A Bacterial Virulence Factor with a Dual Role as an Adhesin and a Solute-binding Protein: The Crystal Structure at 1.5 Å Resolution of the PEB1a Protein from the Food-borne Human

doi:10.1016/j.jmb.2007.06.041 J. Mol. Biol. (2007) 372, 160–171

A Bacterial Virulence Factor with a Dual Role as anAdhesin and a Solute-binding Protein: The CrystalStructure at 1.5 Å Resolution of the PEB1a Protein fromthe Food-borne Human Pathogen Campylobacter jejuni

Axel Müller1, Maria del R. León-Kempis2, Eleanor Dodson1

Keith S. Wilson1, Anthony J. Wilkinson1 and David J. Kelly2⁎

1Structural Biology Laboratory,Department of Chemistry,University of York, Heslington,York YO10 5YW, UK2Department of MolecularBiology and Biotechnology,University of Sheffield,Firth Court, Western Bank,Sheffield S10 2TN, UK

Present address: A. Müller, CaltecMC 114-96, Pasadena, CA 91125, USAbbreviation used: ESR, extracyto

receptor.E-mail address of the correspondi

[email protected]

0022-2836/$ - see front matter © 2007 E

The PEB1a protein is an antigenic factor exposed on the surface of the food-borne human pathogen Campylobacter jejuni, which has a major role inadherence and host colonisation. PEB1a is also the periplasmic bindingprotein component of an aspartate/glutamate ABC transporter essential foroptimal microaerobic growth on these dicarboxylic amino acids. Here, wereport the crystal structure of PEB1a at 1.5 Å resolution. The protein has atypical two-domain α/β structure, characteristic of periplasmic extracyto-plasmic solute receptors and a chain topology related to the type IIsubfamily. An aspartate ligand, clearly defined by electron density in theinterdomain cleft, forms extensive polar interactions with the protein, themajority of which are made with the larger domain. Arg89 and Asp174 formion-pairing interactions with the main chain α-carboxyl and α-amino-groups, respectively, of the ligand, while Arg67, Thr82, Lys19 and Tyr156co-ordinate the ligand side-chain carboxyl group. Lys19 and Arg67 line apositively charged groove, which favours binding of Asp over the neutralAsn. The ligand-binding cleft is of sufficient depth to accommodate aglutamate. This is the first structure of an ABC-type aspartate-bindingprotein, and explains the high affinity of the protein for aspartate andglutamate, and its much weaker binding of asparagine and glutamine.Stopped-flow fluorescence spectroscopy indicates a simple bimolecularmechanism of ligand binding, with high association rate constants.Sequence alignments and phylogenetic analyses revealed PEB1a homo-logues in some Gram-positive bacteria. The alignments suggest a moredistant homology with GltI from Escherichia coli, a known glutamate andaspartate-binding protein, but Lys19 and Tyr156 are not conserved in GltI.Our results provide a structural basis for understanding both the solutetransport and adhesin/virulence functions of PEB1a.

© 2007 Elsevier Ltd. All rights reserved.

*Corresponding author
Keywords: PEB1a; virulence factor; antigen; ABC transporter; Campylobacter
h, 157 Broad Center,A.plasmic solute

ng author:

lsevier Ltd. All rights reserve

Introduction

Campylobacter jejuni, a Gram-negative microaero-philic bacterium, is one of the leading causes of acutehuman gastroenteritis worldwide.1 As a commensalbacterium in the gut of birds, C. jejuni is oftenacquired by humans through ingestion of under-cooked poultry. Although acute enteric diseasecaused by C. jejuni can be severe, it is usually self-limiting, but in some cases infection can lead to

d.

mailto:[email protected]

161Structure of PEB1a from Campylobacter jejuni

serious sequelae, the most important of which isGuillain-Barré syndrome, a form of neuromuscularparalysis.2

The pathogenic mechanisms of C. jejuni afterinfection of the human intestinal tract are relativelypoorly understood, but involve mucosal adherence,host cell invasion and toxin production.3 In inves-tigations of the ability of outer membrane compo-nents of strains of C. jejuni and Campylobacter coli tobind to HeLa cells, Fauchere et al. showed thatseveral proteins with molecular mass in the range of26–30 kDa have an important role in adhesion, asrevealed by a decrease in binding to HeLa cells afterincubation of whole bacterial cells with rabbitantiserum raised against these proteins.4 Someproperties of these proteins had been reported.5–7

Pei et al. purified to homogeneity four differentproteins (PEB1-4) potentially located on the outermembrane, which were released only by an acid-glycine treatment.8 One of these, the 28 kDa proteinPEB1a, which was highly immunogenic and demon-strated by electron microscopy to be surface-exposed,9 was shown to be required for adhesionof C. jejuni to HeLa cells. Importantly, a peb1amutant strain showed decreased adhesion in tissueculture studies, and was markedly deficient in thecolonisation of a mouse model of infection,10

indicating that the PEB1a protein is an importantvirulence factor in C. jejuni.Pei and Blaser showed that PEB1a has homology

with several extracytoplasmic solute receptors(ESRs) for amino acid ABC-transport systems, butthey did not identify a transport function for theprotein.11 PEB1a is encoded by Cj0921c in C. jejuniNCTC 11168.12 Over-expression and purification ofrecombinant Cj0921 followed by steady-state fluor-escence titrations with all 20 commonly occurringamino acids showed that the protein specificallybound L-aspartate and L-glutamate with low Kdvalues.13 Significantly, a Cj0921c mutant was com-pletely deficient in glutamate transport, showed∼20-fold reduction in aspartate transport comparedto the parent strain, and was unable to grow oneither aspartate or glutamate as a carbon source inminimal medium.13 It is thus clear that PEB1a is theperiplasmic component of an aspartate/glutamateABC-transporter, essential for optimal microaerobicgrowth on these dicarboxylic amino acids.The dual role of PEB1a as an adhesin and a

periplasmic solute-binding-protein raises issuesabout its cellular location and its mechanism ofexport across the cytoplasmic and outer membrane.The signal sequence of PEB1a is unusual, as itcontains strongly predicted cleavage sites for signalpeptidases I and II. Differential processing at thesesites was proposed to result in two populations ofthe protein, located in the periplasm and at the cellsurface, respectively.10 Leon-Kempis et al. showedby immunoblotting and fractionation that PEB1awas most abundant in the periplasm, and could bedetected in culture supernatants but not in the inneror outer membrane fractions.13 These data suggestthat the protein is translocated across the outer

membrane rather than being tightly anchored to it,but the associated mechanism has yet to beelucidated. PEB1a is not the only solute-binding-protein with a dual localization in C. jejuni. CjaA(Cj0982) is an immunogenic surface protein that isalso a periplasmic component of an ABC transportsystem. However, CjaA appears to be anchored tothe periplasmic face of the inner membrane, whichmay be related to the presence of a very stronglypredicted signal peptidase II site but an atypicalsignal peptidase I site.14,15 CjaA has been shown bysteady-state fluorescence spectroscopy to bindcysteine tightly and specifically with a Kd of ∼10−7

