ABSTRACT
The Escherichia coli soxRS regulon activates oxidative stress and antibiotic resistance genes in two transcriptional stages. SoxR protein becomes activated in cells exposed to excess superoxide or nitric oxide and then stimulates transcription of the soxS gene, whose product in turn activates >= 10 regulon promoters. Purified SoxR protein is a homodimer containing a pair of [2Fe-2S] centers essential for soxS transcription in vitro. The [2Fe-2S] centers are thought to be anchored by a C-terminal cluster of four cysteine residues in SoxR. Here we analyze mutant SoxR derivatives with individual cysteines replaced by alanine residues (Cys -> Ala). The mutant proteins in cell-free extracts bound the soxS promoter with wild-type affinity, but upon purification lacked Fe or detectable transcriptional activity for soxS in vitro. Electron paramagnetic resonance measurements in vivo indicated that the Cys -> Ala proteins lacked the [2Fe-2S] centers seen for wild-type SoxR. The Cys -> Ala mutant proteins failed to activate soxS expression in vivo in response to paraquat, a superoxide- generating agent. However, when expressed to ~5% of the cell protein, the Cys -> Ala derivatives increased basal soxS transcription 2-4-fold. Overexpression of the Cys119 -> Ala mutant protein strongly interfered with soxS activation by wild-type SoxR in response to paraquat. These studies demonstrate the essential role of the [2Fe-2S] centers for SoxR activation in vivo; the data may also indicate oxidant-independent mechanisms of transcriptional activation by SoxR.
Bacteria regulate many genes in response to imbalances in the production and disposal of reactive oxygen species; such conditions are often called `oxidative stress' (1 ). In Escherichia coli, a set of ~12 dispersed promoters under the control of the soxRS locus is activated when cells are exposed to sublethal levels of compounds, such as paraquat (PQ) (2 ,3 ), which generate intracellular superoxide, or to nitric oxide (4 ,5 ). This soxRS regulon controls antioxidant functions [e.g. superoxide dismutase or glucose-6-phosphate dehydrogenase (1 )], repair of oxidative DNA damage [endonuclease IV (1 )] and antibiotic resistance genes [micF (6 ) or acrAB (7 )]. The soxRS system may also provide resistance to the toxicity of organic solvents and some heavy metals (8 ). The soxRS regulon is switched on in two transcriptional stages: existing SoxR protein is activated by an intracellular redox signal and triggers transcription of the soxS gene; the resulting SoxS protein binds and activates transcription from the various regulon promoters (1 ).
SoxR protein is the master regulator of the soxRS response (9 ,10 ). SoxR in vitro binds and strongly stimulates transcription of the soxS promoter by the exponential-phase RNA polymerase (RNAP) containing the [sigma]70 protein (11 ). The activity of SoxR as a transcription factor is completely dependent on the presence of non-heme iron in the protein (11 ). The metal is present in the active homodimeric SoxR protein as a pair of [2Fe-2S] clusters, which are in the oxidized form when the protein is isolated from cells (12 ,13 ). SoxR activity can be regulated either by the assembly and disassembly of its iron-sulfur clusters (14 ,15 ) or by oxidation-reduction of the [2Fe-2S] centers, with the oxidized protein being the transcriptionally active form in vitro (16 ,17 ). Recent in vivo experiments support the interpretation that reduced SoxR is transcriptionally inactive (18 ).
From the foregoing it is clear that the [2Fe-2S] centers in SoxR are intimately involved in the protein's function as a transcription regulator in vitro, but their in vivo importance has not been validated. Attempts to control iron availability are complicated by the essential roles of Fe in other proteins, and by the complex regulation governing the assimilation and storage of this metal (19 ). Initial spectroscopic analysis indicates that each [2Fe-2S] center in SoxR is anchored by four thiol ligands, which accounts for the four cysteine residues present in each protein monomer (12 ,13 ). In this work we have tested the biological role of the [2Fe-2S] centers by mutating the individual cysteine residues of SoxR. These mutations eliminate the activation of SoxR by oxidative stress in vivo and prevent the assembly of stable iron-sulfur centers in the protein, but still allow a significant level of basal SoxR activity when the mutant proteins are expressed at high levels.
