Routes of phosphoryl group transfer during signal transmission and signal decay in the dimeric sensor histidine kinase ArcB

The Arc (anoxic redox control) two-component system of Escherichia coli, comprising ArcA as the response regulator and ArcB as the sensor histidine kinase, modulates the expression of numerous genes in response to respiratory growth conditions. Under reducing growth conditions, ArcB autophosphorylates at the expense of ATP, and transphosphorylates ArcA via a His292 → Asp576 → His717 → Asp54 phosphorelay, whereas under oxidizing growth conditions, ArcB catalyzes the dephosphorylation of ArcA-P by a reverse Asp54 → His717 → Asp576 → Pi phosphorelay. However, the exact phosphoryl group transfer routes and the molecular mechanisms determining their directions are unclear. Here, we show that, during signal propagation, the His292 → Asp576 and Asp576 → His717 phosphoryl group transfers within ArcB dimers occur intra- and intermolecularly, respectively. Moreover, we report that, during signal decay, the phosphoryl group transfer from His717 to Asp576 takes place intramolecularly. In conclusion, we present a mechanism that dictates the direction of the phosphoryl group transfer within ArcB dimers and that enables the discrimination of the kinase and phosphatase activities of ArcB.

The Arc two-component signal transduction system plays an important role in the transcriptional regulatory network that allows facultative anaerobic bacteria, such as Escherichia coli, to sense and signal changes in respiratory growth conditions and to adapt their gene expression accordingly (1)(2)(3)(4). This system consists of the cytoplasmic response regulator ArcA, and the membrane-anchored sensor kinase ArcB (5,6). ArcA is a typical response regulator, possessing an N-terminal receiver domain with a phosphoryl group-accepting Asp residue at position 54 and a C-terminal helix-turn-helix DNA-binding domain. In contrast, ArcB is an unorthodox sensor kinase that contains three catalytic cytosolic domains: a transmitter domain (H1) with a conserved His 292 residue, a central receiver domain (D1) with a conserved Asp 576 residue, and a C-terminal histidine phosphotransfer domain (H2) with a conserved His 717 residue (6,7). In addition, the ArcB protein contains a functional leucine zipper (8) and a PAS domain (9), both located in the linker region, which is the segment connecting the transmembrane domain with the transmitter domain.
Under reducing growth conditions, ArcB autophosphorylates in an ATP-dependent manner, a process that is enhanced by certain anaerobic metabolites, such as D-lactate, acetate, and pyruvate (10,11), and transphosphorylates ArcA via a His 292 3 Asp 576 3 His 717 3 Asp 54 phosphorelay (12,13). Phosphorylated ArcA (ArcA-P), 2 in turn, represses the expression of many operons involved in respiratory metabolism and activates some others encoding proteins involved in fermentative metabolism (14 -17). Under oxic growth conditions, ArcB acts as a specific ArcA-P phosphatase, catalyzing the dephosphorylation of ArcA-P by a reverse Asp 54 3 His 717 3 Asp 576 3 P i phosphorelay (18,19). The catalytic activity of ArcB has been shown to be set by rotational movements that alter the orientation of the cytosolic portion of ArcB (20). Moreover, the molecular event for ArcB regulation involves the oxidation or reduction of two cytosol-located redox-active cysteine residues that participate in intermolecular disulfide bond formation, a reaction in which the quinol/quinone electron carriers act as the direct oxidants or reductants (21)(22)(23)(24).
the reverse phosphorelay, responsible for the phosphatase activity of these hybrid sensor kinases, has yet to be addressed. In the case of ArcB, the autophosphorylation reaction appears to occur intramolecularly, as the ␥-phosphoryl group of ATP bound to one monomer in the homodimer was shown to be transferred to the histidine of the same monomer (35,38). In contrast, the mode of ϳP transfer in the two subsequent steps in ArcB signaling remains ambiguous, because two independent studies reached dissimilar conclusions (38,39). On one hand, in a report based on in vivo experiments analyzed with mathematical and statistical models, it was suggested that a bimolecular or allosteric mechanism, in which the integrity of all phospho-accepting/donating sites is required and in which no exclusive intra-or intermolecular ϳP transfer mechanisms are identifiable, may be operating during ArcB signaling (39). On the other hand, in a later study based on in vitro complementation analyses, it was proposed that the ϳP transfers from H1 to D1 and from D1 to H2 occur intermolecularly (38).
Here, we present results from in vitro and in vivo experiments addressing the above mentioned discrepancy and extend these studies to include the characterization of the H2 to D1 ϳP transfer step, involved in ArcA-P dephosphorylation during signal decay. Our results demonstrate that the ϳP transfer from H1 to D1 occurs exclusively intramolecularly, whereas the following step (i.e. from D1 to H2) occurs preferentially intermolecularly. Finally, the H2 to D1 ϳP transfer, responsible for signal decay, shows a clear preference for the intramolecular reaction. The consequences of the proposed ϳP transfers between the various ArcB modules for the regulation of signal transmission and signal decay of the Arc two-component system are discussed.