M by steady-state fluorescence spectroscopy.16 Thecrystal structure of CjaA revealed a two-domainprotein with density for a bound cysteine ligand inthe enclosed cavity between the domains.16

Here, we report the crystal structure of theL-aspartate-bound PEB1a protein at 1.5 Å resolution,and an analysis of the ligand-binding mechanism bystopped-flow fluorescence spectroscopy. On thebasis of the PEB1a structure, conserved structuralmotifs and solute-binding features have beenidentified in related ESRs.

Results

PEB1a has a two-domain α/β structure

The crystal structure of PEB1a was solved bymolecular replacement using the coordinates ofthe cysteine-binding-protein Cj0982 as the searchmodel, as described in Materials and Methods. Therefined structure (see Table 1) consists of twocrystallographically independent PEB1a molecules,designated chains A and B, each spanning the entire231 residue polypeptide, two aspartate ligands, 546water molecules and four zinc atoms. Superpositionof the two independent PEB1a molecules by least-squares minimisation gives root-mean-squareddeviations (rmsδ) of 0.37 Å for equivalent Cα

atoms. This compares with rmsd values of 2.1 Åfor 210 equivalent Cα atoms when PEB1a is com-pared with Cj0982, which is the highest-scoringmatch found by a DALI search.PEB1a contains 11 α-helices and 15 β-strands,

assembled as two domains (Figure 1). The largerdomain I has a central five-stranded β-pleated sheetagainst which six α-helices are packed. In thisdomain, which consists of residues 1–99 and 196–231, the strand order is β5–β1–β6–β15–β7 withstrandβ15 running in an anti-parallel direction to theother four strands. A five-stranded β-pleated sheetalso forms the core of domain II (residues 100–193)with five associated α-helices. Its β-sheet has thetopology β11–β10–β12–β9–β13, with strand β9 run-ning in an anti-parallel direction, characteristic oftype II ESRs.17 There are two additional β-sheets; thefirst is an anti-parallel two-stranded sheet formed bystrands β8 and β14, which constitute the segmentsof the polypeptide that join the two domains.The second is a three-stranded anti-parallel sheet

Table 1. PEB1a X-ray data collection and refinementstatistics

A. Data collectionX-ray source ESRF ID23-1Wavelength (Å) 1.2823Collection temperature (K) 100Resolution range (Å) 40.03–1.50Space group P21Unit-cell parameters

a (Å) 53.51b (Å) 75.97c (Å) 61.80β (deg.) 103.34

Matthews coefficient A3 Da−1 2.04Solvent content (%, v/v) 39.33No. unique reflectionsa 75,346 (6545)Completenessa (%) 99.3 (85.9)Redundancya 5.1 (1.7)I/σ(I)a 13.5 (2.0)Rmerge

b (%) 13.5 (32.6)

B. Refinement and model statisticsR-factorc 0.16Rfree

d 0.18Reflections working 70,979Reflections free 3759Outer shelle

R-factorc 0.19Rfree

d 0.25Reflections working 3894Reflections/free) 215

Molecules/asymmetric unit 2No. non-hydrogen atoms 4273No. water molecules 546r.m.s. deviation from targetf

Bond lengths (Å) 0.009Bond angles (deg.) 1.286

Average B-factor (Å2) 15.4Ramachandran plotg

Most favoured regions (%) 361Additionally allowed regions (%) 46Generously allowed regions (%) 3Disallowed regions (%) 0a The outer shell corresponds to 1.55–1.50 Å.b Rmerge=∑hkl∑i|Ii–bIN|/∑hkl∑ibINwhere Ii is the intensity of

the ith measurement of a reflection with indices hkl and bIN is thestatistically weighted average reflection intensity.

c R-factor=∑||Fo|–|Fc||/∑|Fo|, where Fo and Fc arethe observed and calculated structure factor amplitudes,respectively.

d Rfree is the R-factor calculated with 5% of the reflectionschosen at random and omitted from refinement.

e The outer shell for the refinement corresponds to 1.53–1.49 Å.f Root-mean-square deviation of bond lengths and bond angles

from ideal geometry.g According to PROCHECK.

162 Structure of PEB1a from Campylobacter jejuni

(β2–β3–β4) that forms a prominent outcrop fromdomain I. This sheet is a conserved feature in thestructures of a number of amino acid-binding-proteins, including those that bind cysteine(Cj0982), histidine (HisJ), lysine-arginine-ornithine(LAOBP), glutamine (GlnBP) and glutamate (GluR0and GluR2).