This work employed the E.coli K-12 strains carrying single copy operon fusions present in lysogenized phage [lambda]: strain TN521, a derivative of GC4468 (2 ) but [Delta]soxRS::zjc2205-kan [lambda][Phi](soxR+ soxS'::lacZ) (20 ); and strain TN5311, as TN521 except [lambda][Phi]([Delta]soxR soxS'::lacZ) (20 ) and containing F' proAB lacIq lacZ[Delta]M15::Tn10 (21 ), transferred from XL1-blue (Stratagene, La Jolla, CA) by conjugation. The pBluescript plasmid was from Stratagene and does not contain an inducible promoter for recombinant gene expression. The plasmid pSE380 contains the lacIq gene and the lac repressor-regulated trc promoter for controlled expression of recombinant genes (Invitrogen, San Diego, CA). The plasmid pKEN2 [a generous gift of G. Verdine, Harvard University (11 )] contains a lac-regulated derivative of the tac promoter for controlled expression of recombinant genes but does not contain its own lacI gene; expression from pKEN2-derived plasmids is regulated by chromosomal lacI.
Four mutant soxR genes, each encoding an alanine residue substituting for a SoxR cysteine (residues 119, 122, 124 and 130), were constructed by using a plasmid containing the soxRS locus [pBD100; (22 )] as a template for sequential rounds of polymerase chain reaction (PCR). For the first round of amplification, primers T3 (5'-ATTAACCCTCACTAAAG-3') and T7 (5'-AATACGACTCACTATAG-3'), specific for sequences in the pBluescript KS vector were used together with mutagenic primers converting the cysteine codon (TGT or TGC) to an alanine codon (GCT or GCC; underlined below).
a) C119A5'-GAACTGGACGGA
Overlapping PCR products were generated using the pairs of primers, pBD100 (10 ng) as the template, and 1 U cloned Pfu DNA polymerase (Stratagene) and isolated from the reaction mixtures using a commercially available spin column (Qiagen, Chatsworth, CA). Approximately 10 ng of each pair of overlapping soxR fragments were then combined with 50 pmol of the T3 and T7 primers in a 100 [mu]l reaction mixture. After denaturation and reannealing of the overlapping complementary cysteine coding regions, PCR yielded fragments (728 bp) containing the full-length mutated genes. These were purified and digested with EcoRIand HindIII (New England Biolabs, Beverley, MA), electrophoresed in a 1% agarose gel in Tris-acetate/EDTA buffer (23 ) and recovered from gel slices using DEAE membranes (Schleicher and Schuell, Keene, NH). The purified fragments were subcloned into EcoRI/HindIII-digested pBluescript and transformed into XL1-Blue cells. Transformants were selected on LB agar plates (21 ) containing 75 [mu]g/ml ampicillin, 15 [mu]g/ml tetracycline and the LacZ+ indicator 5-bromo-4-chloro-3-indolyl-[beta]-d-galactopyranoside (21 ). White colonies were selected and analyzed by PCR using the T3 and T7 primers to verify the presence of full-length soxR in the plasmids. Plasmid DNA was purified from positive clones and both strands of the soxR inserts completely sequenced by the dideoxy chain termination method (23 ) using Sequenase II (US Biochemical) and the following primers: T3, T7, SRKR (5'-GGGCAACACGCCAAACGCT-3'), SRKF (5'-AGCGTTTGGCGTCTTGCCC-3'), E4 (5'-GCGCGGATCCCAGCGGCGATATAAA-3') and E5K (5'-GCGCGAGCTCGCTTTCG- TCCCAATGG-3').
The four mutant cysteine sequences were subcloned from the pBluescript vectors into pKEN2, generating the plasmids pKENC119A, pKENC122A, pKENC124A and pKEN130A. The C119A mutant gene was also subcloned into pSE380 to generate pSEC119A for subsequent studies. Despite several attempts, stable insertion of the other cysteine mutant genes into pSE380 was unsuccessful.
