended at the C terminus. The rseA gene was amplified from the chromosome using primers RseA-3 and RseA-6, introducing a 5Ј-NcoI site immediately preceding the start codon and a 3Ј-insertion of six histidine codons followed by a stop codon and a BamHI site. rseA His6 was cloned into pBAD-A (Invitrogen) cut with NcoI and BamHI, yielding pBC3. Plasmid pDM2232 encodes His6 E with a six-histidine tag appended at the N terminus of RpoE. rpoE was polymerase chain reaction-amplified from the chromosome using primers RpoE-1 and RseA-5, introducing a 5Ј-BamHI site and a 3Ј-EcoRI site, respectively. This polymerase chain reaction product was cloned in vector pBAD-B (Invitrogen) using the BglII and EcoRI restriction sites. Purified His6 E was cleaved with enterokinase to remove the N-terminal His tag.
Intracellular Concentration and Fractionation of E , RseA, and RseB-E. coli MC4100 or LMG190 were grown at 30°C or 43°C to A 595 nm 0.6. Culture aliquots were removed and plated on agar medium to count the number of colony forming units per milliliter. Cells of 200-ml culture were collected by centrifugation at 3000 ϫ g for 10 min and washed twice with 0.05 M Tris-HCl, 0.15 M NaCl, pH 7.5. Cells were suspended in 20 ml of the same buffer and lysed in a French pressure cell at 14,000 p.s.i. Unbroken cells were removed by centrifugation at 3000 ϫ g for 10 min. The cell lysate was extracted by the addition of octylglucoside to 2% (v/v), and insoluble components were removed using an Ultracentrifuge at 100,000 ϫ g for 30 min at 4°C. The protein concentration in the supernatant was determined using a colorimetric assay. For analysis on 12% SDS-PAGE, samples were heated at 95°C for 10 min and separated by gel electrophoresis. Proteins were revealed by immunoblotting with ␣-E , ␣-RseA, or ␣-RseB antibodies. Immune complexes were detected with 12 Ci of [ 125 I]protein A per blot, and radioactive signals were quantified by PhosphorImager scanning. Each immunoblot was calibrated with a dilution series of known concentrations of E , RseA, or RseB. For fractionation experiments, cells of 50-ml culture were harvested as described above and suspended in 10 mM HEPES, 66 mM KOAc, 10 mM MgOAc, pH 7.6. After lysis in a French pressure cell at 14,000 p.s.i., samples were centrifuged at 100,000 ϫ g for 30 min at 4°C. The supernatant (soluble proteins) and sediment (membrane proteins) were separated and analyzed by 12% SDS-PAGE, followed by immunoblotting.
Pulse-labeling Experiments-E. coli MC4100 was grown at 30°C to A 595 nm 0.7 in M9 minimal media lacking methionine and cysteine. Cells were pulse-labeled for 1 min with 50 Ci of [ 35 S]methionine/cysteine (Express S35 protein labeling mix, PerkinElmer Life Sciences). Incor-poration of radioactive amino acids was quenched by the addition of excess unlabeled amino acids (per milliliter of culture; 100 l of 10% casamino acids, 2% methionine, and 2% cysteine). The stress response was initiated by shifting the temperature from 30 to 43°C. Two kinds of experiments were conducted. Cultures were labeled at 30°C for 1 min and quenched. Immediately after the addition of the quench (chase), cultures were divided into two equal aliquots. One sample was incubated at 30°C, whereas the other was placed at 43°C. During the other experiment, cultures were divided into two equal aliquots prior to labeling. One culture was incubated at 30°C, labeled for 1 min, and quenched. Another culture aliquot was incubated at 43°C, labeled for 1 min, and quenched. Labeled cultures were incubated for various amounts of time, and a 100-l sample was withdrawn and precipitated with trichloroacetic acid. Precipitates were washed in acetone, suspended in 50 l of 0.1 M Tris-HCl, 4% SDS, pH 7.0, by heating to 95°C for 5 min. The samples were centrifuged at 13,000 ϫ g for 10 min to remove any insoluble material. A 5-l aliquot was removed and analyzed in a scintillation counter to account for the total amount of labeled protein. After adjusting for equal counts, aliquots were subjected to immunoprecipitation and samples analyzed by SDS-PAGE and phosphorimaging. Protein half-lives were determined by analyzing data with the exponential decay equation I t ϭ I 0 ϫ e Ϫkt , where I 0 and I t represent the intensity of the radioactive proteins at time 0 min and time t after beginning of chase. Experiments were conducted in triplicate each time. To determine whether temperature shift to 43°C induced the E regulon, we examined the synthesis rate of ␤-galactosidase driven either by the rpoHP3 or the htrL promoters (known E -and 70 -dependent promoters, respectively), as described by Nagai et al. (14) using the anti-␤-galactosidase antibody from 5 Prime 3 3 Prime, Inc. (Boulder, CO).
