Expression, Purification and Characterisation of Full-Length

Home , Histidine kinase

B doi:10.1016/S0022-2836(02)00424-2 available online at http://www.idealibrary.com on w J. Mol. Biol. (2002) 320, 201–213

Expression, Puriﬁcation and Characterisation of Full-length Histidine Protein Kinase RegB from Rhodobacter sphaeroides

Christopher A. Potter1, Alison Ward1, Cedric Laguri2 Michael P. Williamson2, Peter J.F. Henderson1 and Mary K. Phillips-Jones1*

1Division of Microbiology The global redox switch between aerobic and anaerobic growth in School of Biochemistry and Rhodobacter sphaeroides is controlled by the RegA/RegB two-component Molecular Biology, University system, in which RegB is the integral membrane histidine protein kinase, of Leeds, Leeds LS2 9JT, UK and RegA is the cytosolic response regulator. Despite the global regulatory importance of this system and its many homologues, there have been no 2Department of Molecular reported examples to date of heterologous expression of full-length RegB Biology and Biotechnology or any histidine protein kinases. Here, we report the amplified expression University of Sheffield of full-length functional His-tagged RegB in Escherichia coli, its purifi- Sheffield S10 2TN, UK cation, and characterisation of its properties. Both the membrane-bound and purified solubilised RegB protein demonstrate autophosphorylation activity, and the purified protein autophosphorylates at the same rate under both aerobic and anaerobic conditions confirming that an additional regulator is required to control/inhibit autophosphorylation. The intact protein has similar activity to previously characterised soluble forms, but is dephosphorylated more rapidly than the soluble form (half- life ca 30 minutes) demonstrating that the transmembrane segment present in the full-length RegB may be an important regulator of RegB activity. Phosphotransfer from RegB to RegA (overexpressed and purified from E. coli ) by RegB is very rapid, as has been reported for the soluble domain. Dephosphorylation of active RegA by full-length RegB has a rate similar to that observed previously for soluble RegB. q 2002 Elsevier Science Ltd. All rights reserved Keywords: Rhodobacter sphaeroides; RegB; membrane receptor; Ni2þ affinity *Corresponding author purification; phosphorylation kinetics

Introduction in oxygen-responsive regulation of nitrogen fixation genes.4 In Rhodobacter, RegB is the mem- The RegBA two-component system (also known brane-located histidine protein kinase (HPK) as PrrBA) serves as a major transcriptional regula- component of the system, sensing changes in tor of gene expression in several photosynthetic 1–6 redox conditions. Upon anaerobiosis, RegB and nitrogen-fixing bacteria. It has been studied becomes autophosphorylated in an ATP-depen- most intensively in Rhodobacter sphaeroides and dent reaction; RegB , P then transfers the phos- R. capsulatus, in which RegBA is a globally acting, 7 phoryl signal to Asp63 of the partner response redox-responsive system, and in nitrogen-fixing regulator RegA.8,9 Once phosphorylated, Bradyrhizobium japonicum, in which it is involved RegA , P then positively regulates photosynthesis gene expression ( puc, puf and puhA ),1–3 as well as expression of genes involved in carbon dioxide Present address: A. Ward, Astex Technology Ltd, 250 10,11 fixation (cbbI and cbbII operons), nitrogen Cambridge Science Park, Cambridge CB4 0WE, UK. fixation (nifA2 ),7,12 electron transport functions Abbreviations used: HPK, histidine protein kinase; b D ( petABC, cycA, cycY ) and respiratory terminal elec- TMR, transmembrane region; DDM, dodecyl- - - 3,13,14 maltoside. tron functions (cydAB, ccoNOPQ, dorCBA ). E-mail address of the corresponding author: RegA , P also negatively regulates hydrogenase [email protected] expression (hupSLC ).12 Here, we retain the name

Figure 1. SDS-PAGE and Western analysis of mixed membrane proteins from IPTG-induced E. coli NM554 cells carrying pTTQregB (full-length regB gene) and/or pTTQEP6 (truncated regB gene). (a) SDS-PAGE of mixed membranes of E. coli NM554 (pTTQregB ) uninduced or induced with 1 mM IPTG and resolved using 15% polyacrylamide gels and visualised by staining with Coomassie brilliant blue. Lane 1, uninduced mixed membranes; lane 2, IPTG-induced mixed membranes; lane 3, molecular mass markers. (b) Western analysis of mixed membranes from E. coli (pTTQregB ) and E. coli (pTTQEP6 ). Lane 1, molecular mass markers (no His-tag). Lane 2, E. coli NM554 (pTTQEP6 )(1mg of mixed membrane protein). Lane 3, E. coli NM554 (pTTQregB )(1mg of mixed membrane protein). Lane 4, E. coli (pTTQEP6 ) (5 mg of mixed membrane protein). Lane 5, E. coli NM554 (pTTQregB )(5mg of mixed membrane protein).