Plasmids derived from pKEN2 were transformed into E.coli strain TN5311 and grown in 200 ml LB broth (21 ) containing 75 [mu]g/ml ampicillin, 15 [mu]g/ml streptomycin and 15 [mu]g/ml kanamycin to OD600 [approx] 0.5-1. Isopropyl-[beta]-d-thio-galactopyranoside (IPTG) was then added to a final concentration of 0.5 mM, and the incubation was continued for 120 min at 37oC. The cells were harvested by centrifugation at 10 000 g at 4oC, resuspended and washed three times with ice-cold M9 salts (21 ). The final cell pellet was resuspended to a volume of 1.0 ml with 50 mM HEPES-NaOH, pH 7.6, 0.1 M NaCl, and lysed by agitation with glass beads (400 [mu]l beads/ml) in a Mini-bead beater (Biospec Products, Bartlesville, OH) for 3 min. Following centrifugation at 10 000 g for 45 min at 4oC, the supernatants were collected and frozen at -20oC or -80oC until assay. Cell extracts were resolved on 15% SDS-polyacrylamide gels (24 ) and stained with Coomassie blue as previously described (11 ).
SoxR mutant proteins were also extensively purified using DE52-Sepharose and heparin-agarose column chromatography as described previously (11 ).
The Fe content of partially purified SoxR proteins (11 ) was determined by inductively coupled plasma emission spectrometry at the Chemical Analysis Laboratory, Institute of Ecology, University of Georgia (Athens, GA). SoxR concentrations were estimated by scanning densitometry analysis (Visage System, Millipore, Milford, MA) of Coomassie blue-stained SoxR in SDS-polyacrylamide gels (24 ) using a SoxR standard previously quantified by amino acid analysis.
The binding affinity of wild-type and mutant SoxR proteins for the soxS promoter in vitro was analyzed by electrophoretic mobility shift assays, as previously described (9 ,11 ).
TN5311 transformed with the expression plasmid pKEN2 or its soxR-containing derivatives pKOXR [containing the wild-type soxR gene; (22 )], pKENC119A, pKENC122A, pKENC124A or pKENC130A were grown overnight in LB broth containing ampicillin (100 [mu]g/ml). A fresh 125-ml aliquot of the same medium was inoculated with 1.25 ml of overnight culture and incubated at 37oC with shaking at 250 r.p.m. for 110 min. IPTG was added to a final concentration of 0.5 mM, and the incubation continued at 37oC with shaking for an additional 120 min. The cells were harvested by centrifugation and resuspended in 0.5 ml of 50 mM HEPES-NaOH, pH 7.6, 0.1 M NaCl. Freshly dissolved dithionite, 0.1 M dithionite in 1 M HEPES-NaOH pH 7.6, was then added to the cell paste to a final concentration of 1 mM. Aliquots (300-400 [mu]l cell paste) were immediately placed inside 4 mm EPR sample tubes, frozen in liquid nitrogen and kept at -80oC until analysis. The total SoxR concentration in the cells was determined by Western blot analysis of samples lysed directly in sample buffer (24 ), using previously quantified SoxR as a standard.
Wild-type E.coli SoxR protein was purified to near homogeneity as described previously (11 ). The purified protein was emulsified in complete Freund's adjuvant and injected subcutaneously into two New Zealand black rabbits. Starting three weeks after the primary injection, the rabbits received booster injections every 2 weeks with similar SoxR preparations in incomplete Freund's adjuvant. Polyclonal antisera were extracted periodically starting 4 weeks after the first injection, and SoxR-specific antibodies were enriched by affinity chromatography using SoxR-columns generated by coupling purified SoxR protein to HiTrap NHS- activated columns (Pharmacia) according to the conditions recommended by the supplier. Polyclonal serum was applied to the columns and SoxR-specific antibodies eluted following standard procedures (25 ).
For immunoblotting, samples of cell suspensions or purified SoxR were electrophoresed in SDS-polyacrylamide gels, and transferred to nitrocellulose membranes (Schleicher and Schuell, Keene, NH) with a TE series Transphor electrophoresis unit (Hoefer Scientific, San Francisco, CA). The filters were probed with the affinity-purified antisera and bound antibody was detected with alkaline phosphatase-conjugated anti-rabbit IgG antibodies (Promega, Madison, WI).