The ligand-binding domain contains aspartate

There is additional electron density in the spacebetween the two domains that typically forms theligand-binding pocket in periplasmic solute-binding

proteins. This feature cannot be accounted for byprotein atoms and its shape clearly defined theamino acid aspartate (Figure 2(a)). There are ex-tensive interactions between PEB1a and the boundaspartate. The larger domain provides the majorityof these interactions, the smaller lobe contributingonly three of the total of 12 hydrogen bonding/electrostatic interactions (Figure 2(b); Table 2). Theguanidinium group of Arg89 and the carboxylategroup of Asp174 coordinate the α-carboxyl and α-amino groups of the aspartate, respectively. TheArg89 interaction with the α-carboxyl group is two-pronged. The ligand α-carboxyl group makes ad-ditional interactions with the main chain NN-Hgroups of Thr84 and Thr132, and with the side-chainof Arg67. The side-chains of Arg67, Thr82, Lys19and Tyr156 form polar interactions with the boundaspartate side-chain carboxyl group. In PEB1a,Arg67 effectively forms a bridge between the twocarboxylate groups of the ligand. Arg67 is con-served in the cysteine-binding protein Cj0982,16 inwhich it bridges the ligand carboxylate and thiolgroups.As can be seen in Figure 2(c), one face of the

ligand-binding pocket of PEB1a has a stronglypositive electrostatic character that is well suited tobinding a species with two carboxylate groups.PEB1a binds aspartate and glutamate with similaraffinities.13 This pocket must therefore be capableof accommodating glutamate, which contains anadditional methylene group, but how this occurs isunknown, as there is no enclosed water moleculeto be displaced.

Mechanism and kinetics of ligand binding toPEB1a

From analyses of the pre-steady state kinetics ofthe intrinsic protein fluorescence change uponligand addition, two distinct mechanisms of ligandbinding have been observed in periplasmic ESRs. InABC-type ESR proteins, a linear dependence of theobserved rate constant (kobs) on ligand concentrationhas been found for several proteins binding chemi-cally dissimilar ligands,18,19 implying a simplebimolecular association. In contrast, an inversehyperbolic relationship of kobs with ligand concen-tration in some TRAP transporter ESRs has beeninterpreted as revealing a slow isomerisation from aclosed unliganded conformation to an open unli-ganded conformation before ligand binding.20,21

The mechanism of ligand binding to PEB1a wasinvestigated using stopped-flow fluorescence spec-troscopy with either L-aspartate or L-glutamate.PEB1a contains just one tryptophan residue at theextreme C terminus, but nine tyrosine residues thatproduce most of the emission at 345 nm after ex-citation at 280 nm.13 Preliminary attempts to carryout titration experiments in 10mMTris–HCl (pH 8.0)at 20 °C yielded kobs values above 1000 s−1 withmoderate concentrations of either ligand. Therefore,the temperature was lowered to 15 °C and 200 mMNaCl was added to the buffer to ensure kobs values

Figure 1. The overall structure of PEB1a. (a) A stereo worm representation of the PEB1a fold with every tenth residuelabelled. (b) A stereo ribbon representation of PEBla with the two domains coloured yellow (domain I) and blue (domainII). The aspartate ligand is shown in space-filling format and coloured by atom type; carbon, white; oxygen, red; nitrogen,blue.


were in a measurable range at concentrations ofligand that maintained pseudo first-order con-ditions.19,22 Under these conditions, a rapid quench-ing in fluorescence was observed upon addition ofan excess (20 μM) of either of the two dicarboxylicamino acids, and the traces could be fit to a singleexponential (Figure 3(a) and (b)). No further fluor-escence change was observed after 0.02 s, suggestinga single binding event and no additional conforma-tional change. Stopped-flow kobs rates over a range ofligand concentrations that maintained pseudo first-order conditions (4–20 μM) were then obtained.There was a linear relationship between kobs andligand concentration (Figure 3(c) and (d)), implyingthat ligand binding occurs by a single-step, bimole-cular mechanism, and that PEB1a is predominantlyin an open, unliganded form before undergoingrapid closure after ligand binding.19 The associationrate constants (k1) for aspartate and glutamate werecalculated from the gradients shown in Figure 3(c)and (d) (and see Table 3). Both values are in excess of107 M−1 s−1 and well within the range reported forseveral other periplasmic-binding proteins.18–22 Thedissociation rate constants (k−1) were obtained from

the intercepts of the pseudo first-order plots, and arebelow 100 s−1 for each ligand.

Sequence comparisons and phylogeny

BLASTsearches with the PEB1a sequence revealeda large number of homologous proteins in family 3of the bacterial extracytoplasmic solute receptorsuperfamily, with sequence identities of up to 56%for proteins outside of the epsilon group of pro-teobacteria. Family 3 principally comprises aminoacid-binding proteins, although the ligand specifi-city of the majority of these has not been determinedexperimentally.23 Only one dual glutamate/aspar-tate-binding protein has been studied in detail; GltI(YbeJ) from Escherichia coli.24 A sequence alignmentbetween PEB1a and a selection of proteins eitherpredicted or known to bind glutamate, glutamineand/or aspartate is presented in Figure 4. It can beseen that five of the eight amino acids coordinatingthe ligand in PEB1a are conserved in Escherichia coliGltI and close homologues from Pseudomonas andAcinetobacter. However, two of the three non-conserved residues are Lys19 and Tyr156, which

Figure 2. Aspartate binding to PEB1a. (a) A stereo view of the aspartate-binding pocket of PEB1a with the 2Fo–Fcelectron density map contoured at 1.5 σ displayed on the ligand and selected protein side-chains. The aspartate ligand isin stick representation with atoms coloured by element: carbon, green; oxygen, red; nitrogen, blue. (b) A stereo view of thebinding pocket with hydrogen bonding/electrostatic protein–ligand interactions indicated by broken lines. Ligand atomsare coloured by element as in (a). (c) Surface rendering of the binding pocket with positive and negative electrostaticsurface ranging from −0.5 (red) to +0.5 (blue) arbitrary CCP4MG units. Ligand atoms are coloured by element as in (a).


