Expression, purification and activities of the entire family of intact membrane sensor kinases from Enterococcus faecalis

Abstract

Two-component signal transduction systems are the main mechanism by which bacteria sense and respond to their environment, and their membrane-located histidine protein kinases generally constitute the sensory components of these systems. Relatively little is known about their fundamental mechanisms and precise nature of the molecular signals sensed, because of the technical challenges of producing sufficient quantities of these hydrophobic membrane proteins. This study evaluated the heterologous production, purification and activities of the 16 intact membrane sensor kinases of Enterococcus faecalis. Following the cloning of the genes into expression plasmid pTTQ18His, all but one kinase was expressed successfully in Escherichia coli inner membranes. Purification of the hexa-histidine ‘tagged’ recombinant proteins was achieved for 13, and all but one were verified as intact. Thirteen intact kinases possessed autophosphorylation activity with no added signal when assayed in membrane vesicles or as purified proteins. Signal testing of two functionally-characterized kinases, FsrC and VicK, was successful examplifying the potential use of in vitro activity assays of intact proteins for systematic signal identification. Intact FsrC exhibited an approximately 10-fold increase in activity in response to a two-fold molar excess of synthetic GBAP pheromone, whilst glutathione, and possibly redox potential, were identified for the first time as direct modulators of VicK activity in vitro. The impact of DTT on VicK phosphorylation resulted in increased levels of phosphorylated VicR, the downstream response regulator, thereby confirming the potential of this in vitro approach for investigations of modulator effects on the entire signal transduction process of two-component systems.

• Two-component signal transduction systems
• Enterococcus faecalis
• histidine kinase
• membrane receptor
• FsrC and quorum sensing
• VicK and YycG

Abbreviations
DDM	=	dodecyl-β-D-maltoside
DTT	=	dithiothreitol
EDTA	=	ethylenediamine tetraacetic acid
ESI-MS	=	electrospray ionization mass spectrometry
GBAP	=	gelatinase biosynthesis-activating pheromone
GSH	=	reduced glutathione
GSSG	=	oxidized glutathione
HK	=	histidine kinase
IPTG	=	isopropyl -D-1-thiogalactopyranoside
LB	=	Luria-Bertani
LDAO	=	lauryldimethylamine-oxide
Ni2+-NTA	=	nickel-nitrilotriacetic acid
PAS	=	Per-Arnt-Sim
RR	=	response regulator
SDS-PAGE	=	sodium dodecyl sulphate polyacrylamide gel electrophoresis
TCS	=	two-component systems
TM	=	transmembrane

Introduction

Enteroccocus faecalis is a Gram-positive, commensal bacterium commonly found in the gastrointestinal tract of many mammals, including humans [1]. Since the late 1970s enterococci have been increasingly reported as significant causative agents of serious and life-threatening nosocomial infections in both the USA and Europe, including endocarditis, nosocomial bacteraemia, surgical wound infections, and urinary tract infections [2]. This rise in the occurrence of infection accredited to E. faecalis has also been accompanied by intrinsic and increased resistance to most currently approved antibacterial agents, including the ‘last-line’ glycopeptide vancomycin [3], [4].

E. faecalis uses two-component signal transduction systems (TCSs) to monitor and respond to changes in its environment. These systems are found in almost all bacteria and are used to regulate a variety of cellular processes ranging from sporulation, biofilm formation, chemotaxis, and virulence to antibiotic production and resistance [5–8]. TCSs consist of two proteins, a signal-sensing histidine protein kinase (HK) usually in the cell membrane that senses an environmental change, and a cytoplasmic cognate response regulator (RR) that effects an appropriate response, usually regulation of gene expression. The HK responds to the environmental stimulus by autophosphorylating on a conserved histidine residue in the cytoplasmic autokinase domain. The phosphate is then transferred to an aspartate residue on the partner RR, which is then able to effect a response, such as modulation of transcription through increased affinity for its target promoter DNA, thereby eliciting a cellular response to a changing environment (e.g., [9]).

Most bacterial species possess multiple TCS pairs and therefore these species are able to sense and respond to a variety of different environmental cues. In the completed genome sequence of E. faecalis strain V583, 17 TCS pairs have been identified, along with one orphan RR [10]. Using the HK classification scheme described previously [11], Hancock and Perego sub-grouped the 17 HK proteins. All but one of the HKs were predicted to be localized in the membrane [10]. Similarity searches [10], systematic inactivation of RR components [12–15] and investigations using infection models [14], [15] revealed important insights into the potential functions of these two-component signalling systems, some of which are associated with virulence and/or responses to antibacterial agents and other stresses. These are summarized below, and described in more detail in the Supplementary Information which can be found online at http://www.informaworld.com/smpp/title∼content=t713693962.

Some systems of this bacterium have been implicated in virulence using infection models, including EF1704-1703 (HK-RR04), EF3290-3289 (HK-RR05) (also required for intrinsic -lactam resistance), EF1261-1260 (HK-RR06) and EF1051-1050 (HK-RR10) [14], [15]. The systems that have been functionally characterized are: EF1820-1822 (HK-RR15), which corresponds to the Fsr quorum-sensing system that responds to the autoinduced peptide GBAP [16–19], and has roles in both virulence [20–22] and biofilm formation [12], [23], [24]; EF2298-2299 (HK-RR11), VanS_BR_B, that regulates VanB-type vancomycin resistance [10], [25], though in this case the precise molecular signal is unknown; and EF1194-1193 (HK-RR07) VicKR. Similarity and knock-out studies of the response regulator component of this latter system revealed that it is likely to be a member of the YycFG/Vic subfamily of systems that is essential for cell viability [10], [12–14], [26], [27]. The precise signal to which the VicK sensor component responds has not yet been identified, though it may be a redox signal generated within electron transport [27], and the protein to known to be modulated by other membrane proteins in the cell with which it interacts [28–30]. Likewise, the target genes for the VicKR system in E. faecalis remain to be elucidated. Recently we described the crystal structure of the DNA-binding domain of the response regulator component VicR [31].

Amongst the remaining systems of E. faecalis is EF1863-1864 (HK-RR08) which is homologous to VncSR associated with vancomycin tolerance in Streptococcus pneumoniae [32] and is one of the six E. faecalis systems suggested to be involved in responses to oxidative and other cellular stresses [12], and EF3197-3196 (HK-RR02), which is highly homologous to LytSR of Staphylococcus aureus involved in cell wall integrity [33].

Little is known about the roles of the remaining systems, and even in the cases of many of the better-described systems above, the precise roles and in particular the molecular environmental signals sensed by the sensor kinase components and subsequent signal transduction mechanism across the membrane remain incompletely understood. One reason for this is the technical challenges associated with studies of the intact forms of these membrane sensory proteins, which are hydrophobic in nature. Therefore, previous studies of many kinases employed the cytoplasmic HK domains only, in soluble forms often lacking their sensing domains, including a full study of the genome complement of all the TCSs of E. coli [34]. We were the first to demonstrate the successful heterologous overexpression in E. coli and purification of any intact membrane sensor kinase [35]; the RegB (PrrB) redox-responsive sensor kinase of photosynthetic Rhodobacter sphaeroides was active in membranes and when purified retained autophosphorylation, phosphotransfer and phosphatase activities [35]. We demonstrated that intact purified RegB responded to the reducing agent dithiothreitol (DTT), increasing autophosphorylation, using an in vitro approach [36]. To determine how widely applicable such an approach might be, and to learn more about the roles of such kinases in processes such as virulence, antibiotic resistance and redox state, we evaluated the successes of expressing and purifying the full genome complement of membrane sensor kinases of E. faecalis.

Here we evaluate the systematic cloning, expression, purification and activity assays for each of the 16 membrane kinases. Using two of these systems, FsrC and VicK, we also report the successful identification or confirmation of molecular signals for each purified protein, indicating that this in vitro approach for signal identification is feasible and reliable in future systematic screening programs for the intact, heterologously-expressed membrane sensor kinases.

Materials and methods

Overexpression plasmid constructs for 16 HKs

Each HK gene was amplified using Vent polymerase or Pfu Turbo and E. faecalis strain V583 chromosomal DNA as template and the primer pairs listed in Table I. Amplified DNA was confirmed using 1% (w/v) agarose gels and the products were excised and purified before digestion with PstI and EcoRI or NdeI. Plasmid pTTQ18His was digested with the same enzymes. Digested genes and plasmid were visualized in agarose gels, followed by band excision and purification before subsequent ligation with T4 DNA ligase. Competent E. coli DH5α (Invitrogen), Top10 (Invitrogen) or XL10 Gold (Stratagene) cells were transformed and selection of plasmid-containing clones was done using Luria-Bertani (LB) medium [37] containing 50 µg/ml carbenicillin. The EF0373 (HK13) has an internal EcoRI site, and was cloned as a NdeI-, PstI-ended fragment. As the EF1863 (HK08) gene contains an internal EcoRI site and the EF0570 (HK12) gene contains an internal PstI site, a two step cloning process was employed. The correct orientation and sequence of all constructs was confirmed by DNA sequencing. All clones and strains used are described in Table II.

