Protein Expression and Purification 101 (2014) 138–145

Contents lists available at ScienceDirect

Protein Expression and Purification

journal homepage: www.elsevier .com/ locate /yprep

Expression, purification and functional reconstitution of FeoB, theferrous iron transporter from Pseudomonas aeruginosa

http://dx.doi.org/10.1016/j.pep.2014.06.0121046-5928/� 2014 Elsevier Inc. All rights reserved.

⇑ Corresponding author. Tel.: +61 8 83021515; fax: +61 8 83022389.E-mail address: rietie.venter@unisa.edu.au (H. Venter).

Saeed Seyedmohammad a, Diana Born a, Henrietta Venter b,⇑a Department of Pharmacology, University of Cambridge, Tennis Court Road, Cambridge CB2 1PD, UKb School of Pharmacy & Medical Sciences, Sansom Institute for Health Research, University of South Australia, GPO Box 2471, Adelaide 5001, Australia

a r t i c l e i n f o a b s t r a c t

Article history:Received 6 March 2014and in revised form 20 June 2014Available online 30 June 2014

Keywords:Pseudomonas aeruginosaIron transporterMembrane proteinOverexpressionReconstitution

The FeoB Fe(II) transporter from the drug resistant pathogen, Pseudomonas aeruginosa is essential forferrous iron transport and is implicated in virulence and biofilm development. Hence it is an attractivetarget for the development of new anti-infective drugs. FeoB is an intriguing protein that consists of acytosolic N-terminal GTPase domain and an integral membrane domain which most likely acts as ferrousiron permease. Characterisation of FeoB is critical for developing therapeutics aimed at inhibiting thisprotein. However, structural and functional analysis of FeoB is hampered by the lack of high yieldhomogenously pure protein which is monodisperse, stable and active in solution. Here we describe theoptimised procedure for the recombinant expression of FeoB from P. aeruginosa and provide an evalua-tion of the most favourable purification, pH and detergent conditions. The functional reconstitution ofFeoB in liposomes is also described. This represents the first detailed procedure for obtaining a pure,active and stable FeoB solution in milligram quantities which would be amenable to biochemical, bio-physical and structural studies.

� 2014 Elsevier Inc. All rights reserved.

Introduction

Pseudomonas aeruginosa is a Gram-negative pathogen which isassociated with a range of life-threatening hospital acquired infec-tions. It is also the main cause of mortality in patients sufferingfrom Cystic Fibrosis and is characterised by an innate resistanceagainst multiple classes of antimicrobials [1–4]. Antimicrobialresistance amongst clinical isolates of P. aeruginosa is increasingat an alarming rate and has become a major health problem[2,5,6], hence there is an urgent need to discover new treatmentsand therapies against P. aeruginosa. An important way of fightinginfection is the targeting of bacterial iron acquisition. Iron is vitalfor the survival of pathogens as well as being an essential constit-uent of virulence and biofilm formation [7,8]. Ferrous iron isacquired by the Feo transporter which is composed of threeproteins, FeoA, FeoB and FeoC (Fig. 1A and B) [9]. In this transportsystem FeoB is the iron permease protein, which contains a cyto-solic N-terminal domain with GTPase activity and an integralmembrane domain which could function as the Fe(II) permeasedomain. Ferrous iron acquisition through the membrane domainis dependent on the GTPase activity of the soluble domain

[10,11]. Deletion of the feoB gene has a detrimental effect on theability of pathogens to form biofilms and also attenuates their vir-ulence [12–20]. FeoB is therefore a tractable target for inhibition.Despite its important role in bacterial survival and virulence, ourknowledge about this transporter is still in its infancy. Establishingthe structure and molecular mechanism of the FeoB protein wouldbe essential for the development of therapies that target this pro-tein. Although the crystal structures of the N-terminal GTP bindingdomain of FeoB (NFeoB) proteins from various bacteria haverecently been obtained [21–30], this data has yet to be translatedinto structural and functional significance for the full-length FeoBprotein. Due to the difficulties in obtaining high-level overexpres-sion and functional purification of integral membrane proteins, nostructures of full-length FeoB from any organisms has been solvedyet. Key questions remain regarding the structure of the mem-brane domain, the coupling between GTP hydrolysis and irontransport, the iron translocation pathway through the FeoB perme-ase domain and the role of FeoA and FeoC in iron transport by FeoB.Progress on FeoB has been hampered by the difficulty in obtainingsufficient yields of homogenously pure, monodisperse, active full-length protein. In this study we have addressed these issues inorder to obtain purified FeoB which should be amenable tobiochemical, biophysical and structural studies. We have heterolo-gously expressed FeoB from P. aeruginosa in Escherichia coli andhave optimised the conditions of expression and purification. The

Fig. 1. (A) Organisation of the feoABC operon of P. aeruginosa. The predicted mass and subcellular location of each of the feo translation products are indicated. (B) Schematicrepresentation (not to scale) of the proposed structure of the Feo system. FeoB consists of an N-terminal, cytoplasmic GTPase domain, a GDI linker domain and eight putativetransmembrane a-helices. The transmembrane domains may form the pore for iron transport. The roles of FeoA and FeoC are not known, however they both have been shownto interact with FeoB from related species [54,55].

S. Seyedmohammad et al. / Protein Expression and Purification 101 (2014) 138–145 139

stability of the protein over a pH range from 7.0 to 8.0 and in dif-ferent detergents was also assessed and functional reconstitutionof purified protein in liposomes was achieved.

Materials and methods

Materials

Plasmid pET41-a (+) was from Novagen and E. coli C41 (DE3)cells were from Lucigen. Super-competent DH5a cells were fromBioline. Protein molecular weight markers, DNA ladders and othermolecular biology consumables were supplied by Fermentasunless indicated otherwise. n-Dodecyl-b-D-maltopyranoside(DDM)1 and NV10 were from Expedeon, C12E8 from Anapoe andAmphipol A8-35 from Affymetrix. Growth media and kanamycinwere from Formedium and isopropyl b-D-1-thiogalactopyranoside(IPTG) from Melford Labs. General chemicals were from Sigmaunless indicated otherwise.