Table 3. Kinetic parameters of ligand binding to PEB1a

k1 (μM−1 s−1) k−1 (s

−1) Keq (μM−1) Kd (μM)

L-Aspartate 41.2±3.3 78.6±35.6 0.52 1.9L-Glutamate 23.5±1.09 55.4±12.3 0.42 2.4

Table 2. Polar interactions between PEB1a and boundaspartate

Ligand group Protein group

Distance (Å)

MolA MolB

α-NH3+ Thr84 Oγ 2.9 2.9

α-NH3+ Asp174 Oδ2 2.8 2.8

α-NH3+ Thr82 O 2.7 2.7

α-CO2– OXT Arg89 Nη1 2.7 2.8

α-CO2– OXT Thr84 NNH 3.0 3.1

α-CO2– OXT Arg67 Nη2 3.0 2.9

α-CO2– O Thr132 NNH 3.0 3.0

α-CO2– O Arg89 Nη2 2.9 2.9

-OD1 Thr82 Oγ 2.9 2.9-OD1 Arg67 Nη1 2.9 2.8-OD2 Lys19 Nζ 3.0 3.0-OD2 Tyr156 OH 2.6 2.6


bind to the ligand side-chain. They are replaced inGltI by arginine and histidine, respectively, both ofwhich are compatible with the binding of anegatively charged ligand side-chain. In contrast,E. coli GlnH has aspartate and isoleucine at thesepositions. Both Lys19 and Tyr156 are conserved in aPEB1a homologue from the Gram-positive bacter-ium Streptococcus thermophilus, suggesting stronglythat this is also an aspartate/glutamate-bindingprotein, and that similar PEB1-type transport

Figure 3. (a) and (b) Change in protein fluorescence duringstopped-flow fluorescence spectroscopy. Upper traces: 1 μM PEpushed against the same buffer at 15 °C. Lower traces: 1 μML-aspartate or (b) 20 μM L-glutamate. (c) and (d) Plots of thL-aspartate or (d) L-glutamate versus concentration of either ofpoints are the average of three independent titrations. The k1 vand averaged from three independent sets of titrations. The k−best fit (see Table 3).

systems exist in both Gram-negative and Gram-positive bacteria. In contrast to the results obtainedwith PEB1a, a BLAST search with the sequence ofGltI as query produces high-scoring hits only amongGram-negative bacteria. As expected, Asp174 andThr84, which interact with the α-amino and α-carboxyl groups of the ligand, respectively, areconserved in the majority of the sequences.A phylogenetic analysis was performed using a

larger number of proteins that are putative orproven glutamate and/or aspartate-binding pro-teins (Figure 5). Three well-defined groups can beseen in the resulting tree. PEB1a is part of a tightcluster together with proteins from other Campylo-bacter and Helicobacter species in the epsilonsubgroup of the proteobacteria, which are clearlyclosely related to the PEB1a homologue in S.thermophilus. These are well separated from GltIand similarly annotated proteins, which form asecond fairly tight cluster. Glutamine-binding pro-

L-aspartate or L-glutamate binding to PEB1a monitored byB1a in buffer A (10 mM Tris–HCl (pH 8.0), 200 mMNaCl)PEB1a in buffer A pushed against (a) buffer A+20 μM

e rates of association (kobs) between 1 μM PEB1a and (c)these two ligands under pseudo first-order conditions. Thealues were determined from the slope of the line of best fit1 values were determined from the intercepts of the lines of

Figure 4. Sequence alignment of PEB1a with selected amino acid-binding proteins. The secondary structure elementsfor PEB1a are displayed above the sequence and residues that contact the aspartate ligand are indicated by the filledtriangles (see the text for details). Boxed residues are partially conserved, with two completely conserved aspartateresidues boxed in red. The sequences shown are for PEB1a (C. jejuni NCTC 11168), PEB1 (Streptococcus thermophilus), GltI(Escherichia coli K12), PA14_46910 (GltI homologue in Pseudomonas aeruginosa strain PA14), GltI (Acinetobacter sp. ADP1),GlnH (E. coli K12), BztA (Rhodobacter capsulatus) and AapJ (Rhizobium leguminosarum).


teins of E. coli and the Gram-positive genus Bacillusform a third cluster, which is more closely related to,and branches from, the PEB1a clade. The biochemi-cally characterised broad substrate range-bindingproteins BztA from Rhodobacter capsulatus,25 AapJ,26and BraR127 from Rhizobium leguminosarum are at amuch greater phylogenetic distance from the rest ofthe members of the tree, in keeping with the lack ofconservation of many key residues in the alignmentshown in Figure 4.

Discussion

Ligand binding in PEB1a and other dicarboxylicamino acid-binding proteins

The structure of PEB1a is the first to be determinedof a periplasmic aspartate-binding protein from anABC transporter. As observed in structures of mostamino acid-binding proteins, a set of well-conserved

Figure 5. Phylogenetic tree of selected members of putative and known L-glutamate and/or L-aspartate ABC-typebinding proteins homologous with PEB1a, identified using systematic or common names and showing the genus andspecies of each bacterium. The proteins shown in bold are those used for the alignment.


residues binds to the carboxylate and amino groupsof the ligand main chain (Arg89 and Asp174,respectively). Specificity for aspartate is imposedthrough the residues interactingwith the carboxylateside-chain, in this case Lys19, Arg67 and Tyr156. Theresidues Lys19 and Arg67 line a positively chargedgroove, which favours the binding of the negativelycharged ligand over the neutral asparagine. This cleftmust also accommodate the extra methylene groupof glutamate, the alternative ligand of PEB1a. Howthis occurs is unclear but it is notable that there is nonearby water molecule to be displaced upon glu-tamate binding. Expulsion of water molecules isoften associated with versatility in ligand binding byperiplasmic-binding proteins,28 contributing favour-ably to the overall entropy change. In comparisons ofthe mode of ligand binding in PEB1a with thebiochemically characterised and sequenced aspar-tate/glutamate-binding-protein in E. coli (GltI),arginine and histidine replace Lys19 and Tyr156 ofPEB1a. The Kd values for binding of the ligands toPEB1a andGltI are quite similar, despite the differentmode of aspartate and glutamate binding.To date, the structure of only one other bacterial

periplasmic aspartate-binding protein (BugD) hasbeen determined,29 but this is not from an ABCtransporter. BugD is the solute-binding-protein of amember of the novel tripartite tricarboxylate trans-

porter (TTT) family, which consists of large andsmall integral membrane transport proteins inaddition to the periplasmic solute-binding protein.30