Probing the mode of phosphoryl group transfer from the transmitter domain to the receiver domain of ArcB
Autophosphorylation of the tripartite sensor kinase ArcB was shown to be an intramolecular reaction (35,38), whereas contradictory results were reported for the mode of ϳP transfer in the two subsequent steps (38,39). Therefore, we attempted to probe the mode of these ϳP transfer steps by an approach different from the ones used previously.
To this end, we generated N-terminal His 6 -tagged cytosolic WT and mutant ArcB variants (hereafter referred to as His 6 -H1-D1-H2), carrying single or multiple punctual mutations of the conserved phosphorylation sites. In other words, the conserved histidine 292 in H1 was replaced by glutamine (hereafter referred to as H1*), the conserved aspartate 576 in D1 was replaced by alanine (hereafter referred to as D1*), and the conserved histidine 717 in H2 was replaced by glutamine (hereafter referred to as H2*). WT and single-and multiple-mutant His 6 -H1-D1-H2 proteins were purified, and their ability to autophosphorylate and transphosphorylate His 6 -ArcA (hereafter referred to as ArcA) was tested. ArcA was rapidly phosphorylated by His 6 -H1-D1-H2 (Fig. 1A), but not by His 6 -H1*-D1-H2, His 6 -H1-D1*-H2, His 6 -H1-D1-H2*, or His 6 -H1*-D1*-H2 mutant proteins (Fig. S1, A-D), in agreement with previous reports (12,13). The ability of combinations of mutant His 6 -H1-D1-H2 variants to transphosphorylate ArcA was then evaluated. No ArcA phosphorylation was observed in the reaction mixture containing His 6 -H1*-D1-H2 and His 6 -H1-D1*-H2, in which only intermolecular H1 to D1 phosphotransfer is allowed (Fig. 1, B and D). On the other hand, a weak ArcA phosphorylation, ϳ30% of that observed by the WT His 6 -H1-D1-H2, was obtained in the reaction mixture containing His 6 -H1*-D1*-H2 and His 6 -H1-D1-H2*, in which only the intramolecular H1 to D1 ϳP transfer is permitted (Fig. 1, C and D), indicating that the ϳP transfer from H1 to D1 occurs intramolecularly. However, this conclusion is hampered by the fact that none of the combinations was able to restore the phosphorelay at WT ArcB levels, most likely due to the insufficient heterodimer formation in the reaction mixture.
To ensure heterodimer formation, a set of plasmids carrying arabinose-inducible MBP-tagged H1-D1-H2 variants were constructed and transformed in strains harboring plasmids that
Single MBP-tagged proteins, used for heterodimer formation, were purified and subjected to in vitro phosphorylation assays with [␥-32 P]ATP and His 6 -ArcA. Only the MBP-tagged proteins having a WT H1 were able to autophosphorylate (Fig.  S1E), and only the MBP-H1-D1-H2 was able to transphosphorylate ArcA (Fig. 2B), indicating that the purified proteins have the expected activities and that the N-terminal MBP tag does not interfere with the activity of ArcB. Subsequently, the ability of the His 6 -H1*-D1-H2/MBP-H1-D1*-H2 heterodimer, in which only intermolecular phosphoryl group transfer from H1 to D1 is permitted, to functionally complement and restore the phosphorelay to ArcA was tested (Fig. 2C). Although the MBP-H1-D1*-H2 protein was rapidly autophosphorylated, no His 6 -H1*-D1-H2-P was observed, and almost no ArcA-P was formed (Fig. 2, C and E), suggesting that the H1 to D1 phosphotransfer does not occur intermolecularly. In contrast, the His 6 -H1*-D1*-H2/MBP-H1-D1-H2 heterodimer was found to readily transphosphorylate ArcA (Fig. 2, D and E), indicating that the phosphoryl group transfer from H1 to D1 occurs intramolecularly, at least in vitro. It has to be noted that phosphorylation of His 6 -H1*-D1*-H2 was also observed, most likely due to intermolecular phosphoryl group transfer from D1 to H2 or from ArcA to H2.