Reg (rather than Prr) for the R. sphaeroides system, redox sensing.19 – 21 ArcB, like RegB, senses redox for consistency with the homologues identified in signals via the state of components of electron other bacterial species, including R. capsulatus,in transport; oxidised forms of quinone electron which the Reg system was first described.1 carriers serve as negative signals that inhibit auto- RegB senses redox through changes in the phosphorylation of ArcB during aerobiosis.19 In volume of electron flow through the cbb3-type cyto- the case of ArcB, the TMRs serve as membrane chrome c oxidase of the respiratory electron trans- anchors, playing no detectable role in sensing or port chain.15 The precise mechanism is not signal transduction.19 However, this does not elucidated, but may be mediated by the SenC appear to be the case for RegB. Recent in vivo muta- protein.16,17 In vivo studies suggest that under genesis studies of the transmembrane domain of aerobic conditions, the volume of electron flow R. sphaeroides RegB protein revealed the import- through cbb3 oxidase is sufficiently high to repress ance of a central portion of this domain, particu- the default kinase-positive mode of RegB, resulting larly the short second periplasmic loop and in repression of the autophosphorylation of RegB.9 membrane-spanning a-helices 3 and 4, for sensing Anaerobic conditions relieve this repression and and signal transduction.9 Thus, in future studies result in autophosphorylation of RegB and sub- of the signal-sensing and transduction mechanism, sequent signal transfer to RegA.9 full-length versions of RegB that include the trans- R. sphaeroides RegB (462 amino acid residues) membrane regions will be required. However, possesses two domains, an N-terminal trans- previous attempts to express full-length RegB and membrane domain (residues 1–182) predicted to its homologues8,22 – 24 have consistently failed to comprise six membrane-spanning regions, and a obtain this protein in a soluble, folded and func- cytosolic C-terminal histidine kinase domain tionally active form; indeed, of all the membrane (residues 183–462) for autophosphorylation, phos- HPKs reported to date, only Escherichia coli phatase and phosphotransfer reactions.18 It is an KdpD25 and NarX26,27 have been expressed success- atypical HPK because it appears to lack any poten- fully as enriched proteins in E. coli and utilised in tial signal-sensing periplasmic domains between functional assays. Of these two proteins, only the transmembrane regions (TMRs). This implies NarX26 retains functional activity after purification either that RegB signal sensing occurs within the from membranes. KdpD regains activity only after TMRs themselves, or that signals are sensed else- reconstitution into membrane vesicles.25 Other where within the soluble domain, with TMRs studies, including those of RegB, have used trun- merely serving as anchors to keep the protein in cated versions that lack the transmembrane close contact with the source of the signal. This domains.8,23,24 lack of significant periplasmic domains is charac- In order to address this problem, we utilised teristic also of ArcB, another HPK involved in plasmid pTTQ18His, a membrane protein Activities of Full-length RegB 203 expression plasmid that has been used for the successful overexpression of 16 membrane proteins,28 – 30 to amplify expression, and sub- sequently isolate and purify, intact RegB. This is the first successful overexpression of an HPK in an heterologous E. coli host. The overexpressed protein is functional in E. coli inner membranes, as shown by its autophosphorylation activity and its ability to be dephosphorylated by RegA. Impor- tantly, it is functional following purification, since autophosphorylation, phosphotransfer and RegA- dephosphorylation activities were all demonstrable. Our kinetic data obtained for this intact protein reveal important differences compared with the truncated version of soluble RegB from R. sphaeroides,24 demonstrating that the transmembrane region has important regulatory activity. Figure 2. Autophosphorylation of RegB in E. coli inner membranes under aerobic conditions in the presence of Results [g-33P]ATP. Each reaction employed 20 mg of purified inner membranes obtained from IPTG-induced E. coli Expression of RegB NM554 (pTTQEP6 ) (lanes 2 and 4) or NM554 (pTTQregB ) (lanes 3 and 5). Reactions (20 ml final The full-length regB gene was cloned into the volumes) were incubated at 24 8C with 50 mM ATP, pTTQ18His plasmid, to produce plasmid 5 mCi of [g-33P]ATP and 1 mM DTT for ten minutes pTTQregB coding for the C-terminally His-tagged prior to the addition of 3 pmol of RegA (lanes 4 and 5). protein RegB-His6 (referred to as RegB). Following RegA was added to a reaction containing no inner introduction into E. coli NM554, bacterial growth membrane protein (lane 1). After a further ten minutes £ and regB induction with IPTG, membrane incubation, each reaction was added to 4 loading buffer and the 33P-labelled proteins visualised. preparations (mixed inner and outer membranes) were isolated using the procedure of Ward et al.29 SDS-PAGE and Western analysis of the membranes using an anti-RGSH antibody demonstrate that 6 (which does not express RegB) were shown to IPTG-induced E. coli cells carrying pTTQregB syn- possess one major and several minor auto- thesise an additional membrane protein, compared phosphorylating E. coli inner membrane proteins, to cells carrying pTTQEP6 (which possesses a trun- none of these proteins migrated at the same cated version of the regB gene). This additional 46 kDa position observed for full-length RegB protein migrates on SDS/polyacrylamide gels (Figure 2). However, membranes from E. coli with an apparent mass of 46 kDa, contrasting with NM554 carrying IPTG-induced pTTQregB clearly the predicted mass of 52.2 kDa (Figure 1). How- showed the presence of an additional phosphoryl- ever, similar anomalous migration has been ated protein band at this position and we conclude reported for other membrane proteins subjected to that this is induced RegB. Membrane-associated SDS-PAGE analysis.29 After purification, this RegB , 33P was dephosphorylated by the addition protein was identified unequivocally as full-length of RegA, which itself became phosphorylated RegB (see below). From densitometry analysis of (Figure 2). None of the pTTQEP6 autophosphoryl- Coomassie brilliant blue-stained gels, we estimate ating inner membrane proteins was de- that RegB represents 2–10% of the total protein phosphorylated by RegA or could phosphorylate content of mixed membranes of IPTG-induced this protein under the assay conditions employed E. coli NM554 (pTTQregB ) cells (data not shown). (Figure 2). The Western blots revealed an additional protein band migrating with an approximate molecular mass of .98 kDa, suggesting the presence of a dimeric form of RegB (Figure 1(b)). Purification of RegB Mixed membranes were fractionated into outer 29 Autophosphorylation of RegB in E. coli inside- and inner membranes (Ward et al. ) and RegB out inner membrane vesicles was purified by solubilisation of inner membranes in 1% (w/v) dodecyl-b-D-maltoside (DDM), The RegB protein expressed in E. coli inside-out followed by Ni2þ-NTA affinity column chromat- inner membrane vesicles was shown to be capable ography, washing with 20 mM imidazole and elut- of autophosphorylation (Figure 2). Although mem- ing with 60 mM imidazole. Densitometry analysis brane vesicles prepared from E. coli NM554 carry- of Coomassie brilliant blue-stained (Figure 3(a)) ing IPTG-induced plasmid pTTQEP6 (Table 1) and visual analysis of silver-stained (Figure 3(b)) 204 Activities of Full-length RegB

Table 1. Bacterial strains and plasmids E. coli strain Description Reference/source

DH5a supE44 Dlac U169 (f80 lacZDM15) hsdR17 recA1 endA1 gyrA96 thi-1 relA1 Clontech UK 2 2 2 BL21[DE3] F ompT gal [dcm][lon] hsdSb (rB mB ;anE. coli B strain) with DE3, a l prophage carrying Novagen Inc. the T7 RNA polymerase gene NM554 recA 13 araD139 D(ara-leu)7696 D(lac)l7A galU galK hsdR rpsL (Strr) mcrA mcrB Stratagene Plasmid pBluescript-SK Apr Stratagene pBB1 Apr, pBluescript-SK with a 4.5 kb Bam HI fragment containing regA from R. sphaeroides 2 (unpublished plasmid) pTTQ18His Apr, tac-based overexpression system encoding a C-terminal hexahistidine tag 29 pTTQEP6 Apr, pTTQ18His with a 0.9 kb PstI–EcoRI fragment of regB from pREG464 This work pTTQregB Apr, pTTQEP6 with a 0.5 kb EcoRI regB fragment from pREG464 cloned as a translational This work fusion pREG464 Apr, contains a 12 kb fragment of the R. sphaeroides reg cluster 40 pBlregA1.2. Apr, pBluescript-SK with a 0.74 kb Eco RI-BamHI fragment from pREG464, containing the This work R. sphaeroides regA pET14b Apr, T7 polymerase-based expression vector Novagen Inc. pETregA Apr, pET14b containing a 0.74 kb Nde I-Bam HI regA fragment from pBlregA1.2 cloned as a This work translational fusion

SDS/polyacrylamide gel-resolved proteins N-terminal amino acid sequencing of RegB demonstrated that full-length RegB could be puri- demonstrated that the first 11 residues of the pro- fied to .98% purity. Batch cultivation and IPTG tein were indeed as predicted (M-N-S-G-P-D-G-I- induction of E. coli NM554 carrying plasmid L-N-R-D), indicating that the region of the protein pTTQ18regB in shake flasks achieved a yield of up encoding the predicted transmembrane sensing to 1 mg of RegB per litre of culture. Fermentor- domain was intact. Sequencing of the cloned gene scale production (25 l) of RegB was attempted, confirmed the fidelity of the gene sequence in and the cell pellet of 60 g (wet weight) of IPTG- plasmid pTTQregB. Since the His-tag used for induced E. coli NM554 yielded only approximately RegB purification is situated at the C terminus, 0.1 mg of RegB protein per litre. these data collectively confirm that a full-length