Solute transport is driven not by ATP hydrolysisbut by an electrochemical ion gradient, as in themechanistically similar but sequence-unrelated tri-partite ATP-independent periplasmic (TRAP) trans-porters.31,32 BugD was crystallised with an aspartatemolecule bound, but the mode of aspartate bindingis distinctly different from that found here in PEB1a.In BugD, two water molecules bridge the aspartateα-carboxylate group to a common carbonyl groupfrom alanine with each water molecule forming ahydrogen bond with one of the carboxylate oxygenatoms of the ligand.29 Thus, there is no direct ion-pairing with the α-carboxylate, but additional hy-drogen bonds are formed between the ligand α-carboxylate and the backbone amide and side-chainhydroxyl groups of two threonine residues. InPEB1a, the α-carboxyl group makes interactionswith the main chain amide groups of Thr84 andThr132, so in this respect there are some basic simi-larities with BugD. However, the binding of theaspartate side-chain carboxylate in BugD and PEB1ais totally different, with one carboxyl oxygen hy-drogen bonded to a serine side-chain hydroxylgroup and main-chain amide, and the other oxygenatom hydrogen bonded to histidine and alanine in


BugD. Thus, the mode of aspartate binding isdistinctly different in BugD from that seen here inPEB1a.Several structures of glutamate-binding-proteins

are known, and comparisons of ligand binding inPEB1a can be made with those occurring in theL-glutamate-activated gated ion-channel proteins(glutamate receptors or GluRs), which are importantin excitatory synaptic transmission in mammals andhave been shown to have an evolutionary relation-ship to the bacterial periplasmic-binding proteins.33

The structure of GluR2, an AMPA-activated ligand-gated ion channel from rat,34 and GluR0, a pro-karyotic glutamate receptor ion-channel from Syne-chocystis PCC 6803,35 have been determined. InGluR0, L-glutamate bindswith the γ-carboxyl group,interacting directly with an asparagine residue indomain I of the protein, and through hydrogenbonding via a water molecule to a tyrosine residue.An arginine residue interacts with the ligand α-carboxyl group and an aspartate residue interactswith the α-amino-group. However, in GluR2, theconformation of the bound glutamate ligand isdifferent, in that the γ-carboxyl group projectstowards and interacts with a threonine residue atthe base of a helix in domain II.34,35 Thus, the GluRproteins differ substantially from PEB1a in the wayin which the γ-carboxylate group of the ligand isaccommodated, whilst all the proteins use ion-pairing interactions for the α-carboxyl group andα-amino-groups.Previous steady-state ligand-binding studies of

PEB1a revealed an almost equal affinity for aspartateand glutamate, and physiological analysis ofmutants has shown that the PEB1 transporter is im-portant for growth on both aspartate and glutamateas carbon sources.13 Here, we have extended the ki-netic analysis to examine the mechanism of ligand-binding using stopped-flow fluorescence spectro-scopy. As with many such proteins from ABCsystems, an analysis of the ligand concentration-dependence of the time-resolved fluorescencechange clearly revealed a simple bimolecular me-chanism, with a high association rate constant and amoderately low dissociation rate constant for bothaspartate and glutamate, as has been found for ex-ample in sugar-binding and other amino acid-bind-ing-proteins.18,19 The calculated Kd values shown inTable 3 (usingKd=1/Keq) are slightly higher than theKd values obtained by direct steady-state fluores-cence quenching in our previous study;13 this isprobably due to the use of a lower temperature andincreased concentration of salt needed in the presentstudy to ensure kobs values werewithin ameasurablerange.

Antigenicity and adhesion

PEB1a is a major immunogenic protein in C.jejuni.8,9 Whilst the structure presented here doesnot directly identify the relevant antigenic surfaces ofthe protein, it does provide a framework for inter-preting results of future epitope-mapping experi-

ments. The structure features an obvious candidateepitope, a prominently protruding anti-parallel β-ribbon. A similar protruding feature is present inanother immunogenic amino acid-binding proteinfrom C. jejuni, Cj0982,16 and predicted in CjaC(Cj0734) a putative histidine-binding protein,which is also immunogenic and cross-reacts withanti-serum raised against Cj0982.36 Oral immunisa-tion of chickens with an avirulent vaccine strain ex-pressing plasmid-encoded Cj0982 elicits specifichumoral immune responses associated with protec-tion against challenge by wild-type Campylobacter.37

ApaA, a Cj0982 orthologue, is highly immunogenicin Actinobacillus pleuropneumoniae, a bacterium thatcauses a highly contagious and often fatal bronch-opneumonia in pigs.38 In the histidine-bindingprotein HisJ from Salmonella typhimurium, the corre-sponding protruding anti-parallel β-ribbon is impli-cated in interactions with the membrane compo-nents of the histidine transporter that leads to ligandtranslocation to the cytoplasm.39

Aspartate and glutamate uptake systems inother bacteria

While a number of biochemically characterisedbinding protein-dependent glutamate transportershave been characterised from various bacteria, fewsystems solely transporting aspartate have beenidentified. However, dual specificity transportersmay be more widespread. We identified a PEB1ahomologue in the Gram-positive bacterium S.thermophilus, in which all of the key C. jejuni PEB1aligand-binding residues are conserved. Similarproteins are present in other Streptococcus speciesand some other Gram-positive species, indicatingthat the PEB1 type of dual specificity transporter isnot restricted to one just group of Gram-negativebacteria. GltI from E. coli was the first dualglutamate and aspartate-binding protein to becharacterised,24 with binding constants of 0.7 μMand 1.2 μM, respectively. GltI and homologues arerelated to PEB1a but are phylogenetically quitedistinct, and are present in many groups of Gram-negative bacteria. The gene encoding GltI in E. coli isin the gltIJKL operon. gltJ and gltK are the two genesfor the transmembrane subunits and gltL codes forthe ATP-binding protein of the transporter. This isequivalent to the PEB1 operon Cj0921c-Cj0924c in C.jejuni. Despite these similarities, GltI and PEB1adiffer greatly in their importance for the organism.Apart from the additional function of PEB1a as acell-binding factor, transport and growth studies13

suggest that the PEB1-system is the major glutamatetransporter in C. jejuni, in stark contrast to thegltIJKL system, which is complemented by twoindependent secondary transporters that account for85% of the glutamate uptake.40 Aspartate uptake isstrongly reduced in peb1a mutants, but is notentirely lost, suggesting the presence of alternativeaspartate transporters in C. jejuni. Apart from thePEB1 and GltI types of ABC-transporter ESRs, therecently published crystal structure of BugD


provides evidence for the existence of a third routefor binding protein-dependent aspartate uptake.29

In summary, we have crystallised and solved thestructure of one of the key virulence factors of C.jejuni, which plays a major role in host colonisationand in the growth of this bacterium on amino acidcarbon sources. As the first ABC-type periplasmicaspartate-binding protein structure to be deter-mined, our work has also yielded insights intothe mode of ligand-binding in this and relatedproteins.