Purification and Affinity Measurements-RseA His6 and protein complexes containing RseA His6 were purified from cell extracts of E. coli BC1 (pBC3), a ⌬(rseArseB) derivative of strain LMG194. Cells of 8-liter culture were harvested by centrifugation at 3000 ϫ g for 10 min, suspended in buffer A (50 mM Tris-HCl, 150 mM NaCl, 10% glycerol, pH 8.0), and lysed in a French pressure cell at 14,000 p.s.i. Unbroken cells were removed by centrifugation at 3000 ϫ g for 10 min, and 2% octylglucoside was added to the supernatant. The detergent extract was centrifuged at 100,000 ϫ g for 30 min at 4°C, and the supernatant was subjected to chromatography on 1 ml of Ni-NTA (Qiagen), pre-equilibrated with buffer A. Resin was washed with buffer A supplemented with 10 mM imidazole, and proteins were eluted with buffer A and 0.5 M imidazole. Eluted proteins were analyzed by SDS-PAGE followed either by Coomassie Blue staining or immunoblot analysis using ␣-RseA and ␣-E rabbit serum. Purification of His6 E and protein complexes containing His6 E occurred from extracts of E. coli BC1 (pDM2232) and followed the same protocol.
To obtain pure RseA His6 , Ni-NTA resin was precharged with extracts of E. coli BC1(pBC3) and washed with buffer A. Proteins bound to Ni-NTA were treated with two column volumes of buffer B (50 mM Tris-HCl, 150 mM NaCl, 10% glycerol, 2% octylglucoside, 10 mM imidazole, 4 M urea, pH 8.0) and washed with 100 column volumes of buffer C (50 mM Tris-HCl, 150 mM NaCl, 10% glycerol, 2% octylglucoside, 10 mM imidazole, pH 8.0). Treatment with 4 M urea was observed to release proteins interacting with RseA His6 , yielding a 95% pure preparation of RseA His6 . Isolation of pure His6 E was achieved by submitting extracts of E. coli BC1(pDM2232) to affinity chromatography without detergent Encoding malE31 pBAD-A expressing RseA His Amp R This study pDM2232 pBAD-B expressing His E , Amp R This study extraction. Beads containing His6 E were incubated with enterokinase to cleave off the six-histidine tag, and enterokinase was removed as suggested by the manufacturer (Invitrogen). RseB was purified as described previously using plasmid pSR2695 (10). For affinity measurements, a 50% slurry of Ni-NTA resin was prepared that contained bound RseA His6 (0.1 pmol of RseA/l of slurry). RseA His RseB resin was prepared by incubating RseA His :Ni-NTA with a 50-fold excess of RseB for 2 h at 20°C. Beads were washed exhaustively with 20 column volumes of buffer B to remove unbound material. The concentration of bound RseB was found to be equal to the amount of RseA His present on the beads. Increasing concentrations of E (0 -300 nM) were added to a 1-ml suspension of either 50 l of RseA His :Ni-NTA-Sepharose or 50 l of RseA His RseB:Ni-NTA-Sepharose. Samples were incubated for 2 h at 20°C and centrifuged at 15,000 ϫ g for 5 min. The supernatant (850 l) was removed, and proteins were precipitated with trichloroacetic acid. Sediments were washed with acetone, solubilized in 50 l of 0.1 M Tris-HCl, 4% SDS, pH 7.0, and heated at 95°C for 5 min. Samples were separated on 12% SDS-PAGE, electrotransferred to a polyvinylidene difluoride membrane, and immunoblotted with ␣-E , ␣-RseA, or ␣-RseB antibodies. Immune complexes were detected with 12 Ci of [ 125 I]protein A, and radioactive signals were compared with a dilution series of purified protein with known concentration.