Figure 3. Purification of full-length His-tagged RegB. (a) Inner membranes were separated from outer membranes, þ extracted with 1% dodecyl-b-D-maltoside and the RegB protein purified by Ni2 affinity chromatography. Proteins from each of the fractions indicated below were resolved by SDS-PAGE (15% polyacrylamide resolving gel) and visualised by staining with Coomassie brilliant blue. Lane 1, molecular mass markers. Lane 2, solubilised inner membrane proteins. Lane 3, solubilised inner membrane proteins after centrifugation (100,000 g, 40 minutes). Lane 4, solubilised inner membrane proteins after overnight incubation at 4 8C with Ni2þ-NTA resin. Lane 5, 20 mM imidazole wash fraction (first 10 ml wash volume). Lane 6, 20 mM imidazole wash fraction (final 10 ml wash volume). Lane 7, 60 mM imidazole elute fraction. (b) The efficiency of the purification procedure was assessed by subjecting eluted RegB protein to SDS-PAGE (15% polyacrylamide resolving gel) analysis and staining with silver by the method of Heukeshoven & Dernick;42 lane 1, molecular mass markers; lane 2, RegB in 60 mM imidazole elution buffer. All membrane and solubilised samples contain 30 mg of protein, eluted fraction samples contain 2.0 mg of RegB for staining with Coomassie brilliant blue or 1 mg of RegB for staining with silver. Activities of Full-length RegB 205

Autophosphorylation and dephosphorylation of purified full-length RegB RegB/A is a redox-responsive system. Therefore, to examine autophosphorylation kinetics, purified RegB was allowed to autophosphorylate in the presence of [g-33P]ATP under aerobic or anaerobic conditions. Figure 5 shows that RegB displays the expected exponential increase in proportion of phosphorylated protein, with a half-life of 8(^1) minutes. Similar autophosphorylation rates (half- life of seven minutes) have been observed before using the truncated soluble R. sphaeroides RegB.24 Figure 4. CD spectrum of purified full-length RegB Truncation of the protein therefore has little or no sensor HPK protein. Purified protein (1.37 mM) was effect on the autophosphorylation rate. The kinetics solubilised in 0.05% DDM, 10 mM sodium phosphate under aerobic and anaerobic conditions is (pH 7.4). CD spectral analysis of RegB was performed at 1 nm sampling intervals with a scan rate of 50 nm/ indistinguishable (Figure 5(d)). Moreover, when minute. Sensitivity was set at 20 mdeg. with a response RegA was added to the reaction mixture, and time of one second. The spectrum represents an average allowed to mix with RegB for ten minutes prior to of ten scans, from which any solvent contribution was addition of ATP, the rate of autophosphorylation subtracted. (which in this experiment was observed as phosphorylation of RegA rather than RegB, due to the rapid phosphotransfer step, as described in a later section) remained identical (Figure 5(c) and (d)). The autophosphorylation rate is therefore independent of the redox status of the solution, and of the protein has indeed been expressed and purified. presence of RegA. The circular dichroism spectrum between 190 and To examine dephosphorylation of phosphory- 260 nm indicated that a-helix is the predominant lated RegB, the protein was allowed to auto- type of secondary structure in this protein, and phosphorylate and reach equilibrium prior to the that structural integrity is maintained during addition of 1000-fold excess of unlabelled ATP purification (Figure 4). (Figure 6). The dephosphorylation rate of RegB , P follows a first-order decay with a half- life of 34(^5) minutes (Figure 6), which contrasts markedly with the value of 5.5 to six hours Expression and purification of RegA observed for truncated RegB24 (see Discussion). It is of interest to note that only roughly half of the The response regulator protein RegA was cloned RegB , P is dephosphorylated in the initial rapid into the pET14b plasmid, as described in Materials phase (shown here). This does not imply that only and Methods, to produce plasmid pETregA coding half of the RegB is “active”, since addition of for the N-terminal His-tagged protein His -RegA 6 RegA results in rapid loss of all 33P signal (see the (referred to as RegA). This protein was expressed next section). We are currently investigating the with a high level of efficiency (18% of total cellular origin of this observation. protein) by IPTG-induced E. coli cells carrying pET- regA. The resulting protein was purified by Ni2þ affinity chromatography to .99% as determined Phosphorylation and dephosphorylation of by analysis of both Coomassie brilliant blue- purified RegA stained and silver-stained SDS/polyacrylamide gels (data not shown). Elution of the protein from Phosphotransfer from RegB , P to RegA was too the Ni2þ-iminodiacetic acid (IDA) resin in 200 mM rapid to be measured accurately in our assays, sodium acetate (pH 4.0), the addition of 1–10 mM reaching 95–98% of completion within ten seconds DTT and maintenance of the protein concentration of mixing (Figure 7(a) and (b)). Similar rapid rates at ,10 mg/ml during buffer exchange and concen- have been seen in other systems, including the tration were required in order to prevent aggrega- soluble truncated RegB,8,23,24 RegS22 and BarA.31 tion and precipitation of the protein. N-terminal RegA was not phosphorylated in the absence of analysis of RegA demonstrated that the first 26 RegB. We therefore confirm the conclusion reached residues of the recombinant protein were as by Comolli et al.24 that the rate of phosphorylation predicted (G-S-S-H-H-H-H-H-H-S-S-G-L-V-P-R-G- of RegA is limited by the phosphorylation state of S-H-M-A-E-D-L-V-F-E) after post-translational loss RegB. of the N-terminal fMet residue. Electrospray mass Phosphorylated RegA can be dephosphorylated spectrometry analysis gave a mass of 22,515 Da by a number of mechanisms, including intra- for the protein (predicted mass 22,518 Da). The molecularly catalysed hydrolysis, or dephos- protein is functionally active and is able to phorylation by RegB. Back-transfer from RegA to dephosphorylate RegB (Figures 5 and 7). RegB is another formal possibility, though 206 Activities of Full-length RegB

Figure 5. Autophosphorylation of RegB under aerobic and anaerobic conditions, and simultaneous autophosphorylation of RegB and phosphotransfer to RegA under aerobic conditions. Anaerobic and aerobic reactions employed 60 pmol of RegB protein in a ﬁnal reaction volume of 200 ml. This was allowed to autophosphorylate in the presence of [g-33P]ATP for 40 minutes, and 20 ml samples were removed to loading buffer at selected time-points. Aerobic and anaerobic reactions were performed in the presence of 0.1 mM DTT. For the pre-mixed RegB–RegA phosphotransfer reaction, 30 pmol of RegB was incubated in the presence of 75 pmol of RegA and 1 mM DTT for ten minutes prior to the addition of [g-33P] ATP (in a ﬁnal reaction volume of 100 ml). After this addition, 10 ml samples were removed to loading buffer at selected time-points to monitor the phosphotransfer reaction. The quantity of 33P associated with either RegB or His6-RegA ((a) aerobic; (b) anaerobic; (c) pre- mixed) was then determined. (d) Time-course of phosphorylation. The data are also shown graphically (aerobic (B– –) (half-life 8(^1) minutes); anaerobic (O· · ·) (half-life 9(^1.0) minutes); pre-mixed (W—) (half-life 8(^1) minutes)). Because of the rapid phosphotransfer from RegB to RegA, the measured 33P signal was entirely on RegA and is shown by the circles and con- tinuous line.

phosphotransfer reactions between HPKs and hours), and is one of the most important ways response regulators are generally considered to be of controlling the intensity and duration of the effectively unidirectional from HPK to response signal produced by typical two-component regulator. Indeed, we found this to be the case for systems. We therefore investigated the dephos- RegB/A. Using a range of ratios of RegA to RegB, phorylation of RegA , P in the presence of full- phosphotransfer from RegB to RegA is very rapid length RegB. and appears to lie very heavily in the direction of Increasing amounts of RegB resulted in a phosphorylated RegA, since after the initial rapid reduction in the half-life of RegA , P: a 1:1 ratio phosphotransfer, no phosphorylated RegB could (RegB to RegA , P) gave a half-life of 14(^2) min- be detected using ratios in the range 1:10 to 10:1 utes, while a 5:1 ratio gave a half-life of 8(^2) min- (data not shown). Therefore, back-transfer of the utes (Figure 8(a) and (b)). (We assume here that phosphate to RegB appears to be insigniﬁcant in molar ratios are in terms of RegB transphosphory- this system. lating dimers; thus, a 1:1 ratio of RegB to RegA The half-life of phosphorylated response means a 2:1 molar ratio.) We conclude that RegB is regulators is very variable in different two- capable of catalysing loss of phosphate from component systems (from seconds to several RegA , P. Activities of Full-length RegB 207

Figure 6. Stability of RegB , 33P. (a) The reaction employed 60 pmol of RegB, which was allowed to autophosphorylate for 20 minutes in the presence of 1 mM DTT, 50 mM ATP and [g-33P]ATP prior to the addition of a 1000-fold excess of unlabelled ATP in a ﬁnal reaction volume of 200 ml. At each time- point, 6 pmol (20 ml) of RegB was removed to loading buffer and the quantity of 33P associated with RegB determined. (b) Time-course of phosphate transfer presenting the best-ﬁt exponential decay, which has a half-life of 34(^5) minutes, and an asymptote at 46% of initial activity.