Phosphoryl group transfer modes in ArcB
moter were chosen to avoid the overexpression of the sensor kinase. This is because distortion of the ArcB/ArcA concentration balance has been reported to produce unpredictable effects on reporter expression (42). Indeed, Western blot analysis using ArcB polyclonal antibodies, raised against purified His 6 -ArcB 78 -520 (40), revealed that all plasmid-borne arcB alleles produce similar to WT levels of ArcB protein (Fig. S2E). Subsequently, the phenotypic consequences of the above described strains were analyzed by changes in the in vivo levels of phosphorylated ArcA, as indicated by the expression of the ArcA-P-activatable ⌽(cydAЈ-lacZ) reporter (Fig. 3). As expected, arcB-null phenotypes were found for strains IFC2002 (arcB H717Q ) and IFC2001 (arcB D576A ) and when the plasmidborne arcB H717Q,D576A and arcB H292A were expressed in the ⌬arcB mutant strain ECL5004 (Fig. S2, A-D), in agreement with the in vitro results and previous reports (12). Moreover, expression of the plasmid-borne arcB H292Q in IFC2001 (arcB D576A ), where only intermolecular ϳP transfer from H1 to D1 is permitted, resulted in an arcB-null phenotype (Fig. 3, B and C). In contrast, expression of the plasmid-borne arcB H292Q,D576A in IFC2002 (arcB H717Q ), where only intramolecular phosphoryl group transfer from H1 to D1 is permitted, resulted in WT levels of reporter expression (Fig. 3, A and D), in accordance with the above presented in vitro results. Therefore, it appears reasonable to conclude that an intramolecular mode of ϳP transfer operates from H1 to D1. This result also suggests that the in vivo ϳP transfer from D1 to H2 occurs intermolecu-larly, although the intramolecular mode cannot be excluded. Taken together, the above results from the in vitro and in vivo experiments provide strong evidence that the phosphorelay along ArcB comprises an intramolecular H1 to D1 ϳP transfer.
These results indicate that the D1 to H2 ϳP transfer step occurs preferentially intermolecularly in vitro, in agreement with a previous study (38). Indeed, in the above presented in vivo complementation assay, co-expression of the arcB H717Q and arcB H292Q,D576A mutant alleles resulted in WT levels of reporter expression (Fig. 3D), indicating that an intermolecular D1 to H2 phosphotransfer does occur during in vivo ArcB signaling. The direct evaluation of an intramolecular ϳP transfer from D1 to H2 in vivo, requiring the expression of ArcB D576A,H717Q and ArcB H292A , however, is unattainable because the ϳP transfer step from H1 to D1 is exclusively intramolecular, rendering the use of a WT ArcB mandatory, which, in turn, impedes the discrimination between the two modes of ϳP transfer. To circumvent this problem, we reasoned that expression of increasing amounts of arcB D576A,H717Q in an arcB wt background strain should favor ArcB D576A,H717Q / ArcB wt heterodimer formation. Because such heterodimers allow only intramolecular D1 to H2 phosphotransfer, a reduction of the ArcA-P levels should be obtained if an intermolecular D1 to H2 ϳP transfer were to be favored. Because overexpression of a sensor kinase may have unexpected effects on its signaling, increased concentrations of arcB H292Q,D576A in the arcB wt background, favoring the formation of ArcB H292Q,D576A /ArcB wt heterodimers that permit both inter-and intramolecular D1 to H2 phosphotransfer, was used as a control.
To this end, plasmids pBADArcB H292Q,D576A and pBAD ArcB D576D,H717A , carrying, respectively, the arcB H292Q,D576A and arcB D576A,H717Q mutant alleles under the control of the inducible arabinose promoter were constructed and transformed into the ECL5003 strain, which carries the ArcA-P-activatable ⌽(cydAЈ-lacZ) reporter (12). The transformants were grown anaerobically in the presence of various concentrations of arabinose (0 -30 M), and at midexponential phase of growth, A 600 of ϳ0.5, the ␤-gal activities were determined (Fig. 5). Increasing amounts of inductor in the growth medium of the strain carrying the ArcB H292Q,D576A -expressing plasmid resulted in decreasing levels of reporter expression (Fig. 5). Because the intramolecular H1 to D1 phosphotransfer and both modes of phosphotransfer from D1 to H2 are permitted within the At an A 600 of 0.2, one aliquot was withdrawn, to measure the ␤-gal activity (depicted as 0 min), and the rest of the culture was divided in two. One part was kept under aerobic conditions (circles) as a control, whereas the other was shifted to anaerobiosis (squares), and the time course of the ␤-gal activity was followed. The data represent the averages from three independent experiments, and the S.D. values (error bars) are indicated.