MalE31 Aggregates-E. coli pop6499, i.e. strain MC4100 malT1with a nonpolar deletion of malE, was used as a host for plasmids pHCME and pHCM31, encoding wild-type malE and malE31, respectively (16). E. coli pop6499 (pHCME) and E. coli pop6499 (pHCME31) were grown in LB medium supplemented with 0.1 mg/ml ampicillin at 30°C for 3 h. Cells were harvested by centrifugation, and suspensions were normalized to yield similar A 595 nm . The cells were converted to spheroplasts by suspension in 10 mM Tris-HCl, 0.7 M sucrose, 1 mM phenylmethylsulfonyl fluoride (PMSF), 5 mM EDTA, and incubation with lysozyme (0.2 mg/ml) for 10 min on ice. Samples were centrifuged, and the supernatant (periplasmic fraction) was separated from the sediment. The spheroplast sediment was washed and lysed by repeated freeze-thawing under hypo-osmotic conditions. Samples were centrifuged, and the supernatant (cytoplasm) was separated from the sediment. Pellets were washed twice using 10 mM Tris-HCl buffer, pH 7.5, and suspended in 10 mM Tris-HCl, 2% Triton X-100, 1 mM PMSF, pH 7.5. Samples were incubated at 20°C for 30 min prior to centrifugation. Supernatants were removed and the pellets (insoluble fraction containing the MalE31 aggregates) were suspended in 10 mM Tris-HCl, pH 7.5. All samples were heated at 95°C for 10 min and separated on 12% SDS-PAGE. Proteins were electrotransferred to polyvinylidene difluoride membrane and immunoblotted using ␣-MalE (New England BioLabs), ␣-RseA, ␣-RseB, and ␣-DsbC antibodies.

Intracellular
Concentration of E , RseA, and RseB-To measure the cellular concentration of E , RseA, and RseB, E. coli was grown under non-heat shock as well as heat shock conditions. Cells were lysed and extracted with 2% octylglucoside, and insoluble material was removed by centrifugation. Soluble proteins in the sample were separated on SDS-PAGE and analyzed by immunoblotting using specific antisera. Immunoreactive signals of cell extracts were compared with signals generated by a dilution series of purified protein with known concentration. We measured concentrations of 21 (Ϯ 10) fmol of E , 50 (Ϯ 8) fmol of RseA, and 10 (Ϯ 3) fmol of RseB per g of cellular proteins (numbers in parenthesis represent standard deviations obtained in three independent experiments). Similar ratios (2  (15). Under heat shock conditions, the ratio of E , RseA, and RseB remained unaltered even though the cellular concentration of these polypeptides was increased by 1.5-to 2-fold.