Table I.  Primers used in the study.

HK no.	Gene no.	Primer sequence 5'→3'
01	EF2219 Forward	CCGGAATTCGCATATGAAGCGAATCTTCAAAAAACTGT
	EF2219 Reverse	AAAACTGCAGCTGCTACAAACGAAAGACGAATTG
02	EF3197 Forward	CCGGAATTCGCATATGGTCGAATTATTTATTTTAATGATGGAA
	EF3197 Reverse	AAAACTGCAGCACGATTAATACGTGCATCTAATTCCTCCTGT
03	EF2912 Forward	CCGGAATTCGCATATGACCGATCGGATTTCAAGACGCAT
	EF2912 Reverse	AAAACTGCAGCTCCACTAACATTACTTTGATCACTTGC
04	EF1704 Forward	CCGGAATTCGCATATGAAAAAGAGACTGCGGATTGAATATTT
	EF1704 Reverse	AAAACTGCAGCGGCTTCTTTTTTTGTTAAAAGCAAAGTG
05	EF3290 Forward	CCGGAATTCGCATATGCTCGTTAAACCTAAAAAGATTGAAAA
	EF3290 Reverse	AAAACTGCAGCACTCTCTGATTTCTTGTTGGTACGC
06	EF1261 Forward	CCGGAATTCGCATATGAAGTATTTGTATCAACAATTA
	EF1261 Reverse	AAAACTGCAGCGTTTTCAGTAATTTCAACATCAGGAAAT
07	EF1194 Forward	CGATCATGAATTCCAAGAAAAAAGTTCACTTT
	EF1194 Reverse	CGATCCTGCAGCTTCCCACCAATCCTCTTCATA
08	EF1863 Forward	CGATCATGAATTCCAAAAAATTAAAGATATTT
	EF1863 Reverse	CATCCTGCAGCAGAAAATATTTTTATTTTAAAGAT
09	EF0927 Forward	CCGGAATTCGCATATGACCATTCTAAAGTATTTAAAAGAT
	EF0927 Reverse	AAAACTGCAGCTTTCACTTCATTATAATAACTGAGGGA
10	EF1051 Forward	CCGGAATTCGCATATGAAAAGAACGATTAAAAAAGAACTG
	EF1051 Reverse	AAAACTGCAGCTTGTACGTTGCTCTCTTCTTTTATTCCAGCGA
11	EF2298 Forward	CCGGAATTCGCATATGGAAAGAAAAGGGATTTTCATTAAG
	EF2298 Reverse	AAAACTGCAGCTAGTGTTGATGTGGGCGGTAAATC
12	EF0570 Forward	CCGGAATTCGCATATGCAATCTGGCGATGGACGCC
	EF0570 Reverse	AAAACTGCAGCCTTCAAATAAATTCGAACCAATGTTT
13	EF0373 Forward	CCGGAATTCGCATATGACCATTAAACGCCGCTTTTTTA
	EF0373 Reverse	AAAACTGCAGCTTTTTCATCCTCCAACAACGGCAGC
14	EF1209 Forward	CCGGAATTCGCATATGAAGCGTGGCGGGAAATTATGGTGTT
	EF1209 Reverse	AAAACTGCAGCATAATCAATAGGTTCATTTTCCTCTGCTC
15	EF1820 Forward	CCGGAATTCGCATATGATTTTGTCGTTATTAGCTACTAACGTTT
	EF1820 Reverse	AAAACTGCAGCTTCGTTAACAACTTTTTTACTGTTATGATT
16	EF1335 Forward	CCGGAATTCGCATATGCTTGATATAACATTACGAAGCCaps
	EF1335 Reverse	AAAACTGCAGCGTTATCACCGCATTTGTTTCCTTTGTA

Table II.  Bacterial strains and plasmids.

	Description	Reference/source
Bacterial strains
E. coli DH5α	F^− φ80lacZΔM15 Δ(lacZYA-argF) U169 deoR recA1 endA1 hsdR17 (rk^−, mk^+) phoA supE44 λ^− thi-1 gyrA96 relA1	Invitrogen
E. coli Top10	F^− mcrA ?(mrr-hsdRMS-mcrBC) Φ80lacZ?M15 ?lacX74 recA1 araD139 ?(ara-leu)7697 galU galK rpsL (Str^R) endA1 nupG	Invitrogen
E. coli XL10
	Tet^rΔ(mcrA)183 Δ(mcrCB-hsdSMR-mrr)173 endA1 supE44 thi-1 recA1 gyrA96 relA1 lac Hte [F′ proAB lacI^qZΔM15 Tn10 (Tet^r) Amy Cam^r]	Stratagene
E. coli BL21 (DE3)	F^− ompT lon hsdSB(rB^−mB^−) gal dcm (DE3)	Novagen, Merck
E. faecalis V583	Vancomycin-resistant clinical isolate	MS Gilmore, USA
Plasmids
pTTQ18His	Ap^r, tac-based overexpression system encoding a C-terminal hexahistidine tag	Ref [38]
pTTQHK01	Ap^r, pTTQ18His with the 1.7 kb PstI – EcoRI HK01 gene from strain V583.	This work
pTTQHK02	Ap^r, pTTQ18His with the 1.8 kb PstI – EcoRI HK02 gene.	This work
pTTQHK03	Ap^r, pTTQ18His with the 1.1 kb PstI – EcoRI HK03 gene.	This work
pTTQHK04	Ap^r, pTTQ18His with the 1.8 kb PstI – EcoRI HK04 gene.	This work
pTTQHK05	Ap^r, pTTQ18His with the 1.2 kb PstI – EcoRI HK05 gene.	This work
pTTQHK06	Ap^r, pTTQ18His with the 1.5 kb PstI – EcoRI HK06 gene.	This work
pTTQvicK	Ap^r, pTTQ18His with the 1.8 kb PstI – EcoRI HK07 gene.	This work
pTTQHK08	Ap^r, pTTQ18His with the 1.3 kb PstI – EcoRI HK08 gene. Possesses an internal EcoRI site	This work
pTTQHK09	Ap^r, pTTQ18His with the 1.0 kb PstI – EcoRI HK09 gene.	This work
pTTQHK10	Ap^r, pTTQ18His with the 1.5 kb PstI – EcoRI HK10 gene.	This work
pTTQHK11	Ap^r, pTTQ18His with the 1.3 kb PstI – EcoRI HK11 gene.	This work
pTTQHK12	Ap^r, pTTQ18His with the 2.6 kb PstI – EcoRI HK12 gene. Possesses an internal PstI site	This work
pTTQHK13	Ap^r, pTTQ18His with the 1.5 kb PstI – NdeI HK13 gene.	This work
pTTQHK14	Ap^r, pTTQ18His with the 1.6 kb PstI – EcoRI HK14 gene.	This work
pTTQHK15	Ap^r, pTTQ18His with the 1.3 kb PstI – EcoRI HK15 gene.	This work
pTTQHK16	Ap^r, pTTQ18His with the 1.3 kb PstI – EcoRI HK16 gene.	This work

HK gene names follow the nomenclature described previously [10].

The vicR gene was amplified by polymerase chain reaction using upstream primer VICR-F 5'-GATCACATATGAAGAAAATTTTAGTAGTTGA-3' and downstream primer VICR-R 5'-AGCTGGATCCTAAAATTAACAGATTGAAAAAAG-3' and genomic template as described above. The 0.75 kb product was digested with NdeI and BamHI (sites underlined in the primer sequences) and cloned directly into pET14b resulting in pETvicR. The correct sequence of the cloned gene was verified by DNA sequencing, and the expressed protein was verified by ESI-MS, N-terminal sequencing and Western blotting.

Expression trials

E. coli BL21 (DE3) cells were transformed with plasmids derived from single colony-purified clones, and cultured on LB agar plates containing 100 µg/ml carbenicillin overnight at 37°C. Single colonies were used to inoculate 5–10 ml cultures of LB broth, M9 minimal medium or 2TY medium [37] containing 100µg/ml carbenicillin, which were incubated aerobically at 37°C overnight. 1 ml volumes were used to inoculate 50 ml of the same media, again containing 100 µg/ml carbenicillin. These were cultured aerobically at 37°C until they reached an optical density of 0.4–0.6 measured at 600 nm. Expression was induced by the addition of varying amounts, 0–1 mM, of isopropyl -D-1-thiogalactopyranoside (IPTG). Optimal conditions were also investigated using different temperatures and periods of culturing post-induction. Cells were harvested at 4°C by centrifugation (5000 g for 20 min) and the pellets frozen at -80°C.