Phosphoryl group transfer modes in ArcB
ArcB H292Q,D576A /ArcB wt heterodimer, a possible explanation could be that the increasing amount of ArcB H292Q,D576A homodimers may lead to ϳP transfer from ArcA-P to H2, as has been shown earlier (19), thereby lowering its regulatory activity. Nevertheless, the strain carrying the ArcB D576A,H717Q -expressing plasmid exhibited a more notorious effect. A 40% reduction of reporter expression as compared with that of the ArcB H292Q,D576A -carrying strain was observed under all tested conditions (Fig. 5). It thus appears that, in agreement with the in vitro results, the ϳP transfer from D1 to H2 has a clear preference for the intermolecular reaction in vivo.

Probing the mode of the H2 to D1 ϳP transfer for ArcA-P dephosphorylation
The amplitude and duration of an adaptive response depend onthebalancebetweentheratesofphosphorylationanddephosphorylation of the response regulator. The sensor kinase ArcB is a bifunctional enzyme that phosphorylates ArcA under stim-ulatory conditions, and it also dephosphorylates ArcA-P under nonstimulatory conditions (12,13,18,19). The fact that the conserved His 717 and Asp 576 are required for both opposing activities of ArcB (18,19) raises the intriguing possibility that the forward and the reverse phosphorelay do not have the same pattern with respect to inter-and intramolecular phosphoryl group transfers.
The mode of ϳP transfer from H2 to D1 was then probed in vivo, by examining the time lag of ArcA-P dephos-