Discussion (Figure 3) and demonstrated activity in vitro (Figures 5–8). On analysis by SDS-PAGE, the We have expressed in E. coli the full-length RegB protein was found to migrate with an apparent sensor kinase from R. sphaeroides as a C-terminally mass of 46 kDa (compared to an expected mass of His-tagged fusion protein (Figure 1), puriﬁed it 52.2 kDa) (Figures 1 and 3). Similar anomalous

Figure 7. Phospho-transfer of phosphate from RegB , 33Pto RegA. (a) Reactions employed 60 pmol of RegB, which was allowed to autophosphorylate in the presence of [g-33P]ATP for 20 minutes. After this time a 20 ml sample (6 pmol) was removed to loading buffer prior to the addition of 27 pmol of RegA (ﬁnal reaction volume 180 ml) and further incubation for 20 minutes. At selected time-points, 20 ml samples (6 pmol of RegB, 3 pmol of RegA) were removed to loading buffer and the quantity of 33P associated with either RegB or RegA determined. An additional control reaction (RegB-) in which 3 pmol of RegA was incubated in the presence of [g-33P]ATP for 20 minutes prior to the addition of loading buffer was performed. The reaction was performed in the presence of 1 mM DTT and at a RegB to RegA ratio of 1:1, assuming that RegB is present as a functional dimer. (b) Time- course of phosphate transfer. 208 Activities of Full-length RegB

Figure 8. Phosphatase activity of RegB towards RegA , P. (a) RegB (8 pmol)was allowed to autophosphorylate in the presence of [g-33P]ATP for 20 minutes prior to the addition of 40 pmol of RegA. After allowing phosphotransfer from RegB , 33P to RegA for 15 minutes, the ATP was removed from the reaction by Centricon buffer exchange. The reaction was then split into four aliquots, each containing 1.5 pmol of RegB and 7.5 pmol of RegA , 33P in a final reaction volume of 80 ml. To each of these reactions was then added different amounts of RegB: 74.25, 14.25, 2.25 or 0 pmol, representing ratios (RegB to RegA, assuming the presence of a functional RegB dimer) of 5:1, 1:1, 1:5 and 1:10. At selected time-points 20 ml samples were removed from each reaction to loading buffer and the quantity of 33P associated with RegB determined. The reactions were performed in the presence of 1 mM DTT. The data are presented both as (a) autoradiograms, and (b) as a time-course of phosphate transfer with fitted decay curves (1:5 O; 1:1 B; 5:1 V). The data and fitted half lives for the 1:10 RegB to RegA ratio are similar to the 1:5 ratio and are omitted from the Figure for clarity. migration has been reported for other membrane RegB implies that the relatively low proportion of proteins subjected to SDS-PAGE analysis.29 phosphorylated protein does not mean that there Previous attempts to overexpress full-length RegB is a large amount of unfolded or mis-translated have been unsuccessful;8,23,24 indeed no heter- protein. This is supported by the CD spectrum, ologous expression of a full-length HPK has been which indicates the high helical content expected reported. The only other membrane HPKs whose of a correctly folded RegB protein (Figure 4). This amplified homologous expression and purification effect of a high concentration of DTT on RegB has been achieved to date are E. coli KdpD25 and remains to be investigated and could implicate NarX.26,27 The successful application of the either an artefact of the in vitro assay system, or pTTQ18His vector reported here may therefore the reduction of an intermolecular disulphide open the door to more comprehensive studies of bridge as part of the control of the RegB system. full-length HPKs, since it permits reproducible Although DTT is not of significance in vivo, the purification of high-purity functional RegB in existence of additional factors that enhance HPK milligram quantities by shake-flask culture. autophosphorylation rates in response to their Although fermentor-scale production of this stimuli has been noted; for example, FixL rates of protein is possible, optimisation of culture and autophosphorylation are stimulated significantly induction conditions will be required in order to in response to oxygen when manganese is added.32 achieve high yields. Until now, only soluble truncated forms of Quantification of the 33P present in auto- RegB24 or its analogues in R. capsulatus8,23 and phosphorylated RegB suggests that only 25–30% B. japonicum (RegS)22 have been used for in vitro of RegB protein produced is phosphorylated functional analysis. For all these truncated under the conditions used in our assays. However, proteins, a constitutive activity was reported, this is comparable to the fraction of phosphoryl- which was suggested to be due to the absence of ated protein reported previously for the soluble the N-terminal transmembrane domain that could domain.24 A greater fraction (50–60%) of about be required for inhibiting autophosphorylation of twice as much phosphorylated protein can be the catalytic domain. More recently, it has been obtained by carrying out functional assays in the suggested that the regulation of the RegB/A two- presence of high concentrations (100 mM) of DTT. component system by O2 occurs indirectly, through This increase in the fraction of phosphorylated the volume of electron flow through the cbb3-type Activities of Full-length RegB 209 cytochrome c oxidase.9,15 We have observed that increases with decreasing osmolarity.35 The full-length RegB in inner membrane vesicles of mechanism may be similar for RegB in response E. coli (which do not possess any identifiable to redox potential, though there are clear homologue of the RegB/A two-component system) differences between the EnvZ/OmpR and RegB/ demonstrates constitutive autophosphorylation RegA systems; the latter seems to exist in a kinase under aerobic conditions (Figure 2). Furthermore, “on” or “off” state, whereas the former is in active the purified protein has identical autophosphoryla- signalling states under high and low stimulus tion kinetics under both aerobic and anaerobic con- levels. ditions (Figure 5(d)). These results therefore Previous experiments with R. sphaeroides RegA24 confirm that the full-length RegB kinase is active have shown that RegA phosphorylated chemically under both aerobic and anaerobic conditions, and using acetyl phosphate has a half-life of ca 330 that presumably an additional redox-responsive minutes, while the very similar R. capsulatus regulator is required for repression, a potential RegA , P has a half-life of ca 90 minutes.23 Trun- candidate for which is the Rhodobacter protein cated RegB was found to decrease the half-life of SenC,16,17 which may itself be modified by RegA , P to ca 20 minutes.24 Our results show interaction with an oxidised/reduced electron that full-length RegB decreases the half-life of carrier. RegA , P to approximately eight minutes. Taking A recent study on the Sinorhizobium meliloti into account experimental difficulties in obtaining FixL/FixJ system demonstrated that complexation accurate rates in these assays, the results suggest of the histidine protein kinase and response that truncated RegB has phosphatase activity that regulator before phosphorylation increased the displays approximately the same catalytic rate as autophosphorylation rate by a factor of 10, and that of the full-length RegB. The TMR is therefore that autophosphorylation and phosphotransfer not implicated in regulation of the loss of signal were coupled.33 By contrast, the autophosphoryl- from the RegB/RegA system at the level of ation rate of full-length RegB is independent of RegA , P, although, as shown above, it is involved the presence of RegA (Figure 5(d)), indicating in the control of the level of phosphorylation of that complexation of the two components is not RegB. RegA/B regulates major metabolic changes necessary for phosphorylation in this system. in the cell in response to redox potential. This is The half-life of full-length RegB , P was an energy-consuming process and it is therefore approximately 34 minutes. By contrast, the soluble not unreasonable to find that there is more control domain of RegB , P had a half-life of 5.5 to six over the initiation of the signal than over its hours.24 This therefore suggests that the trans- termination. It is noteworthy that in all studies of membrane region has a regulatory role in the the R. sphaeroides Reg system so far, including our stability of the phosphorylated transmitter domain. own of the full-length RegB, the RegB dephos- If so, it implies that redox status will have an effect phorylation activity towards RegA is clearly on the autophosphorylation rate of RegB, and on demonstrable in the absence of ATP. This contrasts its dephosphorylation rate. Because the phospho- with findings for some other two-component transfer step is very rapid compared to both the systems; for example, EnvZ-mediated dephos- autophosphorylation and the dephosphorylation phorylation of OmpR , P requires the presence of RegB, the amount of RegA , P signal is regu- of ATP or some non-hydrolysable analogue, imply- lated tightly by the phosphorylation status of ing that ATP was required as a co-factor, perhaps RegB. The dual effect of redox status on both phos- to permit EnvZ to adopt the correct phorylation and dephosphorylation of RegB would conformation.37,38 therefore accentuate the effectiveness of the redox- In conclusion, we have demonstrated that func- dependent switch; this is reminiscent of the switch- tional full-length RegB can be expressed in E. coli ing on or off of the glycolytic pathway regulated by using the pTTQ18His vector. It has the same func- phosphorylation of a single serine residue on the tion but displays different kinetics compared with tandem enzyme phosphofructokinase 2/fructose soluble versions that lack the transmembrane bisphosphatase 2. Phosphorylation activates the domain, suggesting that the TMR has a regulatory phosphatase activity but inhibits the kinase function. The system has been developed for in activity, thereby exercising a dramatic effect on the vitro studies to elucidate the signal sensing concentration of fructose 2,6-bisphosphate.34 mechanism of RegB, by examining RegB auto- Indeed, switching between kinase- and phos- phosphorylation in combination with candidate phatase-dominant states is widely recognised as interacting proteins such as SenC and components the mechanism by which many bacterial HPKs of the cytochrome c oxidase complex. Targeted respond to environmental signals. For example, mutagenesis will facilitate elucidation of the mutation studies of EnvZ, an HPK that regulates structure-activity relationships of the single RegB, porin expression in response to osmolarity in RegA and regulatory proteins as well as their E. coli, have successfully identified the regions complexes. The ability to produce milligram and residues involved in these distinct activities quantities of highly purified RegB protein is also and support the idea that the kinase-dominant enabling us to undertake 2D/3D crystallisation in state is favoured in high osmolarity, whereas the order to elucidate the 3D structure of this sensor propensity for the phosphatase-dominant state kinase by electron or X-ray diffraction. 210 Activities of Full-length RegB

