Repression of photosynthesis gene expression by formation of a disulfide bond in CrtJ

An S-tagged CrtJ overexpression vector was constructed by polymerase chain amplifying (PCR) the CrtJ coding sequence by using a forward primer containing an NcoI restriction site (5′-CCCAT GGCGACGGGAGGCTTGCA-3′), and a reverse primer containing an XbaI restriction site (5′-CCTCTAGACGGTGC CTTTCCCGTTTCG-3′). The PCR-amplified DNA segment was cloned into pBluescript II SK(±), and subsequently subcloned into NcoI and XbaI sites in pET-29(a)(+) (Novagen) to generate the plasmid, pET29∷crtJ. pET29∷crtJ was transformed into Escherichia coli BL21(DE3) pLysS, and the protein was then overexpressed by induction of a 500-ml culture with 0.1 mM isopropyl-β-d-thiogalactopyranoside at 25°C for 4 h. Cells were harvested and resuspended in 30 ml of ice-cold TPAE buffer composed of 50 mM Tris⋅HCl (pH 8.0), 200 mM potassium acetate, and 1 mM EDTA, and lysed by three passes through a chilled French pressure cell operated at 18,000 psi (1 psi = 6.89 kPa). The lysate was clarified by centrifugation at 28,000 × g for 30 min at 4°C. The supernatant was then transferred to a fresh tube with contaminating proteins precipitated by gradual addition of ammonium sulfate over a 30-min period at 4°C to a final concentration of 1 M. After 30 min of gentle mixing, soluble and insoluble proteins were fractionated by centrifugation at 28,000 × g for 30 min. The supernatant was decanted to a fresh tube to which additional ammonium sulfate was added to bring the final ammonium sulfate concentration to 1.5 M. An enriched CrtJ protein precipitant was then collected by centrifugation at 28,000 × g for 30 min, followed by washing the pellet twice with 1.5 M ammonium sulfate. The protein pellet was resuspended in TPAE buffer containing 10 mM DTT, from which the undissolved proteins were removed by centrifugation. The soluble supernatant fraction was further clarified by filtration through a 0.45 μm Acrodisc (Gelman). The protein preparation was size separated on a FPLC Superose 12 size-filtration column (Pharmacia), equilibrated with TPAE buffer at a flow rate of 1 ml/min, with CrtJ containing fractions collected and dialyzed overnight at 4°C against TPAE buffer containing 20% glycerol. Isolated CrtJ, which was greater than 98% pure as based on Coomassie blue staining of an SDS polyacrylamide gel, was stored at −80°C. For some experiments, a His-tagged variant of CrtJ was also used that was isolated as described by Ponnampalam et al. (4).

Iodoacetamide alkylation of thiol groups was performed with purified CrtJ as described (10). Before alkylation, CrtJ was first treated with either oxygen (air) or 20 mM β-mercaptoethanol at room temperature for 30 min, and then treated with iodoacetamide as described in ref (10). The DNA-binding activity of alkylated CrtJ was then assayed by gel mobility shift assay using a ³²P-labled DNA probe containing the bchC promoter region what was prepared by PCR amplification as described by Ponnampalam et al. (4).

For in vitro AMS modification of free thiol groups, purified CrtJ protein was first treated with oxygen, or freshly prepared hydrogen peroxide (0.001 mM, 0.01 mM, 0.1 mM, and 1.0 mM) or 20 mM β-mercaptoethanol at room temperature for varying times (10–30 min). The protein was then precipitated by the addition of trichloroacetic acid to 10% to fix its redox state and to remove excess hydrogen peroxide or β-mercaptoethanol. The precipitants were then dissolved in 50 mM Tris⋅HCl (pH 8.0) buffer containing 0.1% SDS and 15 mM 4-acetamido-4′-maleimidylstibene-2′,2′-disulfonic acid (AMS) for 1 h. The protein was then separated by nonreducing SDS/PAGE, with CrtJ detected by Western blot analysis using AP monoclonal antibodies (LumiBlot kit; Novagen).