Phosphoryl group transfer modes in ArcB
phorylation after a shift from anaerobic to aerobic growthconditions. To this end, strain ECL5032 (arcB H171Q ) harboring the ArcB H292Q,D576A -expressing plasmid (pEXT22Cm ArcB H292Q,D576A ) and the isogenic WT strain ECL5002, both carrying the ArcA-P-repressible ⌽(lldPЈ-lacZ) reporter (12,40), were grown anaerobically in lysogeny broth (LB) supplemented with 20 mM L-lactate to induce expression of the reporter (43). At an A 600 of ϳ0.2, a sample was withdrawn, and the expression of the reporter was determined (depicted as 0 min in Fig. 7). As expected, in both strains, the expression of the reporter was very low (Fig. 7), indicative of ArcA-P-dependent reporter repression. This result demonstrates that co-expression of the arcB H717Q and arcB H292Q,D576A mutant alleles fully restores ArcB kinase activity and thus proper ArcB H717Q / ArcB H292Q,D576A heterodimer formation. Subsequently, the cultures were shifted to aerobiosis, and the time course of the ␤-gal activity was followed. An almost immediate increase of reporter expression (Fig. 7) was observed for the strain carrying the WT arcB allele, indicating ArcA-P dephosphorylation and release of reporter repression. On the other hand, a retarded and scarce increase of reporter expression was noticed for the

Phosphoryl group transfer modes in ArcB
strain carrying the arcB H717Q and arcB H292Q,D576A mutant alleles (Fig. 7). Taken together, these results strongly suggest that the reverse ϳP transfer during signal decay (i.e. from H2 to D1) operates intramolecularly.
Thus, it seems reasonable to conclude that conformational changes of ArcB may affect the mode of ϳP transfer between the D1 and H2 domains, thereby discriminating between the two opposing activities (i.e. between signal transmission and signal decay).

Discussion
In this study, results from in vitro and in vivo experiments demonstrate that the His 292 to Asp 576 ϳP transfer within the ArcB dimer occurs exclusively intramolecularly, whereas the ϳP transfer from Asp 576 to His 717 occurs preferentially intermolecularly. Finally, the reverse ϳP transfer responsible for signal decay, namely from His 717 to Asp 576 , appears to have a clear preference for the intramolecular reaction. It is therefore tempting to propose a model in which, during signal propagation, the physiological mode of ϳP transfers between the individual ArcB modules occurs intramolecularly from H1 to D1 and intermolecularly from D1 to H2, whereas the reverse ϳP transfer during signal decay (i.e. from H2 to D1) occurs intramolecularly (Fig. 8).
This model differs from that of a previous study, based on in vivo experiments analyzed with mathematical and statistical models, where it was suggested that the mechanism operating in ArcB ϳP transfers is dictated by a cooperative or bimolecular mechanism in which all phosphorylation sites of both monomers appear to be required for proper functioning of the ArcB phosphorelay (39). It is worth noting that in these in vivo exper-iments, the ArcB variants were expressed from medium to high copy number plasmids under the control of IPTG-inducible promoters and with the addition of 0.1 mM IPTG. Surprisingly, all combinations, including those in which a WT ArcB was co-expressed