Stability of RseA under Heat Shock Conditions-Previous work reported degradation of pulse-labeled RseA under stress conditions (11). We examined the stability of RseA by growing E. coli cells (MC4100) in minimal medium, pulse-labeling newly synthesized protein with [ 35 S]methionine/cysteine, and inducing stress via temperature shift to 43°C. At timed intervals before and after heat shock, culture aliquots were precipitated with trichloroacetic acid and proteins solubilized in hot SDS. Immunoprecipitated RseA was separated on SDS-PAGE, and radioactive signals were quantified by phosphorimaging (Fig. 1A). Heat shock stress of E. coli cells caused a decrease in the stability of RseA molecules by 2-to 3-fold, i.e. the t1 ⁄2 of RseA was 30 (Ϯ 3) min at 30°C and 11 (Ϯ 3) min at 43°C. A similar half-life (RseA t1 ⁄2 11 (Ϯ 3) min) was measured when E. coli cultures were incubated for 10 min at 43°C prior to pulselabeling. As a control for the induction of the E response, heat shock increased the expression of the rpoHP3-lacZ reporter gene (Fig. 1B). In contrast, expression of htrL-lacZ, a gene transcribed by 70 -RNA polymerase, was not affected by the heat shock treatment (Fig. 1B). No changes in the stability of E and RseB were observed when E. coli were grown at 30°C and 43°C (data not shown). These data corroborate previous observations that RseA is degraded more rapidly at elevated temperatures (11). Using several different genetic backgrounds, Ades et al. (11) observed a more dramatic decrease in RseA stability when cells were shocked via the overexpression conditions. E. coli cells (MC4100) were grown in minimal medium at 30°C, and newly synthesized proteins were pulse-labeled with [ 35 S]methionine/cysteine. Culture aliquots were either incubated at 30°C (non-heat shock) or 43°C (heat shock). At timed intervals before and after the heat shock signal, culture aliquots were precipitated with trichloroacetic acid and proteins were solubilized in hot SDS. B, synthesis rate of ␤-galactosidase driven from the rpoHP3 E⅐ E -dependent promoter (filled squares) or the htrL E⅐ 70 -dependent promoter (open circles). Cells were grown in minimal medium at 43°C, and proteins were labeled with [ 35 S]methionine/cysteine for 1 min and chased with excess unlabeled methionine/cysteine (200 g/ml) for 1 min prior to trichloroacetic acid precipitation. Immunoprecipitated RseA and ␤-galactosidase were separated on SDS-PAGE, and radioactive signals were quantified using a PhosphorImager (Molecular Dynamics). Experiments were performed in triplicate, and results were averaged. A and B show data that are representative of five independent experiments. of outer membrane protein. This fulminant degradation of RseA appears to be catalyzed by the periplasmic protease DegS and represents a fascinating mechanism of regulation. However, such dramatic changes in RseA concentration are clearly not required for the induction of the E regulon under heat shock conditions.
RseA Promotes Membrane Sequestration of E -RseA binding to E prevents RNA polymerase function in vitro, and deletion of rseA leads to a constitutive activation of the E -dependent response in vivo (9, 10). We wondered how many E and RseB molecules were associated with RseA under normal growth conditions. E. coli cells (MC4100) were lysed in a French pressure cell at 14,000 p.s.i., and unbroken cells were removed by slow speed centrifugation. The supernatant was centrifuged at 100,000 ϫ g for 20 min. Soluble periplasmic and cytoplasmic proteins were separated with the supernatant from the sediment containing membrane proteins. Proteins in both fractions were precipitated with trichloroacetic acid and analyzed by immunoblotting (Fig. 2). RseA sedimented with the membranes, whereas DsbC, the periplasmic protein disulfide reductase, remained soluble in the sample supernatant. In contrast, RseB and E associated in part with the cytoplasmic membrane, because 27% (Ϯ 5) of RseB and 64% (Ϯ 4) of E were found in the sediment. Membrane association of RseB and E absolutely required RseA, because RseB and E failed to sediment during the centrifugation of extracts obtained from rseA mutant E. coli cells (BC38). When extracts of rseB mutant E. coli (BC39) were subjected to centrifugation analysis, 36% (Ϯ 4) of E was found in the membrane, instead of the 64% previously observed in wild-type cells. The reduced membrane association of E was not due to a decrease in the number of RseA molecules, because the concentration of RseA in the rseB deletion mutant was similar to that of wild-type cells. Thus, although RseB was not essential for the binding of E to RseA, association of E :RseA appeared to be increased in the presence of RseB, suggesting that RseB binding may affect the affinity of E :RseA interactions.