X-band EPR spectra were recorded at 20 K on a Bruker model ESP300 spectrometer maintained at constant temperature, with an Oxford Instruments model ESR910 continuous flow cryostat as described previously (14 ,15 ). The amount of reduced SoxR was determined by comparison in the same experiment to standardized Fe-SoxR samples after reduction with dithionite (13 ,14 ). The high EPR background noise of the cell paste was greatly reduced by electronically subtracting from the spectra for SoxR-containing samples the spectrum of TN5311-pKEN2 cells, which do not express SoxR.
The activity of wild-type and mutant SoxR proteins was determined by in vitro transcription of the soxS gene by commercial E.coli RNA polymerase accompanied by the indicated amounts of SoxR, as described previously (13 ,14 ).
The ability of the mutant proteins to stimulate soxS transcription in vivo was assessed using TN521 and TN5311. TN5311 cells were transformed either with the pBluescript-based plasmids for low-level SoxR expression, or with the pKEN2-based plasmids for high-level expression. TN521 cells were also transformed with the pSEC119A plasmid. Overnight cultures of the indicated strains were diluted 1:100 (or in some experiments 1:1000) into fresh LB broth containing 50 [mu]g/ml ampicillin and treated as follows. For TN5311 containing pBluescript-based plasmids, PQ was added to a final concentration of 100 [mu]M to one aliquot at OD600 [approx] 0.4, and incubation at 37oC continued 60 min. For TN5311 containing pKEN2-based plasmids, when the cell density approached OD600 [approx] 0.1, IPTG was added to a final concentration of 0.5 mM and after 60 min PQ was added as above. For TN521 containing pSE380 or pSEC119A, IPTG was added (to 0.5 mM) at OD600 [approx] 0.05. Aliquots of the cultures were removed at 0, 30 or 60 min following IPTG addition, PQ added (final concentration 250 [mu]M) and the incubation continued 30 min before harvesting for the assay. [beta]-Galactosidase activity was assayed in SDS/CHCl3-treated cells as described by Miller (21 ).
Both strands of the four soxR Cys -> Ala mutant genes were sequenced to verify the presence of the mutant codons. Each of the four mutant alleles contained the desired mutations (TGT -> GCT or TGC -> GCC) (Fig. 1 ) that reprogrammed the cysteine codon to an alanine codon. The full-length sequences of the four mutant constructs showed that no additional mutations were introduced by the site-specific procedure.
Our previous work established that the SoxR [2Fe-2S] centers are essential for the protein's transcriptional activity at soxS in vitro (11 ,13 -15 ). Here, we have extended the demonstration of this requirement to the activity of SoxR in vivo by engineering mutant genes that direct the synthesis of SoxR derivatives with individual cysteine residues replaced by alanines. For at least the C119A derivative, high level synthesis interfered with the normal activity of SoxR during activation by PQ. Despite their inability to be activated by oxidative stress, overexpression of the Cys -> Ala SoxR proteins revealed an unexpectedly high level of basal transcriptional activity in vivo that was not detected in vitro.
Removal of the SoxR [2Fe-2S] clusters in vitro by aerobic exposure of the protein to 2-mercaptoethanol (11 ,13 ) or the biological thiol glutathione (15 ) eliminates the transcriptional activity at soxS without any apparent effect on SoxR protein stability, oligomeric state or binding affinity for the soxS promoter. Elimination of detectable metal binding by Cys -> Ala replacements effectively blocked SoxR activation in response to PQ and did not significantly affect protein expression in vivo or soxS binding in vitro. These observations provide new evidence for post-translational activation of SoxR, and show that this activation in vivo depends critically on the integrity of the [2Fe-2S] centers. The results also support the conclusion from EPR spectroscopy that all four SoxR cysteine residues are involved in anchoring the metal center. Efforts to assemble [2Fe-2S] centers into the Cys119 -> Ala protein in vitro were unsuccessful (14 ).