with mutant ArcB variants, failed to fully restore a functional phosphorelay. Therefore, it was suggested that all domains and phosphorylation sites in both monomers have to be present if a functional phosphorelay is to be allowed. However, in our hands, co-expression of a chromosomal arcB H717Q mutant and a low-copy number plasmid-borne arcB H292Q,D576A mutant fully restored the kinase activity of ArcB (Fig. 3D), indicating that the mutant proteins do complement not only the WT ArcB activity but also its regulation. Therefore, we could only speculate that, in the earlier study, a possible overproduction of the ArcB mutant variants might have been the cause of the lower ArcA activation. Indeed, it has been demonstrated previously that significant overexpression of proteins can be achieved with plasmid pCA24N, which was used in the above study, with as little as 50 M IPTG (44). In fact, we have observed that overexpression of a mutant version of ArcB in a WT strain results in a drastic reduction of ArcA-P formation and, thereby, reporter expression (Fig. 5). This, most probably, could be caused by the alteration of the sensor/response regulator concentration balance, in accordance with previous observations for other two-component signal transduction systems (42,45). Moreover, an intermolecular ϳP transfer from H1 to D1 of ArcB, in contrast to our result, was reported in an earlier study. In that study, the mode of ϳP transfers involved in the phosphorelay of ArcB was analyzed by complementation of single-phosphorelay mutant proteins in in vitro phosphorylation assays, followed by separation of phosphorylated proteins by Phos-Tag SDS-PAGE and detection by Coomassie Blue staining (38). However, the high protein concentrations needed for detection of phosphorylated forms of proteins by Coomassie Blue after Phos-Tag SDS-PAGE separation could increase the incidence of nonspecific phosphorylation and, therefore, weaken the resulting conclusions. In contrast, our in vitro assays, using ArcB mutant heterodimers, in addition to the in vivo complementation experiments, clearly discard intermolecular H1 to D1 phosphotransfer and demonstrate that only the intramolecular transfer is allowed.
At any rate, the model proposed herein for ArcB (Fig. 8) appears appealing if we consider that the spatio-steric constraints of a given assembly of an ArcB ␣ /ArcB ␤ homodimer should allow only Asp 576 ␣ or Asp 576 ␤ to be in the vicinity of His 292 ␣ , permitting either an inter-or intramolecular ϳP transfer. Likewise, only one His 717 , the one of either ArcB ␣ or ArcB ␤ , could be in the vicinity of Asp 576 ␣ , permitting only one mode of ϳP transfer. Thus, under anoxic growth conditions, the assembly of the ArcB ␣ /ArcB ␤ dimer would permit the intramolecular ϳP transfer from H1 to D1 (  Under anoxic growth conditions, the ATP-dependent intramolecular autophosphorylation in H1 is followed by a phosphorelay that involves an intramolecular H1 to D1 phosphotransfer and an intermolecular D1 to H2 phosphotransfer. Under oxic growth conditions, disulfide bond-dependent conformational adjustments of the ArcB modules enforce an intramolecular phosphotransfer from H2 to D1 followed by release of ϳP as P i , resulting in ArcA-P dephosphorylation and signal decay.