For in vivo analysis of disulfide bonds, a strain of Rhodobacter capsulatus was used (ES7 CrtJ-flag) that expressed a chromosomally encoded S-Tag variant of CrtJ (11). Spectral and gel mobility shift data indicated that addition of an S tag at the amino terminus does not affect repressing properties of Crt J (data not shown). The ES7 CrtJ-flag construct was grown in RCV+ (12) media with gentamycin added to a final concentration of 5 μg/ml. Cells were grown aerobically in a partially filled Erlenmeyer flask that was shaking at 300 rpm or anaerobically in the dark in a completely filled screw capped test tube to a density of 65 Klett units (≈1.2 × 10⁶ cells per ml). Aerobically grown cells were treated with 1/10 volume of 100% tricloroacetic acid (TCA) and incubated on ice for 20 min. Anaerobically grown screw-capped cultures were transferred to an anaerobic chamber containing an atmosphere of 85% N₂/10% CO₂/5% H₂ and then divided into aliquots, with a portion of the cells kept under strictly anaerobic (reducing) conditions and another portion of the cells anaerobically treated with varying concentrations (0.001 mM, 0.01 mM, 0.1 mM, and 1.0 mM) of freshly prepared hydrogen peroxide (H₂O₂). After various incubation times (2–20 min), the cells were treated with 1/10 volume of 100% TCA and then incubated on ice for 20 min. In one set of assays, anaerobically grown cell samples were transferred to an Erlenmeyer flask and bubbled with O₂ (air) while shaking at 400 rpm for 10 or 30 min. After exposure to O₂ the cells were with 1/10 volume of 100% TCA and incubated on ice for 20 min.

The TCA treated samples were then AMS modified as follows. Precipitated proteins were collected by centrifugation for 10 min at 6,000 × g. After complete removal of the supernatant the pellet was washed with an equal volume of 500 mM Tris⋅HCl (pH 7.9) and centrifuged as before, washed with double distilled H₂O (ddH₂O) to remove excess salt and centrifuged again. Pellet was ultimately resuspended in 25 μl of 50 mM Tris⋅HCl (pH 7.9)/0.1% SDS/15 mM AMS (with the exception of the non-AMS-modified sample which was resuspended in the same buffer without AMS). AMS modifications were performed at 34°C for a minimum of 2 h. Alkylated samples were loaded on SDS/7.5% PAGE gels (Bio-Rad).

After electrophoresis, proteins were blotted to nitrocellulose filters and probed with anti-flag M2 monoclonal antibody (Sigma) according to product instructions. The bound antibodies were detected by using the enhanced chemiluminescence of SuperSignal West Dura substrate (Pierce) according to the manufacturer's instructions.

A measurement of relative percent disulfide bond formation at various redox potentials was measured by assaying fluorescence emission of CrtJ when incubated in the presence of monobromobimane (mBBr). Previous studies (13–15) have indicted that mBBr fluorescence increases significantly when it forms a covalent linkage with a Cys-sulfhydral. Thus, the level of mBBr fluorescence provides a good measurement of free sulfhydrals versus disulfides that are present in protein preparations that have been equilibrated in buffers containing different redox potentials. For this assay, a series of reactions were set up under oxidizing (air) or reducing (argon) conditions at different redox levels by varying ratios of oxidized (GSSG) and reduced (GSH) glutathione at a total glutathione concentration of 5 mM in 100 mM Hepes (pH 7.0 or 8.0) containing 100 μg of CrtJ. E_h values for different GSH/GSSG ratios at pH 7.0 were calculated from the Nernst Equation, by using the ratio of oxidized to reduced glutathione and literature values for the midpoint potential (E_m) of glutathione at pH 7.0 as described (15). E_h values at pH 8.0 were calculated by taking into account the pK_a of the thiol group of glutathione, as described (15). After incubation for 2 h to allow CrtJ to reach equilibrium at the desired redox potential, mBBr (Calbiochem) was then added to a final concentration of 300 mM and allowed to react with the Cys-sulfhydryls for 20 min. The protein was then precipitated by addition of 100 μl of 20% TCA and then incubated on ice for 30 min. The protein was pelleted by centrifugation, washed once with 1% TCA, and then resuspended in 100 mM Tris⋅HCl, pH 8.0/1% SDS. The fluorescence level of the CrtJ–mBBr preparation was measured using an Aminco–Bowman Series 2 Luminescence spectrometer with excitation at 380 nm and emission at 480 nm. The data were plotted and fit to a two-electron Nernst curve by using sigma plot software with maximum fluorescence set at the most reduced sample. E_m values obtained were independent of time, over times ranging from 0.5 to 2.0 h and independent of total glutathione concentration, over a range from 2.5 to 10.0 mM, as would be expected if true redox equilibrium has been attained at all E_h values (13–15).