Mixed E. coli membranes were prepared by water lysis [38] and samples (30 µg) were separated by sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE) using 15% and 4% (w/v) acrylamide in the resolving and stacking gels respectively. Proteins from similar gels were also blotted onto Fluorotrans Transfer membranes (Pall Corp.) for detection of the hexa-Histidine tag of expressed proteins in Western blot analysis using the INDIA^TM HisProbe-HRP (Perbio Science UK Ltd.) and Supersignal West Pico Chemiluminescent substrate (Pierce) detection system.

Preparation of inner membrane vesicles

A total of 4–6 l cultures of E. coli cells harbouring each plasmid were grown aerobically in 500–750 ml batches in 2-l flasks and induced with IPTG. Cells were harvested by centrifugation and stored as described above. Cells were defrosted, resuspended in 20 mM Tris-HCl (pH 7.5) containing 0.5 mM EDTA, and lysed by explosive decompression using a cell disruptor to produce predominantly inside-out membrane vesicles that were purified by separation on sucrose gradients [38]. Freshly-prepared inner membrane vesicles in 20 mM Tris-HCl (pH 7.5), were used in activity assays, or stored at -80°C for subsequent purification of His-tagged protein. Total protein concentration in the inner membrane vesicles was determined by either Schaffner and Weissmann's assay [39], or by BCA assay (Pierce) performed in 96-well plate format using a Fluostar Optima plate reader at a wavelength of 550 nm.

Detergent solubilization trials

100 µg samples of inner membrane proteins were incubated with 100 µl solubilization buffer comprising 20 mM Tris HCl (pH 7.9), 10 mM imidazole, 20% (v/v) glycerol, 300 mM NaCl and 1% (w/v) detergent. The detergents trialed included: n-hexyl-β-D-maltoside, n-octyl-β-D-maltoside, n-nonyl-β-D-maltoside, n-decyl-β-D-maltoside, n-undecyl-β-D-maltoside, n-dodecyl-β-D-maltoside, n-hexyl-β-D-glucopyranoside, n-heptyl-β-D-glucopyranoside, n-octyl-β-D-glucopyranoside, n-nonyl-β-D-glucopyranoside, n-decyl-β-D-glucopyranoside, n-dodecyl-β-D-glucopyranoside, cyclohexyl-n-hexyl-β-D-maltoside, Triton X-100, ethyleneglycol monooctylether (C₈E₁), octaethyleneglycol monododecylether (C₁₂E₈), nonaethyleneglycol monododecylether (C₁₂E₉), N,N-dimethyl dodecylamine oxide (LDAO) and N,N-dimethyl decylamine oxide (DDAO), N,N-dimethyl octylamine oxide (ODAO), N-octyltetraoxyethylene, N-octyl pentaoxyethylene and methyl-6-O-(N-heptylcarbamoyl)-a-D-glucopyranoside (HECAMEG). Following incubation for 1 h at 4°C, with gentle rotation, 2×50 µl samples were tested for successful protein solubilization by retaining one sample on ice whilst the remaining 50 µl were centrifuged at 100,000 g for 30 min to pellet any protein which remained insoluble after treatment. In general, 15 µl of total and soluble fractions were analysed by SDS-PAGE using 12% gels, whilst in the case of lower-expressing proteins Western blotting was also used to visualize expressed proteins. Successful solubilization was judged by the presence of HK protein bands in the lanes containing centrifuged solubilized fractions.

Protein solubilization and purification

Membrane vesicle preparations were incubated for 1–2 h at 4°C in solubilization buffer (20 mM Tris HCl (pH 7.9), 20 mM imidazole (pH 7.9), 2 mM β-mercaptoethanol, 20% v/v glycerol, 250–300 mM NaCl and 1% detergent), to a final concentration of between 2.5–5 mg/ml total inner membrane protein. Protein solutions were then centrifuged at 100,000 g for 30 min to remove all remaining insoluble material, and the soluble proteins incubated between 0.5–15 h with 1 ml nickel-NTA resin (Qiagen) at 4°C with gentle rotation. The resin was centrifuged at 1,000 g for 1 min, resuspended in wash buffer (typically 10 mM Tris HCl (pH 7.9), 20 mM imidazole (pH 7.9), 2 mM β-mercaptoethanol, 10–20% v/v glycerol and 0–0.05% DDM, and in most cases 250 mM NaCl), and transferred to a 5 ml column. Increasing concentrations of imidazole, 0.1 M, 0.2 M, 0.3 M and 0.5 M in wash buffer were trialed to elute some proteins and samples were analysed for the presence of protein by SDS-PAGE. Elution fractions containing protein were concentrated using Vivaspin Centrifugal Concentrators (Sartorius) and exchanged into required buffers for further analyses.

All purified proteins were subjected routinely to verification as intact and undegraded using Western blotting (to verify the presence of the hexa-his tag engineered at the C-terminus of each protein) together with N-terminal sequencing, and/or using ESI-MS.

To obtain soluble recombinant VicR protein, plasmid pPETvicR was transformed into the expression host E. coli BL21 (DE3) and expression and purification carried out as described previously [35]. The protein accounted for 15% of total soluble proteins (data not shown). The integrity of purified VicR was verified by: (i) N-terminal sequence analysis of purified VicR, which gave the predicted sequence G-S-S-H-H-H; and (ii) electrospray mass spectrometry analysis, which identified a mass of 28989 Da that closely matches the predicted mass of 28990 Da.

Autophosphorylation assays for inner membrane vesicles

These were performed as described previously [35], [36]. Briefly, inner membrane proteins (200 µg) were incubated at room temperature (24°C) in assay buffer containing a final concentration of 50 mM Tris-HCl (pH 7.6), 10 mM MgCl₂, 50 mM KCl, 0.005% DDM and 1 mM dithiothreitol (DTT). Total reaction volumes of 200 µl were prepared and the reactions initiated by the addition of 50 µM ATP containing 50 µCi [γ − ³²P] or [γ − ³³P] ATP. Samples (20 µl) were removed at timed intervals between 0 and 60 min, and reactions stopped by the addition of 6 µl 4×SDS-PAGE sample loading buffer (60 mM Tris (pH 6.8), 10% (v/v) glycerol, 2% (w/v) SDS, 0.005% bromophenol blue and 3% β-mercaptoethanol). Samples were prepared for SDS-PAGE analysis by incubation for 30 min at 37°C to ensure full solubilization of proteins prior to loading gels containing 4% and 12% stacking and resolving gels respectively [37]; alternatively samples were stored at -80°C for subsequent analysis. Following gel electrophoresis as described previously [35], gels were briefly washed in deionized water and dried under vacuum at 70–80°C. Labelled protein bands were visualized by autoradiography using BioMax film (Kodak) exposed for 1–7 days. Labelled protein quantitation was undertaken using a Fuji BAS 1000 phospho-imaging analyzer system (Fujifilm Co., Japan) and AIDA 2D analytical software (Raytest).

Assays that included glutathione and accompanying controls employed an additional 500 mM Tris HCl pH 7.6 to maintain assay pH.

Autophosphorylation assays using purified proteins

Assays were carried out as described previously in buffers without added DDM [35]. Briefly, 800 pmoles purified protein (1.5–5 mg/ml stocks) in storage buffer (10 mM Tris-HCl (pH 7.9), 25% glycerol and 2mM β-mercaptoethanol) were incubated at room temperature (24°C) in reaction buffer containing 50 mM Tris-HCl (pH 7.6), 10 mM MgCl₂ and 50 mM KCl. 150 µl total reaction volumes were used for each protein and the reactions were initiated by the addition of 37.5 µCi [γ − ³³P] ATP at a final concentration of 50 µM. 15 µl samples were removed at intervals between 0.3 min and 1 h and terminated by the addition of 5 µl 4×SDS-PAGE sample loading buffer (stop buffer) as described above. For the VicK and FsrC proteins, the effects of individual candidate signal molecules were tested using 60–80 pmoles protein per 15 µl reaction mix. Signal and protein were incubated at 24°C for 20 min, prior to the addition of radiolabelled ATP to initiate the autophosphorylation reaction, which proceeded for 40 or 60 min before reactions were terminated with stop buffer. All samples were analysed by SDS-PAGE as described above for assays of membrane-bound proteins. In the case of VicK, purified protein was stored in buffer without added -mercaptoethanol. Assays using purified proteins with glutathione employed an additional 500 mM Tris HCl pH 7.6 to maintain assay pH. Stock solutions of synthetic GBAP, L-Gln—L-Asn—L-Ser—L-Pro—L-Asn—L-Ile—L-Phe—L-Gly—L-Gln—L-Trp—L-Met possessing a lactone ring formed by C-terminal L-Met and L-Ser and which is identical in structure to natural GBAP, were prepared in 20% (w/v) acetonitrile and stored at -20°C as described previously [40].