Phosphoryl group transfer modes in ArcB
ArcB. Such a model provides some insights within the biochemical mechanism that dictates the direction of ϳP transfer, thereby differentiating the kinase and phosphatase activities of ArcB. A similar modus operandi may also apply to other twocomponent systems comprising a tripartite sensor kinase. Nevertheless, the attainment and analysis of the ArcB crystallographic structure should be needed to elucidate the conformational adjustments that occur upon changes in the redox conditions and to completely understand the bases of the ArcB ϳP transfer pattern.
Bacteria were routinely cultured at 37°C in LB. When necessary, media were supplemented with antibiotics, at the following concentrations: chloramphenicol, 20 g/ml; kanamycin, 50 g/ml; ampicillin, 100 g/ml; and tetracycline, 10 g/ml. For ␤-galactosidase activity assays, the ⌽(cydAЈ-lacZ)bearing strains were grown in LB containing 0.1 M MOPS (pH 7.4) and 20 mM D-xylose, whereas the ⌽(lldPЈ-lacZ)-bearing strains were grown in the above medium supplemented with 20 mM L-lactate as inducer.

Purification of His 6 -and MBP-tagged proteins and phosphorylation assays
Native His 6 -tagged ArcB 78 -778 variants and His 6 -ArcA, used in phosphorylation assays, were prepared as described previously (13,47). MBP-tagged ArcB 78 -778 variants were purified by using amylose resin (New England Biolabs) as affinity matrix according to the instructions provided by the manufacturer. For the co-purification of the His 6 -/MBP-tagged ArcB versions, strain ECL5012 (arcB Ϫ ) (40) carrying both pBADHis-ArcB 78 -778 * and pACT3MBP-ArcB 78 -778 * (where the asterisk refers to specific amino acid replacements) was grown in 1 liter of LB, supplemented with chloramphenicol and ampicillin, in a rotary shaker at 37°C until an A 600 of 0.6. Then the expression of the His 6 -and MBP-tagged ArcB 78 -778 * was induced by the addition of 1 mM L-arabinose and 0.1 mM IPTG, respectively. Cells were harvested 5 h after induction, and the cell pellet was resuspended in 10 ml of lysis buffer (50 mM sodium phosphate, pH 8.0, 300 mM NaCl, 10 mM imidazole). Finally, the proteins were purified under native conditions by Ni-NTA-agarose affinity chromatography, as described previously (13,47). Phosphorylation assays were carried out at room temperature in the presence of 40 M [␥-32 P]ATP (specific activity, 2 Ci mmol Ϫ1 ; New England Nuclear), 33 mM HEPES (pH 7.5), 50 mM KCl, 5 mM MgCl 2 , 1 mM DTT, 0.1 mM EDTA, and 10% glycerol. 5 pmol of purified ArcB 78 -778 * (His 6 -or MBP-tagged) or 10 pmol of copurified His 6 -and MBP-tagged ArcB 78 -778 * and 50 pmol of His 6 -ArcA were used in each phosphorylation assay. The phosphorylation reactions were initiated by the addition of [␥-32 P]ATP, and samples were withdrawn and mixed with SDS sample buffer after 15, 30, 60, 120, and 180 s and subject to Phosphoryl group transfer modes in ArcB analysis by SDS-PAGE on 12% polyacrylamide gels. The radioactivity of proteins resolved in the gels was analyzed by exposing to a Storage Phosphor Screen and scanning by a Typhoon FLA7000 biomolecular imager (GE Healthcare). The intensity of individual band signals was estimated using the ImageQuant version 5.2 software (Molecular Dynamics).

␤-Galactosidase activity assay
In general, for aerobic growth, cells were cultured in 10 -50 ml of medium in 250-ml baffled flasks at 37°C with shaking (300 rpm), whereas for anaerobic growth, cells were cultured in a screw-capped tube filled with medium up to the rim at 37°C and stirred by a magnet. For the aerobiosis to anaerobiosis shift, cells were aerobically grown and, at an A 600 of 0.2, part of the culture was transferred to five prewarmed screw-capped tubes, filled up to the rim, and stirred by a magnet. The rest of the aerobic culture was further incubated with shaking at the same temperature. The time course of the experiment was followed by taking a sample from a filled screw-capped tube (anaerobiosis) and from the baffled flasks (aerobiosis) at each chosen time. For the anaerobiosis-to-aerobiosis shift, cells were cultured in seven screw-capped tubes filled with medium up to the rim at 37°C and stirred by a magnet. At an A 600 of 0.2, the content of a tube was passed to a 250-ml prewarmed baffled flask and incubated with shaking at the same temperature. Samples from aerobic and anaerobic cultures (the rest of the screw-capped tubes) were withdrawn at each chosen time after the shift. ␤-gal activity was assayed and expressed in Miller units as described previously (48).

Western blotting
Cells were harvested from cultures by centrifugation during mid-exponential growth. The cell pellet was resuspended in sample buffer and separated by SDS-PAGE (10% polyacrylamide gel), and the proteins were transferred to a Hybond-ECL filter (Amersham Biosciences). The filter was equilibrated in TTBS buffer (25 mM Tris, 150 mM NaCl, and 0.05% Tween 20) for 10 min and incubated in blocking buffer (1% milk in TTBS) for 1 h at room temperature. ArcB polyclonal antibodies, raised against His 6 -ArcB 78 -520 (12), were added at a dilution of 1:10,000 and incubated for 1 h at room temperature. The bound antibody was detected by using anti-rabbit IgG antibody conjugated to horseradish peroxidase and the ECL detection system (Amersham Biosciences).