A chromosomal deletion of crtJ was constructed to facilitate insertion of individual Cys to Ala mutations into the crtJ allele. Disruption of the chromosomal copy of crtJ was performed by gene-transfer-agent (GTA)-mediated recombination of a plasmid-encoded crtJ∷km construct into the chromosome. The generated crtJ-deleted strain, CD2–4, contains a deletion that extends from 46 bp upstream of CrtJ to 64 bp downstream of CrtJ. Individual Cys to Ala mutations were constructed in CrtJ by using the two-step PCR method (16) using a set of overlapping internal primers that incorporated the appropriate Cys to Ala codon changes. The PCR-generated fragments were cloned into the suicide vector, pZJD3, checked by sequence analysis, and integrated into the chromosome of strain CD2–4 by conjugation. A second recombination event was then identified by screening for loss of the gentamycin-resistance marker, leading to the generation of C22A, C249A, and C420A mutant strains. Proper allelic replacement was also analyzed by PCR and Western blot analysis. Mutant derivatives of the CrtJ overexpression vectors were also constructed by replacement of appropriate restriction sites.

To test whether CrtJ forms a disulfide bond, we undertook several in vivo and in vitro assays for the presence or absence of disulfide bonds when CrtJ is exposed to oxidizing or reducing conditions, respectively. The formation of an intermolecular disulfide bond was ruled out by observing no stable dimers or higher oligomers in oxidized CrtJ preparations as assayed by SDS/PAGE under nonreducing conditions (data not shown). However, in vitro thiol trapping experiments did show evidence for an intramolecular disulfide bond in oxidized CrtJ. In this analysis, Cys sulfhydryl groups were covalently modified with AMS, which retards electrophoretic mobility. Because disulfide bonds are resistant to AMS modification, the presence of an intramolecular disulfide bond in oxidized CrtJ would result in fewer Cys sulfhydryl groups available to react with AMS than would be present in reduced CrtJ. Consequently, if there was an intramolecular disulfide bond forming in CrtJ, then oxidized CrtJ would migrate faster during SDS/PAGE than would reduced CrtJ after AMS modification. As shown in Fig. 1A, AMS treatment of reduced (β-mercaptoethanol exposed) CrtJ resulted in significantly slower electrophoretic migration (lane 1) than is observed with a CrtJ preparation that was exposed to oxygen saturated buffer before treatment with AMS (lane 3). AMS treated oxidized CrtJ also migrates slightly slower than untreated CrtJ (lane 2). Because CrtJ contains three Cys residues, we interpret these results as evidence that oxidized CrtJ may only have one Cys that is available for AMS modification, whereas reduced CrtJ would have three Cys available for AMS modification.

We also undertook in vivo analysis of disulfide bond formation in CrtJ. For this analysis, we first grew R. capsulatus cells under anaerobic (photosynthetic) or aerobic (heterotrophic) growth conditions. At mid-log phase, the aerobically grown cells were treated with 10% TCA to denature the cellular proteins which effectively “fixes ” CrtJ sulfhydryls into their oxidized or reduced states. In parallel, a mid-log phase anaerobic culture was placed in an anaerobic chamber containing no oxygen (85% N₂/5% H₂/10% CO₂) and similarly treated with TCA. Precipitated denatured proteins were then neutralized with a buffer, washed, treated with AMS, and then size fractionated by SDS/PAGE and then analyzed for CrtJ migration by Western blot analysis.

As shown by the results in Fig. 1B, AMS treated anaerobically grown cell extracts (lane 3) exhibited significantly slower migration of CrtJ than did AMS treated aerobic extracts (lane 1). Indeed, the migration patterns of CrtJ in the AMS treated aerobic and anaerobic cell extracts show electrophoretic patterns identical to those observed with purified preparations of AMS treated oxidized and reduced CrtJ (Fig. 1A). These thiol-trapping experiments clearly demonstrate that CrtJ forms an intramolecular disulfide bond and that the disulfide bond is present in aerobically but not anaerobically grown R. capsulatus cells.