Western blotting

Proteins separated by SDS-polyacrylamide gel electrophoresis were transferred to Fluorotrans^TM membrane (Pall BioSupport UK) by semi-dry electroblotting for 2 h at 30 V. Membranes were washed once in TBSTT (20 mM Tris-HCl pH 7.5, containing 500 mM NaCl, 0.05% Tween 20 and 0.2% Triton X100) for 10 min, followed by overnight incubation in TBSTT containing 3% bovine serum albumin. Membranes were then washed twice in TBSTT for 10 min each and incubated for 1 h with 10 ml 1:5000-diluted INDIA^TM HisProbe-HRP (Perbio Science UK Ltd), a nickel-activated derivative of horse radish peroxidase for detection of histidine-tagged proteins. Following four washes with 10 ml TBSTT for 10 min each, membranes were incubated for 5 min in 10 ml Supersignal West Pico Chemiluminescent substrate and exposed to film for 5–60 sec.

Electrospray ionization mass spectrometry (ESI-MS)

50 µl purified HK protein samples (∼3 mg/ml) were prepared for electrospray ionization mass spectrometry by mixing with 150 µl of analytical grade methanol. 50 µl of chloroform were added and mixed to form a single phase, then 150 µl of water was added to initiate phase separation. Following centrifugation at 10,000 g for 2 min the aqueous phase was removed and 100 µl methanol was added. Following a second identical centrifugation step and removal of the solution the pellet of precipitated protein was dried under a flow of nitrogen. Size-exclusion chromatography using Sephadex^TM LH-20 (GE Healthcare) was used to complete the preparation of HK proteins as described previously [41].

N-terminal sequencing

Purified HK proteins were submitted for N-terminal sequencing on Fluorotrans^TM membrane (Pall BioSupport, UK) following electroblot transfer of proteins from SDS-polyacrylamide gels, and visualization using Coomassie Blue G-250 staining. Proteins were excised from the membrane for sequencing using Edman degradation.

Protein determination

Protein determinations were performed as already described above for inner membrane vesicles. Otherwise, spectrophotometry at a wavelength of 280 nm was undertaken and results compared to the absorbance for each HK predicted by the ProtParam tool on the ExPASy website (http://ca.expasy.org/tools/protparam.html).

Results

To clone each full-length HK gene, the polymerase chain reactions we carried out as described in Methods using the specific primer pairs shown in Table I with E. faecalis V583 chromosomal DNA as template, and fragments cloned into the membrane protein expression vector pTTQ18His. Descriptions of the resulting plasmids are shown in Table II. Each purified plasmid was transformed into E. coli BL21 (DE3) for expression, purification and activity trials as described below.

Fifteen out of 16 intact sensor kinases are expressed in E. coli membranes

In general, the expression of most membrane proteins is expected to be significantly lower than that of soluble proteins. Therefore the conditions reported previously [35], [38], [42] for successful culture growth and gene expression of other bacterial membrane proteins were tested initially, and optimized for some proteins, to determine if these were also suitable to obtain significant or detectable amounts of the intact membrane sensor kinases. In general, such conditions (use of LB culture medium, induction of expression using 0.5–1 mM IPTG, post-induction temperature of 30°C) proved successful for the overexpression of the majority of sensors. For proteins exhibiting low expression, optimization of culture and expression conditions were investigated in growth trials using different media (LB, M9 minimal, and 2TY medium), different concentrations of the IPTG inducer for expression of the inserted kinase gene in pTTQ18His, and different time periods and temperatures post-induction prior to cell harvesting. Expression of each sensor kinase was assessed as described in Methods using total E. coli membranes prepared by water lysis and analysed by SDS-PAGE and Western blot analysis to detect the engineered C-terminal hexa-histidine tag possessed by each recombinant protein. The results are summarized in Table III.

Table III.  Expression conditions, solubilization and activities of the 16 intact membrane sensor kinases of E. faecalis.

Kinase	Predicted number TMs^1	Expression medium	Induction temperature (°C)	[IPTG] (mM)	Harvest time post-induction (hours)	Predicted mass with His tag (kDa)^2	Observed mass (kDa) in mixed membranes^3	Proportion of mixed membrane proteins (%)^4	Solubilization detergent	Observed mass (kDa) of intact purified proteins^3	Ref.
EF2219 HK01	2	ND	ND	ND	ND	67.9	ND	ND	NP	NP
EF3197 HK02	6	LB	37	0.5	3	66.8	53	12	DDM	55	[33]
EF2912 HK03	2	LB	37	0.5	3	44.0	38	15	DDM	38	[12]
EF1704 HK04	2	LB	30	1.0	3	69.3	52	11	DDM	54	[12–15]
EF3290 HK05	2	LB	30	1.0	3	46.4	37	14	DDM	38	[12], [15]
EF1261 HK06	2	2TY	30	0.2	3	57.8	52	8	DDM	54	[15]
EF1194 HK07	2	LB	30	0.4	3	71.2	60	11	DDM	60	[12]
EF1863 HK08	2	LB	15	1.0	16	52.6	42	12	LDAO	45	[13], [32]
EF0927 HK09	2	LB	30	1.0	3	41.7	31	6.5	DDM	32	[12]
EF1051 HK10	2	LB	37	0.5	3	60.2	50	16	DDM	50	[13–15]
EF2298 HK11	2	LB,2TY	30	0.5, 1.0	3	52.0	42	10	DDM/UDM	48	[25]
EF0570 HK12	4	LB,2TY	30	1.0	3	98.7	38	10	ND	NP	[12], [14]
EF0373 HK13	2	LB	30	0.5	3	58.1	46	4.5	DDM	44	[13]
EF1209 HK14	2/3	LB	30	0.5	3	61.1	48, 33, 27	16.5	DDM	48	[12]
EF1820 HK15	6	LB	37	0.5	3	53.4	35	3.5	DDM	34	[16–24]
EF1335 HK16	7	LB	15	1.0	16	52.2	32	ND	DDM	NP	[12]

¹Predicted using TMHMM (http://www.cbs.dtu.dk/services/TMHMM) and/or TopPred (http://bioweb.pasteur.fr/seqanal/interfaces/toppred.html); ²Predicted using the ExPASy website (http://ca.expasy.org/tools/protparam.html, including tag possessing hexa-histidine motif; ³Western blotting analysis with INDIA^TM His probe; ⁴Determined by scanning densitometry of SDS-polyacrylamide gel shown in Figure 1. DDM, n-dodecyl-β-D-maltoside; UDM, n-undecyl-β-D-maltoside; LDAO, N,N-dimethyl dodecylamine oxide; ND, included in the trials but none detected or identifiable; NP, not possible to undertake.

Overall, most sensors were optimally expressed in LB media, with only one protein (EF1261 [HK06]) expressed at higher levels in the richer 2TY medium and none of the low-expressing proteins exhibiting better expression in minimal medium compared with the other media (Table III). Suitable inducer concentrations and post-induction temperatures amongst the proteins ranged from 0.2–1 mM IPTG (the highest used in this study) and 15, 30 or 37°C post-induction temperatures (Table III). In general the expression time employed post-induction was 3 h, with only HK08 and HK16 benefiting from overnight incubation at the lower temperature of 15°C.

The observed molecular masses of the majority of recombinant proteins in SDS-PAGE and Western blot analysis appeared less than those predicted (Table III and Figure 1). We therefore verified that the expressed proteins represented intact, undegraded proteins. Using the 13 proteins that could be purified (see below), ESI-MS and/or N-terminal sequencing, together with the positive signals observed in Western analysis that detects an intact C-terminus of the protein through the hexa-histidine tag motif, was undertaken. In all but three cases (EF3197 (HK02) for which no N-terminal sequence or MS data could be obtained, EF0570 (HK12), which could not be solubilized from membranes, and EF1335 (HK16), which was not eluted from Ni^2 +-NTA columns), the presence of intact proteins was successfully confirmed. Therefore the observed mass differences of these proteins are attributable to their anomalous behaviour in SDS-PAGE analysis, as is typically observed for hydrophobic membrane proteins [38]. It is noteworthy that, in general, the greatest mass differences were observed for proteins that possess higher numbers of predicted transmembrane regions (TMs), such as EF1820 (HK15) and EF1335 (HK16) (Table III). These proteins are predicted to possess six and seven putative TMs, respectively, and their apparent masses in SDS-PAGE analysis of total membrane preparations were indeed significantly smaller than expected (35 and 32 kDa compared with 53 and 52 kDa, respectively). Likewise, the apparent mass of EF3197 (HK02) (also six predicted TMs) was 53 kDa compared with a predicted mass of 67 kDa (Table III).