Materials and Methods pREG46439 (isolated from a pSUP202 library of R. sphaeroides chromosomal DNA and possessing 12 kb Bacterial strains and plasmids, DNA manipulation of the reg region) as template. The 0.74 kb product was and reagents digested with EcoRI and BamHI for cloning into EcoRI– BamHI-cut pBluescript-SK to create pBlregA1.2. The Bacterial strains and plasmids are listed in Table 1. All regA fragment was isolated from pBlregA1.2 using NdeI restriction enzymes and phage T4 ligase were obtained and Bam HI, and ligated into pET14b, resulting in the from GibcoBRL; Pfu polymerase was obtained from final expression plasmid pETregA. The expressed RegA g 33 Boehringer Mannheim. [ - P]ATP (3000 Ci/mmol) was protein possesses an additional N-terminal M-G-S-S- obtained from ICN Pharmaceuticals Ltd. Dodecyl-N- (H )-S-S-G-L-V-P-R-G-S-H-M-A-E-L sequence, where maltoside was obtained from Melford Biosciences. 6 2þ A-E-L are the first three amino acid residues in native Agarose-immobilised Ni -nitrilotriacetic acid (NTA) RegA. resin and anti-RGS(H6) monoclonal antibody were obtained from Qiagen. Sepharose 6B fast-flow immobilised Ni2þ-iminodiacetic acid (IDA) resin was obtained DNA sequencing from Sigma Chemical Co. Goat anti-mouse IgG horse radish peroxidase (HRP)-conjugated antibody was DNA sequencing was performed using an ABI obtained from Stratech Scientific Ltd. All media, stock BigDyee deoxy terminator cycle sequencing kit, sup- buffers and procedures for growth of bacterial cultures plied by Applied Biosystems, and sequencing and DNA manipulations followed the methods of reactions were analysed on an Applied Biosystems 377 Sambrook et al.38 All other chemicals and reagents Prism automated sequencer (University of Leicester, employed were of AnalaR or equivalent grade unless UK). otherwise stated. Overexpression of full-length RegB Construction of RegB and RegA E. coli NM554 carrying pTTQ18reg B was cultured overexpression plasmids aerobically at 37 8C in 2 l of Luria–Bertani (LB) medium38 in the presence of 100 mg/ml of carbenicillin, to an To overexpress a full-length His6-tagged version of absorbance at 595 nm (A595) of 0.45. Protein expression RegB that includes the transmembrane sensing domain, was induced through the addition of 1 mM IPTG. After regB was amplified by polymerase chain reaction using a further three hours incubation at 37 8C, and at pBB1 (pBluescript-SK possessing a 4.5 kb BamHI reg A . 1.4, cells were harvested by centrifugation for ten fragment isolated from a pSUP202 plasmid library of 595 2 minutes at 8000 g and 4 8C. Cell pellets were resus- R. sphaeroides chromosomal DNA) as template. The regB pended in 10 mM Tris–HCl (pH 8.0), 0.5 mM EDTA, gene was amplified using the upstream primer 20 mM mercaptoethanol wash buffer at 4 8C, re-centri- 50ATGAGCTGCAATGAATTCCGGTCCCGACG and the 0 fuged and resuspended in wash buffer to a final volume downstream primer 5 -GGCGCCGGCTGCAGTCTG- of 14 ml per litre of original culture. Cell suspensions GATCAGGACG. The PCR product possesses EcoRI and were then stored at 270 8C prior to purification of the Pst I sites for subsequent cloning into pTTQ18His, a RegB protein. plasmid based on expression vector pTTQ18 that we have used previously to successfully overexpress 16 bacterial membrane proteins.29 Both plasmids possess a Overexpression of RegA tac promoter that drives transcription of the gene of interest, but pTTQ18His possesses restriction sites that Recombinant RegA protein encoded by plasmid pET- permit in-frame fusion of the inserted gene with a 30 regA possesses an N-terminal tag that includes R-G-S- 2þ (H6) that facilitates both Ni -NTA affinity chroma- sequence encoding a G-G-R-G-S-(H6) C-terminal tag. Since reg B possesses an internal EcoRI site, the amplified tography purification of the protein and its antibody- product was digested with EcoRI and PstI to produce mediated detection. To obtain RegA protein, plasmid two fragments that were isolated and cloned in turn pPETregA was transformed into the expression host into pTTQ18His. First the 0.9 kb PstI–EcoRI reg B frag- BL21 [DE3] and cultured aerobically at 37 8CinLB ment was cloned into PstI–EcoRI-cut pTTQ18His to medium containing 500 mg/ml of carbenicillin in shake- ¼ make pTTQEP6, a control plasmid possessing only the flasks to A595 0.7. Protein expression was then induced 0.9 kb 30 end of the 1.4 kb full-length reg B gene. The by addition of 0.4 mM IPTG. After a further three hours 0.5 kb EcoRI fragment was then inserted into EcoRI-cut of incubation at 37 8C and at an A595 value of ,1.2, cells pTTQEP6 to create the full-length reg B plasmid were harvested by centrifugation at 8000 g for ten pTTQreg B. The inserted fragments were sequenced to minutes at 4 8C. Cell pellets were washed once in verify correct orientation and sequence. The cloning 10 mM Tris–HCl (pH 8.0), 0.5 mM EDTA at 4 8C, procedure necessitated expression of a reg B product re-harvested and the pellets stored at 270 8C for sub- with N-terminal sequence fM-N-S-G-P-D (rather than sequent purification of RegA protein. the native fM-I-L-G-P-D). The C-terminal region of RegB is predicted to possesses the His6 tag in the sequence Purification of RegB inner membrane fractions and I-Q-T-A-G-G-R-G-S-(H)6, where I-Q-T are the last three soluble protein residues in native RegB and the R-G-S(H)6 sequence can be used in antibody-based detection of the Reg-B Cell suspensions were thawed slowly on ice and dis- protein. rupted in a French pressure cell as described by Ward To overexpress RegA, the regA gene was amplified by et al.29 E. coli inside-out inner membrane vesicles for polymerase chain reaction using upstream primer 50- autophosphorylation assays were purified by the GAGTCGAATTCATATGGCTGAGGATCTGGTATTCG- method of Ward et al.29 modified by including 10 mM 0 AACTCG and downstream primer 5 -GCAGAG- KCl, 10 mM MgCl2 and 0.1 mM DTT in all buffers and GATCCGCCTGCCAATGAAAAAGGCGGCAA using all step-gradients employed. The purified vesicles were Activities of Full-length RegB 211 resuspended in 20 mM Tris–HCl (pH 7.5) containing (10 kDa cutoff) centrifugal filter device employed as per 10 mM KCl, 10 mM MgCl2 and 0.1 mM DTT at a concen- the manufacturer’s instructions (Millipore Co.). tration of approximately 20 mg protein/ml of buffer. Membranes were employed in phosphorylation assays immediately after preparation. Western blotting To purify RegB protein, inner membrane vesicles were Mixed membranes from induced and uninduced again prepared as described by the method of Ward E. coli cultures carrying pTTQEP6 and pTTQreg B were 29 et al. modified by including 20 mM mercaptoethanol in prepared from French-pressed cells according to the all step-gradients and wash buffers. The purified inner method of Ward et al.29 and separated by SDS-PAGE, membranes were resuspended in 20 mM Tris–HCl (pH employing 15% polyacrylamide resolving gels. Proteins 7.5) containing 20 mM mercaptoethanol, at a concen- were then transferred to Fluorotranse membrane (Pall tration of approximately 20 mg protein/ml, flash-frozen BioSupport, UK) by semi-dry electroblotting (350 mA and stored at 270 8C. Affinity chromatography purifi- for one hour). Membranes were incubated for 16 hours cation was then carried out according to the protocols of in 5% (w/v) skimmed milk powder in TBST (10 mM 29,30 Ward et al. Inner membranes were thawed rapidly Tris–HCl (pH 8.0), containing 150 mM NaCl and 0.05% by warming in tepid water and then resuspended at a (v/v) Tween 20). A 1:2000 dilution of mouse anti- final concentration of 2 mg protein/ml in 10 mM Hepes RGS(H6) monoclonal antibody (Qiagen Ltd) was then buffer (pH 7.6) containing 20% (w/v) glycerol, 20 mM prepared in TBST, into which the membrane was imidazole, 1% (w/v) DDM and 20 mM mercaptoethanol immersed and incubated with gentle agitation for one for two hours at 4 8C. The preparation was then centri- hour at room temperature. Following three washes in fuged at 4 8C for 40 minutes at 100,000 g to sediment 100 ml of TBST, a 1:5000 dilution of goat anti-mouse IgG insoluble material. His-tagged RegB protein in the super- horse radish peroxidase conjugate (Stratech Scientific natant fraction was then allowed to mix with and bind to Ltd) in TBST was added and the membrane incubated 0.5 ml of Ni-NTA agarose-immobilised resin (Qiagen) at for one hour at room temperature. Following three 4 8C for 18 hours. The resin was recovered by centrifu- wash steps with large volumes of TBST, the immunoblot