Cloning of the feoB gene and generation of the D123N FeoB mutant

The feoB gene was amplified from genomic DNA of P. aeruginosastrain PAO1 (a gift from Martin Welch, University of Cambridge)with Velocity DNA polymerase (Bioline) using primers 50-GGA-ATTCCATATGACCGCATTGACCCTCGGCC-30 and 50-CGAGCGGGCAGGGAGGATGTCACCTGGTGCCACGCGGTAGTGGGAAGCTTGCGCGC-30.NdeI and HindIII restriction sites (shown in bold) were included.The PCR generated sequence was restriction digested and clonedinto the pET41-a(+) expression vector (Novagen) to yield pFeoBH,which coded for FeoB with a C-terminal 8-histidine tag. The ligatedinsert was thereafter transformed into E. coli DH5a-silver selectcompetent cells. The resulting recombinant construct was con-firmed by restriction digestion at the NdeI/HindIII site and verifiedby nucleotide sequencing.

A D123N mutant of the wild-type FeoB protein was generatedby PCR with pFeoBH as template and using primers 50-GAACATGCTCAACATCGCCCGTAGCCAGCG-30 and 50-CGGGCGATGTTGAGCATGTTCAGCGCGACG-30. The wild-type plasmid DNA template wasremoved by restriction digestion of the PCR product with DpnI(New England Biolabs). The D123N feoB was restriction digestedwith NdeI/HindIII and ligated into restriction digested pET41a(+)

1 Abbreviations used: DDM, Dodecyl-b-D-maltopyranoside; IPTG, isopropyl b-D-1-thiogalactopyranoside; Ni–NTA, Ni2+–nitrilotriacetate; PC, phosphatidylcholine; LBLuria Bertani; TB, Terrific Broth; OG, n-Octyl-b-D-glucopyranoside.

,

to yield plasmid pD123N-FeoBH. The cloned PCR product wassequenced to ensure that only the intended changes wereintroduced.

Preparation of inside-out membrane vesicles

Inside out vesicles were prepared essentially as describedbefore [31] with the following modifications. E. coli C41 (DE3) cellswere transformed with pFeoBH or pD123N-FeoBH, plated on LBagar containing 25 lg/ml kanamycin and incubated at 37 �C. Ter-rific Broth containing 25 lg/ml kanamycin was inoculated withan overnight culture of E. coli C41 (DE3) containing pFeoBH orpD123N-FeoBH. Protein production was induced when the culturereached an OD660 of 0.4 by the addition of IPTG (0.5 mM) and theculture was incubated overnight at 18 �C. Cells were harvestedby centrifugation (6000�g, 10 min, 4 �C). The cell pellet was resus-pended in 100 mM K-HEPES (pH 7.0) followed by centrifugation(6000�g, 10 min, 4 �C). The pellet was resuspended in 100 mM K-HEPES buffer (pH 7.0) in presence of 10 lg/ml DNase. Cells werelysed by three passages through a basic Z 0.75 kW benchtop celldisrupter (Constant Systems) at 0.138 MPa. The lysate was sub-jected to low-speed centrifugation at 13,000�g for 10 min at 4 �Cto remove cell debris. The resulting supernatant was then sub-jected to high-speed centrifugation at 150,000�g for 40 min at4 �C. The pellet containing the inside out (ISO) membrane vesicleswas resuspended in 50 mM K-HEPES (pH 7.0) containing 10% (v/v)glycerol to a protein concentration of approximately 50 mg/ml andstored in liquid nitrogen. The protein concentration of the ISOvesicles was determined by the DC Protein Assay (Bio-Rad Labora-tories) with BSA as a standard.

Protein purification

ISO vesicles containing FeoB or D123N FeoB were solubilised byshaking at 4 �C for 1 h in a volume of solubilisation buffer(20 mM K-HEPES pH 7.4, 20% glycerol, 500 mM NaCl, 2% DDM,10 mM imidazole and 10 mM MgSO4) sufficient to produce a finalprotein concentration of 10 mg/ml. Unsolubilised protein wasremoved by high-speed centrifugation (150,000�g for 40 m at4 �C). Ni–NTA (Ni2+–nitrilotriacetate) resin (Sigma) wasequilibrated by washing with 20 resin volumes of deionized water,gravity sedimentation on ice and resuspended with 5 resin vol-umes of wash buffer (10 mM K-HEPES pH 7.0, 10% glycerol,500 mM NaCl, 0.05% DDM, 10 mM imidazole pH 8.0 and 10 mMMgSO4). Solubilised protein was added to the equilibrated resin

140 S. Seyedmohammad et al. / Protein Expression and Purification 101 (2014) 138–145

and allowed to bind at 4 �C for at least 1 h. The resin was thentransferred to a column (Bio-Rad Laboratories) where the unboundprotein was allowed to drain through. The resin in the column waswashed with a total of 50 resin volumes of wash buffer. His-taggedprotein was eluted in 3 resin volumes of elution buffer (10 mM K-HEPES pH 7.0, 10% glycerol, 200 mM NaCl, 500 mM imidazole,10 mM MgSO4 and different concentrations of detergents; 0.05%DDM, 0.1% C12E8, 1% NV10 and 1% Amphipol A8-35) of which thefirst 0.5 resin volume was discarded. Total purified protein wasassayed using the Micro BCA™ Protein Assay Kit (Thermo Scien-tific) with BSA as a standard.

GTPase assay

The GTPase activity of FeoB and D123N FeoB was determinedusing the QuantiChrom™ GTPase assay kit (BioAssay Systems).The assay was based on the formation of a stable green-colouredcomplex formed between the malachite green colouring reagentand liberated Pi released from the hydrolysis of GTP. Reactionswere prepared to give a final volume of 150 ll, containing theassay buffer (BioAssay Systems), 10 mM MgSO4, 1 mM GTP and20–50 lg/ml of the purified proteins in 20 mM K-HEPES pH 7.0buffer. Reactions were prepared in a 96-well plate (Costar) witheach well containing a reaction volume of 10 ll/well in duplicateswhich typically contained 2 lg of purified protein. The GTPaseassay was performed at 37 �C with continuous shaking after initi-ation of reaction by the addition of GTP. Samples were taken at dif-ferent time intervals, the reaction were terminated by addition ofthe colouring reagent (100 ll) and the A630 values were read after30 min.