Materials and Methods

Cloning and purification

The Cj0921c gene encodes the PEB1a protein in theNCTC 11168 strain of C. jejuni and this was PCR amplifiedusing genomic DNA as template. The oligonucleotidesPEB1aF 5′ATCGGAATTCCATATGGTTTTTAGA-AAATCTTTGTTA 3′; with an EcoRI (in bold) and anNdeI site (underlined) and PEB1aR 5′ATCGGAATTC-GGATCCTCATTATAAACCCCATTTTTTCGC 3′; with anEcoRI site (in bold) and a BamHI site (underlined) wereused as the forward and reverse primers, respectively, toamplify the entire coding region, including the signalpeptide. The amplified gene was then introduced as anNdeI-BamHI fragment into the corresponding sites ofpET21a (Novagen) yielding pMK3. Production of therecombinant protein to be used for stopped-flow spectro-scopy was performed by induction of E. coli BL21 (λDE3)(pMK3) with 1 mM IPTG resulting in overproduction of a25 kDa protein that was purified to homogeneity fromperiplasmic extracts using cation-exchange chromatogra-phy as described.13 The location of the protein in the E. coliperiplasm indicates correct export across the cytoplasmicmembrane, and the single-step purification took advan-tage of the fact that whereas most E. coli proteins did notbind to the cation-exchange column under the conditionsused, the basic PEB1a (pI 8.4) bound tightly andwas elutedat a moderate concentration of salt as a single species. N-terminal amino acid sequencing gave a sequence ofAEGKLESIKSK, which is identical with residues 27–37 ofthe deduced sequence and consistent with cleavage afterthe ANA signal peptidase I recognition site.13 The residuenumbering used here refers to the mature protein.

Crystallisation and data collection

PEB1a was expressed in E. coli ER2566 cells harbouringthe plasmid pMK3 and purified as described above. Theprotein was concentrated to 6 mg ml−1 in 50 mM Mes(pH 6.0), 300 mM NaCl, 5 mM sodium aspartate. Crystal-lization trials employed a Mosquito nanolitre-dispensingrobot with the sitting-drop, vapour-diffusion method inwhich 150 nl of PEB1a was mixed with 150 nl of reservoirsolutions from various commercially available crystal-lisation screens. A crystal that diffracted well wasidentified from a single drop prepared with Pact C12solution (0.1 M Hepes (pH 7), 0.01 M zinc chloride, 20%(w/v) PEG 6000). This crystal was vitrified in a solution ofmother liquor containing 25% (v/v) glycerol. Data werecollected to a resolution of 1.5 Å on beam line ID23-1 at theESRF, Grenoble and processedwithMOSFLMand SCALA(Table 1).41 A Matthews coefficient of 2.04 Å3.Da−1 was

calculated, suggesting the presence of twomolecules in theasymmetric unit.

Structure solution and refinement

The structure was solved by molecular replacementusing the program PHASER42 and the coordinates for thecysteine-binding protein of C. jejuni (PDB entry code1xt8)16 as the search model. Using phases calculated withREFMAC5 as “experimental” phases and a resolution cut-off at 2.0 Å, the program RESOLVE was used to build arevised starting model consisting of 384 of the expected462 residues, including 138 side-chains.43 This resulted inan R-factor of 0.435 and an Rfree value of 0.477. The phaseinformation derived from this model allowed ARP/wARP44 to build 429 out of the 462 residues with allside-chains traced, using all of the diffraction data.Further refinement using iterative cycles of REFMAC5and COOT yielded the entire structure (Table 1).45

Steady-state and stopped-flow fluorescencespectroscopy

Stopped-flow kineticmeasurements weremade using anApplied Photophysics sequential stopped-flow spectro-fluorimeter with a slit width of 1 nm using an excitationwavelength of 280 nm, and monitoring the fluorescenceemission above 305 nm (the emission maximum of PEB1aoccurs at 345 nm). All reactions used a final concentration of1 μMPEB1a at 15 °C in buffer A (10 mMTris–HCl (pH 8.0),200 mM NaCl). L-Aspartate or L-glutamate binding toPEB1a was monitored under pseudo first-order conditionswith an at least fourfold excess of ligand over PEB1aconcentration. A total of 1000 data points were recordedover the course of each reaction, and at least six runs wereaveraged for each measurement. Kinetic traces wereanalysed using the SX.18MV software supplied by AppliedPhotophysics Ltd. The reactions were rapid and mono-phasic, and were fit to a single exponential. The depen-dence of kobs on L-aspartate or L-glutamate concentrationwas analysed using a simple one-step equilibrium process(P+L↔PL).18,19 The “on” and “off” rate constants k1 andk−1 were obtained from the slope and intercept of thestraight-line relationship of ligand concentration versus kobs.

Sequence analyses

Sequences were collected from the NCBI protein data-base. Alignments were performed using CLUSTAL_X46

and ESPript.47 For phylogenetic analyses, the output file ofCLUSTAL_X was used in PHYLIP48 to produce a distancematrix tree, which was viewed in TreeView.49

Protein Data Bank accession number

The co-ordinates have been deposited in the RCSBProtein Data Bank with the accession code 2v25.

Acknowledgements

This work was funded by the European Commis-sion as SPINE, contract no. QLG2-CT-2002-00988


under the RTD programme “Quality of Life andManagement of Living Resources”, and by grantBBS/B/07667 from the UK Biotechnology andBiological Sciences Research Council to D.J.K. (forM.L.K.).

References

1. Friedman, C. R., Neimann, J., Wegener, H. C. & Tauxe,R. V. (2000). Epidemiology of Campylobacter jejuniinfections in the United States and other industrialisednations. InCampylobacter (Nachamkin, I. & Blaser, M. J.eds), 2nd edit. pp. 121–138, ASM Press, Washington,DC.