Isolation of E ⅐RseA⅐RseB and E ⅐RseA Complexes-To examine the molecular architecture of the E ⅐RseA⅐RseB complex in greater detail, we expressed RseA His or His E in wild-type (LMG194), rseA Ϫ (same mutation as BC38 transduced in LMG194) or ⌬(rseArseB) (BC1) mutant strains. E. coli cells were lysed in a French pressure cell, and samples were extracted with octylglucoside and centrifuged at 100,000 ϫ g. Initial experiments determined the amount of detergent required for solubilization of E ⅐RseA⅐RseB. Addition of 2% octylglucoside solubilized the membrane envelope and prevented sedimentation of RseA at 100,000 ϫ g. These conditions were used for the preparation of all cell extracts that were subjected to affinity chromatography on Ni-NTA resin and eluted with 0.5 M imidazole. The load and eluate fractions were analyzed by immunoblotting to reveal the co-purification of various components (Fig. 3). Expression of RseA His in wild-type and rseA Ϫ mutant E. coli led to the co-purification of E and RseB (Fig.  3A). Moreover, expression of RseA His in ⌬(rseArseB) mutant E. coli did not prevent the co-purification of E with RseA His (Fig.  3A, right panel). These data support our observation that RseA alone is sufficient to sequester E in the membrane compartment.
We tested two models for RseA-mediated regulation of E RNA polymerase. RseA might capture E RNA polymerase and arrest the holoenzyme in an inactive state. Alternatively, RseA might sequester E , preventing the assembly of ␣ 2 ␤␤Ј⅐ E holoenzyme. To distinguish between the two models, affinitypurified RseA His complexes were probed by immunoblotting with ␣-RpoA for the presence or absence of the ␣ subunit of RNA polymerase. Expression of RseA His in wild-type or rseA Ϫ mutant E. coli did not lead to the co-purification of RpoA (Fig.  3A). As a control, affinity chromatography of His E resulted in the co-purification of RpoA (Fig. 3B). This result was expected, because His E was able to replace E in run-off transcription assays performed in vitro (data not shown) and hence allowed formation of ␣ 2 ␤␤Ј⅐ His E polymerase. Together our data indicate that binding of E to RseA prevents the association of the sigma factor with RNA polymerase but does not sequester preformed ␣2␤␤Ј⅐ E complexes in the membrane. Co-purification of E , RseB, and RseA His can be explained as the association of E with RseA, or association of RseA with RseB, or the formation of ternary E ⅐RseA⅐RseB complexes. To distinguish between these possibilities, we expressed His E in rseA Ϫ or ⌬(rseArseB) mutant E. coli. His E co-purified with RseA and RseB from wild-type cells, however, RseB was not retained on the affinity column when His E was analyzed in rseA Ϫ cells (Fig. 3B). Thus, co-purification of E ⅐RseA⅐RseB represents a mixture of binary E ⅐RseA and ternary E ⅐RseA⅐RseB complexes.