Far UV circular dichroism spectrophotometry

Samples of purified FeoB (50 lg/ml) protein in 20 mM K-HEPES(pH 7.0), 0.05% DDM were loaded into a Hellma quartz cuvettewith a 1 mm path length. CD spectral measurements were madein a JASCO J-71 circular dichroism spectrophotometer operated at25 �C with constant nitrogen flushing and were measured between189 and 260 nm with an averaging time of 0.5 s. Data were ana-lysed using the Jasco Secondary Structure Estimation software,which used the Yang reference data set.

Reconstitution of purified FeoB in liposomes

For the preparation of liposomes, commercially available E. colitotal lipid extract (Avanti Polar Lipids) were further purified byextraction with acetone–ether to yield E. coli polar lipids. E. colitotal lipid extract (100 mg) was dissolved in 2 ml of chloroformand then slowly dripped into 10 ml of ice-cold nitrogen flushedacetone containing 2 ll of b-mercaptoethanol. The mixture wasstirred at 4 �C overnight and centrifuged in glass tubes for 15 minat 3000g. The pellet was dried under nitrogen gas and dissolvedin 10 ml of nitrogen flushed diethyl ether containing 2 ll of b-mercaptoethanol for 10 min at room temperature. Following cen-trifugation for 10 min at 1600g the supernatant was evaporatedto dryness in a rotary evaporator. The residue was weighed anddissolved in chloroform, and egg phosphatidylcholine (PC) (AvantiPolar Lipids) was added to the polar lipids at a PC to E. coli lipidratio of 1:3 (w/w). Once mixed, the lipids were dried as a thin filmunder nitrogen gas and stored at �20 �C until required. For thereconstitution of purified FeoB or D123N FeoB into Triton X-100-destabilized liposomes, the lipids were hydrated at a concentrationof 20 mg/ml in 50 mM K-HEPES (pH 7.0). The lipids were subjectedto two freeze–thaw cycles and extruded 11 times through a400 nm polycarbonate filter using a 1 ml LiposoFast-Basic extruder(Avestin). The resulting liposomes were diluted to a concentration

of 4 mg/ml in 20 mM KPi (pH 7.0) and destabilized by the additionof 2–3 mM Triton X-100 at 0.25 mM increments until the OD540 ofthe liposome suspension reached a maximum [32]. Destabilisationof liposomes with Triton X-100 in this way resulted in liposomesthat were not completely open. Hence, unidirectional reconstitu-tion will be achieved for membrane proteins that contain a solublehydrophilic domain since such a domain cannot cross the lipidbilayer. In the case of FeoB, the presence of the cytoplasmic, GTPasedomain would force the reconstitution to be inside-out. Afterdestabilisation, purified FeoB in elution buffer was then added toa protein to lipid ratio of 1:200 (w/w). After incubation for30 min at room temperature, 80 mg of hydrated polystyrene Bio-Beads (Bio-Rad) were added per millilitre of liposome suspensionto remove the detergent. Following incubation for 2 h at 4 �C, thesebeads were replaced twice by fresh Bio-Beads (80 mg/ml), whichwere incubated at 4 �C for 2 h and overnight, respectively. Beforeuse, Bio-Beads were hydrated by one wash in methanol, followedby one wash in ethanol and five washes with ultrapure H2O. Theproteoliposomes were collected by centrifugation at 164,000 gfor 30 min and resuspended to 0.5 mg of membrane protein/mlin 50 mM K-HEPES (pH 7.0).

Evaluation of protein aggregation

FeoB was purified as described in ‘Protein Purification’. For eval-uation of different detergents, a detergent-exchange was affectedby including a second wash step with buffer containing the testdetergent/polymer before elution of the protein in buffer contain-ing the same detergent/polymer. The level of aggregation over timewas measured by comparing the amount of soluble protein in thesupernatant after high speed centrifugation (100,000�g, 1 h, 4 �C)of a sample that has been incubated at 4 �C for 4 days, with thatof a sample immediately after purification.

For dynamic light scattering (Malvern), samples were centri-fuged (5000�g, 5 min) to remove any dust particles and 400 llwas transferred to a quartz cuvette for measurement. The solventviscosity was taken as that of water at 20 �C i.e.,0.01002 g cm�3 s�1.

SDS–PAGE and Western blot analysis

Protein expression and purity were evaluated using SDS–PAGEon 4–12% polyacrylamide gels (Expedeon). Proteins were trans-ferred to a PDVF membrane using an iBlot system (Invitrogen)according to the manufacturer’s instructions and immunoblottingwith monoclonal anti-His antibody (Qaigen) was used to confirmthe presence of the histidine-tagged proteins.

Results and discussion

Cloning and expression of FeoB and D123N FeoB

FeoB was overexpressed in E. coli which is the method of choicefor producing bacterial membrane proteins due to its high growthrate and the ease of genetic manipulation [33–35]. The FeoBexpression plasmid was constructed by cloning the feoB gene inthe expression vector pET-41a(+) carrying the T7 promoter andlac-operator system. The resulting plasmid pFeoBH coded for therecombinant FeoB protein with an octa-histidine tag at the C-ter-minal. To examine the GTPase activity of the protein, a mutantwith Asp at position 123 changed to Asn (D123N FeoB) was con-structed. This mutation is in the fourth G protein consensus motifand prevents GTP binding to FeoB [36]. The non-expressing controlplasmid pET-41a(+) and the FeoB expressing plasmids were prop-agated in the E. coli expression strain C41 (DE3) which is especiallyadapted for expression of membrane proteins [34,37]. A band

Fig. 3. Optimal expression of FeoB. (A) The non-expressing cells and cellsexpressing FeoB were grown in different media (LB = Luria Bertani broth;2xYT = Yeast tryptone broth; TB = Terrific broth; AIM = autoinduce medium or (B)at different temperatures. Inside-out membrane vesicles were prepared and thetotal membrane proteins (30 lg/lane) were separated by SDS–PAGE. Expression ofFeoB was visualised by Western blotting and probing with the anti-His antibody.The position of FeoB is indicated by an arrow.