2. Skirrow, M. B. & Blaser, M. J. (2000). Clinical aspects ofCampylobacter infection. In Campylobacter (Nachamkin,I. & Blaser, M. J., eds), pp. 69–88, 2nd edit. ASM Press,Washington DC.

3. Ketley, J. M. (1997). Pathogenesis of enteric infectionby Campylobacter. Microbiology, 143, 5–21.

4. Fauchere, J. L., Kervella, M., Rosenau, A., Mohanna,K. & Veron, M. (1989). Adhesion to HeLa cells ofCampylobacter jejuni and E. coli outer membranecomponents. Res. Microbiol. 140, 379–392.

5. Logan, S. M. & Trust, T. J. (1982). Outer membranecharacteristics of Campylobacter jejuni. Infect. Immun.38, 898–906.

6. Logan, S. M. & Trust, T. J. (1983). Molecularidentification of surface protein antigens of Campylo-bacter jejuni. Infect. Immun. 42, 675–682.

7. Blaser, M. J., Hopkins, J. A. & Vasil, M. L. (1984).Campylobacter jejuni outer membrane proteins areantigenic for humans. Infect. Immun. 43, 986–993.

8. Pei, Z. H., Ellison, R. T., 3rd & Blaser, M. J. (1991).Identification, purification, and characterization ofmajor antigenic proteins of Campylobacter jejuni. J. Biol.Chem. 266, 16363–16369.

9. Kervella, M., Pages, J. M., Pei, Z., Grollier, G., Blaser,M. J. & Fauchere, J. L. (1993). Isolation and character-ization of two Campylobacter glycine-extracted pro-teins that bind to HeLa cell membranes. Infect. Immun.61, 3440–3448.

10. Pei, Z., Burucoa, C., Grignon, B., Baqar, S., Huang,X. Z., Kopecko, D. J. et al. (1998). Mutation in thepeb1A locus of Campylobacter jejuni reduces interac-tions with epithelial cells and intestinal colonizationof mice. Infect. Immun. 66, 938–943.

11. Pei, Z. & Blaser, M. J. (1993). PEB1, the major cell-binding factor of Campylobacter jejuni, is a homolog ofthe binding component in Gram-negative nutrienttransport systems. J. Biol. Chem. 268, 18717–18725.

12. Parkhill, J., Wren, B. W., Mungall, K., Ketley, J. M.,Churcher, C., Basham, D. et al. (2000). The genomesequence of the food-borne pathogen Campylobacterjejuni reveals hypervariable sequences. Nature, 403,665–668.

13. Leon-Kempis M del, R., Guccione, E., Mulholland, F.,Williamson, M. P. & Kelly, D. J. (2006). The Campylo-bacter jejuni PEB1a adhesin is an aspartate/glutamate-binding protein of an ABC transporter essential formicroaerobic growth on dicarboxylic amino acids.Mol. Microbiol. 60, 1262–1275.

14. Pawelec, D., Rozynek, E., Popowski, J. & Jagusztyn-Krynicka, E. K. (1997). Cloning and characterization ofa Campylobacter jejuni 72Dz/92 gene encoding a30 kDa immunopositive protein, component of theABC transport system; expression of the gene in

avirulent Salmonella typhimurium. FEMS Immunol.Med. Microbiol. 19, 137–150.

15. Godlewska, R., Pawowski, M., Wyszynska, A., Buj-nicki, J. & Jagusztyn-Krynicka, E. (2005). Abstracts ofthe 13th International Workshop on Campylobacter,Helicobacter and Related Organisms. September 4–8th,2005. Gold Coast, Australia. F17. pp. 82.

16. Müller, A., Thomas, G. H., Horler, R., Brannigan, J. A.,Blagova, E., Levdikov, V. M. et al. (2005). An ATP-binding cassette-type cysteine transporter in Campy-lobacter jejuni inferred from the structure of an extra-cytoplasmic solute receptor protein.Mol. Microbiol. 57,143–155.

17. Fukami-Kobayashi, K., Tateno, Y. & Nishikawa, K.(1999). Domain dislocation: a change of core structurein periplasmic binding proteins in their evolutionaryhistory. J. Mol. Biol. 286, 279–290.

18. Miller, D. M., 3rd, Olson, J. S., Pflugrath, J. W. &Quiocho, F. A. (1983). Rates of ligand binding toperiplasmic proteins involved in bacterial transportand chemotaxis. J. Biol. Chem. 258, 13665–13672.

19. Miller, D. M., 3rd, Olson, J. S. & Quiocho, F. A. (1980).The mechanism of sugar binding to the periplasmicreceptor for galactose chemotaxis and transport inEscherichia coli. J. Biol. Chem. 255, 2465–2471.

20. Walmsley, A. R., Shaw, J. G. & Kelly, D. J. (1992). Themechanism of ligand binding to the periplasmicC4-dicarboxylate binding protein (DctP) from Rhodo-bacter capsulatus. J. Biol. Chem. 267, 8064–8072.

21. Thomas, G. H., Southworth, T., Leon-Kempis, M. R.,Leech, A. & Kelly, D. J. (2006). Novel ligands for theextracellular solute receptors of two bacterial TRAPtransporters. Microbiology, 152, 187–198.

22. Ledvina, P. S., Tsai, A. L., Wang, Z., Koehl, E. &Quiocho, F. A. (1998). Dominant role of local dipolarinteractions in phosphate binding to a receptor cleftwith an electronegative charge surface: equilibrium,kinetic, and crystallographic studies. Protein Sci. 7,2550–2559.

23. Tam, R. & Saier, M. H., Jr (1993). Structural, functional,and evolutionary relationships among extracellularsolute-binding receptors of bacteria.Microbiol. Rev. 57,320–346.

24. Willis, R. C. & Furlong, C. E. (1975). Purification andproperties of a periplasmic glutamate-aspartate bind-ing protein from Escherichia coli K12 strain W3092.J. Biol. Chem. 250, 2574–2580.

25. Zheng, S. & Haselkorn, R. (1996). A glutamate/glutamine/aspartate/asparagine transport operon inRhodobacter capsulatus. Mol. Microbiol. 20, 1001–1011.