gation at 1000 g and transferred to a 2 ml column (Perbio membrane was incubated with ECL Western blotting Science UK Ltd). The resin bed was washed with 80 ml detection reagent (Amersham Pharmacia Biotech UK of 10 mM Hepes buffer (pH 7.6) containing 0.05% DDM, Ltd) and developed by autoradiography with Xograph 20% glycerol, 20 mM imidazole and 20 mM mercapto- film (Eastman Kodak Co.). ethanol. RegB protein was then eluted in 10 mM Hepes buffer (pH 7.6) containing 20% glycerol, 60 mM imidazole and 20 mM mercaptoethanol. The eluted protein Protein sequencing was concentrated in 10 mM Hepes buffer (pH 7.6) con- m m taining 50% glycerol, 0.05% DDM and 1 mM DTT Approximately 3 g of purified RegB or 2 g RegA through the use of a Centricon YM-10 (10 kDa cutoff) fil- was loaded onto an SDS/15% polyacrylamide gel and e ter (Millipore Co.), employed as per the manufacturer’s transferred to Fluorotrans membrane (Pall BioSupport, instructions. UK) by semi-dry electroblotting. The protein was visualised by staining with Coomassie brilliant blue, excised from the membrane and the N-terminal sequence determined by Edman degradation, courtesy Purification of RegA protein of Dr Arthur Moir (University of Sheffield, UK). The pET system-specified affinity chromatography protocols of Novagen Inc. were used for the affinity chro- Circular dichroism (CD) spectroscopy matography purification of His6-RegA. Cell pellets were thawed slowly on ice and resuspended in 4 ml (per Purified RegB was buffer-exchanged into 10 mM 250 ml original culture) of 20 mM imidazole binding sodium phosphate (pH 7.4) containing 0.05% DDM buffer (20 mM Tris–HC1 (pH 7.9) containing 20 mM using a Centricon YM-40 (40 kDa cutoff) centrifugal filter imidazole, and 500 mM NaCl). Resuspended cells were device (Millipore Co.), employed as per the manu- then disrupted by sonication (eight cycles of ten seconds facturer’s instructions. The concentration of protein in sonication and 15 second intervals) on ice. Sonicated the sample was adjusted to 1.37 mM, and 300 ml was material was then centrifuged at 100,000 g for 40 minutes transferred to a Hellman quartz-glass cell of 1 mm path- at 4 8C. His-tagged RegA protein in the supernatant was length. Circular dichroism measurements were per- allowed to bind to 1 ml of Ni2þ-NTA Sepharose resin formed on a Jasco J-715 spectropolarimeter at 20 8C with (Sigma Chemical Co.) at 4 8C for 40 minutes. The resin constant nitrogen flushing. was recovered by centrifugation at 1000 g for 30 seconds and the supernatant removed. The resin was washed Electrospray mass spectroscopy with ten 10 ml volumes of 20 mM imidazole binding buffer and ten 10 ml volumes of 60 mM imidazole wash Samples of purified RegA were prepared for electro- buffer (20 mM Tris–HC1 (pH 7.9) containing 60 mM spray mass spectroscopy by the method of Hufnagel imidazole, and 500 mM NaCl). The resin was then trans- et al.40 Samples were analysed on a single quadrupole, ferred to a 2 ml column (Perbio Science UK) and the bench-top mass spectrometer (Platform II, Micromass packed resin bed washed with a further 20 ml of 60 mM UK Ltd). The mass spectrometer was fitted with a stan- wash buffer. RegA protein was eluted in 200 mM sodium dard electrospray ionisation source, which was used in acetate buffer (pH 4.0) and stored in 100 mM sodium the positive ionisation mode with a probe tip voltage of acetate buffer (pH 4.0), containing 50 mM MgCl2, and 3.5 kV, and a counter electrode voltage of 0.5 kV. 10 mM DTT. Prior to use in transphosphorylation Nitrogen was employed as both the nebulising and the experiments, RegA protein was concentrated to 7 mg/ drying gas, with flow rates of 20 l per hour and 200 l ml in 20 mM sodium acetate buffer (pH 4.0) containing per hour, respectively. The sampling cone voltage 10 mM MgCl2 and 1 mM DTT using a Centricon YM-10 was set at 40 V. The sample was dissolved in formic 212 Activities of Full-length RegB acid/methanol/water (1:1:1, by volume) and infused References into the ionisation source at a flowrate of 10 ml/minute. Data were acquired over the appropriate m/z range and 1. Sganga, M. & Bauer, C. E. (1992). Regulatory factors were processed using the MassLynx software supplied controlling photosynthetic reaction center and light- with the instrument. The m/z spectrum was transposed harvesting gene expression in Rhodobacter capsulatus. onto a true molecular mass scale for more facile identifi- Cell, 68, 945–954. cation using Maximum Entropy processing techniques. 2. Phillips-Jones, M. K. & Hunter, C. N. (1994). Cloning An external calibration was applied, using horse heart and nucleotide sequence of regA, a putative response myoglobin (16,951.49 Da) as the calibrant. regulator gene of Rhodobacter sphaeroides. FEMS Microbiol. Letters, 116, 269–276. 3. Eraso, J. M. & Kaplan, S. (1994). prrA, A putative In vitro phosphorylation assays response regulator involved in oxygen regulation of Rhodobacter Assays were carried out in an assay buffer containing photosynthesis gene expression in sphaeroides. J. Bacteriol. 176, 32–43. 50 mM Tris–HCl (pH 7.6), 10 mM MgCl2, 50 mM KCl and 0.1–1 mM DTT and were performed at 24 8C (modi- 4. Bauer, E., Kaspar, T., Fischer, H. M. & Hennecke, H. fied from Inoue et al.8 and Bird et al.23). For auto- (1998). Expression of the fixR-nifA operon in phosphorylation assays, 60 pmol of RegB protein was Bradyrhizobium japonicum depends on a new response used, whilst phosphorelay assay mixes employed 30– regulator, RegR. J. Bacteriol. 180, 3853–3863. 300 pmol of RegB and 50–150 pmol of RegA. Unless 5. Tiwari, R. P., Reeve, W. G., Dilworth, M. J. & Glenn, stated otherwise, assays (190 ml) were initiated through A. R. (1996). Acid tolerance in Rhizobium meliloti the addition of 10 ml of radiolabelled ATP (10 mmol ATP strain WSM419 involves a two-component sensor- containing 50 mCi of [g-33P]ATP). Samples (20 ml) were regulator system. Microbiology, 142, 1693–1704. removed at intervals and reactions stopped by the 6. Masuda, S., Matsumoto, Y., Nagashima, K. V., addition of 5 mlof4£ loading buffer (12% glycerol, 3% Shimada, K., Inoue, K., Bauer, C. E. & Matsuura, K. water, 10% (w/v) SDS, 1 M Tris–HCl (pH 7.2) 0.002% (1999). Structural and functional analyses of photo- bromophenol blue, 3% mercaptoethanol). Assays per- synthetic regulatory genes regA and regB from formed under anaerobic conditions were performed in a Rhodovulum sulfidophilum, Roseobacter denitrificans, microflow anaerobic workstation system (Inter Med and Rhodobacter capsulatus. J. Bacteriol. 181, M.D.H) supplied with white-spot nitrogen (British 4205–4215. Oxygen Co.). For anaerobic assays, small volumes of all 7. Joshi, H. M. & Tabita, F. R. (1996). A global two com- reagents were allowed to equilibrate under anaerobic ponent signal transduction system that integrates the conditions for one hour prior to the initiation of phos- control of photosynthesis, carbon dioxide assimila- phorylation experiments. All samples were stored at tion, and nitrogen fixation. Proc. Natl Acad. Sci. USA, 270 8C. 93, 14515–14520. 8. Inoue, K., Kouadio, J.-L. K., Mosley, C. S. & Bauer, C. E. (1995). Isolation and in vitro phosphorylation Quantification of 33P-labelled proteins of sensory transduction components controlling anaerobic induction of light harvesting and reaction 33P-labelled RegB and 33P-labelled RegA proteins were center gene expression in Rhodobacter capsulatus. resolved by SDS-PAGE using 12–15% resolving and Biochemistry, 34, 391–396. 4–5% stacking gels.38 Gels were dried at 80 8C under vac- 9. Oh, J.-I., Ko, I.-J. & Kaplan, S. (2001). The default uum, and the labelled proteins visualised by auto- state of the membrane-localized histidine kinase radiography. Quantification of 33P-label incorporated PrrB of Rhodobacter sphaeroides 2.4.1 is in the kinase- into proteins was determined using a Fugii BAS 1000 J. Bacteriol. 183 phosphorimaging system (Fujifilm Co.) using a series of positive mode. , 6807–6814. diluted [g-33P]ATP standards. Densitometry analysis of 10. Qian, Y. & Tabita, F. R. (1996). A global signal trans- phosphoimager data was performed by means of Aida duction system regulates aerobic and anaerobic CO2 1D/2D analytical software (Raytest). Kinetic parameters fixation in Rhodobacter sphaeroides. J. Bacteriol. 178, were fitted by a least-squares calculation using Excel 12–18. (Microsoft Corp., Seattle). 11. Dubbs, J. M., Bird, T. H., Bauer, C. E. & Tabita, F. R. (2000). Interaction of CbbR and RegAp transcription