Re-insertion in vitro of [2Fe-2S] centers into wild-type apo-SoxR restored full transcriptional activity (14 ). However, the properties of the Cys -> Ala mutant proteins alone do not resolve whether the critical step for SoxR activation in vivo is synthesis of the [2Fe-2S] centers or a subsequent redox reaction. We have previously suggested (13 ,14 ) that these two steps might be linked if the stability of the metal centers is influenced by oxidation and reduction. However, recent experiments (16 ,17 ) indicate that SoxR with reduced [2Fe-2S] centers is stable, but transcriptionally inactive in vitro. In vivo studies of the properties of constitutive forms of SoxR extend this model to living cells (18 ).
For wild-type SoxR, the quantity of reduced [2Fe-2S] centers detected in untreated cells corresponded to only 36% of the total SoxR protein (Fig. 3 ). The EPR-silent forms include both oxidized and apo-SoxR (12 ,13 ), with only the former active in vitro (11 ,16 ,17 ). The in vivo EPR analysis shown here and elsewhere (18 ) thus supports the conclusion that reduced Fe-SoxR is transcriptionally inactive in cells. Some oxidation evidently occurs during sample preparation by the method employed here: using a modified method (H.D. and B.D., manuscript in preparation), we have measured levels of reduced wild-type protein representing >= 95% of the total SoxR.
Two additional features of the SoxR Cys -> Ala mutants merit mention. First, during activation by PQ, none of the substituted forms, when expressed at approximately wild-type levels, competed detectably with wild-type SoxR expressed from a single-copy gene; such interference was observed only upon high level expression of the C119A derivative. In view of their evidently normal stability and DNA binding affinity, the relative ineffectiveness of the mutant proteins in blocking soxS activation by wild-type SoxR is striking. In principle, the mutant proteins could interfere with wild-type SoxR in two ways: by the formation of mixed dimers with a wild-type subunit, or by competition for binding the soxS promoter. It would thus be of interest to determine whether the mixed dimers retain partial or complete function. Regarding the second possible mode of interference, wild-type SoxR and the Cys -> Ala mutant proteins bound the soxS promoter equally well in vitro (Fig. 2 B). However, RNAP binds the soxS promoter cooperatively with activated SoxR (11 ), an effect that could disfavor competition by low levels of the Cys -> Ala mutant proteins.
The second unexpected feature of the Cys -> Ala mutant forms of SoxR was the relatively high basal soxS transcription they exert upon overexpression in vivo, which contrasts with the essentially undetectable activity of these proteins in vitro. It remains possible that the Cys -> Ala mutant proteins cause transcriptional activation higher than that of apo-SoxR, but at a level still undetectable in our in vitro assay. Alternatively, the in vivo activity of the mutant proteins in the absence of a redox signal could suggest the existence of additional controls on soxS activation that are overcome by high concentrations of the Cys -> Ala derivatives but not non-activated wild-type SoxR. The known negative autoregulation by SoxS (27 ) is evidently not responsible for this effect, which we observed in SoxS deficient cells. Previous genetic studies (2 ,3 ) identified regulatory mutations only in soxR and soxS (9 ,10 ), but a smaller effect on soxS expression caused by other mutations might have been missed. One possible negative regulator is Rob protein, which is abundant [~5000 molecules/cell; (28 )] and has DNA binding specificity that overlaps that of SoxS (29 -31 ). MarA protein also exhibits overlapping specificity with SoxS (32 ), but its expression is kept low through repression by the MarR protein (33 ).
We are grateful to our colleagues in the laboratory for advice and discussions, and to Professor J. Stubbe and the MIT Chemistry Department for generosity in sharing their EPR instrument. This work was supported by postdoctoral fellowships to E.H. from the Catalan Government (Commissio Interdepartamental de Recerca i Innovacio Tecnologica) and to H.D. from the US Public Health Service (NRSA, F32 ES05726), and by National Cancer Institute research grant CA37831 to B.D.
*To whom correspondence should be addressed. Tel: +1 617 432 3462; Fax: +1 617 432 2590; Email: demple@mbcrr.harvard.edu
+Permanent address: Department of Biochemistry, Microbiology and Molecular Genetics, University of Rhode Island, Kingston, RI 02881, USA
REFERENCES
Return