S. Seyedmohammad et al. / Protein Expression and Purification 101 (2014) 138–145 141

corresponding to the size of FeoB (83 kDa) could be observed inCoomassie-stained SDS–PAGE of inside-out membrane vesicles(ISO vesicles) prepared from cells expressing FeoB and D123NFeoB. This band was absent in the non-expressing control cells(Fig. 2A). The identification of the �83 kDa protein as FeoB wasreinforced by subjecting the gel to a Western blot analysis, usingan anti-His antibody to detect the C-terminal eight-His tag(Fig. 2B). The expression level of the D123N mutant protein inthe cytoplasmic membrane was similar to that of wild-type FeoB(Fig. 2A and B). As we have successfully cloned and expressedthe full-length FeoB our next aim was to optimise expression ofthe protein to yield sufficient amounts of pure FeoB for biophysicalanalysis.

Optimisation of FeoB expression levels

For structural characterisation of FeoB, a significant amount ofprotein was required and hence, the expression level of FeoB pro-tein was optimised in different bacterial growth media at differenttemperatures. The bacterial growth media investigated includedLuria Bertani (LB) Broth, 2xYT, Terrific Broth (TB) and M9 minimalmedium. The highest level of FeoB expression was obtained in richmedia such as TB. Although autoinduce medium is more conve-nient, the expression level in autoinduce medium was lower thanthat in media which was manually induced (Fig. 3A).

Induction temperature also has a great effect on membrane pro-tein expression, with lower temperatures often favouring higherlevels of expression [38]. To determine the optimal temperaturefor FeoB expression, the cultures were grown and induced by theaddition of IPTG at 37 �C, 30 �C, 25 �C and 18 �C. The expressionlevel of FeoB increased as the temperature was lowered with thehighest level of expression obtained at 18 �C (Fig. 3B).

Purification and reconstitution of FeoB

Once the expression levels of FeoB had been improved, a purifi-cation procedure was developed that would lead to high levels ofhomogeneously pure, active protein. In order to do this, inside outmembrane vesicles were prepared from cells expressing FeoB orD123N FeoB. The concentration of DDM and length of solubilisationwere optimised and the membrane proteins were solubilised withDDM at the optimal conditions determined (2% DDM and 2 h of sol-ubilisation). The solubilised protein was subsequently purifiedusing Ni–NTA resin by exploiting the C-terminal His-tag. Imidazole(500 mM) was sufficient to remove all bound FeoB from the resin.The purification yielded homogenous FeoB and also resulted inequal yields of the D123N mutant of FeoB (Fig. 4A). A faint second

Fig. 2. FeoB and D123N FeoB expressed to equal levels. Inside-out membranevesicles were prepared from the non-expressing control, FeoB and D123N FeoBexpressing cells. Membrane proteins (30 lg/lane) were separated by SDS–PAGE. (A)Coomassie staining showing equal expression of both proteins. The position of FeoBis indicated with an arrow. (B) Western blot probed with anti-His antibodyconfirming equal expression of the D123N mutant and WT FeoB and lack of thisexpression in controls cells.

band running slightly faster than FeoB could sometimes beobserved on a Coomassie stained SDS–PAGE gel of purified FeoB.The addition of protease inhibitor did not prevent the appearanceof this band. We have separated the two bands by a long SDS–PAGErun and have performed MALDI Mass Spectrometry (MS) on a tryp-tic digest of both bands. Identical spectra were obtained for bothbands. Almost 40% of the sequence was covered by the MALDI MSfingerprint of the tryptic digest. Most importantly, both the N-ter-minal as well as the C-terminal peptide signals were present(Fig. 4B), suggesting both bands on the SDS–PAGE gel representthe full-length, intact FeoB and that the lower band was not theresult of proteolytic cleavage of FeoB or degradation. The secondfaint band could represent FeoB with a modification however, giventhat integral membrane proteins often run anomalously on SDS–PAGE due to differential detergent binding [39], the second FeoBband could also represent FeoB with slightly less detergent bound.

The purified FeoB was also active and displayed baseline GTPaseactivity with a kcat of �0.0035 s�1 (Fig. 4C) which is comparable tothe slow GTP hydrolysis rates observed for the soluble GTP bindingdomain of FeoB from related species which yielded kcat values of0.0015 s�1 [36], 0.0024 s�1 [40], 0.0015 s�1 [10] and 0.002 s�1

[29]. The GTPase activity of the D123N FeoB mutant was muchreduced compared to the wild type with a kcat of 0.0012 s�1. Inorder to investigate the structural properties of FeoB, we recordeda far-UV circular-dichroism spectrum. The observed spectrum(Fig. 4D) is characteristic of a protein with a predominantly a-heli-cal structure. Further analysis of the CD data with K2D2 [41] esti-mated an alpha-helical content of approximately 77%.

Although membrane proteins solubilised in detergent are usedfor most biophysical and structural analysis, membrane proteinsthat are functionally reconstituted into liposomes offers a superiortool for mechanistic studies. Membrane proteins which are recon-stituted in the lipid bilayer of a liposome can be studied in an envi-ronment that closely resembles the native membrane, but in theabsence of other proteins. Liposomes also offer the ability to gener-ate highly defined gradients which could mimic the proton-motiveforce that powers many protein mediated transport processes overthe bacterial membrane. Hence a proteoliposome can act as a sur-rogate cell were detailed functional studies can be performed on apurified protein [31,42,43]. We have managed to reconstitute puri-fied FeoB in liposomes and the reconstituted protein was stableover many days. FeoB was reconstituted in the inside-out orienta-tion to allow the measurement of GTPase activity of the exposedGTP binding domain. FeoB was functional in liposomes and dis-played baseline, non-stimulated GTPase activity which yielded a10� higher kcat (0.0046 s�1) compared to the proteoliposomes con-taining the D123N FeoB mutant (Fig. 4E).