Affinity of E for RseA or RseA⅐RseB-The C-terminal RseB binding domain of RseA is positioned in the bacterial periplasm, whereas the N-terminal E binding domain of RseA is located in the cytoplasm. Fusion of the C-terminal domain of RseA to glutathione S-transferase revealed that RseB binding occurred without the N-terminal domain of RseA (data not shown). Presumably, membrane-inserted RseA alternates between two states, unbound (free) RseA and RseA⅐RseB. We   FIG. 2. RseA-mediated sequestration of RseB and E in the bacterial membrane envelope. E. coli cells (MC4100) were lysed in a French pressure cell, and lysates were centrifuged at 100,000 ϫ g to separate soluble proteins in the cytoplasm and periplasm (S, supernatant) from insoluble proteins in the membrane envelope (P, pellet). Samples were precipitated with trichloroacetic acid and analyzed by immunoblotting with ␣-RseA, ␣-RseB, ␣-E , and ␣-DsbC. In wild-type cells, 27% of RseB and 64% of E were sequestered in the membrane envelope. As a control, all RseA (cytoplasmic membrane) sedimented during centrifugation (using an Ultracentrifuge), whereas all DsbC (periplasm) remained soluble. Neither RseB nor E were sequestered within the membrane envelope of rseA mutant E. coli cells. rseB mutants (BC39) contained less sequestered E (36%) than wild-type cells (64%). wondered whether RseA and RseA⅐RseB displayed different affinities for E and asked whether the anti-sigma activity of RseA is modulated by the binding to RseB. RseA His was purified from the crude cellular extracts of E. coli BC1 ⌬(rseArseB), and E was removed by washing the resin with two column volumes of urea-containing buffer, yielding a 95% pure preparation of RseA His chelated to Ni-NTA beads. A 50% slurry of resin was dispensed into buffer containing purified E , and RseA His -E binding was measured as the amount of E that sedimented during slow speed centrifugation. In a second experiment we measured the affinity of E for the RseA His ⅐RseB complex. The results in Fig. 4 indicate that RseA His bound E with 2-to 3-fold higher affinity when complexed to RseB. Variations of the values obtained for the dissociation constant were observed for different preparations of RseA His . However, a 2-to 3-fold difference was reproducibly measured in five independent experiments. Data in Fig. 4 were used to calculate dissociation constants, yielding K d 50 nM for E /RseA His ⅐RseB and 100 nM for the E /RseA His interaction, respectively. The insert panel of Fig. 4 shows a Scatchard plot analysis of the data, revealing a 1:1 stoichiometry of E :RseA in the presence or absence of RseB. Together the data suggest that RseB binding to RseA modulates the affinity of RseA for E and the concentration of E RNA polymerase (Fig. 2).
RseB Binds the Periplasmic Aggregates of Misfolded MalE31-MalE31 is a mutant maltose binding protein that exhibits the remarkable property of aggregating in the periplasmic space of E. coli, after the translocation of MalE31 across the cytoplasmic membrane and signal peptide processing of the precursor. Expression of MalE31, but not of wild-type MalE, leads to the induction of E RNA polymerase activity at the rpoH3 promoter, suggesting that the RseAB signaling device may sense the presence of misfolded (aggregated) MalE31 in the periplasmic space (7,16,17). To examine whether RseB binds to misfolded polypeptide, we asked whether RseB is sequestered within macromolecular aggregates that formed during overproduction of MalE31. E. coli carrying a knockout mutation of the malE gene was transformed with plasmids that allowed overproduction of either wild-type malE or malE31, respectively. After EDTA-and lysozyme-mediated spheroplasting of E. coli cells, samples were centrifuged, and the periplasmic contents in the supernatant (lane labeled Per, Fig. 5) were separated from the spheroplast sediment. Spheroplasts were lysed by multiple freeze thawing, and soluble proteins were separated by centrifugation. The remaining pellets were analyzed by immunoblotting (lane labeled Mb, Fig. 5), and their membrane protein contents were solubilized by extraction with Triton X-100 and separated from insoluble aggregates by a final centrifugation step (lanes labeled