Another issue is the mechanism of disulfide bond formation in CrtJ. In wild-type E. coli, there is little change in the redox-state of the cytoplasm when cells are shifted from aerobic to anaerobic growth conditions. Measurements of oxidation and reduction states of glutathione indicates that the E. coli cellular redox potential remains between −260 and −280 mV in aerobic and in anaerobic cells (6, 7). Consequently, most cytoplasmic proteins remain in a reduced state, even when wild-type E. coli is grown aerobically. This finding has raised the question regarding how a disulfide bond forms in the transcription factor OxyR, which is known to have a midpoint potential for formation of its intramolecular disulfide bond of −185 mV (7). It has been shown that the disulfide bond formation in OxyR is stimulated by trace amounts of hydrogen peroxide, even under reducing conditions (7). Indeed, E. coli mutations that increase hydrogen peroxide levels just 2-fold over those found in wild-type cells cause constitutive in vivo disulfide bond formation in OxyR (7). Because oxidized OxyR is responsible for activating expression of reactive-oxygen defense genes, it appears that it has evolved a mechanism to form a disulfide bond by direct interaction with hydrogen peroxide.

To address the issue of disulfide bond formation in CrtJ, we first undertook an analysis of the redox-state of aerobically and anaerobically grown R. capsulatus cells. For this analysis, the cytosolic redox potential was estimated by measuring the ratio of glutathione (GSH) and glutathione disulfide (GSSG) by using the glutathione reductase recycling method (17, 18). The results of this assay indicate that anaerobically grown cells have a potential of −224 mV, and that aerobically grown cells have a potential of −222 mV. We can conclude, therefore, that R. capsulatus cells grown under either condition both have ambient potentials of approximately −220 mV, which is just slightly less reducing then that reported for E. coli (6, 7). We next measured the midpoint potential of disulfide bond formation of CrtJ in vitro by assaying the amount of fluorescence emission from the covalent adduct formed by the reaction of CrtJ with mBBr. Redox titrations using mBBr fluorescence to monitor the level of disulfide reduction have proven to be a reliable method for measuring midpoint potential (E_m) values (13–15). The data shown in Fig. 2 gives a good fit to an n = 2 Nernst equation for a single component with a E_m midpoint of −180 mV at pH 7.0. These results are precisely as expected for a two-electron event such as that which would be occurring for a reversible cleavage/formation of a single disulfide bond. The −180 E_m value measured for CrtJ at pH 7.0, is very similar to that reported for OxyR (7). A similar redox titration performed at pH 8.0 also gives a similar curve with a good fit to an n = 2 Nernst equation (Fig. 2). The E_m at this pH was a more reducing −226 mV. This pH alteration of CrtJ redox potential is predicted for a reaction that involves an uptake of two protons per disulfide bond reduction (13–15).

We also addressed whether reduced CrtJ that is present in anaerobically grown cells are susceptible to oxidation by the addition of low concentrations of hydrogen peroxide to anaerobic cultures, as is the case for OxyR. In the case of OxyR, it has been observed that addition of as little as 5 μM of H₂O₂ to E. coli cultures will lead to about 50% conversion of reduced OxyR to oxidized OxyR with addition of 10 μM leading to full conversion to an oxidized state (7). For R. capsulatus, we observed that addition of a wide range of hydrogen peroxide concentrations ranging from 0.001 to 1.0 mM to anaerobically grown cells did not lead to detectable in vivo formation of a disulfide bond in CrtJ as measured by AMS modification analysis. An example of the lack of detectable oxidation is shown in Fig. 3A, lanes 2 and 3, where incubation of a high level of H₂O₂ (1 mM) shows no evidence of disulfide bond formation. In contrast, we did observe that addition of oxygen to anaerobically grown cells did result in conversion of CrtJ from a reduced to a oxidized state after a 30-min exposure (Fig. 3A, lane 5).

We also addressed whether disulfide bond formation in isolated CrtJ was stimulated in vitro by exposure to oxygen and/or H₂O₂. For this analysis, we exposed reduced CrtJ to oxygen (air) for varying lengths of time, precipitated the protein with TCA, and then assayed CrtJ for AMS modification. The results of this analysis indicate that exposure of CrtJ to air for ≈30 min (Fig. 3B, lane 5) results in oxidation of CrtJ. In contrast, no oxidation was observed with exposure of CrtJ to H₂O₂ at various concentrations ranging from 10 μM to 1 mM (Fig. 3B, lanes 6–8). Indeed, no oxidation of CrtJ occurred during variation of both H₂O₂ concentration (10 μM to 1 mM) as well as time of exposure (2–30 min). These results indicate that disulfide bond formation in CrtJ differs from that of OxyR in that it appears to response to an increase in molecular oxygen rather than to the presence of an reactive-oxygen species such as H₂O₂.