Fig. 4. High yield purification and reconstitution of homogenous, active FeoB. (A) SDS–PAGE of FeoB and D123N FeoB purified from ISO vesicles prepared from E. coli C41(DE3) cells. The gel was stained with Coomassie Blue. The position of FeoB is indicated with an arrow. (B) Tryptic digestion and MALDI MS of the two bands (upper band andlower band) that were observed on a SDS–PAGE of purified FeoB showed that the N-terminal peptides (left panel) and C-terminal peptides (right panel) were present in bothFeoB bands. (C) GTPase activity of purified FeoB (j) and D123N FeoB (d). The GTPase reaction was initiated by the addition of 4 mM GTP with measurements taken atdifferent time intervals. The GTPase activity of the FeoB proteins was determined by measuring the liberation of Pi from GTP with malachite green (n = 4). (D) Far UV circulardichroism spectrum of FeoB solubilised and purified in DDM. (E) GTPase activity of purified FeoB (j) and D123N FeoB (d) which were reconstituted in liposomes preparedfrom E. coli and egg yolk phosphatidyl choline lipids indicates that functional reconstitution was achieved (n = 4).

142 S. Seyedmohammad et al. / Protein Expression and Purification 101 (2014) 138–145

Optimising the pH for stability and activity

Most biophysical methods require concentrated protein sam-ples with 10 mg/ml being a typical requirement for ESI-MS andcrystallography analysis [44]. An Amicon Ultra 100 kDa MWCO cel-lulose filter device was used to concentrate purified FeoB and toremove excess imidazole. Even though the monomeric size of FeoBis smaller than 100 kDa, the overall size of FeoB in a DDM micelleexceeds 100 kDa. Protein determination and SDS–PAGE confirmedthat no FeoB was lost in the filtrate. A 100 kDa cut-off membrane ismuch preferred over a 50 kDa cut-off membrane for the concentra-tion of samples in the presence of DDM, as the latter will also con-centrate the DDM micelles (micellar size is 50–70 kDa) which oftenlead to protein precipitation [45]. Our first attempts to concentratepurified FeoB resulted in protein aggregation and precipitation. Wetherefore investigated the optimum pH to keep FeoB in solutionduring concentration. It was revealed that the protein aggregatedat pH 7.0 or pH 7.5 and that a pH of 8.0 was needed to ensure pro-tein stability in a concentrated solution.

Crystallisation requires a protein to be stable and monodisperseat 4 or 20 �C for several days or weeks. We therefore investigatedthe stability of FeoB at different pH values when incubated at4 �C for 4 days. FeoB was solubilised and purified as described inthe materials and methods, with the only difference that the resinwas also washed with twenty resin volumes of a second wash buf-fer and that the pH of this buffer and the elution buffer wasadjusted to either pH 7.0, 7.5 or 8.0. We also included a sample

at pH 7.0 where lipids were added throughout the purification asthe addition of lipids can sometimes stabilise membrane proteinsduring purification. Purified fractions of the FeoB protein at differ-ent pH values were subjected to ultracentrifugation to remove theaggregated protein either directly after purification or after 4 daysof incubation at 4 �C. A sample of the supernatant after high speedcentrifugation was run on an SDS–PAGE gel to determine theamount of protein in solution. It is clear that a higher pH is the bestfor protein stability as the sample at pH 8.0 displayed the leastaggregation over the incubation period, while more than half ofthe purified protein was lost due to aggregation for the FeoB sam-ple at pH 7.0. The addition of lipids during the purification processdid not enhance protein stability at pH 7.0 (Fig. 5A).

It was clear that higher pH values favour protein stability overtime hence we also investigated the effect of pH on GTPase activity.The initial reaction rates were similar for all of the pH valuestested, however a significantly lower steady state level wasobserved at pH 8.0 (Fig. 5B). Therefore, FeoB was routinely purifiedin buffer pH 7.0 for activity measurements and in buffer pH 8.0when the protein needed to be concentrated or kept at 4 �C for sev-eral days for biophysical analysis.

Effect of various detergents and polymers on the level of aggregation ofFeoB

The oligomeric state of membrane proteins in solubilised stateis strongly detergent-dependent. In the case of FeoB, DDM is a very

Fig. 5. Effect of pH on FeoB stability and activity. (A) SDS–PAGE of FeoB purified in buffers at different pH values. The gel was stained with Coomassie Blue. Purified proteinwere subjected to high speed spin either immediately after purification (Day 1) or after incubation at 4 �C for 4 days (Day 4) and an aliquot of the supernatant was loaded onthe gel. (B) GTPase activity of purified FeoB at pH 7.0 (j), pH 7.5 (d), pH 8.0 (N) and pH 7.5 in the presence of lipids (.) n = 4.

Fig. 6. Effect of detergents on FeoB stability and activity. (A) SDS–PAGE (10%) of FeoB purified in different detergents. The gel was stained with Coomassie Blue. Purifiedprotein were subjected to high speed spin either immediately after purification (Day 1) or after incubation at 4 �C for 4 days (Day 4) and an aliquot of the supernatant wasloaded on the gel. (B) GTPase activity of purified FeoB in DDM (j), C12E8 (d), NV-10 (N) showing similar activity of FeoB in all three detergents (n = 4).

Table 1Summary of DLS data.

Sample % of total fraction R (nm)

DDM (0.05%) 100 2.8DDM (0.05%) + FeoB 99.5 6.9

C12E8 (0.05%) 100 3.1C12E8 (0.05%) + FeoB 100 5.7

Amphipol A8-35 (1%) 100 2.7Amphipol A8-35 (1%) + FeoB 100 4.6

NV-10 Polymer (1%) 98 7.0NV-10 Polymer (1%) + FeoB 91 8.9

S. Seyedmohammad et al. / Protein Expression and Purification 101 (2014) 138–145 143

effective detergent to solubilise the protein from the bacterialmembrane, but unfortunately the protein exhibits a high level ofaggregation when contained in a DDM micelle. Although DDM isa mild detergent that is often used for the solubilisation of integralmembrane transporter proteins, a high level of aggregation in DDMhas also been reported for other membrane proteins such asMmCDF3 and GlpT from Haemophilus influenzae [34,46]. In orderto preserve protein stability and monodispersity we investigatedother detergents and polymers for their effect on the stabilityand aggregation of FeoB. In addition to DDM, the detergents n-Octyl-b-D-glucopyranoside (OG) and C12E8 were also included.Similar to DDM, these two detergents are in general use for the sol-ubilisation of integral membrane proteins for functional and struc-tural studies [47]. As detergents can sometimes causedestabilisation of proteins which leads to protein unfolding and

aggregation, we also included two polymers in our screen.Amphipols are mild surfactants which have been successfully usedto substitute for detergents and have been shown to enhance thestability of various membrane proteins [48–50]. One polymer,Amphipol A8-35, was able to stabilise outer-membrane proteinsand preserve protein functionally for several months at 4 �C [51].NV10 is a polyfructose polymer that shares some properties ofdetergents and amphipols. It is highly soluble in aqueous solutionsand has been used with great success to refold membrane proteinsafter cell-free synthesis [52,53]. We also found that FeoB could beconcentrated to more than 10 mg/ml in the presence of NV10, evenat pH 7.0 (data not shown).