regulators with the Rhodobacter sphaeroides cbbI Protein determinations promoter-operator region. J. Biol. Chem. 275, 19224–19230. Protein assays for membranes containing RegB, 12. Elsen, S., Dischert, W., Colbeau, A. & Bauer, C. E. puriﬁed RegB and RegA proteins were performed using 41 (2000). Expression of uptake hydrogenase and moly- the method of Schaffner & Weissman. bdenum nitrogenase in Rhodobacter capsulatus is coregulated by the RegB–RegA two-component regulatory system. J. Bacteriol. 182, 2831–2837. 13. Swem, L. R., Elsen, S., Bird, T. H., Swem, D. L., Koch, Acknowledgments H.-G., Myllykallio, H. et al. (2001). The RegB/RegA two-component regulatory system controls synthesis We thank Dr Nick Rutherford for assistance in the of photosynthesis and respiratory electron transfer puriﬁcation of RegB, John O’Reilly for the preparation components in Rhodobacter capsulatus. J. Mol. Biol. of E. coli inner membranes, Dr Arthur Moir for analysis 309, 121–138. of N-terminal amino acid sequences and Dr Alison 14. Kappler, U., Huston, W. M. & McEwan, A. G. (2002). Ashcroft for the electrospray mass spectroscopy analysis Control of dimethylsulfoxide reductase expression in of RegA. This work was supported by BBSRC grant Rhodobacter capsulatus: the role of carbon metabolites 24/B12958, and by the Wellcome Trust, who funded the and the response regulators DorR and RegA. circular dichroism and mass spectrometry instruments. Microbiology, 148, 605–614. Activities of Full-length RegB 213