Figure 1.  Expression of the 16 intact membrane sensor kinases of E. faecalis in E. coli BL21 (DE3). Cells harbouring pTTQ18His containing the recombinant gene for each of the intact membrane kinases were cultured as described in Methods, and mixed membrane proteins prepared by water lysis. Samples (15 µg) were separated and analysed by: (A) SDS-PAGE and visualized using Coomassie blue staining (upper panel); and (B) and (C), Western blotting to detect the engineered hexa-histidine tag at the C-terminus of each protein. Panel C shows a longer exposure required for detection of the successful expression of EF2298 (HK11). M, molecular mass standards. U, mixed membranes of an uninduced culture of E. coli BL21 (DE3)/pTTQHK01.This Figure is reproduced in colour in Molecular Membrane Biology online.

Figure 1 shows that of the 16 intact membrane sensor kinases cloned in this study, 15 were successfully overexpressed. Only EF2219 (HK01) was not detectable in E. coli BL21 (DE3) membranes under any of the test conditions. Partial degradation was observed for EF1209 (HK14), and there was no improvement upon addition of protease inhibitors into the water-lysis preparation procedures (Figure 1).

Eleven out of 15 proteins retain autophosphorylation activity in E. coli membranes in the absence of signal

Inner membrane vesicles expressing each sensor kinase were prepared as described in Methods, and 200 µg of inner membrane protein were used in autophosphorylation assays with radiolabelled ATP. At intervals following ATP addition, 20 µg samples were removed to stop buffer, and the level of phosphorylated protein at each time point evaluated as described in Methods. Figure 2 shows the results for the 11 kinases that produced detectable autophosphorylation activities in these assays. EF0927 (HK09), EF2298 (HK11), EF0570 (HK12) and EF1335 (HK16) showed no detectable activity on at least three attempts, suggesting either inactivity or absence of an activating signal. The membrane-localized kinases exhibit a range of different kinetics in these assays (Figure 2B). Nine of the proteins (EF3197 (HK02), EF1704 (HK04), EF1261 (HK06), EF1194 (HK07), EF1863 (HK08), EF1051 (HK10), EF0373 (HK13), EF1209 (HK14) and EF1820 (HK15) attained maximum autophosphorylation within the initial 20 minutes, and for EF3197 (HK02), EF1704 (HK04), EF1194 (HK07), EF1863 (HK08), EF1051 (HK10), EF0373 (HK13) and EF1209 (HK14), this was followed by rapid dephosphorylation (Figure 2B); EF1051 (HK10), EF0373 (HK13) and EF1209 (HK14) all autophosphorylate particularly rapidly, within the first minute or so following assay initiation.

Figure 2.  Autophosphorylation activities of intact sensor kinases in inner membrane vesicles. Inner membrane proteins (200 µg) were incubated at 24°C as described in Methods, and reactions initiated by the addition of 50 µM ATP containing 50 Ci [γ − ³²P] or [γ − ³³P] ATP. Samples (20 µl) were removed at the time points indicated, reactions terminated by the addition of 6 µl 4×SDS-PAGE sample loading buffer, and proteins separated by SDS-PAGE. Following electrophoresis, gels were washed, dried, and exposed to an image plate for quantitative analysis using a Fuji BAS1000 imaging analyzer system (Fujifilm Co., Japan). (A) Image and (B) quantitative representation of the data for each radiolabelled phosphorylated protein during autophosphorylation. The identities of the phosphorylating proteins (indicated by the arrows) were verified by comparisons with Western data. Graphs show the pixel counts relative to the maximum value attained during the assays. The numbers above each panel indicate the time sampled (minutes) after reaction initiation.

EF3290 (HK05) phosphorylated maximally after approximately 30 min, followed by slow dephosphorylation, still retaining approximately 60% of its maximum phosphorylation level at the end of the 60-min assay period. Uniquely, levels of phosphorylated EF2912 (HK03) increased steadily throughout the assay time and did not reach a maximum within 60 min.

Fourteen out of 15 intact kinases are solubilized from E. coli membranes

DDM has been widely used previously for solubilizing bacterial membrane proteins [38], [42], and was also successful here for solubilizing almost all the proteins in this study, except for EF1863 (HK08) and EF0570 (HK12); LDAO was successfully used for solubilization of EF1863 (HK08), whilst none of the detergents tested for this protein were successful for the solubilization of EF0570 (HK12) from E. coli inner membranes (Table III).

Thirteen of the 15 expressed proteins are successfully purified

Following the solubilization trials described above, the appropriate detergents summarized in Table III were used at 1.0 or 1.2% final concentration to solubilise each intact kinase from E. coli membranes as described in Methods. Purification of the hexa-histidine tagged proteins was performed by nickel affinity purification as described in Methods; only EF1335 (HK16) could not be eluted successfully from the nickel affinity column.

The results of one-step purification are shown in Figure 3. The purities of EF1704 (HK04), EF1261 (HK06) and EF1051 (HK10) were highest reaching 87, 78 and 90%, respectively, as judged by densitometry measurements. The remaining proteins were not as pure, ranging from 19% for EF0373 (HK13) to 74% for EF0927 (HK09) (Figure 3). As described above, all purified proteins were confirmed by Western blotting using the His₆ probe to verify that these were the expected recombinant Histidine-tagged proteins. Their observed masses were in agreement with those obtained in mixed membranes (Figures 1 and 3; Table III).

Figure 3.  One-step purification of the expressed intact membrane sensor kinases of E. faecalis. Following solubilization, proteins were purified by nickel affinity as described in Methods. Samples (10 µg) were analysed by: (A) separation on SDS-polyacrylamide gels and staining with Coomassie Blue, and% purity determined by densitometry; and (B) Western blot analysis using an INDIA^TM His probe. M, molecular mass markers. This Figure is reproduced in colour in Molecular Membrane Biology online.

There was evidence of possible degradation amongst some of the proteins, since the Western blot revealed the presence of minor proteins of lower mass in preparations of EF3197 (HK02), EF2912 (HK03), EF3290 (HK05), EF1863 (HK08), EF1209 (HK14) and EF1820 (HK15) FsrC, and to a lesser extent for EF1051 (HK10) and EF2298 (HK11). The one-step purification of EF1194 (HK07) VicK also resulted in protein of high purity, at 84%. However, with minor changes to the purification protocol such as reduced time for solubilization using higher detergent concentrations of 1.2%, the protein was readily purified to > 95% purity and was therefore used at this purity for subsequent signal identification (see below).

Twelve out of 13 purified proteins retain autophosphorylation activity in the absence of signal

Figure 4 summarizes the autophosphorylation activities obtained for 10 of the purified intact kinases that were active following successful expression and purification from E. coli inner membranes. Two further proteins (EF2298 [HK11] and EF1209 [HK14]) were also active as purified proteins, though only weakly and are therefore not shown. Of the 11 intact kinases that were active when membrane-bound, all but one (EF1863 [HK08]) remained active following purification. In the case of EF1863 (HK08), the loss of detectable activity may be attributable to the presence of 0.1% LDAO in the assays, which was required to prevent aggregation of this protein. We have observed that detergent can be inhibitory for the activities of some purified kinases, and therefore assays are routinely undertaken in the absence of added detergent post-solubilization if possible.

Figure 4.  Autophosphorylation activities of purified intact sensor kinases. Purified proteins (600–1000 pmoles) were added to reaction buffer, and autophosphorylation initiated using radiolabelled ATP in a total reaction volume of 150 µl, as described in Methods. Sample volumes (15 µl) were removed at the time points indicated. (A) Image and (B) quantitative representation of the data for each radiolabelled phosphorylated protein during autophosphorylation. Phosphorylating proteins (shown by the arrows) were verified by mass determinations and comparisons with Western data. Graphs show the pixel counts relative to the maximum value attained during the assays.