The use of OG resulted in almost immediate precipitation ofpurified FeoB. The stability of FeoB in the other detergents andpolymers were assessed by ultracentrifugation with subsequentSDS–PAGE and by dynamic light scattering, which is a quick andnon-invasive way to detect protein aggregation. The detergentC12E8 and the polymer Amphipol A8-35 were both very effectivein preserving FeoB in solution as all of the protein was still foundin the supernatant (soluble fraction) after ultracentrifugation evenafter incubation at 4 �C for 4 days (Fig. 6A). FeoB was also subjectedto DLS immediately after purification. The DLS data correlated wellwith the ultracentrifugation data. The protein-micelle sizes were100% uniform for FeoB in C12E8 and Amphipol A8-35 indicatingthat no aggregation of FeoB occurred in these two compounds(Table 1). A small proportion (0.5% and 9% respectively) of FeoBin DDM or NV-10 polymer was aggregated immediately after puri-fication (Table 1).

144 S. Seyedmohammad et al. / Protein Expression and Purification 101 (2014) 138–145

To ascertain that FeoB retains its activity when solubilised inthe different detergents/polymers, a GTPase assay was performed.The GTPase activity of FeoB in C12E8 and NV10 is comparable withthat of FeoB in DDM (Fig. 6B). It was not possible to measure theactivity of FeoB in Amphipol A8-35 as a component in the polymersolution reacted with the assay kit which resulted in very highbackground readings.

Even though FeoB could be concentrated at pH 7.0 in NV-10without aggregation, NV-10 was not compatible with all the down-stream biophysical applications. Also, Amphipol A8-35 was verygood in preserving a monodisperse protein solution, but as wecould not determine GTPase activity for FeoB in Amphipol A8-35,C12E8 was our detergent of choice to preserve FeoB stability andfunction when in solution.

Conclusion

The microbial ferrous iron transporter FeoB is important forbacterial survival and a major constituent of virulence; hence itis an interesting target for biochemical and structural studies.Our understanding of this protein has been hampered by the gen-eral difficulties in handling membrane proteins and the majorobstacles preventing the purification of functional membrane pro-teins in milligram quantities. In this paper we describe an expres-sion and purification procedure that would deliver high yields ofstable, monodisperse and active FeoB from P. aeruginosa. It willpave the way for studying the full-length protein and answer themany outstanding questions regarding the structure, functionand regulation of FeoB. Moreover, the functional reconstitutionachieved in this work would open up exciting new avenues forthe study of this important drug target in the human pathogen P.aeruginosa and ultimately could aid the design of anti-infectivetherapies.

Acknowledgments

This work was supported by the Royal Society (Grant Number45605 to HV), the University of South Australia and the SansomInstitute for Health Research. S.S. is the recipient of a BBSRC schol-arship and D.B. received an Erasmus Programme Scholarship. Wethank Mr. Len Packman for the MALDI-MS analysis.

References

[1] K. Poole, Pseudomonas aeruginosa: resistance to the max, Front. Microbiol. 2(2011) 65.

[2] J.N. Pendleton, S.P. Gorman, B.F. Gilmore, Clinical relevance of the ESKAPEpathogens, Expert Rev. Anti Infect. Ther. 11 (2013) 297–308.

[3] E.B. Breidenstein, C. de la Fuente-Nunez, R.E. Hancock, Pseudomonasaeruginosa: all roads lead to resistance, Trends Microbiol. 19 (2011) 419–426.

[4] J.B. Goldberg, R.E. Hancock, R.E. Parales, J. Loper, P. Cornelis, Pseudomonas 2007,J. Bacteriol. 190 (2008) 2649–2662.

[5] L.F. Wang, J.L. Li, W.H. Ma, J.Y. Li, Drug resistance analysis of 1914 bacterialstrains isolated from burn patients, Genet. Mol. Res. 13 (2014).

[6] H.S. Sader, D.J. Farrell, R.K. Flamm, R.N. Jones, Antimicrobial susceptibility ofGram-negative organisms isolated from patients hospitalized in intensive careunits in United States and European hospitals (2009–2011), Diagn. Microbiol.Infect. Dis. 78 (2014) 443–448.

[7] P.K. Singh, M.R. Parsek, E.P. Greenberg, M.J. Welsh, A component of innateimmunity prevents bacterial biofilm development, Nature 417 (2002) 552–555.

[8] I.L. Lamont, P.A. Beare, U. Ochsner, A.I. Vasil, M.L. Vasil, Siderophore-mediatedsignaling regulates virulence factor production in Pseudomonas aeruginosa,Proc. Natl. Acad. Sci. USA 99 (2002) 7072–7077.

[9] M.L. Cartron, S. Maddocks, P. Gillingham, C.J. Craven, S.C. Andrews, Feo-transport of ferrous iron into bacteria, Biometals 19 (2006) 143–157.

[10] E.T. Eng, A.R. Jalilian, K.A. Spasov, V.M. Unger, Characterization of a novelprokaryotic GDP dissociation inhibitor domain from the G protein coupledmembrane protein FeoB, J. Mol. Biol. 375 (2008) 1086–1097.

[11] K. Hantke, Is the bacterial ferrous iron transporter FeoB a living fossil?, TrendsMicrobiol 11 (2003) 192–195.