We next addressed whether formation of an intramolecular disulfide bond affects DNA-binding properties of CrtJ. For this analysis, we performed gel mobility shift assays with a DNA segment of the bchC promoter region that contains two CrtJ DNA-binding palindromes (5). Lanes 2–4 in Fig. 4A are control reactions that show the reversible effect of redox on the DNA-binding activity of CrtJ. Specifically, lane 2 shows that exposure of CrtJ to atmospheric oxygen results in the ability of CrtJ to form a stable bchC–CrtJ complex. In contrast, lane 4 shows that reduction of similar amounts of CrtJ with 20 mM β-mercaptoethanol inhibits the formation of a stable DNA–CrtJ complex. Lane 3 demonstrates that the inhibitory effect of β-mercaptoethanol is reversible, because DNA-binding activity is restored when the β-mercaptoethanol-treated CrtJ is subsequently exposed to pure oxygen for 20 min before addition of the DNA probe.

The involvement of Cys in affecting DNA-binding was also addressed by assaying whether alkylation of CrtJ Cys sulfhydryls affects DNA-binding activity. Specifically, Fig. 4A, lane 5, is a DNA-binding assay in which CrtJ was first reduced by exposure to 20 mM β-mercaptoethanol, alkylated with iodoacetamide, and finally oxidized by exposure to pure oxygen. Unlike lane 3, where reduced CrtJ is capable of being reactivated by exposure to oxygen, the reduced and alkylated CrtJ can no longer be activated. Importantly, DNA-binding activity is not affected when CrtJ is first oxidized and then treated with iodoacetamide (lane 6). Because iodoacetamide does not react with disulfide bonds, these results indicate that oxidized CrtJ is capable of forming a disulfide bond that subsequently promotes DNA binding.

We also undertook mutational analysis of Cys residues in CrtJ to determine which Cys residues form an intramolecular disulfide bond, as well as to address the functional importance of forming a disulfide bond in vivo. Sequence analysis indicates that R. capsulatus CrtJ contains three Cys residues located at amino acids 22, 249, and 420. Cys-249 and -420 were identified as attractive targets for mutational analysis because they are also conserved in a CrtJ homolog from Rhodobacter sphaeroides (19). Furthermore, Cys-420 is also very near the C-terminal helix-turn-helix DNA-binding motif located at amino acid residues 426–463 (19). For our in vivo studies, we constructed individual Cys to Ala mutants at all three locations and recombined the mutations into the chromosomal copy of crtJ. Analysis of expression patterns of CrtJ repressed genes using β-galactosidase-based reporter plasmids indicates that the Cys to Ala mutation at position 22 (strain C22A) had no effect on CrtJ-mediated aerobic repression of the pucB, bchC, and crtI promoters, as evidenced by aerobic expression levels that are the same as that observed with wild-type R. capsulatus (strain SB1003) (Fig. 5). In contrast, a Cys to Ala mutation at codon 420 (strain C420A) elevated expression of the reporter plasmids to a level that was the same as observed with the crtJ deleted strain CD2–4. This finding indicates that the Cys to Ala mutation at this position abolished CrtJ in vivo repressing activity (Fig. 5). The Cys to Ala mutation at codon 249 (strain C249A) also elevates aerobic expression of the reporter plasmids, but not nearly to the same extent as does the C420A mutation. A similar pattern is observed with OxyR, in which a mutation in one of the disulfide bond forming Cys (Cys-199) exhibits a much more severe in vivo phenotype than does a mutation in the second disulfide bond forming Cys (Cys-208) (9). Western blot analysis also indicates that there are comparable levels of CrtJ in wild type and in the C420A and C249A mutant strains, indicating that the C420A and C249A mutations are not simply affecting protein stability (data not shown).

To confirm the in vivo expression analysis of photosynthesis genes, we also conducted in vitro DNA-binding studies with purified C249A and C420A mutant CrtJ proteins. As shown in the gel mobility shift assays in Fig. 4B, oxidized C249A and C420A mutant CrtJ protein preparations both exhibit defects in binding to the bchC promoter probe, relative to that observed with oxidized wild-type CrtJ. Furthermore, acetylation of the purified mutant proteins with AMS showed no evidence of altered electrophoretic mobility with protein that was pre-exposed to β-mercaptoethanol or to oxygen (data not shown). Together with the in vivo mutational data, these results implicate the conserved Cys-249 and Cys-420 residues as being involved in the formation of an intramolecular disulfide bond.