Sol and Ins, respectively, Fig. 5). Wild-type MalE was found soluble in the periplasmic space of fractionated E. coli cells (Fig. 5). In contrast, the periplasmic aggregates of MalE31 remained insolu- ble throughout this fractionation procedure and sedimented even after Triton X-100 extraction. As a control, Triton X-100 extraction of such samples resulted in solubilization of RseA. When examined in malE mutant E. coli or in strains overproducing wild-type MalE, RseB was found in the periplasm as well as associated with the fraction containing the membrane proteins. However, overproduction of MalE31 sequestered most RseB molecules within the Triton X-100-insoluble aggregate. We think it is likely that the sequestration of RseB in MalE31 aggregates prevents the association of RseB with RseA, thereby triggering the release of E . This hypothesis is also corroborated by the co-purification of RseB and MalE31, but not of RseA or E , during sedimentation centrifugation and affinity chromatography (data not shown). Overproduction of MalE31 did not alter the overall concentration of RseA, RseB, or E (Fig. 5; and data not shown) consistent with our earlier observations on the stability of these factors under non-heat shock and heat shock conditions ( Fig. 2; and data not shown). As a control, periplasmic DsbC was not captured within the aggregates of MalE31. DISCUSSION We have examined the role of RseA⅐RseB in signaling the presence of misfolded polypeptides in the periplasm of E. coli. Our current model predicts that RseA functions as an antisigma factor, sequestering E from core RNA polymerase (␣ 2 ␤␤Ј) (Fig. 6). Calculating the amounts of free and sequestered E from the dissociation constant and cellular concentrations of E and RseA, 99% of the sigma factor should be tethered to RseA. This is clearly not the case as 36% of E in wild-type and 64% in rseB mutant cells remained soluble within the bacterial cytoplasm. These soluble E molecules are shown to be associated with RNA polymerase (␣ 2 ␤␤Ј⅐ E ) (Fig.  2), and we presume that very little sigma factor exists in an unbound (free) state. Thus, the concentration of soluble sigma factor must be accounted for by the affinity of E for RseA, RseA⅐RseB, as well as RNA polymerase core enzyme (␣ 2 ␤␤Ј). We assume that E displays greater affinity for RNA polymerase (␣ 2 ␤␤Ј) than for RseA. Overproduction of RseA should shift the physiological balance toward the sequestration of E , preventing the formation of ␣ 2 ␤␤Ј⅐ E RNA polymerase. This prediction is fulfilled as increased expression of RseA, similar to a FIG. 5. RseB is captured within aggregates of misfolded MalE31. E. coli cells carrying a deletion of malE (pop6499) were transformed with plasmids encoding wild-type MalE or MalE31. Cellular fractions were analyzed by immunoblotting with ␣-RseA, ␣-RseB, ␣-MalE, and ␣-DsbC. During fractionation, E. coli cells were treated with EDTA/lysozyme and centrifuged to separate proteins soluble within the periplasm (Per) from the spheroplast sediment. Spheroplasts were lysed and pellets (Mb) further extracted with Triton X-100 and centrifuged to separate solubilized membrane proteins in the supernatant (Sol) from insoluble aggregates in the sediment (Ins). Wild-type MalE and DsbC were found in the periplasmic space, whereas MalE31 formed Triton X-100-insoluble aggregates. RseA sedimented into the membrane fraction, but could be solubilized by Triton X-100 extraction. RseB was located within the insoluble aggregates of MalE31-producing cells. However, in the absence of MalE31, RseB was located predominantly in the soluble periplasmic-or Triton X-100-extractable fractions.
FIG. 6. Model for the sensing of misfolded polypeptides in the envelope of E. coli. In the absence of misfolded polypeptide (i), some of the extracellular sigma factor was sequestered from RNA polymerase in a ternary, high affinity E ⅐RseA⅐RseB complex. An increase in the concentration of misfolded polypeptide (ii) resulted in the release of RseB and in the formation of a low affinity E ⅐RseA complex. The sigma factor was released from the low affinity complex (iii) to associate with RNA polymerase (␣ 2 ␤␤Ј⅐ E ) and initiate transcription of heat shock promoters.