26. Walshaw, D. L. & Poole, P. S. (1996). The generalL-amino acid permease of Rhizobium leguminosarum isan ABC uptake system that also influences efflux ofsolutes. Mol. Microbiol. 21, 1239–1252.

27. Hosie, A. H., Allaway, D., Galloway, C. S., Dunsby,H. A. & Poole, P. S. (2002). Rhizobium leguminosarumhas a second general amino acid permease withunusually broad substrate specificity and high simi-larity to branched-chain amino acid transporters (Bra/LIV) of the ABC family. J. Bacteriol. 184, 4071–4080.

28. Sleigh, S. H., Tame, J. R., Dodson, E. J. & Wilkinson,A. J. (1997). Peptide binding in OppA, the crystalstructures of the periplasmic oligopeptide bindingprotein in the unliganded form and in complex withlysyllysine. Biochemistry, 36, 9747–9758.

29. Huvent, I., Belrhali, H., Antoine, R., Bompard, C.,Locht, C., Jacob-Dubuisson, F. & Villeret, V. (2006).Crystal structure of Bordetella pertussis BugD solutereceptor unveils the basis of ligand binding in a new


family of periplasmic binding proteins. J. Mol. Biol.356, 1014–1026.

30. Winnen, B., Hvorup, R. N. & Saier, M. H., Jr (2003).The tripartite tricarboxylate transporter (TTT) family.Res. Microbiol. 154, 457–465.

31. Forward, J. A., Behrendt, M. C., Wyborn, N. R., Cross,R. & Kelly, D. J. (1997). TRAP transporters: a newfamily of periplasmic solute transport systemsencoded by the dctPQM genes of Rhodobacter capsula-tus and by homologs in diverse gram-negativebacteria. J. Bacteriol. 179, 5482–5493.

32. Kelly, D. J. & Thomas, G. H. (2001). The tripartite ATP-independent periplasmic (TRAP) transporters of bac-teria and archaea. FEMS Microbiol. Rev. 25, 405–424.

33. Chiu, J., DeSalle, R., Lam, H. M., Meisel, L. & Coruzzi,G. (1999). Molecular evolution of glutamate receptors:a primitive signaling mechanism that existed beforeplants and animals diverged. Mol. Biol. Evol. 16,826–838.

34. Armstrong, N. & Gouaux, E. (2000). Mechanisms foractivation and antagonism of an AMPA-sensitiveglutamate receptor: crystal structures of the GluR2ligand binding core. Neuron, 28, 165–181.

35. Mayer, M. L., Olson, R. & Gouaux, E. (2001).Mechanisms for ligand binding to GluR0 ion channels:crystal structures of the glutamate and serine com-plexes and a closed apo state. J. Mol. Biol. 311, 815–836.

36. Pawelec, D., Jakubowska-Mroz, J. & Jagusztyn-Kry-nicka, E. K. (1998). Campylobacter jejuni 72Dz/92 cjaCgene coding 28 kDa immunopositive protein, a homo-logue of the solute-binding components of the ABCtransport system. Letters Appl. Microbiol. 26, 69–76.

37. Wyszynska, A., Raczko, A., Lis, M. & Jagusztyn-Krynicka, E. K. (2004). Oral immunization of chickenswith avirulent Salmonella vaccine strain carrying C.jejuni 72Dz/92 cjaA gene elicits specific humoralimmune response associated with protection againstchallenge with wild-type Campylobacter. Vaccine, 22,1379–1389.

38. Martin, P. R. & Mulks, M. H. (1999). Cloning andcharacterization of a gene encoding an antigenic

membrane protein from Actinobacillus pleuropneumo-niae with homology to ABC transporters. FEMSImmunol. Med. Microbiol. 25, 245–254.

39. Liu, C. E., Liu, P. Q., Wolf, A., Lin, E. & Ames, G. F.(1999). Both lobes of the soluble receptor of theperiplasmic histidine permease, an ABC transporter(traffic ATPase), interact with the membrane-boundcomplex. Effect of different ligands and consequencesfor the mechanism of action. J. Biol. Chem. 274,739–747.

40. Schellenberg, G. D. & Furlong, C. E. (1977). Resolutionof the multiplicity of the glutamate and aspartatetransport systems of Escherichia coli. J. Biol. Chem. 252,9055–9064.

41. Leslie, A. G. W. (1999). Integration of macromoleculardiffraction data. Acta Crystallog. sect. D, 55, 1696–1702.

42. Storoni, L. C., McCoy, A. J. & Read, R. J. (2004).Likelihood-enhanced fast rotation functions. ActaCrystallog. sect. D, 60, 432–438.

43. Terwilliger, T. C. (2003). SOLVE and RESOLVE:automated structure solution and density modifica-tion. Methods Enzymol. 374, 22–37.

44. Lamzin, V. S. & Wilson, K. S. (1993). Automatedrefinement of protein models. Acta Crystallog. sect. D,49, 129–147.

45. Emsley, P. & Cowtan, K. (2004). COOT: model-building tools for molecular graphics. Acta Crystallog.sect. D, 60, 2126–2132.

46. Thompson, J. D., Gibson, T. J., Plewniak, F., Jeanmou-gin, F. & Higgins, D. G. (1997). The CLUSTAL_Xwindows interface: flexible strategies for multiplesequence alignment aided by quality analysis tools.Nucl. Acids Res. 25, 4876–4882.

47. Gouet, P., Courcelle, E., Stuart, D. I. &Metoz, F. (1999).ESPript: multiple sequence alignments in PostScript.Bioinformatics, 15, 305–308.

48. Felsenstein, J. (1989). PHYLIP - phylogeny inferencepackage (version 3.2). Cladistics, 5, 164–166.

49. Page, R. D. (1996). TreeView: an application to displayphylogenetic trees on personal computers. Comput.Appl. Biosci. 12, 357–358.

Edited by R. Huber

(Received 3 April 2007; received in revised form 13 June 2007; accepted 14 June 2007)Available online 19 June 2007

Documents

A Bacterial Virulence Factor with a Dual Role as an Adhesin and a Solute-binding Protein: The Crystal Structure at 1.5 Å Resolution of the PEB1a Protein from the Food-borne Human