15. Oh, J. I. & Kaplan, S. (2000). Redox signaling: amplified expression, identification, purification, globalization of gene expression. EMBO J. 19, assay and properties of hexahistidine-tagged 4237–4247. bacterial membrane transport proteins. In Membrane 16. Buggy, J. & Bauer, C. E. (1995). Cloning and charac- Transport—A Practical Approach (Baldwin, S. A., ed.), terization of senC, a gene involved in both aerobic pp. 141–166, Oxford University Press, Oxford. respiration and photosynthesis gene expression in 30. Ward, A., Hoyle, C., Palmer, S., O’Reilly, J., Griffith, Rhodobacter capsulatus. J. Bacteriol. 177, 6958–6965. J., Pos, M. et al. (2001). Prokaryote multidrug efflux 17. Eraso, J. M. & Kaplan, S. (2000). From redox flow to proteins of the major facilitator superfamily: ampli- gene regulation: role of the PrrC protein of fied expression, purification and characterisation. Rhodobacter sphaeroides 2.4.1. Biochemistry, 39, Mol. Microbiol. Biotechnol. 3, 193–200. 2052–2062. 31. Pernestig, A. K., Melefors, O. & Georgellis, D. (2001). 18. Ouchane, S. & Kaplan, S. (1999). Topological analysis Identification of UvrY as the cognate response regu- of the membrane-localized redox-responsive sensor lator for the BarA sensor kinase in Escherichia coli. kinase PrrB from Rhodobacter sphaeroides 2.4.1. J. Biol. J. Biol. Chem. 276, 225–231. Chem. 274, 17290–17296. 32. Gilles-Gonzalez, M. A. & Gonzalez, G. (1993). 19. Georgellis, D., Kwon, O. & Lin, E. C. C. (2001). Regulation of the kinase activity of heme protein Quinones as the redox signal for the Arc two- FixL from the two component system FixL/FixJ of component system of bacteria. Science, 292, Rhizobium meliloti. J. Biol. Chem. 268, 16293–16297. 2314–2316. 33. Tuckerman, J. R., Gonzalez, G. & Gilles-Gonzalez, 20. Kwon, O., Georgellis, D. & Lin, E. C. (2000). Phos- M. A. (2001). Complexation precedes phosphoryl- phorelay as the sole physiological route of signal ation for two-component regulatory system FixL/ transmission by the Arc two-component system of FixJ of Sinorhizobium meliloti. J. Mol. Biol. 308, Escherichia coli. J. Bacteriol. 182, 3858–3862. 449–455. 21. Taylor, B. L. & Zhulin, I. B. (1999). PAS domains: 34. Van Schaftingen, E. & Hers, H. G. (1986). Purification internal sensors of oxygen, redox potential, and and properties of phosphofructokinase 2/fructose light. Microbiol. Mol. Biol. Rev. 63, 479–506. 2,6-bisphosphatase from chicken liver and from 22. Emmerich, R., Panglungtshang, K., Stehler, P., pigeon muscle. Eur. J. Biochem. 159, 359–365. Hennecke, H. & Fischer, H.-M. (1999). Phosphoryl- 35. Russo, F. & Silhavy, T. J. (1991). EnvZ controls the ation, dephosphorylation and DNA-binding of the concentration of phosphorylated OmpR to mediate Bradyrhizobium japonicum RegSR two-component osmoregulation of the porin genes. J. Mol. Biol. 222, regulatory proteins. Eur. J. Biochem. 263, 455–463. 567–580. 23. Bird, T. H., Du, S. & Bauer, C. E. (1999). Auto- 36. Aiba, H., Mizuno, T. & Mizushima, S. (1989). Trans- phosphorylation, phosphotransfer, and DNA-bind- fer of phosphoryl group between two regulatory ing properties of the RegB/RegA two-component proteins involved in osmoregulatory expression of regulatory system in Rhodobacter capsulatus. J. Biol. the ompF and ompC genes in Escherichia coli. J. Biol. Chem. 274, 16343–16348. Chem. 264, 8563–8567. 24. Comolli, J. C., Carl, A. J., Hall, C. & Donohue, T. 37. Igo, M. M., Ninfa, A. J., Stock, J. B. & Silhavy, T. J. (2002). Transcriptional activation of the Rhodobacter (1989). Phosphorylation and dephosphorylation of a sphaeroides cytochrome c2 gene P2 promoter by the bacterial transcriptional regulator by a trans- response regulator PrrA. J. Bacteriol. 184, 390–399. membrane receptor. Genes Dev. 3, 1725–1734. 25. Stallkamp, I., Dowhan, W., Altendorf, K. & Jung, K. 38. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989). (1999). Negatively charged phospholipids influence Molecular Cloning: A Laboratory Manual (Nolan, C., the activity of the sensor kinase KdpD of Escherichia ed.), Cold Spring Harbor Laboratory Press, Cold coli. Arch. Microbiol. 172, 295–302. Spring Harbor, NY. 26. Walker, M. S. & DeMoss, J. A. (1993). Phosphoryl- 39. Hunter, C. N. & Turner, G. (1988). Transfer of genes ation and dephosphorylation catalyzed in vitro by coding for apoproteins of reaction center and light- purified components of the nitrate sensing system, harvesting LH1 complexes to Rhodobacter sphaeroides. NarX and NarL. J. Biol. Chem. 268, 8391–8933. J. Gen. Microbiol. 134, 1471–1480. 27. Lee, A. I., Delgado, A. & Gunsalus, R. P. (1999). 40. Hufnagel, P., Schweiger, U., Eckerskorn, C. & Signal-dependent phosphorylation of the mem- Oesterhelt, D. (1996). Electrospray ionization mass brane-bound NarX two-component sensor- spectrometry of genetically and chemically modified transmitter protein of Escherichia coli: nitrate elicits a bacteriorhodopsins. Anal. Biochem. 243, 46–54. superior anion ligand response compared to nitrite. 41. Schaffner, W. & Weissmann, C. (1973). A rapid, sensi- J. Bacteriol. 181, 5309–5316. tive, and specific method for the determination of 28. Henderson, P. J. F., Hoyle, C. K. & Ward, A. (2000). protein in dilute solution. Anal. Biochem. 56, 502–514. Expression, purification and properties of multidrug 42. Heukeshoven, J. & Dernick, R. (1988). Improved efflux proteins. Trans. Biochem. Soc. 28, 513–517. silver staining procedure for fast staining in Phast- 29. Ward, A., Sanderson, N. M., O’Reilly, J., Rutherford, System development Unit. I. Staining of sodium N. G., Poolman, B. & Henderson, P. J. F. (1999). The dodecyl sulfate gels. Electrophoresis, 9, 28–32.

Edited by B. Holland

(Received 25 February 2002; received in revised form 22 April 2002; accepted 29 April 2002)