EF0927 (HK09), which previously produced no detectable autophosphorylation when assayed in membranes, produced detectable levels of autophosphorylation activities when purified protein was used (Figure 4). In the case of EF2298 (HK11) VanS_B, in which activity levels were low, there was no detectable increase upon addition into the assays of a 10-fold molar excess of vancomycin, a candidate signal for this kinase (data not shown). We therefore conclude that vancomycin is not an activating signal for this kinase under these conditions.

Comparisons of autophosphorylation kinetics exhibited by each kinase throughout the sixty minute assay when membrane-bound compared with purified proteins, revealed that only EF2912 (HK03) and EF1261 (HK06) possessed similar kinetics using both purified and membrane-bound versions. In the case of most other kinases, specifically EF3197 (HK02), EF1704 (HK04), EF3290 (HK05), EF1194 (HK07), EF1051 (HK10), EF0373 (HK13) and EF1820 (HK15), autophosphorylation activities of purified forms reached maximum levels later in the assays, with little subsequent dephosphorylation observed in the time period compared to that observed for the proteins when membrane-bound (Figures 2 and 4). Although the different concentrations present will presumably affect kinetics, these data may also suggest that autophosphorylation rates are faster when the proteins are bound into a membrane environment or have not undergone the extraction and purification procedures. Alternatively, it is also possible that the membrane environment or even some interfering membrane-bound E. coli phosphatases present in the vesicles assayed, may dephosphorylate or promote the intrinsic phosphatase activities that are typically associated with many sensor proteins.

EF1820 (HK15) FsrC and EF1194 (HK07) VicK are the kinase components of two of the three systems with established roles in this bacterium (the third being EF2298 (HK11) which exhibits low activity as described above) [12], [16–25]. Therefore these two intact active kinases (hence forward referred to as FsrC and VicK, respectively) were selected for further studies to determine if signal identification and signal confirmation may be achieved using heterologously-expressed intact membrane sensor kinases.

Signal identification: The peptide pheromone GBAP stimulates autophosphorylation of purified intact FsrC

FsrCA (EF1820-EF1822, HK-RR15) is a quorum-sensing TCS that positively regulates the expression of infection-related extracellular proteases, gelatinase and a serine protease [16–19] and regulates biofilm formation in E. faecalis [23], [24]. An activating signal for the sensor is the peptide gelatinase biosynthesis-activating pheromone, GBAP [16]. GBAP has been synthesized chemically in an active form, as demonstrated by its ability to activate gelatinase biosynthesis in E. faecalis cells [40]. To demonstrate that synthetic GBAP induces activation through direct interactions with the FsrC sensor kinase, and indeed to demonstrate that intact FsrC is able to respond to its expected GBAP signal in vitro, purified FsrC was pre-incubated with different concentrations of synthetic GBAP for 20 min prior to initiation of autophosphorylation assays (Materials and methods). GBAP stimulated FsrC autophosphorylation by approximately 10-fold (Figure 5A) in response to increasing concentrations of GBAP up to approximately 10 µM, a two-fold molar excess of GBAP over FsrC. Higher GBAP concentrations up to 85.3 µM elicited little further increase in levels of phosphorylated FsrC (Figure 5A).

Figure 5.  Effect of the GBAP peptide pheromone on autophosphorylation activity of FsrC and other purified sensor kinases. Reactions contained 80 pmoles purified protein preincubated in reaction buffer for 20 min in the presence or absence of synthetic GBAP, prior to the initiation of autophosphorylation reactions in a total reaction volume of 15 µl. After 60 min, samples were removed to 5 µl stop buffer. The quantity of ³²P associated with each protein was determined by SDS-PAGE followed by autoradiography and phosphoimager analysis. All reactions (including controls) contained a final concentration of 3.4% acetonitrile (used to dissolve GBAP). (A) FsrC-P levels in response to increasing GBAP. Means of duplicate data are presented (except for 0.7, 21.3 and 170.7 µM GBAP datapoints, for which single measurements were made), and averaged datapoints varied by±0.07–5.43%. (B) Comparison of responses to 10.7 µM GBAP by six purified sensors. Single data measurements are shown for all proteins except FsrC (EF1820) which shows the mean and standard deviation derived from triplicate data. Autoradiographs of individual data are also shown.

To determine if GBAP exhibits specificity for the FsrC protein or whether it stimulates the activities of other intact E. faecalis sensor kinases, we also tested the effects of GBAP on VicK and five other purified intact kinases which have no known or predicted roles in quorum sensing. Using 10.7 µM GBAP, there was very little change in the activities of any of the other proteins (Figure 5B), confirming that GBAP exerts a strong, significant effect specifically on FsrC. These data therefore provide the first evidence for direct and specific interactions between GBAP and FsrC, and indeed for any peptide pheromone with its cognate quorum sensor kinase.

Signal identification: Reductants dithiothreitol (DTT) and glutathione significantly increase autophosphorylation of VicK

Ng & Winkler [27] reported a link between the class I YycFG/VicKR TCSs and the capacity for electron transport, and suggested that YycG may directly (or indirectly through YycH [28]) sense the redox state of electron transport. To identify definitively the signals for intact VicK (a Class I member), we firstly investigated whether redox potential acts as a direct modulator of intact VicK activity using DTT as chemical reductant as well as reduced glutathione as a physiologically-relevant reducing agent. DTT was used as in vitro reducing agent since it has been used extensively in previous studies of redox-sensing proteins (e.g., [35], [36], [43]). Using both purified and membrane-bound VicK proteins in which the predicted intracellular PAS, ATP-binding and kinase domains are all exposed in the ATP-requiring assays, levels of VicK-P increased significantly in response to increasing concentrations of DTT (up to 50 mM) in the assays (Figure 6A, 6B), suggesting that VicK responds to redox signals in vitro and may therefore be a redox sensor in vivo. Low but detectable levels of VicK phosphorylation occurred under air-oxidizing conditions in the absence of DTT, but purified VicK-P levels were increased by 432% and 347% in response to 5 and 20 mM DTT respectively, whilst VicK-P levels in membranes increased by 170% and 296% with 5 and 25 mM DTT (Figure 6A, 6B). Hence, the purified version of the intact protein was more responsive to the membrane-permeant signalling agent than the membrane-bound form, possibly indicating either an increase in signal sensitivity upon protein purification, or a barrier role played by the membrane in DTT access to the protein. It is also possible that it is the extracellular sensing domain that senses DTT and that this difference reflects an incomplete equilibration of DTT across the membrane and into the interior of the vesicles.

Figure 6.  Effects of dithiothreitol (DTT) and reduced (GSH) and oxidized (GSSG) glutathione on autophosphorylation activity of intact VicK. Each reaction contained either membrane-bound VicK in E. coli membranes (20 µg) (A), (C) & (E), or 60–80 pmoles purified VicK (B) and (D), which were incubated in reaction buffer for 20 min in the presence or absence of DTT (A) and (B), or 5 and 25 mM concentrations of GSH or GSSG (C), (D) & (E), prior to the initiation of autophosphorylation reactions in a reaction volume of 15 µl. Reactions were terminated with stop buffer after 2 min (membrane-bound VicK [A] and [C]), or 40 min (purified protein [B] and [D]) when VicK was autophosphorylating, or after 40 min for membrane-bound protein, when VicK is dephosphorylating (E). The quantity of ³³P associated with VicK was then determined by SDS-PAGE followed by autoradiography and phosphoimager analysis. Reactions were performed in triplicate. Autoradiographs show one individual blot; numbers beneath are the phosphorimager data presented as a mean of the triplicate data.

To determine whether the effects of redox potential towards VicK autophosphorylation could function in vivo, DTT was replaced with the alternative and more physiologically-relevant reductant, glutathione (GSH). Whether using membrane-bound VicK in vesicles or as purified VicK, the presence of reduced GSH at the physiologically-relevant concentration of 5 mM, resulted in modest but significant increases in VicK-P levels in the autophosphorylation assays (Figure 6C, 6D). The largest effect was observed with purified protein, with a 60% increase in VicK-P levels using 5 mM GSH compared with the 37% increase obtained using membrane-bound protein, suggesting that the protein may be slightly less responsive to GSH when membrane-bound. At the higher concentration of 25 mM, the effects of GSH were less pronounced (Figure 6C, 6D). Unexpectedly, oxidized glutathione (GSSG) also exerted significant increases in purified VicK autophosphorylation rates, and these were either similar to, or slightly greater than those obtained using GSH. This suggests that glutathione itself, rather than its reduced/oxidized state, exerts a positive effect on VicK activity. Indeed, the effects of both GSH and GSSG on membrane-bound VicK were significantly more pronounced when the assays were terminated after 40 min (instead of 2 min). At this point, VicK is significantly dephosphorylated (Figure 2), contrasting with the autophosphorylating protein assayed in Figure 6A–D. Under such conditions, the physiologically-relevant concentration of 5 mM GSH elicited a 360% increase in VicK-P, whilst GSSG produced a 226% increase (Figure 6E), again confirming an effect of glutathione per se.