[12] C.A. Thomas-Charles, H. Zheng, L.E. Palmer, P. Mena, D.G. Thanassi, M.B. Furie,FeoB-mediated uptake of iron by Francisella tularensis, Infect. Immun. 81(2013) 2828–2837.

[13] J.D. Fetherston, I. Mier Jr., H. Truszczynska, R.D. Perry, The Yfe and Feotransporters are involved in microaerobic growth and virulence of Yersiniapestis in bubonic plague, Infect. Immun. 80 (2012) 3880–3891.

[14] A. Pandey, R.V. Sonti, Role of the FeoB protein and siderophore in promotingvirulence of Xanthomonas oryzae pv. oryzae on rice, J. Bacteriol. 192 (2010)3187–3203.

[15] L. Zhao, X. Chen, X. Zhu, W. Yang, L. Dong, X. Xu, S. Gao, X. Liu, Prevalence ofvirulence factors and antimicrobial resistance of uropathogenic Escherichia coliin Jiangsu province (China), Urology 74 (2009) 702–707.

[16] M. Sabri, M. Caza, J. Proulx, M.H. Lymberopoulos, A. Bree, M. Moulin-Schouleur, R. Curtiss 3rd, C.M. Dozois, Contribution of the SitABCD, MntH,and FeoB metal transporters to the virulence of avian pathogenic Escherichiacoli O78 strain chi7122, Infect. Immun. 76 (2008) 601–611.

[17] H. Naikare, K. Palyada, R. Panciera, D. Marlow, A. Stintzi, Major role for FeoB inCampylobacter jejuni ferrous iron acquisition, gut colonization, andintracellular survival, Infect. Immun. 74 (2006) 5433–5444.

[18] M. Robey, N.P. Cianciotto, Legionella pneumophila feoAB promotes ferrous ironuptake and intracellular infection, Infect. Immun. 70 (2002) 5659–5669.

[19] J. Velayudhan, N.J. Hughes, A.A. McColm, J. Bagshaw, C.L. Clayton, S.C. Andrews,D.J. Kelly, Iron acquisition and virulence in Helicobacter pylori: a major role forFeoB, a high-affinity ferrous iron transporter, Mol. Microbiol. 37 (2000) 274–286.

[20] Y. Wang, J.C. Wilks, T. Danhorn, I. Ramos, L. Croal, D.K. Newman, Phenazine-1-carboxylic acid promotes bacterial biofilm development via ferrous ironacquisition, J. Bacteriol. 193 (2011) 3606–3617.

[21] S. Koster, M. Wehner, C. Herrmann, W. Kuhlbrandt, O. Yildiz, Structure andfunction of the FeoB G-domain from Methanococcus jannaschii, J. Mol. Biol. 392(2009) 405–419.

[22] M. Hattori, Y. Jin, H. Nishimasu, Y. Tanaka, M. Mochizuki, T. Uchiumi, R.Ishitani, K. Ito, O. Nureki, Structural basis of novel interactions between thesmall-GTPase and GDI-like domains in prokaryotic FeoB iron transporter,Structure 17 (2009) 1345–1355.

[23] A. Guilfoyle, M.J. Maher, M. Rapp, R. Clarke, S. Harrop, M. Jormakka, Structuralbasis of GDP release and gating in G protein coupled Fe2+ transport, EMBO J.28 (2009) 2677–2685.

[24] K.W. Hung, Y.W. Chang, E.T. Eng, J.H. Chen, Y.C. Chen, Y.J. Sun, C.D. Hsiao, G.Dong, K.A. Spasov, V.M. Unger, T.H. Huang, Structural fold, conservation andFe(II) binding of the intracellular domain of prokaryote FeoB, J. Struct. Biol. 170(2010) 501–512.

[25] C.N. Deshpande, A.P. McGrath, J. Font, A.P. Guilfoyle, M.J. Maher, M. Jormakka,Structure of an atypical FeoB G-domain reveals a putative domain-swappeddimer, Acta Crystallogr. Sect. F Struct. Biol. Cryst. Commun. 69 (2013) 399–404.

[26] K.W. Hung, J.Y. Tsai, T.H. Juan, Y.L. Hsu, C.D. Hsiao, T.H. Huang, Crystalstructure of the Klebsiella pneumoniae NFeoB/FeoC complex and roles of FeoCin regulation of Fe2+ transport by the bacterial Feo system, J. Bacteriol. 194(2012) 6518–6526.

[27] M.R. Ash, M.J. Maher, J.M. Guss, M. Jormakka, The structure of an N11A mutantof the G-protein domain of FeoB, Acta Crystallogr. Sect. F Struct. Biol. Cryst.Commun. 67 (2011) 1511–1515.

[28] M.R. Ash, M.J. Maher, J.M. Guss, M. Jormakka, A suite of Switch I and Switch IImutant structures from the G-protein domain of FeoB, Acta Crystallogr. D Biol.Crystallogr. 67 (2011) 973–980.

[29] M.R. Ash, M.J. Maher, J.M. Guss, M. Jormakka, The initiation of GTP hydrolysisby the G-domain of FeoB: insights from a transition-state complex structure,PLoS ONE 6 (2011) e23355.

[30] N. Petermann, G. Hansen, C.L. Schmidt, R. Hilgenfeld, Structure of the GTPaseand GDI domains of FeoB, the ferrous iron transporter of Legionellapneumophila, FEBS Lett. 584 (2010) 733–738.

[31] A. Welch, C.U. Awah, S. Jing, H.W. van Veen, H. Venter, Promiscuous partneringand independent activity of MexB, the multidrug transporter protein fromPseudomonas aeruginosa, Biochem. J. 430 (2010) 355–364.

[32] J. Knol, K. Sjollema, B. Poolman, Detergent-mediated reconstitution ofmembrane proteins, Biochemistry 37 (1998) 16410–16415.

[33] A. Ward, J. O’Reilly, N.G. Rutherford, S.M. Ferguson, C.K. Hoyle, S.L. Palmer, J.L.Clough, H. Venter, H. Xie, G.J. Litherland, G.E. Martin, J.M. Wood, P.E. Roberts,M.A. Groves, W.J. Liang, A. Steel, B.J. McKeown, P.J. Henderson, Expression ofprokaryotic membrane transport proteins in Escherichia coli, Biochem. Soc.Trans. 27 (1999) 893–899.