FIG. 4. Affinity of RseA His and RseA-His -RseB for E . Ni-NTA resin precharged with either RseA His (0.1 nmol/ml resin, closed triangles) or RseA His -RseB (both at 0.1 nmol/ml resin, closed circles) were incubated with purified E . The binding of E to RseA His or RseA His -RseB was assessed by co-sedimentation of E with Ni-NTA-Sepharose beads during slow speed centrifugation. The concentration of E in the supernatant was measured by immunoblotting using [ 125 I]protein A and PhosphorImager analysis. Increasing amounts of E were added to the beads and plotted against the concentration of E that co-sedimented with Ni-NTA. The inset shows a Scatchard plot analysis of the data, revealing a 1:1 association between E and RseA His both in the presence or absence of RseB. Dissociation constants K d 50 nM ( E / RseA His ⅐RseB) and K d 100 nM ( E /RseA-His ) were calculated.
reduction of E expression, is toxic for E. coli cells, causing the rapid accumulation of suppressor mutations.
To alter the availability of E for core RNA polymerase under stress conditions, cells could either modify the affinity of the RseA-sigma factor interaction or reduce the relative concentration of the anti-sigma factor without altering that of E or RNA polymerase. In fact, both mechanisms appear to exist. Ades et al. (11) reported specific, rapid degradation of all RseA molecules by the DegS protease in response to overproduction of outer membrane proteins. Due to its location in the periplasm, the DegS protease likely cuts the C-terminal portion of RseA, without affecting the N-terminal domain. DegS-initiated degradation is presumably completed by other proteases that cleave the N-terminal sigma factor binding domain of RseA. It appears that the massive degradation of RseA represents a state of extreme stress, because we did not observe massive RseA degradation when E. coli cells were exposed to heat shock or overproduction of MalE31. Under these conditions, the overall concentration of E , RseA, and RseB was unaltered even though transcription from ␣ 2 ␤␤Ј⅐ E RNA polymerase-dependent promoters was increased by about 3-to 4-fold, respectively. Our data suggest that the association and disassociation of RseB with the C-terminal domain of RseA modulates the affinity of the anti-sigma factor complex for E (Fig. 6). We propose further that RseB binding to misfolded polypeptide affects its ability to interact with the C-terminal domain of RseA, thereby reducing the affinity of the anti-sigma factor and allowing for the association of E with core RNA polymerase. If so, one could view RseA⅐RseB complexes as representing a switch with an affinity for E that is equal or slightly above that of the sigma factor for RNA polymerase. Dissociation of RseB from the ternary E ⅐RseA⅐RseB complex may flip the switch and promote release of the sigma factor from RseA. In this manner, the E ⅐RseA⅐RseB complex may respond rapidly and reversibly to changes in the cellular environment that affect the folding of polypeptides in the bacterial envelope.
Regulatory mechanisms that provide for a cellular response to environmental stress are conserved in other microorganisms. E belongs to a family of sigma factors, members of which regulate extra-cytoplasmic functions and are referred to as extra-cytoplasmic factors (18,19). Most Gram-negative organisms contain one or more rpoErseABC like operon (20). Extracytoplasmic factors are also present in Gram-positive bacteria along with a cognate anti-sigma factor (20 -22). Pseudomonas aeruginosa regulates the expression of alginate, an extracellular polysaccharide virulence factor, via AlgU/AlgT (RpoE) RNA polymerase (23,24). algU is the first gene of an operon, includ-ing algUmucABC. mucABC represent homologs of rseABC (25,26). Deletion of mucA or mucB causes an increase in transcription of AlgU-dependent promoters, resulting in massive expression and secretion of alginate and a hypermucoid colony phenotype (25). Alginate production is not accompanied by a degradation of the MucA anti-sigma factor (27). Future work will need to examine the interaction of E with RNA polymerase to reveal the molecular mechanisms underlying the release of E from RseA⅐RseB complexes in the presence of misfolded polypeptide.