The effects of 10 other agents that could potentially serve as signals to modulate activity of VicK in vivo were also tested, particularly as VicK possesses two putative PAS domains that could be involved in sensing more than one signal. These included respiratory quinones [43], [44] and reduced and oxidized versions of menaquinone (vitamin K) (in 50-fold molar excess) (included because E. faecalis possesses dimethylmenaquinones [45–47]), 10 mM NAD⁺ or NADH, 10 mM sodium phosphate as a source of a 1875-fold molar excess of phosphate ions [48], 10 mM ammonium chloride as a source of ammonium ions [49] and 10 mM lactate, acetate, fumarate or pyruvate [50], [51]. In all cases these agents brought about no significant change in VicK-P levels in autophosphorylation assays (data not shown).

Sensor responses exert an effect on downstream signal transduction in vitro: Elevated VicK-P levels resulting from DTT result in higher levels of VicR-P during phosphotransfer

When VicK was permitted to autophosphorylate for 15 min in the presence of 0.5 mM or 100 mM DTT and VicR was then added at a final concentration of 1.5:1 VicK:VicR, in both cases phosphotransfer resulted in a steady loss of VicK-P and increasing levels of VicR-P throughout the 2 h of the assay (Figure 7). The presence of higher DTT resulted in higher levels of VicR-P levels throughout the phosphotransfer assays, thereby demonstrating that the increased levels of phosphorylated VicK induced by DTT has a concomitant impact on downstream levels of VicR-P. Figure 7 also shows that the kinetics of phosphotransfer or maintenance of the VicR-P signalling state in the presence of low DTT differ with those observed previously for PrrB; as described above, levels of VicR-P increase steadily with time during phosphotransfer (Figure 7), unlike PrrA-P levels which actually decrease with time [35].

Figure 7.  Effect of redox potential on phosphotransfer from VicK to VicR in vitro using DTT as reducing agent. 420 pmoles VicK were permitted to autophosphorylate for 15 min in the presence of either 0.5 mM or 100 mM DTT in 130 µl reaction volumes. 13 µl samples were removed to stop buffer prior to the addition of 252 pmoles VicR to the remaining 378 pmoles VicK (1.5:1 VicK:VicR). Samples (50 pmoles) were removed to stop buffer at regular intervals up to 2 h. Autoradiograph and graphical representation of quantitative data obtained using (A) 0.5 mM DTT, and (B) 100 mM DTT.

Discussion

To understand how changes in a myriad of environmental conditions are perceived by bacterial cells and how these signals are networked and coordinated into overall cellular responses requires a fundamental knowledge of all the sensory proteins present in individual cells. In the main, these sensory proteins are histidine protein kinases (or sensor kinases) and most bacteria possess many different HKs to match the varied environmental cues encountered. Laboratory production of these membrane sensory proteins has previously been one of the bottlenecks in determining their structures and mechanisms. Here we have evaluated the heterologous production and activities of the 16 intact membrane sensor kinases of E. faecalis. Following the systematic cloning of the genes for each protein into the membrane protein expression vector pTTQ18His, we demonstrated that of the 16 proteins, all but one (EF2219 [HK01]) were successfully expressed in E. coli membranes (Figure 1). The success of pTTQ18 as an expression plasmid for membrane proteins has been well documented previously [38], [42] and in a recent comparison of eight vectors for expression of 37 membrane protein targets, pTTQ18 showed the greatest number of targets expressed [52].

The levels of E. faecalis HK expression varied, depending on the protein, from 3.5–16.5% of mixed membrane proteins, significantly higher than the estimated 100 or so copies per cell of each sensory kinases in vivo, and therefore to be regarded as an amplified level of expression. In this study the choice of host strain, expression medium, induction temperature, concentration of inducer and post-induction time were based on our previous experiences with expression of intact RegB [35], other bacterial membrane proteins [42], and on limited trials to optimise conditions for this set of sensor kinases. In general, LB broth at induction temperatures of 30 or 37°C with 0.5 or 1 mM IPTG and 3 h post-induction time were successful for the overexpression of the majority of E. faecalis sensors (Table III). High success rates were also obtained for the proportion of intact sensor kinases that could be subsequently solubilized from E. coli membranes and purified by nickel affinity purification. Fourteen out of 15 proteins were successfully solubilized and 13 purified.

Producing active proteins is crucial for in vitro studies of signal identification and interactions. Unlike truncated soluble versions, some full-length sensor kinases that possess intact signal sensing domains might be expected to phosphorylate only in the presence of their environmental signal. For other intact sensor kinases the converse may hold true, and the presence of the signal may be required to repress the autophosphorylation activities of the kinase or enhance any phosphatase activity of the sensor towards its cognate response regulator, as is the case for RegB (PrrB) [53]. Indeed, it is also possible that the activities of some kinases such as VicK that belong to essential two-component systems, are never switched off entirely in vivo, and rather they are up- or down-regulated from a constitutive basal level, so activities are present even in the absence of activating signals.

In the majority of cases, the nature of the signal(s) is unknown. Therefore, in this study we determined the autophosphorylation activities displayed by the membrane-bound and purified sensor kinases in the absence of known signals, as a starting point for assessing activities of these overexpressed intact proteins. Eleven of the 15 membrane-bound proteins and 12 of the 13 purified proteins were active in autophosphorylation assays. Remarkably, 13 of the 15 expressed proteins were active in E. coli membranes and/or as purified proteins. Their activities may be attributable to absence in these assays of any repressive ‘signals’ required to regulate these sensors in vivo, as discussed above. However, this is unlikely to hold true for all these sensors, since some are predicted/known to respond to signals that activate kinase activity, rather than repress it, as in the case of FsrC, which senses the GBAP pheromone (Figures 2 and 4) [16], [19], [40]. Indeed, in our present studies with FsrC and VicK, we demonstrated that levels of autophosphorylation were significantly increased upon addition into activity assays of activating signals (Figures 5 and 6). These findings not only confirm direct interactions between each sensor and its signal in this in vitro approach, but also demonstrate the potential of such an approach in future studies of signal identification using other sensor kinases.

In the case of FsrC, the strong stimulatory effect on autophosphorylation activity produced by a synthetic version of the peptide inducer GBAP was also a specific effect, since five other kinases included in the assays exhibited little or no response to the pheromone (Figure 5). Signal specificity is assumed to be a fundamental feature of two-component systems. Interestingly, the Fsr system, unlike its Agr counterparts in Staphylococcus aureus, is particularly highly conserved amongst E. faecalis strains [18].

Of the 12 agents tested as potential signals for intact VicK, the findings that only increasing concentrations of DTT and of physiological concentrations of glutathione significantly increase activities of the intact kinase (Figure 6), provide the first direct in vitro signal identification for this intact kinase. The responses of VicK to the reducing agent DTT were similar to those observed using other redox-responsive proteins including PrrB [35], [36] and OxyR [54], and therefore our data indicate that VicK may also sense redox potential. It is possible that the chelation properties of DTT might also exert an effect on VicK activity, though the only predicted role for metals is in the Mg^2 +dependent ATPase functions which would be expected to reduce VicK activity, not increase it as observed here. Finally, we also demonstrated that the increased levels of VicK-P caused by DTT resulted in concomitant increases in phosphotransfer and VicR-P levels (Figure 7), demonstrating that these in vitro assays can also be used to examine signal effects on subsequent signal transduction through TCS pathways.

Reduced glutathione (GSH) was included in VicK activity assays as an alternative reducing agent. The results with GSH supported the DTT data; the presence of the physiologically-relevant concentration of 5 mM GSH also significantly increased VicK autophosphorylation activity, confirming a possible role for redox potential in modulating VicK, a protein that possesses two putative PAS domains (Figure 6). Unusually for many Gram-positive bacteria, E. faecalis possesses significant levels of glutathione [55], suggesting that redox control of VicK might be exerted at least in part by GSH itself in vivo. However, the findings here (Figure 6) that oxidized glutathione (GSSG), in addition to GSH, also significantly increased VicK∼P levels suggest a different or additional role for glutathione itself in VicK modulation. Glutathione is linked with protection against oxidative and other stresses in the cell in vivo [56], [57]. Glutathione may therefore play a protective role towards this essential kinase, possibly under stress conditions. The protective role for glutathione in bacteria has been attributed, at least in part, to the direct glutathionylation of proteins by GSSG [57–60]. Since glutathione is located intracellularly in vivo, the region of VicK involved in glutathione-sensing must presumably be located in the cytoplasmic portion. This is supported by the membrane vesicle data in Figure 6, in which the inside-out orientation supported increased VicK activity in response to membrane-impermeant glutathione, though we cannot exclude the possibility that in the case of the reduced form, GSH may enter the interior of these inside-out vesicles via the E. coli transporter CydDC, giving access to the putative extracellular sensing domain of the protein [61].