[34] D.N. Wang, M. Safferling, M.J. Lemieux, H. Griffith, Y. Chen, X.D. Li, Practicalaspects of overexpressing bacterial secondary membrane transporters forstructural studies, Biochim. Biophys. Acta 1610 (2003) 23–36.

[35] E.R. Geertsma, B. Poolman, Production of membrane proteins in Escherichia coliand Lactococcus lactis, Methods Mol. Biol. 601 (2010) 17–38.

[36] T.C. Marlovits, W. Haase, C. Herrmann, S.G. Aller, V.M. Unger, The membraneprotein FeoB contains an intramolecular G protein essential for Fe(II) uptake inbacteria, Proc. Natl. Acad. Sci. USA 99 (2002) 16243–16248.

[37] B. Miroux, J.E. Walker, Over-production of proteins in Escherichia coli: mutanthosts that allow synthesis of some membrane proteins and globular proteinsat high levels, J. Mol. Biol. 260 (1996) 289–298.

[38] N. Gul, D.M. Linares, F.Y. Ho, B. Poolman, Evolved Escherichia coli strains foramplified, functional expression of membrane proteins, J. Mol. Biol. 426 (2014)136–149.

S. Seyedmohammad et al. / Protein Expression and Purification 101 (2014) 138–145 145

[39] A. Rath, M. Glibowicka, V.G. Nadeau, G. Chen, C.M. Deber, Detergent bindingexplains anomalous SDS–PAGE migration of membrane proteins, Proc. Natl.Acad. Sci. USA 106 (2009) 1760–1765.

[40] M.R. Ash, A. Guilfoyle, R.J. Clarke, J.M. Guss, M.J. Maher, M. Jormakka,Potassium-activated GTPase reaction in the G Protein-coupled ferrous irontransporter B, J. Biol. Chem. 285 (2010) 14594–14602.

[41] C. Perez-Iratxeta, M.A. Andrade-Navarro, K2D2: estimation of proteinsecondary structure from circular dichroism spectra, BMC Struct. Biol. 8(2008) 25.

[42] H. Venter, R.A. Shilling, S. Velamakanni, L. Balakrishnan, H.W. Van Veen, AnABC transporter with a secondary-active multidrug translocator domain,Nature 426 (2003) 866–870.

[43] H. Venter, S. Velamakanni, L. Balakrishnan, H.W. van Veen, On the energy-dependence of Hoechst 33342 transport by the ABC transporter LmrA,Biochem. Pharmacol. 75 (2008) 866–874.

[44] N.P. Barrera, S.C. Isaacson, M. Zhou, V.N. Bavro, A. Welch, T.A. Schaedler, M.A.Seeger, R.N. Miguel, V.M. Korkhov, H.W. van Veen, H. Venter, A.R. Walmsley,C.G. Tate, C.V. Robinson, Mass spectrometry of membrane transporters revealssubunit stoichiometry and interactions, Nat. Methods 6 (2009) 585–587.

[45] H. Venter, A.E. Ashcroft, J.N. Keen, P.J. Henderson, R.B. Herbert, Moleculardissection of membrane-transport proteins: mass spectrometry and sequencedetermination of the galactose-H+ symport protein, GalP, of Escherichia coliand quantitative assay of the incorporation of [ring-2-13C]histidine and(15)NH(3), Biochem. J. 363 (2002) 243–252.

[46] D. Russell, O. Kolaj-Robin, T. Soulimane, Maricaulis maris cation diffusionfacilitator: achieving homogeneity through a mixed-micelle approach, ProteinExpr. Purif. 85 (2012) 173–180.

[47] M. le Maire, P. Champeil, J.V. Moller, Interaction of membrane proteins andlipids with solubilizing detergents, Biochem. Biophys. Acta 1508 (2000) 86–111.

[48] C.L. Pocanschi, T. Dahmane, Y. Gohon, F. Rappaport, H.J. Apell, J.H.Kleinschmidt, J.L. Popot, Amphipathic polymers: tools to fold integralmembrane proteins to their active form, Biochemistry 45 (2006) 13954–13961.

[49] C.L. Pocanschi, J.L. Popot, J.H. Kleinschmidt, Folding and stability of outermembrane protein A (OmpA) from Escherichia coli in an amphipathic polymer,amphipol A8-35, Eur. Biophys. J. 42 (2013) 103–118.

[50] C. Bechara, G. Bolbach, P. Bazzaco, K.S. Sharma, G. Durand, J.L. Popot, F. Zito, S.Sagan, MALDI-TOF mass spectrometry analysis of amphipol-trappedmembrane proteins, Anal. Chem. 84 (2012) 6128–6135.

[51] A.C. Leney, L.M. McMorran, S.E. Radford, A.E. Ashcroft, Amphipathic polymersenable the study of functional membrane proteins in the gas phase, Anal.Chem. 84 (2012) 9841–9847.

[52] C. Klammt, M.H. Perrin, I. Maslennikov, L. Renault, M. Krupa, W. Kwiatkowski,H. Stahlberg, W. Vale, S. Choe, Polymer-based cell-free expression of ligand-binding family B G-protein coupled receptors without detergents, Protein Sci.20 (2011) 1030–1041.

[53] K. Guild, Y. Zhang, R. Stacy, E. Mundt, S. Benbow, A. Green, P.J. Myler, Wheatgerm cell-free expression system as a pathway to improve protein yield andsolubility for the SSGCID pipeline, Acta Crystallogr. Sect. F Struct. Biol. Cryst.Commun. 67 (2011) 1027–1031.

[54] E.A. Weaver, E.E. Wyckoff, A.R. Mey, R. Morrison, S.M. Payne, FeoA and FeoC areessential components of the Vibrio cholerae ferrous iron uptake system, andFeoC interacts with FeoB, J. Bacteriol. 195 (2013) 4826–4835.

[55] H. Kim, H. Lee, D. Shin, The FeoA protein is necessary for the FeoB transporterto import ferrous iron, Biochem. Biophys. Res. Commun. 423 (2012) 733–738.