In this study, all the intact kinases of virulence-associated systems (FsrCA, EF1704-1703 [HK-RR04] [PhoPR], EF3290-3289 [HK-RR05] [CroSR], EF1261-1260 [HK-RR06] and EF1051-1050 [HK-RR10] [EtaSR]) were successfully overexpressed and active in membranes or as purified proteins. The putative VncS homologue (EF1863 [HK08]) was also successfully expressed and active, and EF2298 (HK11) VanS_B and EF1209 (HK14) were also observed to exhibit low autophosphorylation as purified proteins. The nature of the signals that these kinases sense can therefore be fully investigated in future in vitro experiments using the full-length versions of these membrane proteins. For example, EF2298-2299 (HK-RR11) VanS_BR_B and EF1863-1864 (HK-RR08) may sense changes in cell wall intermediates and components; successful expression of their intact sensory components here paves the way for future studies of mechanisms of vancomycin resistance and tolerance in this bacterium, and expands on previous studies with soluble versions of the kinase (e.g., [62]).

Given the successful demonstration of GBAP as a direct and specific activating signal for FsrC, and the first successful identification of modulators of VicK autophosphorylation activities with subsequent downstream effects on levels of phospho-signalling to its partner response regulator VicR, further work is now underway using both membrane-bound and purified proteins to identify the signals for each of the kinases of E. faecalis to determine their full roles in virulence and antibiotic resistance in this pathogenic bacterium.

Supplementary information

In the completed genome sequence of E. faecalis strain V583, 17 two-component system (TCS) pairs have previously been identified, along with one orphan RR [1]. Using the histidine kinase (HK) classification scheme described previously [2], Hancock and Perego sub-grouped the 17 HK proteins, identifying two HKs as belonging to the group I sub-family, one group II member, 10 HKs belonging to group IIIa (which appears to be the most abundant class of kinases in the Gram-positive pathogens whose genomes have currently been sequenced), one in group IV and three that do not belong to any of the previously described classes of HKs [1]. All but one of the HKs were predicted to be membrane localized [1].

Similarity searches, comparisons with known TCSs in other bacteria [1], systematic inactivation of the response regulator (RR) components [65–68], and investigations using mouse peritonitis and peritonitis macrophage models [3], [6] revealed important insights into the potential functions of many systems, some of which are associated with virulence or responses to antibacterial agents and other stresses. However, in most cases the precise roles and in particular the molecular environmental signals to which these systems respond remain incompletely understood. The following is a summary of the main points of what is currently known or suggested about each system.

EF2219-2218 (HK-RR01) is homologous to a B. subtilis system located next to putative sugar and galactan utilisation genes, and whilst its role remains unknown, inactivation of the response regulator component RR01 was shown to result in increased sensitivity to cell wall active agent bacitracin [5]. The TCS EF3197-3196 (HK-RR02) is highly homologous to LytSR of Staphylococcus aureus; the HK proteins possess 52% identity [1], [7], suggesting a possible role for the enterococcal system in cell wall integrity mediated through effects on the intrinsic murein hydrolase activity of the cells. However, Hancock and Perego found no change in autolysis rates in an RR02 mutant, and therefore such a role remains unconfirmed, though alterations in uptake of antibiotics such as erythromycin were observed in the mutant, suggesting some involvement in cell wall-related permeability functions [5]. Mutations in the RR component of the TCS EF2912-2911 (HK-RR03) resemble those of EF2218 (RR01), exhibiting increased sensitivity to bacitracin; this TCS most closely resembles B. subtilis YcqE and Staphylococcus aureus VraS [5], but again, the precise role remains unknown. The TCS EF1704-1703 (HK-RR04) is homologous to B. subtilis PhoPR [1], [3], [8]; disruption of the err04 gene of this system resulted in increased heat sensitivity in strain JH2-2 (a phenotype also shared by two other RR mutants, err08 and err18) [4] and a growth defect in strain V583 [5]. This system was recently shown to be induced in phosphate deprivation conditions, and to be required for full survival in macrophages hence confirming a role in virulence [6]. The EF3290-3289 (HK-RR05) system, CroSR, is required for intrinsic β-lactam resistance [5], [9]; a croR mutation caused significant growth defects and cell morphology alterations in strain JH2-2 [4], and is implicated in virulence since the mutation also impairs survival in macrophages [6]. EF1261-1260 (HK-RR06) is also linked with virulence, since mutations in the RR component RR06 result in decreased survival in macrophages and susceptibility to oxidative stress [6]; this may be linked with membrane functions and architecture, since a RR06 mutation was also found to be more susceptible than wild type to heat and SDS [5]. EF1194-1193 (HK-RR07) (VicKR) has been reported to be an essential system for E. faecalis, with high similarity to the essential YycFG systems found in B. subtilis, Staphylococcus aureus, Streptococcus pneumoniae and other low%GC Gram-positive species [3–5], [72–74]. Target genes have been identified for many species and include those involved in cell wall biosynthesis or cell membrane composition [8], [75–79], cell division (ftsAZ) [18], virulence and genetic competence [17], [19], [20] and biofilms [20]. EF1863-1864 (HK-RR08) is most homologous to the VncSR system of S. pneumoniae implicated in vancomycin tolerance [21], is linked to mobile genetic elements [5] and in common with EF1704-1703 and EF3329 mutations, its disruption leads to heat sensitivity [4]. EF0927-0926 (HK-RR09) is less well understood and shows no particular homology with any known systems in the database. Using both a mouse peritonitis model and peritoneal macrophages, the EtaSR (EF1051-1050 or HK-RR10) system has been shown to contribute to virulence [3], [6] and is most homologous to the LisKR and CsrSR systems of other Gram-positive bacteria. Mutants display increased acid sensitivity [3] and heat resistance correlated with increased DnaK and GroEL levels in these strains [4]. EF2298-2299 (HK-RR11), VanS_BR_B, regulates VanB-type vancomycin resistance and its inactivation renders E. faecalis significantly more sensitive to vancomycin [5], [22]. EF0570-0571 (HK-RR12) is similar to KdpDE found in several Gram-positive and Gram-negative organisms; in E. faecalis it is often located on a pathogenicity island though it is not present in all strains of E. faecalis suggesting that it is not essential for a commensal life style [3], [5], [23]. EF0373-0372 (HK-RR13) expression was only detectable at a higher temperature of 50°C, suggesting a possible involvement in the E. faecalis stress response [4]. The RR of the EF1209-1210 (HK-RR14) system is most similar to the CitBT family of response regulators involved in regulating citrate and/or C4 dicarboxylate levels; EF1210 inactivation resulted in a growth defect [5] and increased sensitivity at 50°C, implicating it is the stress response [4]. EF1820-1822 (HK-RR15) corresponds to the Fsr quorum-sensing system of E. faecalis [86–89] and has a role in virulence [90–92] and biofilm formation [5], [93–95]. Little is known of EF1335-1336 (HK-RR16), though in common with several other RR knockout strains of E. faecalis, the RR16 mutant displays faster doubling time during growth in the presence of subinhibitory concentrations of erythromycin, suggesting an effect in uptake of this class of drug [5]. EF1632-1633 (HK-RR17) is the only system with a cytoplasmic histidine kinase and exhibits no significant similarity to known TCSs [1]. Inactivation of orphan response regulator EF3329 (RR18) results in increased heat sensitivity [4], and a phenotype similar to that of RR16 [5].

Acknowledgements

We are grateful to Professor Michael S. Gilmore, Department of Microbiology & Immunology, University of Oklahoma Health Sciences Center, USA, for provision of E. faecalis V583, and to Halina Norbertczak and Peter Roach for preparations of EF1051 (HK10)-harbouring E. coli membranes in a fermenter facility. We thank Lianne Davis, Claire Harris, Tacita Nye and Rida Saeed for optimization and activity data for some of the kinases, and Denise Ashworth, Jeff Keen, Alison Ashcroft and Lynsey Jones (University of Leeds) for DNA sequencing, N-terminal sequencing and ESI-mass spectrometry services respectively, provided by the Wellcome Trust. MKP-J and PJFH are grateful to the BBSRC (BB/D001641/1), and PJFH to the European Union (EMeP grant LSHG-CT-2004-504601) for financial support. Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.